diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14815 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 2465,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 2.5998,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 2.4334,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 2.5643,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 2.5764,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 2.5187,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 2.4022,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 2.245,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 2.152,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 2.2341,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.8867,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 1.8761,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.9464,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 1.7282,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.766,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 1.7577,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.6377,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 1.6865,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.6057,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 1.6538,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.6081,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 1.5874,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.5722,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 1.5543,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.5481,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 1.5706,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.5649,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 1.5326,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.5119,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 1.4722,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.4811,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 1.4661,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.5548,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 1.4536,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.4453,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 1.4591,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.4022,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.5014,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.4654,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 1.4518,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.4513,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 1.4281,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.4458,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 1.4574,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 1.437,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 1.3969,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.4186,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 1.4087,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 1.3955,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 1.445,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.4683,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 1.4439,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 1.5076,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 1.4462,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 1.4006,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 1.4353,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 1.3908,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 1.4384,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.4318,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 1.4051,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.4305,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 1.4098,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.4415,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 1.3764,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.4164,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 1.3503,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.3842,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 1.4,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 1.371,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 1.3661,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.4277,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 1.4204,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.355,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 1.3862,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.4194,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991368017825e-05,
+      "loss": 1.3531,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999965472086192e-05,
+      "loss": 1.4356,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999922312249814e-05,
+      "loss": 1.3551,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999861888583198e-05,
+      "loss": 1.4084,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999784201190658e-05,
+      "loss": 1.3774,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999689250206324e-05,
+      "loss": 1.3966,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999577035794106e-05,
+      "loss": 1.3677,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999447558147738e-05,
+      "loss": 1.4145,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999300817490745e-05,
+      "loss": 1.4277,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999136814076465e-05,
+      "loss": 1.3876,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998955548188027e-05,
+      "loss": 1.4235,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998757020138372e-05,
+      "loss": 1.3687,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998541230270237e-05,
+      "loss": 1.4489,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998308178956164e-05,
+      "loss": 1.3877,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998057866598487e-05,
+      "loss": 1.3634,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997790293629344e-05,
+      "loss": 1.3839,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997505460510675e-05,
+      "loss": 1.3249,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997203367734218e-05,
+      "loss": 1.3372,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996884015821492e-05,
+      "loss": 1.3415,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999654740532384e-05,
+      "loss": 1.3372,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996193536822378e-05,
+      "loss": 1.2969,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999582241092803e-05,
+      "loss": 1.3397,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995434028281493e-05,
+      "loss": 1.3609,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999502838955328e-05,
+      "loss": 1.3553,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999460549544368e-05,
+      "loss": 1.373,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999416534668278e-05,
+      "loss": 1.3588,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993707944030448e-05,
+      "loss": 1.3197,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323328827634e-05,
+      "loss": 1.3521,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992741380239905e-05,
+      "loss": 1.3704,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999223222077037e-05,
+      "loss": 1.3319,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991705810746748e-05,
+      "loss": 1.3435,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991162151077822e-05,
+      "loss": 1.3563,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990601242702174e-05,
+      "loss": 1.3479,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999002308658815e-05,
+      "loss": 1.3702,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998942768373388e-05,
+      "loss": 1.3437,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998881503516726e-05,
+      "loss": 1.341,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988185141945973e-05,
+      "loss": 1.3558,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987538005157453e-05,
+      "loss": 1.3472,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986873625918923e-05,
+      "loss": 1.34,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998619200537736e-05,
+      "loss": 1.3656,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998549314470952e-05,
+      "loss": 1.3862,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99847770451219e-05,
+      "loss": 1.3741,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984043707850782e-05,
+      "loss": 1.3164,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983293134162195e-05,
+      "loss": 1.3554,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982525325351924e-05,
+      "loss": 1.377,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998174028274551e-05,
+      "loss": 1.3527,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980938007698257e-05,
+      "loss": 1.326,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99801185015952e-05,
+      "loss": 1.3549,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979281765851133e-05,
+      "loss": 1.3383,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99784278019106e-05,
+      "loss": 1.3864,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977556611247872e-05,
+      "loss": 1.3192,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997666819536698e-05,
+      "loss": 1.3464,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975762555801674e-05,
+      "loss": 1.3147,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974839694115444e-05,
+      "loss": 1.3251,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973899611901527e-05,
+      "loss": 1.3287,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972942310782868e-05,
+      "loss": 1.3717,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971967792412154e-05,
+      "loss": 1.4284,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970976058471782e-05,
+      "loss": 1.3689,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969967110673887e-05,
+      "loss": 1.3607,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968940950760308e-05,
+      "loss": 1.3365,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967897580502606e-05,
+      "loss": 1.3364,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996683700170205e-05,
+      "loss": 1.3834,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965759216189617e-05,
+      "loss": 1.3322,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964664225825997e-05,
+      "loss": 1.3451,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963552032501578e-05,
+      "loss": 1.3354,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996242263813644e-05,
+      "loss": 1.2889,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961276044680374e-05,
+      "loss": 1.347,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960112254112847e-05,
+      "loss": 1.3693,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958931268443028e-05,
+      "loss": 1.3276,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995773308970976e-05,
+      "loss": 1.3307,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956517719981587e-05,
+      "loss": 1.4271,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955285161356708e-05,
+      "loss": 1.3484,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954035415963008e-05,
+      "loss": 1.3205,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995276848595805e-05,
+      "loss": 1.3689,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995148437352905e-05,
+      "loss": 1.3289,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99501830808929e-05,
+      "loss": 1.3422,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994886461029615e-05,
+      "loss": 1.3652,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947528964014992e-05,
+      "loss": 1.2959,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946176144355292e-05,
+      "loss": 1.3218,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944806153652545e-05,
+      "loss": 1.3017,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99434189942719e-05,
+      "loss": 1.3178,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994201466860815e-05,
+      "loss": 1.374,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994059317908571e-05,
+      "loss": 1.3309,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993915452815864e-05,
+      "loss": 1.347,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993769871831062e-05,
+      "loss": 1.3947,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936225752054956e-05,
+      "loss": 1.3387,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934735631934568e-05,
+      "loss": 1.3344,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933228360521997e-05,
+      "loss": 1.3611,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993170394041939e-05,
+      "loss": 1.3204,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930162374258496e-05,
+      "loss": 1.3236,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992860366470068e-05,
+      "loss": 1.306,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927027814436885e-05,
+      "loss": 1.3316,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925434826187657e-05,
+      "loss": 1.3208,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992382470270312e-05,
+      "loss": 1.2951,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992219744676299e-05,
+      "loss": 1.2952,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920553061176557e-05,
+      "loss": 1.2888,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918891548782677e-05,
+      "loss": 1.3657,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917212912449782e-05,
+      "loss": 1.3285,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915517155075865e-05,
+      "loss": 1.3711,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913804279588473e-05,
+      "loss": 1.3513,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912074288944708e-05,
+      "loss": 1.3792,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910327186131226e-05,
+      "loss": 1.3474,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990856297416421e-05,
+      "loss": 1.3244,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906781656089395e-05,
+      "loss": 1.3053,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990498323498204e-05,
+      "loss": 1.3456,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903167713946937e-05,
+      "loss": 1.36,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901335096118386e-05,
+      "loss": 1.325,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989948538466022e-05,
+      "loss": 1.275,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897618582765773e-05,
+      "loss": 1.3043,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989573469365788e-05,
+      "loss": 1.2712,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893833720588885e-05,
+      "loss": 1.35,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891915666840623e-05,
+      "loss": 1.319,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988998053572441e-05,
+      "loss": 1.3638,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888028330581052e-05,
+      "loss": 1.3408,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988605905478083e-05,
+      "loss": 1.3571,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988407271172349e-05,
+      "loss": 1.3524,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988206930483826e-05,
+      "loss": 1.3327,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880048837583797e-05,
+      "loss": 1.3401,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878011313448244e-05,
+      "loss": 1.2731,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987595673594916e-05,
+      "loss": 1.3552,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873885108633577e-05,
+      "loss": 1.3561,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871796435077932e-05,
+      "loss": 1.3202,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986969071888811e-05,
+      "loss": 1.3389,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986756796369941e-05,
+      "loss": 1.3677,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986542817317655e-05,
+      "loss": 1.3522,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986327135101366e-05,
+      "loss": 1.3774,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986109750093426e-05,
+      "loss": 1.3402,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858906626691285e-05,
+      "loss": 1.3617,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985669873206705e-05,
+      "loss": 1.2735,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985447382087326e-05,
+      "loss": 1.2837,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985223189695099e-05,
+      "loss": 1.3445,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849972964170687e-05,
+      "loss": 1.3188,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984769702643217e-05,
+      "loss": 1.3441,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984540408766461e-05,
+      "loss": 1.3543,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843094151826522e-05,
+      "loss": 1.2795,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840767222905778e-05,
+      "loss": 1.3167,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838423304919575e-05,
+      "loss": 1.3206,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836062401914447e-05,
+      "loss": 1.3366,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833684517966247e-05,
+      "loss": 1.3108,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983128965718015e-05,
+      "loss": 1.3249,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828877823690623e-05,
+      "loss": 1.3043,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826449021661466e-05,
+      "loss": 1.3289,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824003255285735e-05,
+      "loss": 1.3078,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821540528785802e-05,
+      "loss": 1.2881,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819060846413314e-05,
+      "loss": 1.3123,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816564212449173e-05,
+      "loss": 1.3171,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981405063120357e-05,
+      "loss": 1.3113,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811520107015935e-05,
+      "loss": 1.3587,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980897264425496e-05,
+      "loss": 1.3432,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806408247318576e-05,
+      "loss": 1.2806,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803826920633947e-05,
+      "loss": 1.31,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980122866865747e-05,
+      "loss": 1.2418,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798613495874754e-05,
+      "loss": 1.2823,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979598140680062e-05,
+      "loss": 1.3153,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793332405979107e-05,
+      "loss": 1.2996,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790666497983433e-05,
+      "loss": 1.2931,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787983687416017e-05,
+      "loss": 1.2865,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978528397890845e-05,
+      "loss": 1.2836,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97825673771215e-05,
+      "loss": 1.3009,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779833886745098e-05,
+      "loss": 1.3045,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777083512498336e-05,
+      "loss": 1.3405,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774316259129444e-05,
+      "loss": 1.2954,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771532131415806e-05,
+      "loss": 1.3143,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768731134163924e-05,
+      "loss": 1.3035,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976591327220943e-05,
+      "loss": 1.319,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763078550417073e-05,
+      "loss": 1.2983,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760226973680706e-05,
+      "loss": 1.3635,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757358546923286e-05,
+      "loss": 1.3352,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754473275096845e-05,
+      "loss": 1.2943,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751571163182506e-05,
+      "loss": 1.3266,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748652216190477e-05,
+      "loss": 1.2917,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745716439160007e-05,
+      "loss": 1.3231,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742763837159412e-05,
+      "loss": 1.3212,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973979441528606e-05,
+      "loss": 1.2958,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736808178666342e-05,
+      "loss": 1.325,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733805132455693e-05,
+      "loss": 1.3052,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730785281838554e-05,
+      "loss": 1.376,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727748632028394e-05,
+      "loss": 1.2786,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972469518826767e-05,
+      "loss": 1.336,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972162495582783e-05,
+      "loss": 1.3231,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971853794000932e-05,
+      "loss": 1.335,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715434146141553e-05,
+      "loss": 1.2935,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712313579582907e-05,
+      "loss": 1.2602,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709176245720713e-05,
+      "loss": 1.3257,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706022149971257e-05,
+      "loss": 1.3222,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702851297779755e-05,
+      "loss": 1.306,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699663694620355e-05,
+      "loss": 1.3124,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969645934599613e-05,
+      "loss": 1.2529,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693238257439056e-05,
+      "loss": 1.3394,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969000043451e-05,
+      "loss": 1.2641,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686745882798733e-05,
+      "loss": 1.3406,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683474607923902e-05,
+      "loss": 1.2976,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680186615533022e-05,
+      "loss": 1.3545,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676881911302476e-05,
+      "loss": 1.3167,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967356050093749e-05,
+      "loss": 1.2897,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670222390172134e-05,
+      "loss": 1.3552,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966686758476931e-05,
+      "loss": 1.27,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966349609052075e-05,
+      "loss": 1.2866,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660107913246977e-05,
+      "loss": 1.3472,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965670305879734e-05,
+      "loss": 1.279,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965328153304996e-05,
+      "loss": 1.3485,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964984334191175e-05,
+      "loss": 1.2921,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646388491318394e-05,
+      "loss": 1.3219,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642916987234327e-05,
+      "loss": 1.2858,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963942883565274e-05,
+      "loss": 1.2486,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635924042595578e-05,
+      "loss": 1.3343,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963240261411349e-05,
+      "loss": 1.345,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962886455628587e-05,
+      "loss": 1.3315,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962530987522079e-05,
+      "loss": 1.2929,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621738577055056e-05,
+      "loss": 1.2514,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618150667954137e-05,
+      "loss": 1.3249,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961454615411219e-05,
+      "loss": 1.2699,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961092504175203e-05,
+      "loss": 1.2964,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607287337125133e-05,
+      "loss": 1.2859,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603633046511623e-05,
+      "loss": 1.3541,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599962176220256e-05,
+      "loss": 1.2441,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596274732588404e-05,
+      "loss": 1.282,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959257072198206e-05,
+      "loss": 1.3441,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958885015079581e-05,
+      "loss": 1.2833,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585113025452837e-05,
+      "loss": 1.3236,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581359352404907e-05,
+      "loss": 1.3109,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577589138132338e-05,
+      "loss": 1.3429,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573802389144023e-05,
+      "loss": 1.3003,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569999111977384e-05,
+      "loss": 1.3273,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566179313198393e-05,
+      "loss": 1.2804,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562342999401532e-05,
+      "loss": 1.266,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95584901772098e-05,
+      "loss": 1.3056,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554620853274698e-05,
+      "loss": 1.2829,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550735034276213e-05,
+      "loss": 1.3287,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546832726922803e-05,
+      "loss": 1.2759,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542913937951405e-05,
+      "loss": 1.2981,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538978674127397e-05,
+      "loss": 1.2394,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953502694224461e-05,
+      "loss": 1.3035,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531058749125295e-05,
+      "loss": 1.2572,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952707410162013e-05,
+      "loss": 1.3141,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523073006608187e-05,
+      "loss": 1.2854,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519055470996954e-05,
+      "loss": 1.332,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515021501722286e-05,
+      "loss": 1.2617,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510971105748412e-05,
+      "loss": 1.3134,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506904290067922e-05,
+      "loss": 1.3515,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502821061701752e-05,
+      "loss": 1.2777,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498721427699174e-05,
+      "loss": 1.3654,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494605395137778e-05,
+      "loss": 1.331,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949047297112347e-05,
+      "loss": 1.2538,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486324162790455e-05,
+      "loss": 1.2857,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482158977301217e-05,
+      "loss": 1.2548,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947797742184652e-05,
+      "loss": 1.316,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473779503645386e-05,
+      "loss": 1.2491,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469565229945083e-05,
+      "loss": 1.2723,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946533460802112e-05,
+      "loss": 1.2805,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461087645177227e-05,
+      "loss": 1.3511,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945682434874535e-05,
+      "loss": 1.3341,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452544726085625e-05,
+      "loss": 1.275,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448248784586374e-05,
+      "loss": 1.2664,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94439365316641e-05,
+      "loss": 1.2789,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943960797476346e-05,
+      "loss": 1.2557,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943526312135726e-05,
+      "loss": 1.31,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430901978946436e-05,
+      "loss": 1.3282,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426524555060053e-05,
+      "loss": 1.2366,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422130857255277e-05,
+      "loss": 1.3083,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941772089311737e-05,
+      "loss": 1.3641,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413294670259688e-05,
+      "loss": 1.283,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408852196323638e-05,
+      "loss": 1.2961,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404393478978688e-05,
+      "loss": 1.3311,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399918525922364e-05,
+      "loss": 1.2939,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93954273448802e-05,
+      "loss": 1.357,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939091994360575e-05,
+      "loss": 1.2905,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386396329880593e-05,
+      "loss": 1.2824,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938185651151426e-05,
+      "loss": 1.3042,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937730049634429e-05,
+      "loss": 1.2906,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937272829223617e-05,
+      "loss": 1.2737,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368139907083333e-05,
+      "loss": 1.3328,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936353534880715e-05,
+      "loss": 1.33,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358914625356914e-05,
+      "loss": 1.3324,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935427774470983e-05,
+      "loss": 1.3198,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349624714870994e-05,
+      "loss": 1.3147,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344955543873373e-05,
+      "loss": 1.2836,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934027023977781e-05,
+      "loss": 1.2388,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335568810672995e-05,
+      "loss": 1.2902,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330851264675462e-05,
+      "loss": 1.3519,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326117609929568e-05,
+      "loss": 1.2688,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321367854607472e-05,
+      "loss": 1.3022,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316602006909132e-05,
+      "loss": 1.3308,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311820075062303e-05,
+      "loss": 1.2646,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307022067322484e-05,
+      "loss": 1.2789,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930220799197294e-05,
+      "loss": 1.1968,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929737785732468e-05,
+      "loss": 1.3752,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292531671716423e-05,
