diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.7735849056603774,
+  "global_step": 800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 0.5934,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 0.6041,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.5272,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 0.4536,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.4569,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.479,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7450980392156867e-06,
+      "loss": 0.4895,
+      "step": 7
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.4636,
+      "step": 8
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.4308,
+      "step": 9
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.4242,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.4305,
+      "step": 11
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.4359,
+      "step": 12
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.4432,
+      "step": 13
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 0.4306,
+      "step": 14
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.4278,
+      "step": 15
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.4061,
+      "step": 16
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3843,
+      "step": 17
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.414,
+      "step": 18
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.4017,
+      "step": 19
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.4138,
+      "step": 20
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.3523,
+      "step": 21
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.3927,
+      "step": 22
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 0.4004,
+      "step": 23
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.3774,
+      "step": 24
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.4201,
+      "step": 25
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.368,
+      "step": 26
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.3638,
+      "step": 27
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0980392156862747e-05,
+      "loss": 0.4056,
+      "step": 28
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1372549019607844e-05,
+      "loss": 0.3971,
+      "step": 29
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.354,
+      "step": 30
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 0.4311,
+      "step": 31
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 0.3519,
+      "step": 32
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.3972,
+      "step": 33
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.378,
+      "step": 34
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 0.3766,
+      "step": 35
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.3864,
+      "step": 36
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 0.3933,
+      "step": 37
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 0.3491,
+      "step": 38
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.3682,
+      "step": 39
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 0.3915,
+      "step": 40
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 0.37,
+      "step": 41
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.3648,
+      "step": 42
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 0.387,
+      "step": 43
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 0.3751,
+      "step": 44
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.3872,
+      "step": 45
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8039215686274513e-05,
+      "loss": 0.379,
+      "step": 46
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 0.3497,
+      "step": 47
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.3757,
+      "step": 48
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 0.3743,
+      "step": 49
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.3795,
+      "step": 50
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3625,
+      "step": 51
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9999981763654435e-05,
+      "loss": 0.385,
+      "step": 52
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9999927054684256e-05,
+      "loss": 0.3674,
+      "step": 53
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9999835873288997e-05,
+      "loss": 0.3378,
+      "step": 54
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.999970821980122e-05,
+      "loss": 0.3337,
+      "step": 55
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9999544094686517e-05,
+      "loss": 0.3885,
+      "step": 56
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99993434985435e-05,
+      "loss": 0.4232,
+      "step": 57
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9999106432103785e-05,
+      "loss": 0.3807,
+      "step": 58
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9998832896232022e-05,
+      "loss": 0.3791,
+      "step": 59
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.999852289192587e-05,
+      "loss": 0.3935,
+      "step": 60
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9998176420316002e-05,
+      "loss": 0.3309,
+      "step": 61
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9997793482666093e-05,
+      "loss": 0.3667,
+      "step": 62
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9997374080372813e-05,
+      "loss": 0.3394,
+      "step": 63
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9996918214965843e-05,
+      "loss": 0.322,
+      "step": 64
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.999642588810784e-05,
+      "loss": 0.3304,
+      "step": 65
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9995897101594454e-05,
+      "loss": 0.3509,
+      "step": 66
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9995331857354318e-05,
+      "loss": 0.376,
+      "step": 67
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9994730157449024e-05,
+      "loss": 0.3461,
+      "step": 68
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999409200407313e-05,
+      "loss": 0.3331,
+      "step": 69
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9993417399554163e-05,
+      "loss": 0.3922,
+      "step": 70
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999270634635258e-05,
+      "loss": 0.3382,
+      "step": 71
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9991958847061786e-05,
+      "loss": 0.4079,
+      "step": 72
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999117490440811e-05,
+      "loss": 0.3469,
+      "step": 73
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9990354521250803e-05,
+      "loss": 0.3527,
+      "step": 74
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9989497700582025e-05,
+      "loss": 0.3449,
+      "step": 75
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998860444552683e-05,
+      "loss": 0.3743,
+      "step": 76
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998767475934316e-05,
+      "loss": 0.3372,
+      "step": 77
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9986708645421828e-05,
+      "loss": 0.3399,
+      "step": 78
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9985706107286515e-05,
+      "loss": 0.4058,
+      "step": 79
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9984667148593745e-05,
+      "loss": 0.3814,