+      "loss": 1.3139,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928766944351461e-05,
+      "loss": 1.2842,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282791181113377e-05,
+      "loss": 1.3023,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277896892934536e-05,
+      "loss": 1.355,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927298658742757e-05,
+      "loss": 1.333,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926806027306961e-05,
+      "loss": 1.3034,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263117958365428e-05,
+      "loss": 1.3001,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258159651847424e-05,
+      "loss": 1.3139,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253185362075593e-05,
+      "loss": 1.2983,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248195097637538e-05,
+      "loss": 1.2835,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924318886714843e-05,
+      "loss": 1.2303,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238166679251006e-05,
+      "loss": 1.3059,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233128542615558e-05,
+      "loss": 1.2933,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228074465939904e-05,
+      "loss": 1.3104,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223004457949382e-05,
+      "loss": 1.2525,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217918527396842e-05,
+      "loss": 1.2498,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212816683062612e-05,
+      "loss": 1.3036,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207698933754498e-05,
+      "loss": 1.3394,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202565288307764e-05,
+      "loss": 1.31,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197415755585117e-05,
+      "loss": 1.3163,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192250344476694e-05,
+      "loss": 1.2515,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918706906390004e-05,
+      "loss": 1.3157,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91818719228001e-05,
+      "loss": 1.2771,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91766589301492e-05,
+      "loss": 1.2918,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171430094947036e-05,
+      "loss": 1.3033,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166185426220645e-05,
+      "loss": 1.2782,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160924933024403e-05,
+      "loss": 1.2802,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155648624440012e-05,
+      "loss": 1.2921,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915035650957647e-05,
+      "loss": 1.2432,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145048597570067e-05,
+      "loss": 1.3168,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913972489758436e-05,
+      "loss": 1.3478,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913438541881017e-05,
+      "loss": 1.3436,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129030170465545e-05,
+      "loss": 1.2682,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912365916179578e-05,
+      "loss": 1.2684,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118272402073353e-05,
+      "loss": 1.2897,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112869900597955e-05,
+      "loss": 1.2726,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107451666696443e-05,
+      "loss": 1.2705,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910201770972284e-05,
+      "loss": 1.2503,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096568039058297e-05,
+      "loss": 1.3235,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909110266411112e-05,
+      "loss": 1.2746,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908562159431671e-05,
+      "loss": 1.3297,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080124839137563e-05,
+      "loss": 1.27,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074612408063257e-05,
+      "loss": 1.2985,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069084310610434e-05,
+      "loss": 1.2558,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063540556322787e-05,
+      "loss": 1.2548,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057981154771026e-05,
+      "loss": 1.2442,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905240611555289e-05,
+      "loss": 1.3064,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046815448293095e-05,
+      "loss": 1.3292,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041209162643365e-05,
+      "loss": 1.2474,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035587268282357e-05,
+      "loss": 1.2643,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029949774915698e-05,
+      "loss": 1.2781,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902429669227593e-05,
+      "loss": 1.2757,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018628030122526e-05,
+      "loss": 1.2552,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012943798241833e-05,
+      "loss": 1.2912,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007244006447093e-05,
+      "loss": 1.3363,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001528664578402e-05,
+      "loss": 1.2695,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995797782502715e-05,
+      "loss": 1.2385,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990051370113796e-05,
+      "loss": 1.3433,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898428943733224e-05,
+      "loss": 1.3122,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897851199410542e-05,
+      "loss": 1.3003,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972719050407494e-05,
+      "loss": 1.3035,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966910616239387e-05,
+      "loss": 1.24,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961086701628753e-05,
+      "loss": 1.2801,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955247316629975e-05,
+      "loss": 1.2825,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894939247132415e-05,
+      "loss": 1.3342,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943522175819065e-05,
+      "loss": 1.2953,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893763644024917e-05,
+      "loss": 1.2841,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931735274775587e-05,
+      "loss": 1.2675,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925818689586057e-05,
+      "loss": 1.3026,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891988669489496e-05,
+      "loss": 1.2711,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913939300943266e-05,
+      "loss": 1.3072,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907976517998534e-05,
+      "loss": 1.3019,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890199835635489e-05,
+      "loss": 1.2874,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896004826333018e-05,
+      "loss": 1.3067,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889995938280124e-05,
+      "loss": 1.2788,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883971702569924e-05,
+      "loss": 1.2873,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877932129602646e-05,
+      "loss": 1.2921,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871877229804987e-05,
+      "loss": 1.2958,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88658070136301e-05,
+      "loss": 1.2817,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859721491557582e-05,
+      "loss": 1.3038,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885362067409346e-05,
+      "loss": 1.2806,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847504571770167e-05,
+      "loss": 1.2834,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884137319514652e-05,
+      "loss": 1.2631,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88352265548077e-05,
+      "loss": 1.2801,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829064661365247e-05,
+      "loss": 1.3125,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822887525457038e-05,
+      "loss": 1.3051,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816695157747244e-05,
+      "loss": 1.2442,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810487568926362e-05,
+      "loss": 1.2382,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804264769711145e-05,
+      "loss": 1.2973,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879802677084461e-05,
+      "loss": 1.3268,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879177358309602e-05,
+      "loss": 1.266,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878550521726085e-05,
+      "loss": 1.2906,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877922168416079e-05,
+      "loss": 1.2884,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772922994643707e-05,
+      "loss": 1.279,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766609159583635e-05,
+      "loss": 1.3092,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876028018988076e-05,
+      "loss": 1.2463,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753936096461388e-05,
+      "loss": 1.2928,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747576890277944e-05,
+      "loss": 1.2865,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741202582308932e-05,
+      "loss": 1.2831,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734813183558942e-05,
+      "loss": 1.264,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728408705058604e-05,
+      "loss": 1.2814,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721989157864594e-05,
+      "loss": 1.2774,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715554553059587e-05,
+      "loss": 1.2297,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709104901752263e-05,
+      "loss": 1.2348,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702640215077283e-05,
+      "loss": 1.2686,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696160504195253e-05,
+      "loss": 1.2733,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689665780292727e-05,
+      "loss": 1.2503,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868315605458217e-05,
+      "loss": 1.278,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676631338301955e-05,
+      "loss": 1.2799,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867009164271632e-05,
+      "loss": 1.2811,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663536979115376e-05,
+      "loss": 1.3028,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656967358815073e-05,
+      "loss": 1.2087,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650382793157177e-05,
+      "loss": 1.2165,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643783293509262e-05,
+      "loss": 1.2872,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863716887126468e-05,
+      "loss": 1.2827,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630539537842544e-05,
+      "loss": 1.2871,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862389530468771e-05,
+      "loss": 1.274,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617236183270764e-05,
+      "loss": 1.2905,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610562185087985e-05,
+      "loss": 1.3051,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860387332166134e-05,
+      "loss": 1.3382,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859716960453846e-05,
+      "loss": 1.266,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590451045292624e-05,
+      "loss": 1.2211,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583717655522717e-05,
+      "loss": 1.2826,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576969446853248e-05,
+      "loss": 1.2982,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570206430934297e-05,
+      "loss": 1.2174,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856342861944151e-05,
+      "loss": 1.2363,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855663602407608e-05,
+      "loss": 1.2736,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549828656564717e-05,
+      "loss": 1.2443,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543006528659634e-05,
+      "loss": 1.2887,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536169652138532e-05,
+      "loss": 1.2406,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852931803880457e-05,
+      "loss": 1.2398,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522451700486348e-05,
+      "loss": 1.2533,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515570649037885e-05,
+      "loss": 1.2847,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508674896338606e-05,
+      "loss": 1.3033,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501764454293315e-05,
+      "loss": 1.2482,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494839334832176e-05,
+      "loss": 1.2877,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848789954991069e-05,
+      "loss": 1.2094,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848094511150967e-05,
+      "loss": 1.2353,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847397603163524e-05,
+      "loss": 1.2254,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466992322318794e-05,
+      "loss": 1.2621,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459993995616984e-05,
+      "loss": 1.2639,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452981063611696e-05,
+      "loss": 1.2573,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445953538410026e-05,
+      "loss": 1.2764,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438911432144274e-05,
+      "loss": 1.26,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431854756971905e-05,
+      "loss": 1.2559,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424783525075538e-05,
+      "loss": 1.2916,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417697748662918e-05,
+      "loss": 1.2367,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410597439966915e-05,
+      "loss": 1.2441,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403482611245468e-05,
+      "loss": 1.2963,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839635327478159e-05,
+      "loss": 1.2826,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389209442883355e-05,
+      "loss": 1.2521,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382051127883836e-05,
+      "loss": 1.3524,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374878342141126e-05,
+      "loss": 1.2186,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83676910980383e-05,
+      "loss": 1.2319,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360489407983385e-05,
+      "loss": 1.2674,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353273284409356e-05,
+      "loss": 1.2181,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346042739774105e-05,
+      "loss": 1.2052,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338797786560416e-05,
+      "loss": 1.3586,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331538437275954e-05,
+      "loss": 1.2719,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324264704453228e-05,
+      "loss": 1.2552,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316976600649584e-05,
+      "loss": 1.241,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309674138447187e-05,
+      "loss": 1.3277,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302357330452972e-05,
+      "loss": 1.3036,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295026189298658e-05,
+      "loss": 1.2192,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287680727640695e-05,
+      "loss": 1.2083,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280320958160268e-05,
+      "loss": 1.2739,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272946893563247e-05,
+      "loss": 1.2211,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265558546580205e-05,
+      "loss": 1.2375,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258155929966344e-05,
+      "loss": 1.264,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825073905650152e-05,
+      "loss": 1.1998,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243307938990203e-05,
+      "loss": 1.3312,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823586259026144e-05,
+      "loss": 1.2745,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228403023168856e-05,
+      "loss": 1.2941,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220929250590625e-05,
+      "loss": 1.2748,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213441285429435e-05,
+      "loss": 1.2579,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205939140612485e-05,
+      "loss": 1.3042,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198422829091457e-05,
+      "loss": 1.3261,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190892363842474e-05,
+      "loss": 1.2939,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183347757866112e-05,
+      "loss": 1.2822,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817578902418735e-05,
+      "loss": 1.2947,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168216175855557e-05,
+      "loss": 1.2764,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160629225944473e-05,
+      "loss": 1.3044,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815302818755218e-05,
+      "loss": 1.247,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814541307380109e-05,
+      "loss": 1.3146,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137783897837897e-05,
+      "loss": 1.2736,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813014067283359e-05,
+      "loss": 1.3067,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122483411983403e-05,
+      "loss": 1.2847,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811481212850681e-05,
+      "loss": 1.2938,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810712683564748e-05,
+      "loss": 1.2774,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099427546673277e-05,
+      "loss": 1.2057,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809171427487623e-05,
+      "loss": 1.3193,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80839870335725e-05,
+      "loss": 1.2454,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076245836102368e-05,
+      "loss": 1.2655,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806849069583021e-05,
+      "loss": 1.2774,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060721626144477e-05,
+      "loss": 1.3159,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805293864045766e-05,
+      "loss": 1.2653,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045141752206277e-05,
+      "loss": 1.2936,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037330974850852e-05,
+      "loss": 1.2495,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029506321875875e-05,
+      "loss": 1.2486,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802166780678981e-05,
+      "loss": 1.27,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801381544312503e-05,
+      "loss": 1.2576,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005949244437836e-05,
+      "loss": 1.2843,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998069224308407e-05,
+      "loss": 1.2909,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799017539634077e-05,
+      "loss": 1.2601,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982267774162813e-05,
+      "loss": 1.2096,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974346371426225e-05,
+      "loss": 1.2947,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796641120180648e-05,
+      "loss": 1.2426,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958462279002834e-05,
+      "loss": 1.2035,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950499616738276e-05,
+      "loss": 1.3082,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942523228759516e-05,
+      "loss": 1.2358,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793453312883697e-05,
+      "loss": 1.2768,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926529330764707e-05,
+      "loss": 1.2737,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791851184836046e-05,
+      "loss": 1.2801,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791048069546559e-05,
+      "loss": 1.2581,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902435885945037e-05,
+      "loss": 1.2474,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894377433687337e-05,
+      "loss": 1.2783,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788630535260458e-05,
+      "loss": 1.2996,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878219656632368e-05,
+      "loss": 1.283,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870120359729822e-05,
+      "loss": 1.2889,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786200747587954e-05,
+      "loss": 1.2124,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853881019087572e-05,
+      "loss": 1.2676,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845741003383412e-05,
+      "loss": 1.2323,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837587442819946e-05,
+      "loss": 1.2979,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829420351473456e-05,
+      "loss": 1.2589,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782123974344358e-05,
+      "loss": 1.3089,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813045632853284e-05,
+      "loss": 1.219,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804838033848863e-05,
+      "loss": 1.2478,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796616960599874e-05,
+      "loss": 1.2142,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788382427299158e-05,
+      "loss": 1.2464,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780134448162777e-05,
+      "loss": 1.2899,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771873037430018e-05,
+      "loss": 1.2934,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776359820936335e-05,
+      "loss": 1.2505,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755309978248404e-05,
+      "loss": 1.2317,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747008358393958e-05,
+      "loss": 1.2781,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738693364131893e-05,
+      "loss": 1.2745,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730365009817186e-05,
+      "loss": 1.2664,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722023309827885e-05,
+      "loss": 1.2359,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713668278565062e-05,
+      "loss": 1.2581,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770529993045282e-05,
+      "loss": 1.305,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696918279938242e-05,
+      "loss": 1.2144,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768852334149138e-05,
+      "loss": 1.2052,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7680115129605228e-05,
+      "loss": 1.2261,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671693658795687e-05,
+      "loss": 1.2374,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766325894360156e-05,
+      "loss": 1.2992,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765481099858451e-05,
+      "loss": 1.2695,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646349838329035e-05,
+      "loss": 1.2445,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763787547744245e-05,
+      "loss": 1.2821,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762938793055487e-05,
+      "loss": 1.2462,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620887212319163e-05,
+      "loss": 1.2838,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761237333741093e-05,
+      "loss": 1.2394,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603846320528507e-05,
+      "loss": 1.2494,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595306176392894e-05,
+      "loss": 1.2288,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586752919747776e-05,
+      "loss": 1.21,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578186565359458e-05,
+      "loss": 1.2629,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756960712801686e-05,
+      "loss": 1.257,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75610146225315e-05,
+      "loss": 1.2881,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755240906373745e-05,
+      "loss": 1.2006,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754379046649131e-05,
+      "loss": 1.2824,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75351588456722e-05,
+      "loss": 1.2892,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752651421618171e-05,
+      "loss": 1.2371,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751785659294391e-05,
+      "loss": 1.2669,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509185990905285e-05,
+      "loss": 1.3145,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500502425034728e-05,
+      "loss": 1.2429,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749180591032352e-05,
+      "loss": 1.2373,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483096461785295e-05,
+      "loss": 1.2801,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474374094456007e-05,
+      "loss": 1.2626,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465638823393926e-05,
+      "loss": 1.2502,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456890663679593e-05,
+      "loss": 1.2234,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448129630415793e-05,
+      "loss": 1.2246,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743935573872755e-05,
+      "loss": 1.284,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430569003762077e-05,
+      "loss": 1.2018,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421769440688757e-05,
+      "loss": 1.227,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412957064699134e-05,
+      "loss": 1.2617,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404131891006852e-05,
+      "loss": 1.2572,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739529393484767e-05,
+      "loss": 1.2721,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386443211479395e-05,
+      "loss": 1.1909,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737757973618189e-05,
+      "loss": 1.217,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368703524257023e-05,
+      "loss": 1.2494,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359814591028658e-05,
+      "loss": 1.2571,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735091295184262e-05,
+      "loss": 1.2824,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341998622066657e-05,
+      "loss": 1.2458,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333071617090444e-05,
+      "loss": 1.2745,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732413195232553e-05,
+      "loss": 1.3129,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315179643205323e-05,
+      "loss": 1.2808,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306214705185048e-05,
+      "loss": 1.2773,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297237153741754e-05,
+      "loss": 1.3087,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288247004374244e-05,
+      "loss": 1.2891,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279244272603083e-05,
+      "loss": 1.2667,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727022897397056e-05,
+      "loss": 1.2512,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261201124040646e-05,
+      "loss": 1.2676,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252160738398998e-05,
+      "loss": 1.2469,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72431078326529e-05,
+      "loss": 1.2629,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234042422431258e-05,
+      "loss": 1.229,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722496452338456e-05,
+      "loss": 1.3131,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721587415118486e-05,
+      "loss": 1.2284,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206771321525745e-05,
+      "loss": 1.2743,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719765605012231e-05,
+      "loss": 1.2922,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718852835271113e-05,
+      "loss": 1.26,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717938824505022e-05,
+      "loss": 1.297,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717023574291903e-05,
+      "loss": 1.2917,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716107086211842e-05,
+      "loss": 1.235,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715189361847059e-05,
+      "loss": 1.3189,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142704027819115e-05,
+      "loss": 1.2769,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133502106028864e-05,
+      "loss": 1.2297,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124287868986004e-05,
+      "loss": 1.2448,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711506133259796e-05,
+      "loss": 1.248,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105822512793392e-05,
+      "loss": 1.2445,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709657142552217e-05,
+      "loss": 1.2788,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708730808675533e-05,
+      "loss": 1.2706,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078032512485073e-05,
+      "loss": 1.2602,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068744718724713e-05,
+      "loss": 1.2517,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059444721508667e-05,
+      "loss": 1.2349,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050132536892415e-05,
+      "loss": 1.2287,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040808180952485e-05,
+      "loss": 1.2968,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031471669786402e-05,
+      "loss": 1.2777,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70221230195127e-05,
+      "loss": 1.261,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701276224627084e-05,
+      "loss": 1.3257,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700338936622123e-05,
+      "loss": 1.2018,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699400439554519e-05,
+      "loss": 1.2344,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984607350444885e-05,
+      "loss": 1.1981,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697519824714335e-05,
+      "loss": 1.2702,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965777101884423e-05,
+      "loss": 1.2165,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956343930932733e-05,
+      "loss": 1.2724,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694689875057368e-05,
+      "loss": 1.2422,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693744157711338e-05,
+      "loss": 1.2068,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927972426878674e-05,
+      "loss": 1.2007,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918491316217063e-05,
+      "loss": 1.2204,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908998261496704e-05,
+      "loss": 1.2474,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899493279106373e-05,
+      "loss": 1.2286,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889976385455435e-05,
+      "loss": 1.24,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880447596973826e-05,
+      "loss": 1.2873,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687090693011201e-05,
+      "loss": 1.1807,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686135440134096e-05,
+      "loss": 1.1958,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685179002715213e-05,
+      "loss": 1.2636,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842213824057414e-05,
+      "loss": 1.2246,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832625808589144e-05,
+      "loss": 1.2832,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823025997300034e-05,
+      "loss": 1.2307,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681341440676316e-05,
+      "loss": 1.2732,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680379105357194e-05,
+      "loss": 1.205,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794155954340098e-05,
+      "loss": 1.2888,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784509125701634e-05,
+      "loss": 1.22,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774850584310795e-05,
+      "loss": 1.2056,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765180346842057e-05,
+      "loss": 1.2629,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755498429990082e-05,
+      "loss": 1.2702,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745804850469694e-05,
+      "loss": 1.2576,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736099625015862e-05,
+      "loss": 1.2276,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726382770383645e-05,
+      "loss": 1.2789,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671665430334819e-05,
+      "loss": 1.2591,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706914240704682e-05,
+      "loss": 1.2464,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697162599268342e-05,
+      "loss": 1.249,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687399395874356e-05,
+      "loss": 1.2832,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677624647377893e-05,
+      "loss": 1.2576,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666783837065404e-05,
+      "loss": 1.2387,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665804058259779e-05,
+      "loss": 1.2255,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664823130012401e-05,
+      "loss": 1.224,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663841054016741e-05,
+      "loss": 1.2108,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628578319682515e-05,
+      "loss": 1.2505,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618734655643637e-05,
+      "loss": 1.2339,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660887956504484e-05,
+      "loss": 1.2713,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659901306489992e-05,
+      "loss": 1.1898,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589135172242363e-05,
+      "loss": 1.2744,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579245904125335e-05,
+      "loss": 1.2136,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569345277621625e-05,
+      "loss": 1.2174,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655943330982365e-05,
+      "loss": 1.2315,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654951001784338e-05,
+      "loss": 1.2276,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653957541881237e-05,
+      "loss": 1.1954,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652962952988166e-05,
+      "loss": 1.2255,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519672368221808e-05,
+      "loss": 1.2505,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509703951022813e-05,
+      "loss": 1.2521,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499724295494125e-05,
+      "loss": 1.2681,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489733418864578e-05,
+      "loss": 1.209,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647973133838239e-05,
+      "loss": 1.2499,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469718071315114e-05,
+      "loss": 1.2549,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459693634949624e-05,
+      "loss": 1.2104,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449658046592065e-05,
+      "loss": 1.2934,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439611323567844e-05,
+      "loss": 1.2486,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429553483221584e-05,
+      "loss": 1.2231,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641948454291711e-05,
+      "loss": 1.2852,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409404520037406e-05,
+      "loss": 1.203,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399313431984583e-05,
+      "loss": 1.1871,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638921129617986e-05,
+      "loss": 1.1715,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637909813006353e-05,
+      "loss": 1.2439,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636897395109492e-05,
+      "loss": 1.2257,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635883877675239e-05,
+      "loss": 1.2789,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348692624533258e-05,
+      "loss": 1.2337,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338535511953807e-05,
+      "loss": 1.2333,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328367456549245e-05,
+      "loss": 1.2039,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318188475873657e-05,
+      "loss": 1.197,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630799858750001e-05,
+      "loss": 1.2612,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629779780902008e-05,
+      "loss": 1.2612,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287586158044467e-05,
+      "loss": 1.2039,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627736365220252e-05,
+      "loss": 1.2119,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267130309142337e-05,
+      "loss": 1.2274,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625688614653073e-05,
+      "loss": 1.2565,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246631182053177e-05,
+      "loss": 1.2539,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623636543341382e-05,
+      "loss": 1.2305,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226088918335406e-05,
+      "loss": 1.2458,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215801654559276e-05,
+      "loss": 1.2283,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205503659845328e-05,
+      "loss": 1.215,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619519495197198e-05,
+      "loss": 1.215,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184875548736142e-05,
+      "loss": 1.2491,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617454546795321e-05,
+      "loss": 1.2871,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616420472745699e-05,
+      "loss": 1.2256,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153853345099694e-05,
+      "loss": 1.2525,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143491338751922e-05,
+      "loss": 1.2252,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133118726302603e-05,
+      "loss": 1.2667,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122735525658974e-05,
+      "loss": 1.2284,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112341754746553e-05,
+      "loss": 1.2494,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101937431509118e-05,
+      "loss": 1.2198,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609152257390865e-05,
+      "loss": 1.2375,
+      "step": 771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081097199925325e-05,
+      "loss": 1.2129,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070661327557465e-05,
+      "loss": 1.2472,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060214974821532e-05,
+      "loss": 1.2812,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049758159752062e-05,
+      "loss": 1.2528,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039290900401674e-05,
+      "loss": 1.2756,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028813214841e-05,
+      "loss": 1.2515,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601832512115868e-05,
+      "loss": 1.18,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007826637461323e-05,
+      "loss": 1.234,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997317781873474e-05,
+      "loss": 1.2143,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986798572537584e-05,
+      "loss": 1.1755,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976269027613975e-05,
+      "loss": 1.1912,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965729165280822e-05,
+      "loss": 1.2246,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59551790037341e-05,
+      "loss": 1.192,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944618561187573e-05,
+      "loss": 1.2346,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593404785587275e-05,
+      "loss": 1.2591,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592346690603886e-05,
+      "loss": 1.2364,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912875729952817e-05,
+      "loss": 1.2439,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590227434589919e-05,
+      "loss": 1.1754,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5891662772180173e-05,
+      "loss": 1.2029,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588104102711554e-05,
+      "loss": 1.2492,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870409129042645e-05,
+      "loss": 1.2877,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859767096316354e-05,
+      "loss": 1.2349,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849114947309033e-05,
+      "loss": 1.2471,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838452700410517e-05,
+      "loss": 1.2581,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827780374028065e-05,
+      "loss": 1.2729,
+      "step": 796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817097986586355e-05,
+      "loss": 1.2261,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580640555652741e-05,
+      "loss": 1.2214,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795703102310613e-05,
+      "loss": 1.213,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784990642412635e-05,
+      "loss": 1.2156,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774268195327432e-05,
+      "loss": 1.2538,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57635357795662e-05,
+      "loss": 1.2796,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752793413657336e-05,
+      "loss": 1.2352,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742041116146433e-05,
+      "loss": 1.2184,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731278905596215e-05,
+      "loss": 1.1891,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720506800586522e-05,
+      "loss": 1.1997,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709724819714277e-05,
+      "loss": 1.2088,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698932981593456e-05,
+      "loss": 1.2342,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688131304855047e-05,
+      "loss": 1.2509,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677319808147032e-05,
+      "loss": 1.2611,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666498510134335e-05,
+      "loss": 1.2088,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655667429498803e-05,
+      "loss": 1.2196,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644826584939185e-05,
+      "loss": 1.225,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563397599517107e-05,
+      "loss": 1.2056,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562311567892688e-05,
+      "loss": 1.2829,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561224565495582e-05,
+      "loss": 1.2498,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560136594202387e-05,
+      "loss": 1.2278,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559047655891372e-05,
+      "loss": 1.1795,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557957752442477e-05,
+      "loss": 1.2368,
+      "step": 819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556866885737307e-05,
+      "loss": 1.2947,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557750576591297e-05,
+      "loss": 1.2475,
+      "step": 821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546822700928744e-05,
+      "loss": 1.244,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553588524925125e-05,
+      "loss": 1.2871,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552493824044119e-05,
+      "loss": 1.2486,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551398169339745e-05,
+      "loss": 1.2028,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503015627035365e-05,
+      "loss": 1.2265,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549204006028672e-05,
+      "loss": 1.287,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481055012099686e-05,
+      "loss": 1.2091,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547006050143882e-05,
+      "loss": 1.2238,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459056547284998e-05,
+      "loss": 1.2092,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448043168635413e-05,
+      "loss": 1.2436,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543702038450352e-05,
+      "loss": 1.2426,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425988213919016e-05,
+      "loss": 1.203,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54149466759278e-05,
+      "loss": 1.2608,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403895789591943e-05,
+      "loss": 1.214,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392835573989656e-05,
+      "loss": 1.2465,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538176604821526e-05,
+      "loss": 1.1802,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370687231379135e-05,
+      "loss": 1.1842,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359599142607722e-05,
+      "loss": 1.2874,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534850180104345e-05,
+      "loss": 1.2041,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337395225844733e-05,
+      "loss": 1.2304,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326279436185923e-05,
+      "loss": 1.2239,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315154451257276e-05,
+      "loss": 1.2092,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530402029026494e-05,
+      "loss": 1.2408,
+      "step": 844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529287697243087e-05,
+      "loss": 1.2561,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5281724516992863e-05,
+      "loss": 1.196,
+      "step": 846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527056294320448e-05,
+      "loss": 1.2538,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259392270335017e-05,
+      "loss": 1.1747,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524821251766948e-05,
+      "loss": 1.2267,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237023704508564e-05,
+      "loss": 1.2716,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522582585016859e-05,
+      "loss": 1.2275,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214618973981496e-05,
+      "loss": 1.2339,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203403095294794e-05,
+      "loss": 1.2094,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519217823347153e-05,
+      "loss": 1.2438,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180944407890275e-05,
+      "loss": 1.1846,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169701637945057e-05,
+      "loss": 1.2618,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158449943045361e-05,
+      "loss": 1.2028,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147189342616072e-05,
+      "loss": 1.1873,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135919856097445e-05,
+      "loss": 1.1632,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512464150294509e-05,
+      "loss": 1.1899,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113354302629904e-05,
+      "loss": 1.191,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510205827463808e-05,
+      "loss": 1.2195,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090753438471034e-05,
+      "loss": 1.2226,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50794398136454e-05,
+      "loss": 1.2309,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068117419692976e-05,
+      "loss": 1.2154,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056786276160703e-05,
+      "loss": 1.1916,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045446402610626e-05,
+      "loss": 1.2307,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034097818619863e-05,
+      "loss": 1.2702,
+      "step": 868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022740543780572e-05,
+      "loss": 1.2346,
+      "step": 869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011374597699911e-05,
+      "loss": 1.1746,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.183,
+      "step": 871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988616770317915e-05,
+      "loss": 1.2253,
+      "step": 872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977224928305623e-05,
+      "loss": 1.2425,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496582449362995e-05,
+      "loss": 1.2999,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954415485972573e-05,
+      "loss": 1.2645,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942997925029961e-05,
+      "loss": 1.2373,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493157183051335e-05,
+      "loss": 1.2426,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492013722214871e-05,
+      "loss": 1.2401,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908694119676703e-05,
+      "loss": 1.2061,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897242542852667e-05,
+      "loss": 1.2287,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885782511446562e-05,
+      "loss": 1.22,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874314045242941e-05,
+      "loss": 1.2224,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862837164040928e-05,
+      "loss": 1.2458,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851351887654168e-05,
+      "loss": 1.254,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839858235910802e-05,
+      "loss": 1.2544,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482835622865343e-05,
+      "loss": 1.2123,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816845885739071e-05,
+      "loss": 1.1965,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805327227039147e-05,
+      "loss": 1.1995,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793800272439427e-05,
+      "loss": 1.2668,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782265041839999e-05,
+      "loss": 1.2397,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770721555155251e-05,
+      "loss": 1.2688,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759169832313818e-05,
+      "loss": 1.2621,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747609893258546e-05,
+      "loss": 1.2392,
+      "step": 893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736041757946477e-05,
+      "loss": 1.2271,
+      "step": 894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724465446348801e-05,
+      "loss": 1.2166,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712880978450818e-05,
+      "loss": 1.2513,
+      "step": 896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701288374251912e-05,
+      "loss": 1.2135,
+      "step": 897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468968765376551e-05,
+      "loss": 1.2297,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678078837019064e-05,
+      "loss": 1.2242,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666461944053984e-05,
+      "loss": 1.2048,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465483699492564e-05,
+      "loss": 1.252,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643204009703298e-05,
+      "loss": 1.2243,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631563008470102e-05,
+      "loss": 1.2737,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619914011323042e-05,
+      "loss": 1.2788,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608257038372894e-05,
+      "loss": 1.2382,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596592109744225e-05,
+      "loss": 1.2325,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584919245575323e-05,
+      "loss": 1.2698,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573238466018173e-05,
+      "loss": 1.2001,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561549791238438e-05,
+      "loss": 1.2117,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549853241415404e-05,
+      "loss": 1.1744,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453814883674195e-05,
+      "loss": 1.2546,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452643659742452e-05,
+      "loss": 1.2254,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514716543683083e-05,
+      "loss": 1.2671,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45029886957511e-05,
+      "loss": 1.2226,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491253073875477e-05,
+      "loss": 1.2105,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479509698316556e-05,
+      "loss": 1.2524,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467758589348065e-05,
+      "loss": 1.1855,
+      "step": 917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455999767257068e-05,
+      "loss": 1.1997,
+      "step": 918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444233252343951e-05,
+      "loss": 1.2732,
+      "step": 919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432459064922393e-05,
+      "loss": 1.2493,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420677225319301e-05,
+      "loss": 1.2339,
+      "step": 921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440888775387481e-05,
+      "loss": 1.2378,
+      "step": 922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397090670942208e-05,
+      "loss": 1.2533,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438528599688795e-05,
+      "loss": 1.2285,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373473752091575e-05,
+      "loss": 1.2694,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361653956945708e-05,
+      "loss": 1.1898,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349826631855993e-05,
+      "loss": 1.1779,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337991797241082e-05,
+      "loss": 1.2194,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326149473532598e-05,
+      "loss": 1.149,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314299681175085e-05,
+      "loss": 1.1935,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302442440625974e-05,
+      "loss": 1.2147,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290577772355575e-05,
+      "loss": 1.2494,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278705696847002e-05,
+      "loss": 1.2737,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266826234596163e-05,
+      "loss": 1.1874,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254939406111723e-05,
+      "loss": 1.2564,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243045231915058e-05,
+      "loss": 1.2364,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423114373254023e-05,
+      "loss": 1.2542,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421923492853394e-05,
+      "loss": 1.1722,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420731884045551e-05,
+      "loss": 1.1909,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195395488876833e-05,
+      "loss": 1.2462,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183464894382335e-05,
+      "loss": 1.2062,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171527077568953e-05,
+      "loss": 1.2087,
+      "step": 942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159582059046097e-05,
+      "loss": 1.226,
+      "step": 943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147629859435601e-05,
+      "loss": 1.1946,
+      "step": 944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135670499371698e-05,
+      "loss": 1.2044,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123703999500987e-05,
+      "loss": 1.2047,
+      "step": 946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111730380482387e-05,
+      "loss": 1.205,
+      "step": 947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099749662987118e-05,
+      "loss": 1.1955,
+      "step": 948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408776186769864e-05,
+      "loss": 1.1973,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075767015312644e-05,
+      "loss": 1.3095,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063765126537005e-05,
+      "loss": 1.2135,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051756222091735e-05,
+      "loss": 1.2075,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039740322708961e-05,
+      "loss": 1.187,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027717449132897e-05,
+      "loss": 1.2304,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015687622119785e-05,
+      "loss": 1.2127,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003650862437874e-05,
+      "loss": 1.1795,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991607190867385e-05,
+      "loss": 1.2435,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397955662820047e-05,
+      "loss": 1.2159,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967499195241176e-05,
+      "loss": 1.1949,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955434912805413e-05,
+      "loss": 1.2201,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943363801720918e-05,
+      "loss": 1.2381,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931285882827208e-05,
+      "loss": 1.2888,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919201176975566e-05,
+      "loss": 1.2135,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907109705028979e-05,
+      "loss": 1.2314,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895011487862123e-05,
+      "loss": 1.2547,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882906546361322e-05,
+      "loss": 1.1997,
+      "step": 966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870794901424493e-05,
+      "loss": 1.2688,
+      "step": 967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858676573961147e-05,
+      "loss": 1.24,
+      "step": 968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846551584892318e-05,
+      "loss": 1.2359,
+      "step": 969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834419955150545e-05,
+      "loss": 1.2159,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822281705679826e-05,
+      "loss": 1.1884,
+      "step": 971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810136857435591e-05,
+      "loss": 1.2307,
+      "step": 972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797985431384672e-05,
+      "loss": 1.2467,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785827448505243e-05,
+      "loss": 1.2161,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773662929786795e-05,
+      "loss": 1.2942,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761491896230119e-05,
+      "loss": 1.1754,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749314368847242e-05,
+      "loss": 1.2159,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737130368661398e-05,
+      "loss": 1.2631,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724939916707013e-05,
+      "loss": 1.2621,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712743034029628e-05,
+      "loss": 1.2137,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700539741685903e-05,
+      "loss": 1.2061,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368833006074356e-05,
+      "loss": 1.2678,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676114012281345e-05,
+      "loss": 1.2681,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663891617389005e-05,
+      "loss": 1.1937,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651662897167234e-05,
+      "loss": 1.2038,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639427872727654e-05,
+      "loss": 1.2224,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627186565192767e-05,
+      "loss": 1.2425,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614938995695921e-05,
+      "loss": 1.2403,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602685185381282e-05,
+      "loss": 1.1842,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590425155403777e-05,
+      "loss": 1.2813,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578158926929081e-05,
+      "loss": 1.2029,
+      "step": 991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565886521133572e-05,
+      "loss": 1.2498,
+      "step": 992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553607959204276e-05,
+      "loss": 1.2103,
+      "step": 993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541323262338868e-05,
+      "loss": 1.2421,
+      "step": 994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529032451745608e-05,
+      "loss": 1.2056,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516735548643296e-05,
+      "loss": 1.2378,
+      "step": 996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350443257426127e-05,
+      "loss": 1.1846,
+      "step": 997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492123549839339e-05,
+      "loss": 1.2534,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479808496627758e-05,
+      "loss": 1.2138,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467487435887192e-05,
+      "loss": 1.2259,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455160388888678e-05,
+      "loss": 1.2045,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442827376913581e-05,
+      "loss": 1.2277,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430488421253576e-05,
+      "loss": 1.2241,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418143543210585e-05,
+      "loss": 1.2373,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405792764096768e-05,
+      "loss": 1.2149,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393436105234462e-05,
+      "loss": 1.2345,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381073587956154e-05,
+      "loss": 1.2066,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336870523360446e-05,
+      "loss": 1.2545,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356331063532058e-05,
+      "loss": 1.2198,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343951099101671e-05,
+      "loss": 1.2573,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331565361686023e-05,
+      "loss": 1.2531,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319173872667812e-05,
+      "loss": 1.2174,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330677665343966e-05,
+      "loss": 1.2654,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294373725404077e-05,
+      "loss": 1.2215,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281965109973439e-05,
+      "loss": 1.2734,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326955082856993e-05,
+      "loss": 1.1704,
+      "step": 1016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257130902625528e-05,
+      "loss": 1.2808,
+      "step": 1017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244705353581946e-05,
+      "loss": 1.2034,
+      "step": 1018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232274202890602e-05,
+      "loss": 1.2268,
+      "step": 1019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219837472012598e-05,
+      "loss": 1.2018,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320739518241866e-05,
+      "loss": 1.1844,
+      "step": 1021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319494735558911e-05,
+      "loss": 1.1852,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182494013013836e-05,
+      "loss": 1.1789,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317003517619224e-05,
+      "loss": 1.2675,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157570866633216e-05,
+      "loss": 1.2095,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31451011058551e-05,
+      "loss": 1.249,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132625915385647e-05,
+      "loss": 1.1795,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120145316761979e-05,
+      "loss": 1.2369,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310765933153056e-05,
+      "loss": 1.2865,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095167981247143e-05,
+      "loss": 1.2313,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308267128747676e-05,
+      "loss": 1.2092,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070169271793653e-05,
+      "loss": 1.253,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057661955781258e-05,
+      "loss": 1.1952,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045149361032161e-05,
+      "loss": 1.1611,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032631509148062e-05,
+      "loss": 1.2058,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020108421739735e-05,
+      "loss": 1.2295,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007580120426993e-05,
+      "loss": 1.1871,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299504662683865e-05,
+      "loss": 1.2324,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298250796261249e-05,
+      "loss": 1.21,
+      "step": 1039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969964149395212e-05,
+      "loss": 1.2051,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957415208842412e-05,
+      "loss": 1.2437,
+      "step": 1041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944861162618536e-05,
+      "loss": 1.1786,
+      "step": 1042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932302032396849e-05,
+      "loss": 1.1942,
+      "step": 1043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919737839859382e-05,
+      "loss": 1.2522,
+      "step": 1044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907168606696916e-05,
+      "loss": 1.1958,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289459435460893e-05,
+      "loss": 1.2021,
+      "step": 1046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882015105303569e-05,
+      "loss": 1.157,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869430880497599e-05,
+      "loss": 1.2172,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856841701916389e-05,
+      "loss": 1.196,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844247591293846e-05,
+      "loss": 1.2608,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831648570372397e-05,
+      "loss": 1.183,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819044660902953e-05,
+      "loss": 1.2048,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806435884644853e-05,
+      "loss": 1.1992,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793822263365843e-05,
+      "loss": 1.2054,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781203818842038e-05,
+      "loss": 1.2434,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768580572857874e-05,
+      "loss": 1.2519,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755952547206079e-05,
+      "loss": 1.2244,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743319763687627e-05,
+      "loss": 1.216,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730682244111715e-05,
+      "loss": 1.2134,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718040010295706e-05,
+      "loss": 1.2012,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705393084065114e-05,
+      "loss": 1.236,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692741487253546e-05,
+      "loss": 1.1757,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680085241702669e-05,
+      "loss": 1.2342,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667424369262186e-05,
+      "loss": 1.1822,
+      "step": 1064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654758891789782e-05,
+      "loss": 1.2067,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642088831151086e-05,
+      "loss": 1.1771,
+      "step": 1066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262941420921965e-05,
+      "loss": 1.1798,
+      "step": 1067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616735047876897e-05,
+      "loss": 1.2111,
+      "step": 1068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604051369012083e-05,
+      "loss": 1.2091,
+      "step": 1069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591363194522268e-05,
+      "loss": 1.1791,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257867054631227e-05,
+      "loss": 1.2052,
+      "step": 1071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565973446294632e-05,
+      "loss": 1.1783,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553271916389587e-05,
+      "loss": 1.2311,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540565978525e-05,
+      "loss": 1.1891,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527855654636367e-05,
+      "loss": 1.2063,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515140966666742e-05,
+      "loss": 1.2411,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502421936566713e-05,
+      "loss": 1.1631,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489698586294376e-05,
+      "loss": 1.2058,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476970937815274e-05,
+      "loss": 1.2057,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464239013102373e-05,
+      "loss": 1.2616,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451502834136025e-05,
+      "loss": 1.182,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438762422903921e-05,
+      "loss": 1.2747,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426017801401062e-05,
+      "loss": 1.2508,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413268991629719e-05,
+      "loss": 1.2074,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400516015599389e-05,
+      "loss": 1.1937,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238775889532677e-05,
+      "loss": 1.1811,
+      "step": 1086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23749976528357e-05,
+      "loss": 1.2451,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362232310157149e-05,
+      "loss": 1.1847,
+      "step": 1088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349462889329158e-05,
+      "loss": 1.173,
+      "step": 1089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336689412396811e-05,
+      "loss": 1.2289,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232391190141219e-05,
+      "loss": 1.1724,
+      "step": 1091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311130378434346e-05,
+      "loss": 1.1885,
+      "step": 1092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298344865529254e-05,
+      "loss": 1.2361,
+      "step": 1093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228555538476978e-05,
+      "loss": 1.2442,
+      "step": 1094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272761958235634e-05,
+      "loss": 1.2065,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259964608013346e-05,
+      "loss": 1.1868,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247163356196212e-05,
+      "loss": 1.1926,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223435822488427e-05,
+      "loss": 1.2018,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221549236184255e-05,
+      "loss": 1.1902,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208736412209557e-05,
+      "loss": 1.23,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219591977508019e-05,
+      "loss": 1.1874,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218309934692275e-05,
+      "loss": 1.209,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170275149870374e-05,
+      "loss": 1.173,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215744720606272e-05,
+      "loss": 1.1968,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144615537645898e-05,
+      "loss": 1.188,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131780166772458e-05,
+      "loss": 1.2074,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118941115601334e-05,
+      "loss": 1.2078,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106098406297822e-05,
+      "loss": 1.2301,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093252061033532e-05,
+      "loss": 1.2037,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080402101986337e-05,
+      "loss": 1.2421,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067548551340373e-05,
+      "loss": 1.21,
+      "step": 1111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054691431285962e-05,
+      "loss": 1.2135,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041830764019584e-05,
+      "loss": 1.2287,
+      "step": 1113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028966571743851e-05,
+      "loss": 1.2379,
+      "step": 1114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016098876667465e-05,
+      "loss": 1.1677,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003227701005163e-05,
+      "loss": 1.2259,
+      "step": 1116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19903530669777e-05,
+      "loss": 1.187,
+      "step": 1117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977474996811793e-05,
+      "loss": 1.2123,
+      "step": 1118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964593512740103e-05,
+      "loss": 1.1554,
+      "step": 1119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951708637001173e-05,
+      "loss": 1.1852,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938820391839408e-05,
+      "loss": 1.2363,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925928799505033e-05,
+      "loss": 1.2302,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913033882254042e-05,
+      "loss": 1.1714,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900135662348173e-05,
+      "loss": 1.244,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887234162054872e-05,
+      "loss": 1.192,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874329403647237e-05,
+      "loss": 1.1878,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861421409404006e-05,
+      "loss": 1.2248,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848510201609487e-05,
+      "loss": 1.2068,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835595802553542e-05,
+      "loss": 1.1989,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822678234531549e-05,
+      "loss": 1.2362,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809757519844344e-05,
+      "loss": 1.1638,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796833680798212e-05,
+      "loss": 1.2021,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783906739704819e-05,
+      "loss": 1.2136,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770976718881185e-05,
+      "loss": 1.1788,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758043640649659e-05,
+      "loss": 1.1635,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745107527337855e-05,
+      "loss": 1.1783,
+      "step": 1136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732168401278638e-05,
+      "loss": 1.1728,
+      "step": 1137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171922628481007e-05,
+      "loss": 1.1727,
+      "step": 1138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706281200275368e-05,
+      "loss": 1.1939,
+      "step": 1139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693333170022884e-05,
+      "loss": 1.1899,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680382216406052e-05,
+      "loss": 1.188,
+      "step": 1141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166742836178335e-05,
+      "loss": 1.1683,
+      "step": 1142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165447162851827e-05,
+      "loss": 1.2164,
+      "step": 1143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641512038979265e-05,
+      "loss": 1.1878,
+      "step": 1144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162854961553973e-05,
+      "loss": 1.2033,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615584380577941e-05,
+      "loss": 1.2113,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602616356477036e-05,
+      "loss": 1.2025,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589645565624968e-05,
+      "loss": 1.1962,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157667203041446e-05,
+      "loss": 1.183,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563695773242978e-05,
+      "loss": 1.2589,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550716816512686e-05,
+      "loss": 1.1769,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537735182630409e-05,
+      "loss": 1.2426,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524750894007599e-05,
+      "loss": 1.1617,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511763973060273e-05,
+      "loss": 1.1969,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498774442209018e-05,
+      "loss": 1.1934,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485782323878907e-05,
+      "loss": 1.2161,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472787640499488e-05,
+      "loss": 1.1721,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459790414504735e-05,
+      "loss": 1.1711,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446790668333012e-05,
+      "loss": 1.2361,
+      "step": 1159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433788424427038e-05,
+      "loss": 1.1823,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420783705233843e-05,
+      "loss": 1.2067,
+      "step": 1161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407776533204715e-05,
+      "loss": 1.1851,
+      "step": 1162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394766930795202e-05,
+      "loss": 1.1932,
+      "step": 1163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381754920465033e-05,
+      "loss": 1.2046,
+      "step": 1164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136874052467809e-05,
+      "loss": 1.2009,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355723765902388e-05,
+      "loss": 1.1893,
+      "step": 1166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134270466661001e-05,
+      "loss": 1.1655,
+      "step": 1167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329683249277076e-05,
+      "loss": 1.1953,
+      "step": 1168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316659536383723e-05,
+      "loss": 1.2156,
+      "step": 1169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303633550414039e-05,
+      "loss": 1.1651,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290605313856045e-05,
+      "loss": 1.2156,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277574849201632e-05,
+      "loss": 1.1867,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126454217894655e-05,
+      "loss": 1.2291,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251507325590366e-05,
+      "loss": 1.1827,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238470311636392e-05,
+      "loss": 1.2224,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225431159591684e-05,
+      "loss": 1.1649,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212389891966993e-05,
+      "loss": 1.1498,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199346531276714e-05,
+      "loss": 1.1995,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118630110003886e-05,
+      "loss": 1.1944,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173253620775009e-05,
+      "loss": 1.1291,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160204116010291e-05,
+      "loss": 1.1667,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147152608273329e-05,
+      "loss": 1.2066,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134099120096187e-05,
+      "loss": 1.1605,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121043674014367e-05,
+      "loss": 1.2093,
+      "step": 1184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107986292566746e-05,
+      "loss": 1.1984,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094926998295536e-05,
+      "loss": 1.1856,
+      "step": 1186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081865813746259e-05,
+      "loss": 1.1472,
+      "step": 1187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10688027614677e-05,
+      "loss": 1.1661,
+      "step": 1188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055737864011861e-05,
+      "loss": 1.1724,
+      "step": 1189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042671143933941e-05,
+      "loss": 1.2084,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029602623792268e-05,
+      "loss": 1.1985,
+      "step": 1191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016532326148301e-05,
+      "loss": 1.1526,
+      "step": 1192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003460273566546e-05,
+      "loss": 1.1998,
+      "step": 1193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990386488614553e-05,
+      "loss": 1.1733,
+      "step": 1194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977310993862856e-05,
+      "loss": 1.1659,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964233811884942e-05,
+      "loss": 1.2009,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951154965257214e-05,
+      "loss": 1.1923,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938074476558939e-05,
+      "loss": 1.2249,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924992368372236e-05,
+      "loss": 1.1671,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911908663282007e-05,
+      "loss": 1.2281,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898823383875909e-05,
+      "loss": 1.2126,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885736552744325e-05,
+      "loss": 1.1268,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872648192480314e-05,
+      "loss": 1.2141,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859558325679574e-05,
+      "loss": 1.1874,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846466974940403e-05,
+      "loss": 1.2365,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833374162863664e-05,
+      "loss": 1.1994,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820279912052741e-05,
+      "loss": 1.1548,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807184245113505e-05,
+      "loss": 1.1961,
+      "step": 1208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794087184654258e-05,
+      "loss": 1.1821,
+      "step": 1209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780988753285732e-05,
+      "loss": 1.2035,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767888973621006e-05,
+      "loss": 1.2434,
+      "step": 1211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754787868275489e-05,
+      "loss": 1.1676,
+      "step": 1212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074168545986689e-05,
+      "loss": 1.1792,
+      "step": 1213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728581771015154e-05,
+      "loss": 1.2105,
+      "step": 1214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715476824342448e-05,
+      "loss": 1.2233,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702370642473104e-05,
+      "loss": 1.2111,
+      "step": 1216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689263248033581e-05,
+      "loss": 1.1908,
+      "step": 1217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676154663652449e-05,
+      "loss": 1.2061,
+      "step": 1218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663044911960311e-05,
+      "loss": 1.1647,
+      "step": 1219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649934015589805e-05,
+      "loss": 1.2724,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636821997175528e-05,
+      "loss": 1.2066,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0623708879354025e-05,
+      "loss": 1.2011,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610594684763737e-05,
+      "loss": 1.1714,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597479436044959e-05,
+      "loss": 1.2005,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584363155839815e-05,
+      "loss": 1.2285,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571245866792204e-05,
+      "loss": 1.1718,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055812759154776e-05,
+      "loss": 1.2213,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545008352753834e-05,
+      "loss": 1.1706,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531888173059428e-05,
+      "loss": 1.2122,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051876707511518e-05,
+      "loss": 1.1708,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05056450815733e-05,
+      "loss": 1.2367,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049252221508755e-05,
+      "loss": 1.1687,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047939849831321e-05,
+      "loss": 1.1568,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466273953907009e-05,
+      "loss": 1.2214,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453148604527112e-05,
+      "loss": 1.2333,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440022472833085e-05,
+      "loss": 1.2374,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426895581485826e-05,
+      "loss": 1.1506,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413767953147558e-05,
+      "loss": 1.1337,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400639610481768e-05,
+      "loss": 1.1694,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387510576153185e-05,
+      "loss": 1.2163,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374380872827725e-05,
+      "loss": 1.1291,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036125052317246e-05,
+      "loss": 1.2027,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348119549855577e-05,
+      "loss": 1.1437,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033498797554635e-05,
+      "loss": 1.1211,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.032185582291507e-05,
+      "loss": 1.1419,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308723114633048e-05,
+      "loss": 1.2233,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029558987337254e-05,
+      "loss": 1.1766,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282456121806728e-05,
+      "loss": 1.1627,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269321882609675e-05,
+      "loss": 1.22,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025618717845628e-05,
+      "loss": 1.1875,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243052032022257e-05,
+      "loss": 1.231,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229916465984074e-05,
+      "loss": 1.2004,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216780503018919e-05,
+      "loss": 1.2124,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020364416580468e-05,
+      "loss": 1.245,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190507477019878e-05,
+      "loss": 1.227,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177370459343648e-05,
+      "loss": 1.2205,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164233135455686e-05,
+      "loss": 1.18,
+      "step": 1257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151095528036226e-05,
+      "loss": 1.1711,
+      "step": 1258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137957659765985e-05,
+      "loss": 1.2521,
+      "step": 1259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124819553326133e-05,
+      "loss": 1.1843,
+      "step": 1260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111681231398246e-05,
+      "loss": 1.2282,
+      "step": 1261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098542716664282e-05,
+      "loss": 1.1439,
+      "step": 1262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085404031806524e-05,
+      "loss": 1.1875,
+      "step": 1263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0072265199507543e-05,
+      "loss": 1.1705,
+      "step": 1264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059126242450187e-05,
+      "loss": 1.1978,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045987183317494e-05,
+      "loss": 1.1881,
+      "step": 1266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003284804479269e-05,
+      "loss": 1.1961,
+      "step": 1267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019708849559142e-05,
+      "loss": 1.1702,
+      "step": 1268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006569620300303e-05,
+      "loss": 1.1558,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993430379699698e-06,
+      "loss": 1.1998,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980291150440861e-06,
+      "loss": 1.2164,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96715195520731e-06,
+      "loss": 1.1811,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954012816682511e-06,
+      "loss": 1.1816,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940873757549815e-06,
+      "loss": 1.1309,
+      "step": 1274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927734800492457e-06,
+      "loss": 1.1936,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914595968193481e-06,
+      "loss": 1.1503,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90145728333572e-06,
+      "loss": 1.1962,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888318768601758e-06,
+      "loss": 1.1506,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875180446673872e-06,
+      "loss": 1.1506,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862042340234019e-06,
+      "loss": 1.1776,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848904471963776e-06,
+      "loss": 1.2006,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835766864544315e-06,
+      "loss": 1.1567,
+      "step": 1282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822629540656357e-06,
+      "loss": 1.1461,
+      "step": 1283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809492522980123e-06,
+      "loss": 1.2348,
+      "step": 1284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796355834195323e-06,
+      "loss": 1.1644,
+      "step": 1285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783219496981083e-06,
+      "loss": 1.1917,
+      "step": 1286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77008353401593e-06,
+      "loss": 1.2401,
+      "step": 1287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756947967977744e-06,
+      "loss": 1.1923,
+      "step": 1288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74381282154372e-06,
+      "loss": 1.1482,
+      "step": 1289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730678117390328e-06,
+      "loss": 1.1679,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717543878193276e-06,
+      "loss": 1.1281,
+      "step": 1291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704410126627464e-06,
+      "loss": 1.1885,
+      "step": 1292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691276885366956e-06,
+      "loss": 1.1429,
+      "step": 1293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678144177084933e-06,
+      "loss": 1.2364,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665012024453656e-06,
+      "loss": 1.1722,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651880450144423e-06,
+      "loss": 1.1912,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638749476827542e-06,
+      "loss": 1.1524,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625619127172278e-06,
+      "loss": 1.2492,
+      "step": 1298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612489423846816e-06,
+      "loss": 1.163,
+      "step": 1299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599360389518234e-06,
+      "loss": 1.1865,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586232046852447e-06,
+      "loss": 1.1302,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573104418514179e-06,
+      "loss": 1.1248,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559977527166918e-06,
+      "loss": 1.2017,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54685139547289e-06,
+      "loss": 1.2324,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533726046092995e-06,
+      "loss": 1.1689,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520601501686794e-06,
+      "loss": 1.2044,
+      "step": 1306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507477784912451e-06,
+      "loss": 1.1992,
+      "step": 1307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494354918426704e-06,
+      "loss": 1.1278,
+      "step": 1308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481232924884827e-06,
+      "loss": 1.1732,
+      "step": 1309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468111826940572e-06,
+      "loss": 1.1521,
+      "step": 1310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454991647246168e-06,
+      "loss": 1.1472,
+      "step": 1311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441872408452242e-06,
+      "loss": 1.1807,
+      "step": 1312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428754133207799e-06,
+      "loss": 1.1801,
+      "step": 1313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415636844160188e-06,
+      "loss": 1.1193,
+      "step": 1314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402520563955043e-06,
+      "loss": 1.2173,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389405315236266e-06,
+      "loss": 1.1704,
+      "step": 1316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37629112064598e-06,
+      "loss": 1.2393,
+      "step": 1317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363178002824474e-06,
+      "loss": 1.2018,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350065984410198e-06,
+      "loss": 1.2327,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33695508803969e-06,
+      "loss": 1.1823,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323845336347554e-06,
+      "loss": 1.1846,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310736751966422e-06,
+      "loss": 1.2335,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297629357526901e-06,
+      "loss": 1.2445,
+      "step": 1323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284523175657556e-06,
+      "loss": 1.1474,
+      "step": 1324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271418228984846e-06,
+      "loss": 1.1589,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258314540133111e-06,
+      "loss": 1.19,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245212131724513e-06,
+      "loss": 1.1923,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232111026378997e-06,
+      "loss": 1.1997,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21901124671427e-06,
+      "loss": 1.1599,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205912815345743e-06,
+      "loss": 1.098,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192815754886502e-06,
+      "loss": 1.1491,
+      "step": 1331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179720087947259e-06,
+      "loss": 1.2136,
+      "step": 1332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166625837136338e-06,
+      "loss": 1.2033,
+      "step": 1333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153533025059598e-06,
+      "loss": 1.2282,
+      "step": 1334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14044167432043e-06,
+      "loss": 1.16,
+      "step": 1335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12735180751969e-06,
+      "loss": 1.1571,
+      "step": 1336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114263447255677e-06,
+      "loss": 1.1888,
+      "step": 1337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101176616124096e-06,
+      "loss": 1.1785,
+      "step": 1338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.088091336717998e-06,
+      "loss": 1.2356,
+      "step": 1339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075007631627764e-06,
+      "loss": 1.209,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06192552344106e-06,
+      "loss": 1.1631,
+      "step": 1341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048845034742788e-06,
+      "loss": 1.2139,
+      "step": 1342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03576618811506e-06,
+      "loss": 1.1735,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022689006137148e-06,
+      "loss": 1.1968,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00961351138545e-06,
+      "loss": 1.2018,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996539726433458e-06,
+      "loss": 1.1926,
+      "step": 1346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9834676738517e-06,
+      "loss": 1.1929,
+      "step": 1347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970397376207732e-06,
+      "loss": 1.1405,
+      "step": 1348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957328856066062e-06,
+      "loss": 1.1717,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94426213598814e-06,
+      "loss": 1.1417,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.931197238532301e-06,
+      "loss": 1.2035,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918134186253743e-06,
+      "loss": 1.1783,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905073001704466e-06,
+      "loss": 1.139,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892013707433255e-06,
+      "loss": 1.2307,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878956325985636e-06,
+      "loss": 1.2481,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865900879903817e-06,
+      "loss": 1.1348,
+      "step": 1356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852847391726676e-06,
+      "loss": 1.1848,
+      "step": 1357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83979588398971e-06,
+      "loss": 1.2206,
+      "step": 1358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826746379224994e-06,
+      "loss": 1.1747,
+      "step": 1359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813698899961147e-06,
+      "loss": 1.151,
+      "step": 1360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800653468723293e-06,
+      "loss": 1.1866,
+      "step": 1361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787610108033007e-06,
+      "loss": 1.1666,
+      "step": 1362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774568840408318e-06,
+      "loss": 1.1231,
+      "step": 1363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761529688363612e-06,
+      "loss": 1.1176,
+      "step": 1364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748492674409637e-06,
+      "loss": 1.2047,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735457821053451e-06,
+      "loss": 1.1622,
+      "step": 1366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722425150798373e-06,
+      "loss": 1.1698,
+      "step": 1367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709394686143962e-06,
+      "loss": 1.1694,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69636644958596e-06,
+      "loss": 1.1404,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683340463616277e-06,
+      "loss": 1.2018,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670316750722925e-06,
+      "loss": 1.1926,
+      "step": 1371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657295333389995e-06,
+      "loss": 1.1535,
+      "step": 1372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644276234097613e-06,
+      "loss": 1.1909,
+      "step": 1373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631259475321913e-06,
+      "loss": 1.1701,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618245079534972e-06,
+      "loss": 1.124,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6052330692048e-06,
+      "loss": 1.2018,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592223466795287e-06,
+      "loss": 1.1794,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579216294766162e-06,
+      "loss": 1.1766,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566211575572965e-06,
+      "loss": 1.1966,
+      "step": 1379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55320933166699e-06,
+      "loss": 1.2195,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54020958549527e-06,
+      "loss": 1.1927,
+      "step": 1381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527212359500518e-06,
+      "loss": 1.157,
+      "step": 1382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514217676121098e-06,
+      "loss": 1.1706,
+      "step": 1383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501225557790983e-06,
+      "loss": 1.145,
+      "step": 1384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488236026939727e-06,
+      "loss": 1.2287,
+      "step": 1385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475249105992404e-06,
+      "loss": 1.2009,
+      "step": 1386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462264817369593e-06,
+      "loss": 1.1631,
+      "step": 1387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449283183487317e-06,
+      "loss": 1.1935,
+      "step": 1388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436304226757025e-06,
+      "loss": 1.1421,
+      "step": 1389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423327969585547e-06,
+      "loss": 1.1777,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410354434375034e-06,
+      "loss": 1.1854,
+      "step": 1391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397383643522965e-06,
+      "loss": 1.1409,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38441561942206e-06,
+      "loss": 1.1927,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371450384460272e-06,
+      "loss": 1.2164,
+      "step": 1394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358487961020737e-06,
+      "loss": 1.1986,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345528371481734e-06,
+      "loss": 1.1574,
+      "step": 1396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332571638216652e-06,
+      "loss": 1.1241,
+      "step": 1397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319617783593948e-06,
+      "loss": 1.1604,
+      "step": 1398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306666829977119e-06,
+      "loss": 1.1738,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293718799724635e-06,
+      "loss": 1.1872,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280773715189934e-06,
+      "loss": 1.114,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267831598721365e-06,
+      "loss": 1.1389,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254892472662148e-06,
+      "loss": 1.1823,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241956359350344e-06,
+      "loss": 1.1595,
+      "step": 1404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229023281118817e-06,
+      "loss": 1.1583,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216093260295183e-06,
+      "loss": 1.1861,
+      "step": 1406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203166319201791e-06,
+      "loss": 1.1654,
+      "step": 1407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190242480155657e-06,
+      "loss": 1.2076,
+      "step": 1408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177321765468456e-06,
+      "loss": 1.165,
+      "step": 1409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164404197446463e-06,
+      "loss": 1.2218,
+      "step": 1410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151489798390518e-06,
+      "loss": 1.1715,
+      "step": 1411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138578590595998e-06,
+      "loss": 1.1644,
+      "step": 1412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125670596352761e-06,
+      "loss": 1.1952,
+      "step": 1413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11276583794513e-06,
+      "loss": 1.124,
+      "step": 1414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099864337651829e-06,
+      "loss": 1.2069,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086966117745961e-06,
+      "loss": 1.1662,
+      "step": 1416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074071200494969e-06,
+      "loss": 1.1537,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061179608160594e-06,
+      "loss": 1.1611,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048291362998832e-06,
+      "loss": 1.192,
+      "step": 1419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0354064872599e-06,
+      "loss": 1.1632,
+      "step": 1420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022525003188208e-06,
+      "loss": 1.1864,
+      "step": 1421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009646933022302e-06,
+      "loss": 1.1103,
+      "step": 1422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996772298994839e-06,
+      "loss": 1.2245,
+      "step": 1423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983901123332538e-06,
+      "loss": 1.1735,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97103342825615e-06,
+      "loss": 1.1668,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958169235980423e-06,
+      "loss": 1.1376,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945308568714043e-06,
+      "loss": 1.2084,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932451448659629e-06,
+      "loss": 1.1675,
+      "step": 1428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919597898013663e-06,
+      "loss": 1.1555,
+      "step": 1429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906747938966473e-06,
+      "loss": 1.2091,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89390159370218e-06,
+      "loss": 1.1736,
+      "step": 1431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881058884398667e-06,
+      "loss": 1.1829,
+      "step": 1432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868219833227546e-06,
+      "loss": 1.1606,
+      "step": 1433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855384462354105e-06,
+      "loss": 1.1466,
+      "step": 1434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842552793937279e-06,
+      "loss": 1.1444,
+      "step": 1435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829724850129626e-06,
+      "loss": 1.2511,
+      "step": 1436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816900653077254e-06,
+      "loss": 1.1767,
+      "step": 1437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804080224919813e-06,
+      "loss": 1.1869,
+      "step": 1438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791263587790446e-06,
+      "loss": 1.1781,
+      "step": 1439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778450763815746e-06,
+      "loss": 1.1846,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765641775115731e-06,
+      "loss": 1.1868,
+      "step": 1441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752836643803787e-06,
+      "loss": 1.1471,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740035391986654e-06,
+      "loss": 1.1032,
+      "step": 1443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727238041764367e-06,
+      "loss": 1.2139,
+      "step": 1444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714444615230223e-06,
+      "loss": 1.1584,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701655134470747e-06,
+      "loss": 1.1294,
+      "step": 1446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688869621565657e-06,
+      "loss": 1.1869,
+      "step": 1447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676088098587813e-06,
+      "loss": 1.2129,
+      "step": 1448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663310587603192e-06,
+      "loss": 1.1749,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650537110670842e-06,
+      "loss": 1.1528,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637767689842853e-06,
+      "loss": 1.1661,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625002347164303e-06,
+      "loss": 1.212,
+      "step": 1452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612241104673235e-06,
+      "loss": 1.1586,
+      "step": 1453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599483984400615e-06,
+      "loss": 1.1527,
+      "step": 1454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586731008370286e-06,
+      "loss": 1.1859,
+      "step": 1455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573982198598941e-06,
+      "loss": 1.1579,
+      "step": 1456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5612375770960814e-06,
+      "loss": 1.1933,
+      "step": 1457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5484971658639774e-06,
+      "loss": 1.1533,
+      "step": 1458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535760986897629e-06,
+      "loss": 1.1907,
+      "step": 1459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523029062184728e-06,
+      "loss": 1.1862,
+      "step": 1460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510301413705625e-06,
+      "loss": 1.1874,
+      "step": 1461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497578063433289e-06,
+      "loss": 1.2309,
+      "step": 1462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4848590333332635e-06,
+      "loss": 1.1582,
+      "step": 1463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472144345363636e-06,
+      "loss": 1.1446,
+      "step": 1464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4594340214750006e-06,
+      "loss": 1.195,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446728083610417e-06,
+      "loss": 1.1618,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434026553705369e-06,
+      "loss": 1.2087,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421329453687732e-06,
+      "loss": 1.2184,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408636805477734e-06,
+      "loss": 1.1501,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395948630987921e-06,
+      "loss": 1.1739,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383264952123108e-06,
+      "loss": 1.1662,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370585790780352e-06,
+      "loss": 1.1719,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357911168848916e-06,
+      "loss": 1.2004,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345241108210222e-06,
+      "loss": 1.2084,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332575630737817e-06,
+      "loss": 1.1659,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319914758297333e-06,
+      "loss": 1.1684,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307258512746458e-06,
+      "loss": 1.199,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294606915934891e-06,
+      "loss": 1.1245,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281959989704295e-06,
+      "loss": 1.2153,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2693177558882895e-06,
+      "loss": 1.1737,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256680236312377e-06,
+      "loss": 1.178,
+      "step": 1481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244047452793925e-06,
+      "loss": 1.1825,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231419427142129e-06,
+      "loss": 1.1565,
+      "step": 1483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218796181157965e-06,
+      "loss": 1.1843,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206177736634159e-06,
+      "loss": 1.1759,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193564115355149e-06,
+      "loss": 1.2388,
+      "step": 1486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180955339097049e-06,
+      "loss": 1.1398,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1683514296276046e-06,
+      "loss": 1.1733,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155752408706158e-06,
+      "loss": 1.2124,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143158298083615e-06,
+      "loss": 1.1718,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130569119502404e-06,
+      "loss": 1.2205,
+      "step": 1491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117984894696437e-06,
+      "loss": 1.1881,
+      "step": 1492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105405645391075e-06,
+      "loss": 1.1617,
+      "step": 1493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092831393303085e-06,
+      "loss": 1.1461,
+      "step": 1494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08026216014062e-06,
+      "loss": 1.2233,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067697967603155e-06,
+      "loss": 1.1513,
+      "step": 1496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055138837381465e-06,
+      "loss": 1.147,
+      "step": 1497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042584791157591e-06,
+      "loss": 1.1432,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030035850604792e-06,
+      "loss": 1.1438,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017492037387514e-06,
+      "loss": 1.1797,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00495337316135e-06,
+      "loss": 1.122,
+      "step": 1501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9924198795730084e-06,
+      "loss": 1.1748,
+      "step": 1502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979891578260268e-06,
+      "loss": 1.2062,
+      "step": 1503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96736849085194e-06,
+      "loss": 1.2131,
+      "step": 1504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95485063896784e-06,
+      "loss": 1.1734,
+      "step": 1505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942338044218744e-06,
+      "loss": 1.1443,
+      "step": 1506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929830728206351e-06,
+      "loss": 1.1535,
+      "step": 1507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917328712523242e-06,
+      "loss": 1.1224,
+      "step": 1508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9048320187528575e-06,
+      "loss": 1.1714,
+      "step": 1509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892340668469442e-06,
+      "loss": 1.2102,
+      "step": 1510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879854683238023e-06,
+      "loss": 1.1893,
+      "step": 1511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867374084614356e-06,
+      "loss": 1.1816,
+      "step": 1512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854898894144903e-06,
+      "loss": 1.1652,
+      "step": 1513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84242913336679e-06,
+      "loss": 1.2144,
+      "step": 1514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829964823807764e-06,
+      "loss": 1.1933,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817505986986166e-06,
+      "loss": 1.1348,
+      "step": 1516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80505264441089e-06,
+      "loss": 1.1437,
+      "step": 1517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792604817581341e-06,
+      "loss": 1.147,
+      "step": 1518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780162527987404e-06,
+      "loss": 1.2073,
+      "step": 1519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767725797109401e-06,
+      "loss": 1.2198,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7552946464180605e-06,
+      "loss": 1.1586,
+      "step": 1521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742869097374478e-06,
+      "loss": 1.1822,
+      "step": 1522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7304491714300705e-06,
+      "loss": 1.1488,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7180348900265655e-06,
+      "loss": 1.1266,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705626274595926e-06,
+      "loss": 1.1609,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693223346560344e-06,
+      "loss": 1.1457,
+      "step": 1526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680826127332192e-06,
+      "loss": 1.1274,
+      "step": 1527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66843463831398e-06,
+      "loss": 1.1609,
+      "step": 1528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656048900898333e-06,
+      "loss": 1.1356,
+      "step": 1529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643668936467944e-06,
+      "loss": 1.2017,
+      "step": 1530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631294766395541e-06,
+      "loss": 1.1298,
+      "step": 1531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618926412043847e-06,
+      "loss": 1.1567,
+      "step": 1532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606563894765543e-06,
+      "loss": 1.1931,
+      "step": 1533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594207235903236e-06,
+      "loss": 1.1344,
+      "step": 1534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581856456789419e-06,
+      "loss": 1.1427,
+      "step": 1535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569511578746428e-06,
+      "loss": 1.1904,
+      "step": 1536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557172623086421e-06,
+      "loss": 1.1656,
+      "step": 1537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544839611111324e-06,
+      "loss": 1.1863,
+      "step": 1538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532512564112809e-06,
+      "loss": 1.227,
+      "step": 1539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520191503372244e-06,
+      "loss": 1.1379,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507876450160664e-06,
+      "loss": 1.156,
+      "step": 1541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495567425738734e-06,
+      "loss": 1.155,
+      "step": 1542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483264451356709e-06,
+      "loss": 1.182,
+      "step": 1543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470967548254399e-06,
+      "loss": 1.193,
+      "step": 1544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4586767376611315e-06,
+      "loss": 1.2011,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446392040795726e-06,
+      "loss": 1.2366,
+      "step": 1546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4341134788664325e-06,
+      "loss": 1.1415,
+      "step": 1547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421841073070922e-06,
+      "loss": 1.1663,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409574844596227e-06,
+      "loss": 1.156,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397314814618722e-06,
+      "loss": 1.1569,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385061004304083e-06,
+      "loss": 1.1644,
+      "step": 1551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3728134348072356e-06,
+      "loss": 1.1255,
+      "step": 1552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360572127272349e-06,
+      "loss": 1.1489,
+      "step": 1553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348337102832768e-06,
+      "loss": 1.1302,
+      "step": 1554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336108382610999e-06,
+      "loss": 1.1212,
+      "step": 1555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323885987718657e-06,
+      "loss": 1.178,
+      "step": 1556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311669939256443e-06,
+      "loss": 1.139,
+      "step": 1557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2994602583140985e-06,
+      "loss": 1.1718,
+      "step": 1558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287256965970377e-06,
+      "loss": 1.131,
+      "step": 1559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275060083292991e-06,
+      "loss": 1.1338,
+      "step": 1560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262869631338602e-06,
+      "loss": 1.2019,
+      "step": 1561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250685631152762e-06,
+      "loss": 1.1371,
+      "step": 1562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238508103769885e-06,
+      "loss": 1.1679,
+      "step": 1563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226337070213208e-06,
+      "loss": 1.2261,
+      "step": 1564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214172551494764e-06,
+      "loss": 1.1518,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2020145686153334e-06,
+      "loss": 1.1824,
+      "step": 1566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189863142564407e-06,
+      "loss": 1.163,
+      "step": 1567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177718294320178e-06,
+      "loss": 1.2163,
+      "step": 1568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16558004484946e-06,
+      "loss": 1.1553,
+      "step": 1569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153448415107684e-06,
+      "loss": 1.1586,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141323426038857e-06,
+      "loss": 1.1386,
+      "step": 1571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12920509857551e-06,
+      "loss": 1.2284,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117093453638685e-06,
+      "loss": 1.195,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104988512137877e-06,
+      "loss": 1.1207,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092890294971024e-06,
+      "loss": 1.2104,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0807988230244385e-06,
+      "loss": 1.1157,
+      "step": 1576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068714117172793e-06,
+      "loss": 1.1125,
+      "step": 1577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056636198279085e-06,
+      "loss": 1.2017,
+      "step": 1578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0445650871945895e-06,
+      "loss": 1.1771,
+      "step": 1579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032500804758827e-06,
+      "loss": 1.1876,
+      "step": 1580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020443371799534e-06,
+      "loss": 1.1367,
+      "step": 1581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008392809132616e-06,
+      "loss": 1.1162,
+      "step": 1582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996349137562128e-06,
+      "loss": 1.1337,
+      "step": 1583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98431237788022e-06,
+      "loss": 1.0932,
+      "step": 1584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972282550867106e-06,
+      "loss": 1.1164,
+      "step": 1585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96025967729104e-06,
+      "loss": 1.1586,
+      "step": 1586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948243777908272e-06,
+      "loss": 1.1788,
+      "step": 1587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936234873462999e-06,
+      "loss": 1.1509,
+      "step": 1588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924232984687355e-06,
+      "loss": 1.1727,
+      "step": 1589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912238132301361e-06,
+      "loss": 1.2122,
+      "step": 1590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900250337012885e-06,
+      "loss": 1.1426,
+      "step": 1591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888269619517615e-06,
+      "loss": 1.1776,
+      "step": 1592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876296000499017e-06,
+      "loss": 1.1561,
+      "step": 1593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864329500628305e-06,
+      "loss": 1.1246,
+      "step": 1594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852370140564403e-06,
+      "loss": 1.1594,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840417940953902e-06,
+      "loss": 1.1587,
+      "step": 1596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828472922431047e-06,
+      "loss": 1.1267,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816535105617669e-06,
+      "loss": 1.195,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80460451112317e-06,
+      "loss": 1.1433,
+      "step": 1599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7926811595444935e-06,
+      "loss": 1.1635,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7807650714660644e-06,
+      "loss": 1.1623,
+      "step": 1601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768856267459774e-06,
+      "loss": 1.1292,
+      "step": 1602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756954768084947e-06,
+      "loss": 1.1089,
+      "step": 1603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74506059388828e-06,
+      "loss": 1.1822,
+      "step": 1604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7331737654038365e-06,
+      "loss": 1.1281,
+      "step": 1605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721294303153e-06,
+      "loss": 1.1532,
+      "step": 1606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709422227644428e-06,
+      "loss": 1.1792,
+      "step": 1607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697557559374025e-06,
+      "loss": 1.1591,
+      "step": 1608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6857003188249204e-06,
+      "loss": 1.1405,
+      "step": 1609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673850526467406e-06,
+      "loss": 1.2273,
+      "step": 1610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6620082027589196e-06,
+      "loss": 1.2226,
+      "step": 1611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65017336814401e-06,
+      "loss": 1.1732,
+      "step": 1612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638346043054297e-06,
+      "loss": 1.172,
+      "step": 1613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626526247908428e-06,
+      "loss": 1.1917,
+      "step": 1614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614714003112051e-06,
+      "loss": 1.1725,
+      "step": 1615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602909329057794e-06,
+      "loss": 1.1687,
+      "step": 1616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591112246125198e-06,
+      "loss": 1.1639,
+      "step": 1617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579322774680697e-06,
+      "loss": 1.1631,
+      "step": 1618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567540935077609e-06,
+      "loss": 1.1817,
+      "step": 1619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555766747656051e-06,
+      "loss": 1.1779,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544000232742934e-06,
+      "loss": 1.188,
+      "step": 1621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532241410651937e-06,
+      "loss": 1.1298,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.520490301683446e-06,
+      "loss": 1.2453,
+      "step": 1623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5087469261245245e-06,
+      "loss": 1.1418,
+      "step": 1624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497011304248907e-06,
+      "loss": 1.1697,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485283456316919e-06,
+      "loss": 1.0777,
+      "step": 1626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473563402575484e-06,
+      "loss": 1.0867,
+      "step": 1627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461851163258052e-06,
+      "loss": 1.1321,
+      "step": 1628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450146758584599e-06,
+      "loss": 1.1479,
+      "step": 1629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438450208761565e-06,
+      "loss": 1.1607,
+      "step": 1630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426761533981829e-06,
+      "loss": 1.0875,
+      "step": 1631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415080754424684e-06,
+      "loss": 1.1826,
+      "step": 1632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403407890255776e-06,
+      "loss": 1.1574,
+      "step": 1633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391742961627109e-06,
+      "loss": 1.151,
+      "step": 1634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3800859886769605e-06,
+      "loss": 1.1075,
+      "step": 1635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3684369915299e-06,
+      "loss": 1.1314,
+      "step": 1636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356795990296708e-06,
+      "loss": 1.1542,
+      "step": 1637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345163005074363e-06,
+      "loss": 1.1579,
+      "step": 1638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333538055946019e-06,
+      "loss": 1.1575,
+      "step": 1639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32192116298094e-06,
+      "loss": 1.1758,
+      "step": 1640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310312346234489e-06,
+      "loss": 1.1398,
+      "step": 1641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298711625748092e-06,
+      "loss": 1.1368,
+      "step": 1642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2871190215491865e-06,
+      "loss": 1.146,
+      "step": 1643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275534553651201e-06,
+      "loss": 1.2156,
+      "step": 1644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263958242053524e-06,
+      "loss": 1.0969,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252390106741458e-06,
+      "loss": 1.1425,
+      "step": 1646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240830167686186e-06,
+      "loss": 1.124,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229278444844748e-06,
+      "loss": 1.1464,
+      "step": 1648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217734958160002e-06,
+      "loss": 1.1415,
+      "step": 1649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206199727560579e-06,
+      "loss": 1.1773,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194672772960856e-06,
+      "loss": 1.1511,
+      "step": 1651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183154114260932e-06,
+      "loss": 1.1104,
+      "step": 1652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171643771346577e-06,
+      "loss": 1.174,
+      "step": 1653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1601417640892004e-06,
+      "loss": 1.1852,
+      "step": 1654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148648112345831e-06,
+      "loss": 1.2014,
+      "step": 1655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137162835959073e-06,
+      "loss": 1.1828,
+      "step": 1656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125685954757058e-06,
+      "loss": 1.1828,
+      "step": 1657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114217488553441e-06,
+      "loss": 1.1607,
+      "step": 1658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102757457147336e-06,
+      "loss": 1.2233,
+      "step": 1659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091305880323298e-06,
+      "loss": 1.1589,
+      "step": 1660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0798627778512955e-06,
+      "loss": 1.1694,
+      "step": 1661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068428169486652e-06,
+      "loss": 1.125,
+      "step": 1662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057002074970042e-06,
+      "loss": 1.174,
+      "step": 1663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045584514027429e-06,
+      "loss": 1.2409,
+      "step": 1664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0341755063700525e-06,
+      "loss": 1.1596,
+      "step": 1665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0227750716943835e-06,
+      "loss": 1.1489,
+      "step": 1666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011383229682085e-06,
+      "loss": 1.1537,
+      "step": 1667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.1605,
+      "step": 1668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988625402300097e-06,
+      "loss": 1.1774,
+      "step": 1669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977259456219427e-06,
+      "loss": 1.1872,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965902181380137e-06,
+      "loss": 1.1041,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954553597389378e-06,
+      "loss": 1.1296,
+      "step": 1672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9432137238393e-06,
+      "loss": 1.1017,
+      "step": 1673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931882580307028e-06,
+      "loss": 1.1162,
+      "step": 1674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9205601863546046e-06,
+      "loss": 1.1775,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909246561528967e-06,
+      "loss": 1.1762,
+      "step": 1676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89794172536192e-06,
+      "loss": 1.1481,
+      "step": 1677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8866456973700975e-06,
+      "loss": 1.1527,
+      "step": 1678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875358497054915e-06,
+      "loss": 1.1802,
+      "step": 1679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864080143902555e-06,
+      "loss": 1.1599,
+      "step": 1680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852810657383931e-06,
+      "loss": 1.2291,
+      "step": 1681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841550056954641e-06,
+      "loss": 1.1518,
+      "step": 1682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8302983620549436e-06,
+      "loss": 1.1004,
+      "step": 1683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819055592109726e-06,
+      "loss": 1.1673,
+      "step": 1684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807821766528471e-06,
+      "loss": 1.176,
+      "step": 1685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796596904705212e-06,
+      "loss": 1.2061,
+      "step": 1686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7853810260185055e-06,
+      "loss": 1.165,
+      "step": 1687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774174149831413e-06,
+      "loss": 1.1725,
+      "step": 1688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76297629549144e-06,
+      "loss": 1.1448,
+      "step": 1689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751787482330521e-06,
+      "loss": 1.1034,
+      "step": 1690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74060772966499e-06,
+      "loss": 1.118,
+      "step": 1691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729437056795523e-06,
+      "loss": 1.1689,
+      "step": 1692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718275483007136e-06,
+      "loss": 1.1337,
+      "step": 1693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707123027569132e-06,
+      "loss": 1.1367,
+      "step": 1694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695979709735067e-06,
+      "loss": 1.1567,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684845548742723e-06,
+      "loss": 1.1865,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67372056381408e-06,
+      "loss": 1.1909,
+      "step": 1697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662604774155273e-06,
+      "loss": 1.135,
+      "step": 1698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651498198956554e-06,
+      "loss": 1.1155,
+      "step": 1699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640400857392281e-06,
+      "loss": 1.1806,
+      "step": 1700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629312768620867e-06,
+      "loss": 1.1603,
+      "step": 1701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6182339517847465e-06,
+      "loss": 1.0865,
+      "step": 1702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607164426010346e-06,
+      "loss": 1.1489,
+      "step": 1703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59610421040806e-06,
+      "loss": 1.1506,
+      "step": 1704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585053324072205e-06,
+      "loss": 1.1337,
+      "step": 1705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574011786080983e-06,
+      "loss": 1.1298,
+      "step": 1706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56297961549648e-06,
+      "loss": 1.1518,
+      "step": 1707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551956831364589e-06,
+      "loss": 1.1556,
+      "step": 1708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540943452715002e-06,
+      "loss": 1.1183,
+      "step": 1709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529939498561183e-06,
+      "loss": 1.1689,
+      "step": 1710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518944987900318e-06,
+      "loss": 1.1229,
+      "step": 1711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507959939713283e-06,
+      "loss": 1.1805,
+      "step": 1712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4969843729646375e-06,
+      "loss": 1.1621,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486018306602552e-06,
+      "loss": 1.1553,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475061759558812e-06,
+      "loss": 1.1663,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464114750748752e-06,
+      "loss": 1.1539,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45317729907126e-06,
+      "loss": 1.1486,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442249423408707e-06,
+      "loss": 1.2155,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431331142626936e-06,
+      "loss": 1.1404,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420422475575233e-06,
+      "loss": 1.1718,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4095234410862785e-06,
+      "loss": 1.1413,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398634057976129e-06,
+      "loss": 1.1773,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387754345044179e-06,
+      "loss": 1.138,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376884321073124e-06,
+      "loss": 1.1877,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366024004828931e-06,
+      "loss": 1.1725,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355173415060817e-06,
+      "loss": 1.1624,
+      "step": 1726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3443325705012e-06,
+      "loss": 1.1776,
+      "step": 1727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333501489865669e-06,
+      "loss": 1.1403,
+      "step": 1728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322680191852969e-06,
+      "loss": 1.1512,
+      "step": 1729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3118686951449525e-06,
+      "loss": 1.1662,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301067018406548e-06,
+      "loss": 1.1524,
+      "step": 1731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290275180285725e-06,
+      "loss": 1.1543,
+      "step": 1732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279493199413482e-06,
+      "loss": 1.173,
+      "step": 1733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268721094403791e-06,
+      "loss": 1.1742,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2579588838535676e-06,
+      "loss": 1.1413,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247206586342663e-06,
+      "loss": 1.1505,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236464220433805e-06,
+      "loss": 1.1098,
+      "step": 1737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225731804672574e-06,
+      "loss": 1.2026,
+      "step": 1738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2150093575873685e-06,
+      "loss": 1.1637,
+      "step": 1739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204296897689393e-06,
+      "loss": 1.1538,
+      "step": 1740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19359444347259e-06,
+      "loss": 1.1726,
+      "step": 1741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182902013413649e-06,
+      "loss": 1.152,
+      "step": 1742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172219625971934e-06,
+      "loss": 1.1394,
+      "step": 1743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161547299589487e-06,
+      "loss": 1.1576,
+      "step": 1744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150885052690968e-06,
+      "loss": 1.166,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140232903683649e-06,
+      "loss": 1.1247,
+      "step": 1746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1295908709573585e-06,
+      "loss": 1.1011,
+      "step": 1747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1189589728844605e-06,
+      "loss": 1.1648,
+      "step": 1748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108337227819833e-06,
+      "loss": 1.1354,
+      "step": 1749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097725654100811e-06,
+      "loss": 1.1571,
+      "step": 1750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087124270047186e-06,
+      "loss": 1.1853,
+      "step": 1751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076533093961141e-06,
+      "loss": 1.1687,
+      "step": 1752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065952144127255e-06,
+      "loss": 1.1636,
+      "step": 1753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055381438812433e-06,
+      "loss": 1.137,
+      "step": 1754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044820996265905e-06,
+      "loss": 1.1335,
+      "step": 1755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034270834719184e-06,
+      "loss": 1.1552,
+      "step": 1756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02373097238603e-06,
+      "loss": 1.1509,
+      "step": 1757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0132014274624175e-06,
+      "loss": 1.1787,
+      "step": 1758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0026822181265276e-06,
+      "loss": 1.1043,
+      "step": 1759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99217336253868e-06,
+      "loss": 1.1727,
+      "step": 1760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9816748788413215e-06,
+      "loss": 1.1274,
+      "step": 1761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971186785159003e-06,
+      "loss": 1.1567,
+      "step": 1762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960709099598331e-06,
+      "loss": 1.1608,
+      "step": 1763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950241840247939e-06,
+      "loss": 1.2006,
+      "step": 1764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93978502517847e-06,
+      "loss": 1.1813,
+      "step": 1765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929338672442536e-06,
+      "loss": 1.1133,
+      "step": 1766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91890280007468e-06,
+      "loss": 1.1358,
+      "step": 1767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908477426091351e-06,
+      "loss": 1.198,
+      "step": 1768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898062568490885e-06,
+      "loss": 1.1106,
+      "step": 1769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887658245253449e-06,
+      "loss": 1.1579,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8772644743410305e-06,
+      "loss": 1.1229,
+      "step": 1771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866881273697398e-06,
+      "loss": 1.0963,
+      "step": 1772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856508661248078e-06,
+      "loss": 1.1549,
+      "step": 1773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846146654900305e-06,
+      "loss": 1.1307,
+      "step": 1774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835795272543014e-06,
+      "loss": 1.1687,
+      "step": 1775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8254545320467925e-06,
+      "loss": 1.1203,
+      "step": 1776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815124451263857e-06,
+      "loss": 1.1192,
+      "step": 1777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804805048028025e-06,
+      "loss": 1.1445,
+      "step": 1778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944963401546776e-06,
+      "loss": 1.1092,
+      "step": 1779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784198345440725e-06,
+      "loss": 1.1798,
+      "step": 1780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.773911081664594e-06,
+      "loss": 1.1144,
+      "step": 1781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7636345665861816e-06,
+      "loss": 1.1851,
+      "step": 1782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7533688179468263e-06,
+      "loss": 1.1693,
+      "step": 1783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7431138534692736e-06,
+      "loss": 1.1526,
+      "step": 1784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7328696908576656e-06,
+      "loss": 1.1555,
+      "step": 1785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7226363477974857e-06,
+      "loss": 1.1222,
+      "step": 1786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7124138419555357e-06,
+      "loss": 1.1623,
+      "step": 1787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7022021909799176e-06,
+      "loss": 1.1342,
+      "step": 1788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692001412499994e-06,
+      "loss": 1.1231,
+      "step": 1789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6818115241263463e-06,
+      "loss": 1.1243,
+      "step": 1790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6716325434507593e-06,
+      "loss": 1.1623,
+      "step": 1791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661464488046197e-06,
+      "loss": 1.1609,
+      "step": 1792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651307375466745e-06,
+      "loss": 1.1383,
+      "step": 1793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.641161223247611e-06,
+      "loss": 1.133,
+      "step": 1794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6310260489050785e-06,
+      "loss": 1.124,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6209018699364752e-06,
+      "loss": 1.1481,
+      "step": 1796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6107887038201418e-06,
+      "loss": 1.1779,
+      "step": 1797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6006865680154203e-06,
+      "loss": 1.1326,
+      "step": 1798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5905954799625977e-06,
+      "loss": 1.13,
+      "step": 1799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5805154570828893e-06,
+      "loss": 1.1619,
+      "step": 1800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5704465167784177e-06,
+      "loss": 1.1804,
+      "step": 1801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560388676432158e-06,
+      "loss": 1.1028,
+      "step": 1802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5503419534079384e-06,
+      "loss": 1.1051,
+      "step": 1803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5403063650503776e-06,
+      "loss": 1.1904,
+      "step": 1804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5302819286848875e-06,
+      "loss": 1.1678,
+      "step": 1805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520268661617614e-06,
+      "loss": 1.2086,
+      "step": 1806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5102665811354244e-06,
+      "loss": 1.1844,
+      "step": 1807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50027570450588e-06,
+      "loss": 1.1218,
+      "step": 1808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4902960489771875e-06,
+      "loss": 1.2043,
+      "step": 1809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480327631778194e-06,
+      "loss": 1.139,
+      "step": 1810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4703704701183416e-06,
+      "loss": 1.1209,
+      "step": 1811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.460424581187635e-06,
+      "loss": 1.0779,
+      "step": 1812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4504899821566186e-06,
+      "loss": 1.135,
+      "step": 1813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440566690176356e-06,
+      "loss": 1.147,
+      "step": 1814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430654722378378e-06,
+      "loss": 1.16,
+      "step": 1815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4207540958746686e-06,
+      "loss": 1.1715,
+      "step": 1816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4108648277576373e-06,
+      "loss": 1.1532,
+      "step": 1817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4009869351000836e-06,
+      "loss": 1.1495,
+      "step": 1818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391120434955164e-06,
+      "loss": 1.147,
+      "step": 1819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381265344356366e-06,
+      "loss": 1.14,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3714216803174894e-06,
+      "loss": 1.1598,
+      "step": 1821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361589459832596e-06,
+      "loss": 1.132,
+      "step": 1822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3517686998759946e-06,
+      "loss": 1.1574,
+      "step": 1823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341959417402213e-06,
+      "loss": 1.1428,
+      "step": 1824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3321616293459635e-06,
+      "loss": 1.1195,
+      "step": 1825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322375352622108e-06,
+      "loss": 1.1393,
+      "step": 1826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3126006041256454e-06,
+      "loss": 1.1073,
+      "step": 1827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3028374007316633e-06,
+      "loss": 1.1781,
+      "step": 1828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2930857592953167e-06,
+      "loss": 1.123,
+      "step": 1829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283345696651814e-06,
+      "loss": 1.1402,
+      "step": 1830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273617229616356e-06,
+      "loss": 1.1395,
+      "step": 1831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26390037498414e-06,
+      "loss": 1.1622,
+      "step": 1832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2541951495303047e-06,
+      "loss": 1.1334,
+      "step": 1833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2445015700099204e-06,
+      "loss": 1.1339,
+      "step": 1834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2348196531579466e-06,
+      "loss": 1.1246,
+      "step": 1835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225149415689207e-06,
+      "loss": 1.1467,
+      "step": 1836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2154908742983713e-06,
+      "loss": 1.074,
+      "step": 1837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2058440456599038e-06,
+      "loss": 1.1663,
+      "step": 1838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196208946428062e-06,
+      "loss": 1.0979,
+      "step": 1839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1865855932368417e-06,
+      "loss": 1.1009,
+      "step": 1840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1769740026999696e-06,
+      "loss": 1.1946,
+      "step": 1841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167374191410856e-06,
+      "loss": 1.1429,
+      "step": 1842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1577861759425876e-06,
+      "loss": 1.1378,
+      "step": 1843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1482099728478765e-06,
+      "loss": 1.1324,
+      "step": 1844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1386455986590436e-06,
+      "loss": 1.1735,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1290930698879916e-06,
+      "loss": 1.1258,
+      "step": 1846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119552403026177e-06,
+      "loss": 1.1483,
+      "step": 1847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1100236145445683e-06,
+      "loss": 1.1285,
+      "step": 1848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10050672089363e-06,
+      "loss": 1.0913,
+      "step": 1849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910017385032997e-06,
+      "loss": 1.1181,
+      "step": 1850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081508683782942e-06,
+      "loss": 1.1786,
+      "step": 1851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.072027573121329e-06,
+      "loss": 1.1498,
+      "step": 1852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062558422886619e-06,
+      "loss": 1.121,
+      "step": 1853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0531012494263233e-06,
+      "loss": 1.0958,
+      "step": 1854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04365606906727e-06,
+      "loss": 1.1689,
+      "step": 1855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0342228981155797e-06,
+      "loss": 1.1146,
+      "step": 1856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024801752856653e-06,
+      "loss": 1.1316,
+      "step": 1857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0153926495551177e-06,
+      "loss": 1.1127,
+      "step": 1858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0059956044548133e-06,
+      "loss": 1.1148,
+      "step": 1859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996610633778768e-06,
+      "loss": 1.1309,
+      "step": 1860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9872377537291654e-06,
+      "loss": 1.1198,
+      "step": 1861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9778769804873042e-06,
+      "loss": 1.1722,
+      "step": 1862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968528330213598e-06,
+      "loss": 1.116,
+      "step": 1863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95919181904752e-06,
+      "loss": 1.1358,
+      "step": 1864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.949867463107585e-06,
+      "loss": 1.1117,
+      "step": 1865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940555278491337e-06,
+      "loss": 1.1461,
+      "step": 1866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9312552812752926e-06,
+      "loss": 1.1799,
+      "step": 1867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9219674875149317e-06,
+      "loss": 1.1729,
+      "step": 1868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9126919132446717e-06,
+      "loss": 1.1436,
+      "step": 1869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9034285744778334e-06,
+      "loss": 1.1624,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89417748720661e-06,
+      "loss": 1.11,
+      "step": 1871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8849386674020407e-06,
+      "loss": 1.0849,
+      "step": 1872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8757121310139992e-06,
+      "loss": 1.1697,
+      "step": 1873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8664978939711397e-06,
+      "loss": 1.1856,
+      "step": 1874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8572959721808847e-06,
+      "loss": 1.1322,
+      "step": 1875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8481063815294084e-06,
+      "loss": 1.1712,
+      "step": 1876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838929137881584e-06,
+      "loss": 1.0947,
+      "step": 1877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829764257080968e-06,
+      "loss": 1.1522,
+      "step": 1878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8206117549497846e-06,
+      "loss": 1.1436,
+      "step": 1879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8114716472888758e-06,
+      "loss": 1.0739,
+      "step": 1880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8023439498776904e-06,
+      "loss": 1.1922,
+      "step": 1881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7932286784742544e-06,
+      "loss": 1.1219,
+      "step": 1882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7841258488151433e-06,
+      "loss": 1.0706,
+      "step": 1883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7750354766154466e-06,
+      "loss": 1.1751,
+      "step": 1884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7659575775687462e-06,
+      "loss": 1.1343,
+      "step": 1885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756892167347104e-06,
+      "loss": 1.0813,
+      "step": 1886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7478392616010052e-06,
+      "loss": 1.1021,
+      "step": 1887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.738798875959353e-06,
+      "loss": 1.1542,
+      "step": 1888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729771026029443e-06,
+      "loss": 1.1583,
+      "step": 1889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7207557273969165e-06,
+      "loss": 1.0741,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7117529956257594e-06,
+      "loss": 1.1202,
+      "step": 1891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702762846258249e-06,
+      "loss": 1.1104,
+      "step": 1892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6937852948149534e-06,
+      "loss": 1.1226,
+      "step": 1893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6848203567946806e-06,
+      "loss": 1.1276,
+      "step": 1894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6758680476744715e-06,
+      "loss": 1.1254,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6669283829095593e-06,
+      "loss": 1.095,
+      "step": 1896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6580013779333456e-06,
+      "loss": 1.1575,
+      "step": 1897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649087048157384e-06,
+      "loss": 1.1182,
+      "step": 1898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640185408971343e-06,
+      "loss": 1.1352,
+      "step": 1899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6312964757429793e-06,
+      "loss": 1.1488,
+      "step": 1900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622420263818112e-06,
+      "loss": 1.1441,
+      "step": 1901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613556788520607e-06,
+      "loss": 1.1868,
+      "step": 1902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6047060651523347e-06,
+      "loss": 1.1216,
+      "step": 1903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595868108993149e-06,
+      "loss": 1.122,
+      "step": 1904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5870429353008686e-06,
+      "loss": 1.1359,
+      "step": 1905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5782305593112445e-06,
+      "loss": 1.1408,
+      "step": 1906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5694309962379283e-06,
+      "loss": 1.1366,
+      "step": 1907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5606442612724524e-06,
+      "loss": 1.1298,
+      "step": 1908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5518703695842105e-06,
+      "loss": 1.0966,
+      "step": 1909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543109336320414e-06,
+      "loss": 1.1566,
+      "step": 1910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5343611766060762e-06,
+      "loss": 1.139,
+      "step": 1911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5256259055439936e-06,
+      "loss": 1.1519,
+      "step": 1912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5169035382147088e-06,
+      "loss": 1.0748,
+      "step": 1913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50819408967648e-06,
+      "loss": 1.1471,
+      "step": 1914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499497574965274e-06,
+      "loss": 1.0786,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4908140090947197e-06,
+      "loss": 1.1557,
+      "step": 1916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4821434070560913e-06,
+      "loss": 1.096,
+      "step": 1917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4734857838182913e-06,
+      "loss": 1.1719,
+      "step": 1918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464841154327804e-06,
+      "loss": 1.1984,
+      "step": 1919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4562095335086933e-06,
+      "loss": 1.131,
+      "step": 1920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4475909362625504e-06,
+      "loss": 1.1786,
+      "step": 1921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438985377468499e-06,
+      "loss": 1.1323,
+      "step": 1922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4303928719831415e-06,
+      "loss": 1.1631,
+      "step": 1923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4218134346405463e-06,
+      "loss": 1.1156,
+      "step": 1924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413247080252228e-06,
+      "loss": 1.1269,
+      "step": 1925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404693823607106e-06,
+      "loss": 1.0972,
+      "step": 1926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3961536794714935e-06,
+      "loss": 1.1082,
+      "step": 1927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3876266625890697e-06,
+      "loss": 1.1409,
+      "step": 1928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3791127876808406e-06,
+      "loss": 1.1111,
+      "step": 1929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3706120694451296e-06,
+      "loss": 1.1532,
+      "step": 1930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3621245225575497e-06,
+      "loss": 1.1137,
+      "step": 1931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3536501616709708e-06,
+      "loss": 1.1651,
+      "step": 1932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345189001415493e-06,
+      "loss": 1.1465,
+      "step": 1933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3367410563984403e-06,
+      "loss": 1.1322,
+      "step": 1934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328306341204316e-06,
+      "loss": 1.1188,
+      "step": 1935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3198848703947783e-06,
+      "loss": 1.15,
+      "step": 1936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3114766585086234e-06,
+      "loss": 1.1794,
+      "step": 1937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3030817200617616e-06,
+      "loss": 1.2016,
+      "step": 1938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2947000695471854e-06,
+      "loss": 1.1672,
+      "step": 1939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28633172143494e-06,
+      "loss": 1.1059,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2779766901721166e-06,
+      "loss": 1.1084,
+      "step": 1941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269634990182815e-06,
+      "loss": 1.1046,
+      "step": 1942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2613066358681112e-06,
+      "loss": 1.1405,
+      "step": 1943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252991641606045e-06,
+      "loss": 1.0829,
+      "step": 1944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2446900217515977e-06,
+      "loss": 1.1219,
+      "step": 1945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236401790636651e-06,
+      "loss": 1.157,
+      "step": 1946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2281269625699842e-06,
+      "loss": 1.1324,
+      "step": 1947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2198655518372237e-06,
+      "loss": 1.1394,
+      "step": 1948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211617572700846e-06,
+      "loss": 1.101,
+      "step": 1949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2033830394001264e-06,
+      "loss": 1.1396,
+      "step": 1950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1951619661511412e-06,
+      "loss": 1.1572,
+      "step": 1951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1869543671467177e-06,
+      "loss": 1.2037,
+      "step": 1952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1787602565564223e-06,
+      "loss": 1.093,
+      "step": 1953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170579648526546e-06,
+      "loss": 1.1344,
+      "step": 1954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162412557180057e-06,
+      "loss": 1.1185,
+      "step": 1955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1542589966165906e-06,
+      "loss": 1.1058,
+      "step": 1956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146118980912427e-06,
+      "loss": 1.1185,
+      "step": 1957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1379925241204626e-06,
+      "loss": 1.138,
+      "step": 1958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1298796402701815e-06,
+      "loss": 1.0907,
+      "step": 1959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1217803433676353e-06,
+      "loss": 1.106,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113694647395426e-06,
+      "loss": 1.168,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1056225663126662e-06,
+      "loss": 1.1684,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097564114054964e-06,
+      "loss": 1.162,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089519304534414e-06,
+      "loss": 1.1222,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08148815163954e-06,
+      "loss": 1.1547,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0734706692352947e-06,
+      "loss": 1.164,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0654668711630345e-06,
+      "loss": 1.1449,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0574767712404865e-06,
+      "loss": 1.1862,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049500383261728e-06,
+      "loss": 1.1527,
+      "step": 1969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0415377209971686e-06,
+      "loss": 1.1131,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033588798193523e-06,
+      "loss": 1.1396,
+      "step": 1971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0256536285737817e-06,
+      "loss": 1.1627,
+      "step": 1972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017732225837189e-06,
+      "loss": 1.1183,
+      "step": 1973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0098246036592316e-06,
+      "loss": 1.0883,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0019307756915996e-06,
+      "loss": 1.1154,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9940507555621646e-06,
+      "loss": 1.119,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9861845568749728e-06,
+      "loss": 1.1704,
+      "step": 1977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978332193210194e-06,
+      "loss": 1.1452,
+      "step": 1978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970493678124125e-06,
+      "loss": 1.1616,
+      "step": 1979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962669025149151e-06,
+      "loss": 1.0986,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954858247793725e-06,
+      "loss": 1.1209,
+      "step": 1981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9470613595423405e-06,
+      "loss": 1.1568,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9392783738555245e-06,
+      "loss": 1.1276,
+      "step": 1983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931509304169792e-06,
+      "loss": 1.1315,
+      "step": 1984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9237541638976355e-06,
+      "loss": 1.1341,
+      "step": 1985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916012966427503e-06,
+      "loss": 1.1553,
+      "step": 1986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908285725123773e-06,
+      "loss": 1.1403,
+      "step": 1987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9005724533267256e-06,
+      "loss": 1.1272,
+      "step": 1988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8928731643525222e-06,
+      "loss": 1.1297,
+      "step": 1989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851878714931937e-06,
+      "loss": 1.0937,
+      "step": 1990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8775165880165992e-06,
+      "loss": 1.1112,
+      "step": 1991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8698593271664134e-06,
+      "loss": 1.1219,
+      "step": 1992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8622161021621055e-06,
+      "loss": 1.1082,
+      "step": 1993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8545869261989157e-06,
+      "loss": 1.1582,
+      "step": 1994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469718124478198e-06,
+      "loss": 1.1137,
+      "step": 1995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839370774055529e-06,
+      "loss": 1.1295,
+      "step": 1996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8317838241444464e-06,
+      "loss": 1.1207,
+      "step": 1997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8242109758126525e-06,
+      "loss": 1.1239,
+      "step": 1998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8166522421338916e-06,
+      "loss": 1.1552,
+      "step": 1999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8091076361575277e-06,
+      "loss": 1.0828,
+      "step": 2000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8015771709085483e-06,
+      "loss": 1.1032,
+      "step": 2001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7940608593875154e-06,
+      "loss": 1.1287,
+      "step": 2002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7865587145705687e-06,
+      "loss": 1.1513,
+      "step": 2003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7790707494093807e-06,
+      "loss": 1.1318,
+      "step": 2004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771596976831146e-06,
+      "loss": 1.1259,
+      "step": 2005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7641374097385645e-06,
+      "loss": 1.1802,
+      "step": 2006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7566920610097993e-06,
+      "loss": 1.1329,
+      "step": 2007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492609434984808e-06,
+      "loss": 1.1083,
+      "step": 2008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741844070033658e-06,
+      "loss": 1.1836,
+      "step": 2009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7344414534197995e-06,
+      "loss": 1.1439,
+      "step": 2010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7270531064367546e-06,
+      "loss": 1.1034,
+      "step": 2011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7196790418397357e-06,
+      "loss": 1.1231,
+      "step": 2012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7123192723593075e-06,
+      "loss": 1.1005,
+      "step": 2013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7049738107013435e-06,
+      "loss": 1.1396,
+      "step": 2014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6976426695470272e-06,
+      "loss": 1.1992,
+      "step": 2015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.690325861552815e-06,
+      "loss": 1.1344,
+      "step": 2016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6830233993504174e-06,
+      "loss": 1.1564,
+      "step": 2017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6757352955467754e-06,
+      "loss": 1.147,
+      "step": 2018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6684615627240509e-06,
+      "loss": 1.1475,
+      "step": 2019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6612022134395865e-06,
+      "loss": 1.1856,
+      "step": 2020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6539572602258968e-06,
+      "loss": 1.1657,
+      "step": 2021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6467267155906442e-06,
+      "loss": 1.1184,
+      "step": 2022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6395105920166165e-06,
+      "loss": 1.0777,
+      "step": 2023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6323089019617045e-06,
+      "loss": 1.1059,
+      "step": 2024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6251216578588747e-06,
+      "loss": 1.1568,
+      "step": 2025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6179488721161674e-06,
+      "loss": 1.1656,
+      "step": 2026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6107905571166494e-06,
+      "loss": 1.1441,
+      "step": 2027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6036467252184085e-06,
+      "loss": 1.1282,
+      "step": 2028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965173887545326e-06,
+      "loss": 1.1592,
+      "step": 2029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5894025600330876e-06,
+      "loss": 1.1093,
+      "step": 2030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.582302251337081e-06,
+      "loss": 1.1595,
+      "step": 2031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5752164749244657e-06,
+      "loss": 1.1442,
+      "step": 2032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5681452430280997e-06,
+      "loss": 1.1241,
+      "step": 2033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5610885678557287e-06,
+      "loss": 1.1486,
+      "step": 2034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5540464615899764e-06,
+      "loss": 1.1618,
+      "step": 2035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.547018936388307e-06,
+      "loss": 1.1846,
+      "step": 2036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5400060043830178e-06,
+      "loss": 1.1331,
+      "step": 2037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5330076776812053e-06,
+      "loss": 1.1461,
+      "step": 2038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5260239683647616e-06,
+      "loss": 1.1757,
+      "step": 2039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5190548884903334e-06,
+      "loss": 1.1025,
+      "step": 2040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5121004500893144e-06,
+      "loss": 1.109,
+      "step": 2041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5051606651678264e-06,
+      "loss": 1.1399,
+      "step": 2042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4982355457066866e-06,
+      "loss": 1.114,
+      "step": 2043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4913251036613952e-06,
+      "loss": 1.1883,
+      "step": 2044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4844293509621167e-06,
+      "loss": 1.1181,
+      "step": 2045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4775482995136558e-06,
+      "loss": 1.1554,
+      "step": 2046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4706819611954315e-06,
+      "loss": 1.1786,
+      "step": 2047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.463830347861469e-06,
+      "loss": 1.1029,
+      "step": 2048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4569934713403678e-06,
+      "loss": 1.1527,
+      "step": 2049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.450171343435286e-06,
+      "loss": 1.1042,
+      "step": 2050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4433639759239203e-06,
+      "loss": 1.1579,
+      "step": 2051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4365713805584902e-06,
+      "loss": 1.1504,
+      "step": 2052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4297935690657062e-06,
+      "loss": 1.1301,
+      "step": 2053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230305531467537e-06,
+      "loss": 1.0797,
+      "step": 2054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4162823444772855e-06,
+      "loss": 1.1255,
+      "step": 2055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4095489547073815e-06,
+      "loss": 1.1245,
+      "step": 2056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4028303954615398e-06,
+      "loss": 1.1394,
+      "step": 2057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3961266783386595e-06,
+      "loss": 1.1523,
+      "step": 2058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.389437814912018e-06,
+      "loss": 1.1412,
+      "step": 2059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.38276381672924e-06,
+      "loss": 1.1293,
+      "step": 2060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3761046953122913e-06,
+      "loss": 1.1655,
+      "step": 2061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3694604621574604e-06,
+      "loss": 1.1313,
+      "step": 2062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3628311287353247e-06,
+      "loss": 1.1814,
+      "step": 2063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3562167064907405e-06,
+      "loss": 1.1684,
+      "step": 2064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3496172068428259e-06,
+      "loss": 1.1407,
+      "step": 2065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3430326411849304e-06,
+      "loss": 1.1562,
+      "step": 2066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3364630208846253e-06,
+      "loss": 1.1144,
+      "step": 2067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3299083572836835e-06,
+      "loss": 1.11,
+      "step": 2068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32336866169805e-06,
+      "loss": 1.1572,
+      "step": 2069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3168439454178307e-06,
+      "loss": 1.1882,
+      "step": 2070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3103342197072754e-06,
+      "loss": 1.1097,
+      "step": 2071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3038394958047496e-06,
+      "loss": 1.0831,
+      "step": 2072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973597849227192e-06,
+      "loss": 1.1415,
+      "step": 2073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2908950982477364e-06,
+      "loss": 1.1265,
+      "step": 2074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2844454469404167e-06,
+      "loss": 1.1811,
+      "step": 2075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.278010842135411e-06,
+      "loss": 1.0641,
+      "step": 2076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715912949413967e-06,
+      "loss": 1.1111,
+      "step": 2077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651868164410609e-06,
+      "loss": 1.0905,
+      "step": 2078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258797417691071e-06,
+      "loss": 1.1775,
+      "step": 2079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2524231097220585e-06,
+      "loss": 1.1686,
+      "step": 2080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2460639035386135e-06,
+      "loss": 1.1331,
+      "step": 2081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2397198101192432e-06,
+      "loss": 1.1988,
+      "step": 2082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2333908404163653e-06,
+      "loss": 1.1328,
+      "step": 2083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2270770053562953e-06,
+      "loss": 1.1467,
+      "step": 2084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2207783158392127e-06,
+      "loss": 1.1817,
+      "step": 2085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2144947827391507e-06,
+      "loss": 1.1166,
+      "step": 2086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2082264169039837e-06,
+      "loss": 1.1694,
+      "step": 2087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201973229155391e-06,
+      "loss": 1.1386,
+      "step": 2088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195735230288857e-06,
+      "loss": 1.1308,
+      "step": 2089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1895124310736384e-06,
+      "loss": 1.1128,
+      "step": 2090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183304842252756e-06,
+      "loss": 1.166,
+      "step": 2091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1771124745429685e-06,
+      "loss": 1.1183,
+      "step": 2092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1709353386347544e-06,
+      "loss": 1.1062,
+      "step": 2093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1647734451923031e-06,
+      "loss": 1.1401,
+      "step": 2094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158626804853482e-06,
+      "loss": 1.1385,
+      "step": 2095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1524954282298317e-06,
+      "loss": 1.1361,
+      "step": 2096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463793259065391e-06,
+      "loss": 1.1263,
+      "step": 2097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1402785084424206e-06,
+      "loss": 1.1498,
+      "step": 2098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1341929863699042e-06,
+      "loss": 1.1155,
+      "step": 2099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1281227701950158e-06,
+      "loss": 1.0784,
+      "step": 2100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1220678703973554e-06,
+      "loss": 1.1574,
+      "step": 2101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1160282974300761e-06,
+      "loss": 1.1433,
+      "step": 2102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1100040617198782e-06,
+      "loss": 1.0994,
+      "step": 2103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1039951736669829e-06,
+      "loss": 1.1206,
+      "step": 2104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0980016436451113e-06,
+      "loss": 1.1358,
+      "step": 2105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0920234820014697e-06,
+      "loss": 1.1286,
+      "step": 2106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.086060699056739e-06,
+      "loss": 1.1545,
+      "step": 2107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.080113305105045e-06,
+      "loss": 1.1231,
+      "step": 2108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0741813104139441e-06,
+      "loss": 1.1081,
+      "step": 2109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0682647252244149e-06,
+      "loss": 1.103,
+      "step": 2110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0623635597508297e-06,
+      "loss": 1.1392,
+      "step": 2111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056477824180937e-06,
+      "loss": 1.15,
+      "step": 2112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506075286758488e-06,
+      "loss": 1.1488,
+      "step": 2113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044752683370026e-06,
+      "loss": 1.1607,
+      "step": 2114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0389132983712512e-06,
+      "loss": 1.1345,
+      "step": 2115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0330893837606149e-06,
+      "loss": 1.0661,
+      "step": 2116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027280949592504e-06,
+      "loss": 1.0734,
+      "step": 2117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0214880058945819e-06,
+      "loss": 1.0913,
+      "step": 2118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015710562667762e-06,
+      "loss": 1.0734,
+      "step": 2119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0099486298862048e-06,
+      "loss": 1.1361,
+      "step": 2120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0042022174972888e-06,
+      "loss": 1.144,
+      "step": 2121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.98471335421598e-07,
+      "loss": 1.1358,
+      "step": 2122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927559935529107e-07,
+      "loss": 1.1635,
+      "step": 2123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870562017581687e-07,
+      "loss": 1.1237,
+      "step": 2124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.813719698774771e-07,
+      "loss": 1.1306,
+      "step": 2125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.757033077240673e-07,
+      "loss": 1.088,
+      "step": 2126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700502250843035e-07,
+      "loss": 1.1465,
+      "step": 2127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644127317176454e-07,
+      "loss": 1.134,
+      "step": 2128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58790837356639e-07,
+      "loss": 1.1181,
+      "step": 2129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531845517069061e-07,
+      "loss": 1.1839,
+      "step": 2130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475938844471166e-07,
+      "loss": 1.0376,
+      "step": 2131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420188452289747e-07,
+      "loss": 1.2111,
+      "step": 2132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.364594436772157e-07,
+      "loss": 1.0856,
+      "step": 2133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.309156893895677e-07,
+      "loss": 1.1316,
+      "step": 2134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.253875919367461e-07,
+      "loss": 1.1045,
+      "step": 2135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.198751608624412e-07,
+      "loss": 1.0935,
+      "step": 2136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.143784056832939e-07,
+      "loss": 1.1233,
+      "step": 2137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.088973358888809e-07,
+      "loss": 1.1681,
+      "step": 2138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.034319609417031e-07,
+      "loss": 1.1409,
+      "step": 2139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.979822902771651e-07,
+      "loss": 1.1308,
+      "step": 2140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.925483333035578e-07,
+      "loss": 1.112,
+      "step": 2141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.871300994020449e-07,
+      "loss": 1.1756,
+      "step": 2142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.817275979266471e-07,
+      "loss": 1.1679,
+      "step": 2143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.763408382042238e-07,
+      "loss": 1.1415,
+      "step": 2144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.709698295344549e-07,
+      "loss": 1.1691,
+      "step": 2145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656145811898331e-07,
+      "loss": 1.1575,
+      "step": 2146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60275102415642e-07,
+      "loss": 1.1751,
+      "step": 2147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.549514024299343e-07,
+      "loss": 1.1851,
+      "step": 2148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496434904235306e-07,
+      "loss": 1.1338,
+      "step": 2149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443513755599897e-07,
+      "loss": 1.0882,
+      "step": 2150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.390750669755976e-07,
+      "loss": 1.1226,
+      "step": 2151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.33814573779359e-07,
+      "loss": 1.1653,
+      "step": 2152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.285699050529672e-07,
+      "loss": 1.1062,
+      "step": 2153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.233410698508005e-07,
+      "loss": 1.1584,
+      "step": 2154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181280771999e-07,
+      "loss": 1.1429,
+      "step": 2155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129309360999627e-07,
+      "loss": 1.1163,
+      "step": 2156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.0774965552331e-07,
+      "loss": 1.1164,
+      "step": 2157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025842444148857e-07,
+      "loss": 1.1446,
+      "step": 2158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.974347116922398e-07,
+      "loss": 1.1451,
+      "step": 2159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.923010662455066e-07,
+      "loss": 1.1675,
+      "step": 2160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87183316937391e-07,
+      "loss": 1.1497,
+      "step": 2161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.820814726031589e-07,
+      "loss": 1.152,
+      "step": 2162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.769955420506181e-07,
+      "loss": 1.1306,
+      "step": 2163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719255340600995e-07,
+      "loss": 1.1563,
+      "step": 2164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668714573844438e-07,
+      "loss": 1.038,
+      "step": 2165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618333207489959e-07,
+      "loss": 1.167,
+      "step": 2166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.568111328515737e-07,
+      "loss": 1.0741,
+      "step": 2167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.518049023624629e-07,
+      "loss": 1.1109,
+      "step": 2168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.468146379244079e-07,
+      "loss": 1.1471,
+      "step": 2169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.418403481525793e-07,
+      "loss": 1.1245,
+      "step": 2170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368820416345724e-07,
+      "loss": 1.0871,
+      "step": 2171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.31939726930393e-07,
+      "loss": 1.1461,
+      "step": 2172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270134125724349e-07,
+      "loss": 1.152,
+      "step": 2173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.221031070654661e-07,
+      "loss": 1.1542,
+      "step": 2174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.172088188866255e-07,
+      "loss": 1.1516,
+      "step": 2175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123305564853899e-07,
+      "loss": 1.1434,
+      "step": 2176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074683282835792e-07,
+      "loss": 1.0948,
+      "step": 2177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.026221426753221e-07,
+      "loss": 1.1388,
+      "step": 2178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.977920080270606e-07,
+      "loss": 1.14,
+      "step": 2179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.929779326775187e-07,
+      "loss": 1.1525,
+      "step": 2180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.881799249376997e-07,
+      "loss": 1.1181,
+      "step": 2181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.833979930908685e-07,
+      "loss": 1.1236,
+      "step": 2182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.786321453925316e-07,
+      "loss": 1.1664,
+      "step": 2183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.73882390070435e-07,
+      "loss": 1.1284,
+      "step": 2184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691487353245385e-07,
+      "loss": 1.1541,
+      "step": 2185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.644311893270072e-07,
+      "loss": 1.1212,
+      "step": 2186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.597297602221942e-07,
+      "loss": 1.0896,
+      "step": 2187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550444561266311e-07,
+      "loss": 1.1094,
+      "step": 2188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.503752851290102e-07,
+      "loss": 1.145,
+      "step": 2189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.457222552901698e-07,
+      "loss": 1.1287,
+      "step": 2190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410853746430856e-07,
+      "loss": 1.1499,
+      "step": 2191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.364646511928529e-07,
+      "loss": 1.135,
+      "step": 2192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318600929166719e-07,
+      "loss": 1.0976,
+      "step": 2193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272717077638324e-07,
+      "loss": 1.1506,
+      "step": 2194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226995036557104e-07,
+      "loss": 1.1343,
+      "step": 2195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.181434884857407e-07,
+      "loss": 1.127,
+      "step": 2196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.136036701194114e-07,
+      "loss": 1.1037,
+      "step": 2197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.090800563942479e-07,
+      "loss": 1.117,
+      "step": 2198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.045726551198039e-07,
+      "loss": 1.1138,
+      "step": 2199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.000814740776384e-07,
+      "loss": 1.131,
+      "step": 2200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.956065210213114e-07,
+      "loss": 1.1936,
+      "step": 2201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911478036763663e-07,
+      "loss": 1.144,
+      "step": 2202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.867053297403136e-07,
+      "loss": 1.1215,
+      "step": 2203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.822791068826294e-07,
+      "loss": 1.1526,
+      "step": 2204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.778691427447247e-07,
+      "loss": 1.1252,
+      "step": 2205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.734754449399494e-07,
+      "loss": 1.1171,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.690980210535646e-07,
+      "loss": 1.0573,
+      "step": 2207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.647368786427432e-07,
+      "loss": 1.1397,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.603920252365425e-07,
+      "loss": 1.0985,
+      "step": 2209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.560634683359012e-07,
+      "loss": 1.1017,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.517512154136284e-07,
+      "loss": 1.1026,
+      "step": 2211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.474552739143779e-07,
+      "loss": 1.1199,
+      "step": 2212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.431756512546526e-07,
+      "loss": 1.1265,
+      "step": 2213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389123548227726e-07,
+      "loss": 1.1281,
+      "step": 2214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346653919788825e-07,
+      "loss": 1.1423,
+      "step": 2215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.304347700549206e-07,
+      "loss": 1.1917,
+      "step": 2216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262204963546169e-07,
+      "loss": 1.1401,
+      "step": 2217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.220225781534815e-07,
+      "loss": 1.1152,
+      "step": 2218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178410226987851e-07,
+      "loss": 1.0825,
+      "step": 2219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.136758372095452e-07,
+      "loss": 1.1262,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095270288765298e-07,
+      "loss": 1.1087,
+      "step": 2221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053946048622238e-07,
+      "loss": 1.1738,
+      "step": 2222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012785723008285e-07,
+      "loss": 1.117,
+      "step": 2223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971789382982506e-07,
+      "loss": 1.1016,
+      "step": 2224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93095709932081e-07,
+      "loss": 1.128,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890288942515897e-07,
+      "loss": 1.1413,
+      "step": 2226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84978498277715e-07,
+      "loss": 1.118,
+      "step": 2227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809445290030479e-07,
+      "loss": 1.058,
+      "step": 2228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.769269933918152e-07,
+      "loss": 1.1279,
+      "step": 2229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.729258983798757e-07,
+      "loss": 1.0716,
+      "step": 2230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.689412508747082e-07,
+      "loss": 1.0874,
+      "step": 2231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.649730577553935e-07,
+      "loss": 1.2149,
+      "step": 2232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6102132587260373e-07,
+      "loss": 1.1779,
+      "step": 2233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5708606204859617e-07,
+      "loss": 1.15,
+      "step": 2234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5316727307719745e-07,
+      "loss": 1.1552,
+      "step": 2235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4926496572378907e-07,
+      "loss": 1.1205,
+      "step": 2236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4537914672530214e-07,
+      "loss": 1.1607,
+      "step": 2237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415098227902004e-07,
+      "loss": 1.1243,
+      "step": 2238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3765700059846926e-07,
+      "loss": 1.1837,
+      "step": 2239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3382068680160816e-07,
+      "loss": 1.0687,
+      "step": 2240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3000088802261695e-07,
+      "loss": 1.0802,
+      "step": 2241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.261976108559795e-07,
+      "loss": 1.0648,
+      "step": 2242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224108618676626e-07,
+      "loss": 1.1152,
+      "step": 2243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.186406475950966e-07,
+      "loss": 1.1184,
+      "step": 2244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.148869745471651e-07,
+      "loss": 1.1017,
+      "step": 2245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1114984920419323e-07,
+      "loss": 1.1573,
+      "step": 2246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0742927801794563e-07,
+      "loss": 1.0937,
+      "step": 2247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037252674115999e-07,
+      "loss": 1.1586,
+      "step": 2248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.000378237797453e-07,
+      "loss": 1.0557,
+      "step": 2249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.963669534883763e-07,
+      "loss": 1.1222,
+      "step": 2250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.927126628748679e-07,
+      "loss": 1.1246,
+      "step": 2251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8907495824797246e-07,
+      "loss": 1.1017,
+      "step": 2252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8545384588781433e-07,
+      "loss": 1.1625,
+      "step": 2253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8184933204586494e-07,
+      "loss": 1.116,
+      "step": 2254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.782614229449455e-07,
+      "loss": 1.1469,
+      "step": 2255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7469012477921007e-07,
+      "loss": 1.1428,
+      "step": 2256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7113544371413566e-07,
+      "loss": 1.1085,
+      "step": 2257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6759738588651117e-07,
+      "loss": 1.1214,
+      "step": 2258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6407595740442394e-07,
+      "loss": 1.1796,
+      "step": 2259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6057116434725867e-07,
+      "loss": 1.1359,
+      "step": 2260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570830127656766e-07,
+      "loss": 1.1453,
+      "step": 2261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5361150868160833e-07,
+      "loss": 1.1906,
+      "step": 2262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5015665808824983e-07,
+      "loss": 1.1199,
+      "step": 2263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4671846695004006e-07,
+      "loss": 1.123,
+      "step": 2264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432969412026621e-07,
+      "loss": 1.0781,
+      "step": 2265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.398920867530231e-07,
+      "loss": 1.1352,
+      "step": 2266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.365039094792544e-07,
+      "loss": 1.1921,
+      "step": 2267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3313241523069137e-07,
+      "loss": 1.1279,
+      "step": 2268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2977760982786934e-07,
+      "loss": 1.1368,
+      "step": 2269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2643949906251414e-07,
+      "loss": 1.145,
+      "step": 2270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.23118088697526e-07,
+      "loss": 1.1474,
+      "step": 2271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198133844669782e-07,
+      "loss": 1.134,
+      "step": 2272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1652539207610025e-07,
+      "loss": 1.1562,
+      "step": 2273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132541172012693e-07,
+      "loss": 1.1263,
+      "step": 2274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.099995654900023e-07,
+      "loss": 1.1112,
+      "step": 2275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0676174256094703e-07,
+      "loss": 1.1135,
+      "step": 2276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035406540038699e-07,
+      "loss": 1.1621,
+      "step": 2277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0033630537964376e-07,
+      "loss": 1.1475,
+      "step": 2278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.971487022202468e-07,
+      "loss": 1.094,
+      "step": 2279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.939778500287471e-07,
+      "loss": 1.0979,
+      "step": 2280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9082375427929023e-07,
+      "loss": 1.0879,
+      "step": 2281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8768642041709595e-07,
+      "loss": 1.1187,
+      "step": 2282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8456585385844836e-07,
+      "loss": 1.103,
+      "step": 2283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8146205999068035e-07,
+      "loss": 1.1151,
+      "step": 2284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.783750441721711e-07,
+      "loss": 1.0933,
+      "step": 2285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7530481173233316e-07,
+      "loss": 1.1081,
+      "step": 2286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7225136797160656e-07,
+      "loss": 1.1049,
+      "step": 2287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.692147181614446e-07,
+      "loss": 1.1706,
+      "step": 2288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6619486754430936e-07,
+      "loss": 1.1142,
+      "step": 2289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.631918213336593e-07,
+      "loss": 1.1396,
+      "step": 2290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6020558471394284e-07,
+      "loss": 1.0947,
+      "step": 2291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.572361628405884e-07,
+      "loss": 1.1318,
+      "step": 2292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.542835608399952e-07,
+      "loss": 1.1547,
+      "step": 2293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5134778380952573e-07,
+      "loss": 1.0778,
+      "step": 2294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4842883681749365e-07,
+      "loss": 1.108,
+      "step": 2295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.455267249031601e-07,
+      "loss": 1.0784,
+      "step": 2296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.426414530767196e-07,
+      "loss": 1.134,
+      "step": 2297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.397730263192943e-07,
+      "loss": 1.0657,
+      "step": 2298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3692144958292863e-07,
+      "loss": 1.1646,
+      "step": 2299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3408672779057117e-07,
+      "loss": 1.1325,
+      "step": 2300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3126886583607845e-07,
+      "loss": 1.1066,
+      "step": 2301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2846786858419566e-07,
+      "loss": 1.0871,
+      "step": 2302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2568374087055588e-07,
+      "loss": 1.1305,
+      "step": 2303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229164875016654e-07,
+      "loss": 1.1026,
+      "step": 2304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2016611325490268e-07,
+      "loss": 1.1318,
+      "step": 2305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1743262287850175e-07,
+      "loss": 1.1374,
+      "step": 2306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1471602109155222e-07,
+      "loss": 1.0931,
+      "step": 2307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.120163125839847e-07,
+      "loss": 1.1324,
+      "step": 2308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0933350201656765e-07,
+      "loss": 1.1392,
+      "step": 2309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.066675940208951e-07,
+      "loss": 1.1278,
+      "step": 2310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0401859319937988e-07,
+      "loss": 1.0969,
+      "step": 2311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0138650412524941e-07,
+      "loss": 1.1636,
+      "step": 2312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9877133134253211e-07,
+      "loss": 1.1398,
+      "step": 2313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9617307936605213e-07,
+      "loss": 1.1314,
+      "step": 2314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935917526814235e-07,
+      "loss": 1.1335,
+      "step": 2315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9102735574504038e-07,
+      "loss": 1.113,
+      "step": 2316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8847989298406698e-07,
+      "loss": 1.128,
+      "step": 2317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.859493687964331e-07,
+      "loss": 1.1272,
+      "step": 2318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.834357875508286e-07,
+      "loss": 1.1578,
+      "step": 2319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8093915358669001e-07,
+      "loss": 1.1404,
+      "step": 2320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7845947121419739e-07,
+      "loss": 1.1176,
+      "step": 2321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.759967447142652e-07,
+      "loss": 1.1388,
+      "step": 2322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7355097833853807e-07,
+      "loss": 1.1329,
+      "step": 2323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7112217630937512e-07,
+      "loss": 1.1328,
+      "step": 2324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6871034281985444e-07,
+      "loss": 1.103,
+      "step": 2325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6631548203375537e-07,
+      "loss": 1.1402,
+      "step": 2326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6393759808555509e-07,
+      "loss": 1.1538,
+      "step": 2327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.615766950804276e-07,
+      "loss": 1.0957,
+      "step": 2328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5923277709422592e-07,
+      "loss": 1.0975,
+      "step": 2329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5690584817347975e-07,
+      "loss": 1.1503,
+      "step": 2330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.545959123353935e-07,
+      "loss": 1.1159,
+      "step": 2331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5230297356783164e-07,
+      "loss": 1.1415,
+      "step": 2332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5002703582931433e-07,
+      "loss": 1.1207,
+      "step": 2333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4776810304901412e-07,
+      "loss": 1.0892,
+      "step": 2334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4552617912674373e-07,
+      "loss": 1.1181,
+      "step": 2335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.433012679329515e-07,
+      "loss": 1.0721,
+      "step": 2336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.410933733087161e-07,
+      "loss": 1.1189,
+      "step": 2337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3890249906574171e-07,
+      "loss": 1.113,
+      "step": 2338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367286489863451e-07,
+      "loss": 1.1531,
+      "step": 2339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3457182682344972e-07,
+      "loss": 1.1078,
+      "step": 2340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3243203630059042e-07,
+      "loss": 1.144,
+      "step": 2341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.303092811118911e-07,
+      "loss": 1.1262,
+      "step": 2342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.282035649220692e-07,
+      "loss": 1.1217,
+      "step": 2343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2611489136642562e-07,
+      "loss": 1.1442,
+      "step": 2344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2404326405083934e-07,
+      "loss": 1.0796,
+      "step": 2345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2198868655176054e-07,
+      "loss": 1.0679,
+      "step": 2346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1995116241620308e-07,
+      "loss": 1.1041,
+      "step": 2347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1793069516174427e-07,
+      "loss": 1.1004,
+      "step": 2348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1592728827650835e-07,
+      "loss": 1.1559,
+      "step": 2349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1394094521917199e-07,
+      "loss": 1.1157,
+      "step": 2350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1197166941894988e-07,
+      "loss": 1.1226,
+      "step": 2351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.100194642755914e-07,
+      "loss": 1.0819,
+      "step": 2352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.080843331593795e-07,
+      "loss": 1.105,
+      "step": 2353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0616627941111623e-07,
+      "loss": 1.1308,
+      "step": 2354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0426530634212173e-07,
+      "loss": 1.1683,
+      "step": 2355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0238141723423079e-07,
+      "loss": 1.138,
+      "step": 2356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0051461533978179e-07,
+      "loss": 1.1199,
+      "step": 2357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.866490388161564e-08,
+      "loss": 1.1187,
+      "step": 2358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.683228605306682e-08,
+      "loss": 1.1517,
+      "step": 2359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.50167650179601e-08,
+      "loss": 1.1618,
+      "step": 2360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.321834391060602e-08,
+      "loss": 1.0883,
+      "step": 2361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.143702583579106e-08,
+      "loss": 1.0971,
+      "step": 2362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.967281386877636e-08,
+      "loss": 1.1419,
+      "step": 2363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.792571105529113e-08,
+      "loss": 1.1308,
+      "step": 2364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.619572041152935e-08,
+      "loss": 1.0992,
+      "step": 2365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.44828449241375e-08,
+      "loss": 1.1504,
+      "step": 2366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.278708755021902e-08,
+      "loss": 1.1209,
+      "step": 2367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.110845121732546e-08,
+      "loss": 1.1237,
+      "step": 2368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.944693882344645e-08,
+      "loss": 1.1298,
+      "step": 2369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.780255323701081e-08,
+      "loss": 1.1259,
+      "step": 2370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.6175297296881e-08,
+      "loss": 1.1255,
+      "step": 2371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.45651738123454e-08,
+      "loss": 1.1771,
+      "step": 2372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.297218556311602e-08,
+      "loss": 1.161,
+      "step": 2373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.139633529932077e-08,
+      "loss": 1.1621,
+      "step": 2374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.983762574150344e-08,
+      "loss": 1.1622,
+      "step": 2375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.82960595806137e-08,
+      "loss": 1.1167,
+      "step": 2376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.677163947800602e-08,
+      "loss": 1.1506,
+      "step": 2377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.526436806543523e-08,
+      "loss": 1.1502,
+      "step": 2378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.37742479450465e-08,
+      "loss": 1.0896,
+      "step": 2379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.230128168938088e-08,
+      "loss": 1.129,
+      "step": 2380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.08454718413598e-08,
+      "loss": 1.1557,
+      "step": 2381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.940682091428951e-08,
+      "loss": 1.1079,
+      "step": 2382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.798533139185103e-08,
+      "loss": 1.1426,
+      "step": 2383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6581005728099106e-08,
+      "loss": 1.1388,
+      "step": 2384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.519384634745773e-08,
+      "loss": 1.1239,
+      "step": 2385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.382385564471126e-08,
+      "loss": 1.1465,
+      "step": 2386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.247103598500891e-08,
+      "loss": 1.1411,
+      "step": 2387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1135389703853564e-08,
+      "loss": 1.1211,
+      "step": 2388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9816919107098514e-08,
+      "loss": 1.1044,
+      "step": 2389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.851562647094965e-08,
+      "loss": 1.0756,
+      "step": 2390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7231514041951033e-08,
+      "loss": 1.1173,
+      "step": 2391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.596458403699156e-08,
+      "loss": 1.0801,
+      "step": 2392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4714838643293844e-08,
+      "loss": 1.1224,
+      "step": 2393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.348228001841426e-08,
+      "loss": 1.1495,
+      "step": 2394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.226691029023733e-08,
+      "loss": 1.1117,
+      "step": 2395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1068731556973554e-08,
+      "loss": 1.1048,
+      "step": 2396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.988774588715383e-08,
+      "loss": 1.1078,
+      "step": 2397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.872395531962836e-08,
+      "loss": 1.1607,
+      "step": 2398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.757736186355998e-08,
+      "loss": 1.104,
+      "step": 2399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6447967498424166e-08,
+      "loss": 1.0604,
+      "step": 2400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.533577417400347e-08,
+      "loss": 1.1027,
+      "step": 2401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42407838103842e-08,
+      "loss": 1.1302,
+      "step": 2402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.316299829795311e-08,
+      "loss": 1.1912,
+      "step": 2403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.210241949739623e-08,
+      "loss": 1.1514,
+      "step": 2404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.105904923969338e-08,
+      "loss": 1.1007,
+      "step": 2405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0032889326114814e-08,
+      "loss": 1.1353,
+      "step": 2406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9023941528218967e-08,
+      "loss": 1.0848,
+      "step": 2407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8032207587849192e-08,
+      "loss": 1.077,
+      "step": 2408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7057689217133696e-08,
+      "loss": 1.1294,
+      "step": 2409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6100388098474482e-08,
+      "loss": 1.0971,
+      "step": 2410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5160305884556202e-08,
+      "loss": 1.1568,
+      "step": 2411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.423744419832952e-08,
+      "loss": 1.1012,
+      "step": 2412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3331804633023314e-08,
+      "loss": 1.0916,
+      "step": 2413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2443388752128036e-08,
+      "loss": 1.1078,
+      "step": 2414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1572198089402362e-08,
+      "loss": 1.1467,
+      "step": 2415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0718234148867644e-08,
+      "loss": 1.1636,
+      "step": 2416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9881498404802357e-08,
+      "loss": 1.1255,
+      "step": 2417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9061992301745435e-08,
+      "loss": 1.0843,
+      "step": 2418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8259717254489605e-08,
+      "loss": 1.1435,
+      "step": 2419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.747467464807917e-08,
+      "loss": 1.147,
+      "step": 2420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6706865837807784e-08,
+      "loss": 1.0631,
+      "step": 2421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5956292149219565e-08,
+      "loss": 1.156,
+      "step": 2422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5222954878101327e-08,
+      "loss": 1.1858,
+      "step": 2423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.450685529048368e-08,
+      "loss": 1.1315,
+      "step": 2424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.380799462263882e-08,
+      "loss": 1.1511,
+      "step": 2425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3126374081078308e-08,
+      "loss": 1.1474,
+      "step": 2426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2461994842547508e-08,
+      "loss": 1.1062,
+      "step": 2427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1814858054030043e-08,
+      "loss": 1.1465,
+      "step": 2428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1184964832738898e-08,
+      "loss": 1.1303,
+      "step": 2429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0572316266120874e-08,
+      "loss": 1.1139,
+      "step": 2430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.97691341184992e-09,
+      "loss": 1.121,
+      "step": 2431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.398757297827133e-09,
+      "loss": 1.1594,
+      "step": 2432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.837848922179648e-09,
+      "loss": 1.1457,
+      "step": 2433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.294189253257313e-09,
+      "loss": 1.1704,
+      "step": 2434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.767779229631566e-09,
+      "loss": 1.1273,
+      "step": 2435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258619760095453e-09,
+      "loss": 1.1436,
+      "step": 2436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.76671172366139e-09,
+      "loss": 1.1526,
+      "step": 2437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2920559695556215e-09,
+      "loss": 1.0866,
+      "step": 2438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.834653317222661e-09,
+      "loss": 1.1204,
+      "step": 2439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.394504556320845e-09,
+      "loss": 1.171,
+      "step": 2440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.971610446722341e-09,
+      "loss": 1.141,
+      "step": 2441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.565971718508699e-09,
+      "loss": 1.1484,
+      "step": 2442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.177589071974186e-09,
+      "loss": 1.1089,
+      "step": 2443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.806463177621345e-09,
+      "loss": 1.1491,
+      "step": 2444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4525946761598816e-09,
+      "loss": 1.1119,
+      "step": 2445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1159841785066704e-09,
+      "loss": 1.1646,
+      "step": 2446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.796632265785748e-09,
+      "loss": 1.1024,
+      "step": 2447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4945394893260976e-09,
+      "loss": 1.17,
+      "step": 2448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2097063706572053e-09,
+      "loss": 1.0966,
+      "step": 2449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942133401515722e-09,
+      "loss": 1.151,
+      "step": 2450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6918210438376936e-09,
+      "loss": 1.1226,
+      "step": 2451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4587697297618886e-09,
+      "loss": 1.1816,
+      "step": 2452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2429798616275802e-09,
+      "loss": 1.0985,
+      "step": 2453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0444518119734348e-09,
+      "loss": 1.1429,
+      "step": 2454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.631859235364026e-10,
+      "loss": 1.1238,
+      "step": 2455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.991825092561577e-10,
+      "loss": 1.1412,
+      "step": 2456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.524418522639963e-10,
+      "loss": 1.1368,
+      "step": 2457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2296420589504894e-10,
+      "loss": 1.1496,
+      "step": 2458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1074979367828884e-10,
+      "loss": 1.1329,
+      "step": 2459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.157988093409724e-10,
+      "loss": 1.135,
+      "step": 2460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.381114168041986e-10,
+      "loss": 1.1103,
+      "step": 2461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.768775018845987e-11,
+      "loss": 1.0644,
+      "step": 2462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.452791381031162e-11,
+      "loss": 1.0737,
+      "step": 2463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.631982177931265e-12,
+      "loss": 1.0918,
+      "step": 2464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.2356,
+      "step": 2465
+    },
+    {
+      "epoch": 1.0,
+      "step": 2465,
+      "total_flos": 1.360752174736343e+18,
+      "train_loss": 1.2143683295220922,
+      "train_runtime": 14582.0067,
+      "train_samples_per_second": 10.816,
+      "train_steps_per_second": 0.169
+    }
+  ],
+  "max_steps": 2465,
+  "num_train_epochs": 1,
+  "total_flos": 1.360752174736343e+18,
+  "trial_name": null,
+  "trial_params": null
+}