+      "step": 80
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9983591773132885e-05,
+      "loss": 0.3552,
+      "step": 81
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9982479984826112e-05,
+      "loss": 0.3435,
+      "step": 82
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9981331787728416e-05,
+      "loss": 0.3445,
+      "step": 83
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.998014718602759e-05,
+      "loss": 0.3148,
+      "step": 84
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9978926184044187e-05,
+      "loss": 0.4291,
+      "step": 85
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9977668786231536e-05,
+      "loss": 0.3768,
+      "step": 86
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.99763749971757e-05,
+      "loss": 0.3654,
+      "step": 87
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9975044821595472e-05,
+      "loss": 0.3718,
+      "step": 88
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9973678264342372e-05,
+      "loss": 0.3509,
+      "step": 89
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9972275330400597e-05,
+      "loss": 0.3596,
+      "step": 90
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.997083602488702e-05,
+      "loss": 0.3535,
+      "step": 91
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9969360353051182e-05,
+      "loss": 0.3356,
+      "step": 92
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9967848320275253e-05,
+      "loss": 0.3594,
+      "step": 93
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.996629993207402e-05,
+      "loss": 0.3377,
+      "step": 94
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9964715194094876e-05,
+      "loss": 0.3896,
+      "step": 95
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9963094112117786e-05,
+      "loss": 0.318,
+      "step": 96
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9961436692055273e-05,
+      "loss": 0.3208,
+      "step": 97
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 0.3211,
+      "step": 98
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9958012861986715e-05,
+      "loss": 0.3343,
+      "step": 99
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9956246464468294e-05,
+      "loss": 0.3519,
+      "step": 100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9954443753839666e-05,
+      "loss": 0.3549,
+      "step": 101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.99526047366758e-05,
+      "loss": 0.3556,
+      "step": 102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9950729419684084e-05,
+      "loss": 0.3347,
+      "step": 103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.99488178097043e-05,
+      "loss": 0.3477,
+      "step": 104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.994686991370861e-05,
+      "loss": 0.3695,
+      "step": 105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.994488573880152e-05,
+      "loss": 0.3368,
+      "step": 106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9942865292219837e-05,
+      "loss": 0.3655,
+      "step": 107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.994080858133268e-05,
+      "loss": 0.3782,
+      "step": 108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.993871561364143e-05,
+      "loss": 0.3475,
+      "step": 109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.99365863967797e-05,
+      "loss": 0.3613,
+      "step": 110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9934420938513313e-05,
+      "loss": 0.3598,
+      "step": 111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.993221924674028e-05,
+      "loss": 0.3599,
+      "step": 112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9929981329490774e-05,
+      "loss": 0.341,
+      "step": 113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9927707194927067e-05,
+      "loss": 0.354,
+      "step": 114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9925396851343545e-05,
+      "loss": 0.3725,
+      "step": 115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9923050307166655e-05,
+      "loss": 0.3686,
+      "step": 116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9920667570954874e-05,
+      "loss": 0.357,
+      "step": 117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.991824865139868e-05,
+      "loss": 0.3271,
+      "step": 118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.991579355732053e-05,
+      "loss": 0.3085,
+      "step": 119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9913302297674805e-05,
+      "loss": 0.3721,
+      "step": 120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9910774881547803e-05,
+      "loss": 0.3449,
+      "step": 121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.990821131815769e-05,
+      "loss": 0.3325,
+      "step": 122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.990561161685447e-05,
+      "loss": 0.3543,
+      "step": 123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.990297578711996e-05,
+      "loss": 0.3531,
+      "step": 124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9900303838567732e-05,
+      "loss": 0.3776,
+      "step": 125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9897595780943104e-05,
+      "loss": 0.3228,
+      "step": 126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9894851624123092e-05,
+      "loss": 0.3007,
+      "step": 127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9892071378116378e-05,
+      "loss": 0.3977,
+      "step": 128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.988925505306326e-05,
+      "loss": 0.336,
+      "step": 129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9886402659235638e-05,
+      "loss": 0.3184,
+      "step": 130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.988351420703696e-05,
+      "loss": 0.3719,
+      "step": 131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9880589707002186e-05,
+      "loss": 0.3495,
+      "step": 132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9877629169797756e-05,
+      "loss": 0.3397,
+      "step": 133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9874632606221547e-05,
+      "loss": 0.3449,
+      "step": 134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.987160002720283e-05,
+      "loss": 0.3362,
+      "step": 135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.986853144380224e-05,
+      "loss": 0.3358,
+      "step": 136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9865426867211724e-05,
+      "loss": 0.3788,
+      "step": 137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9862286308754514e-05,
+      "loss": 0.3188,
+      "step": 138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.985910977988506e-05,
+      "loss": 0.3245,
+      "step": 139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.985589729218903e-05,
+      "loss": 0.3097,
+      "step": 140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9852648857383224e-05,
+      "loss": 0.3536,
+      "step": 141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.984936448731556e-05,
+      "loss": 0.3285,
+      "step": 142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.984604419396502e-05,
+      "loss": 0.3453,
+      "step": 143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9842687989441605e-05,
+      "loss": 0.3642,
+      "step": 144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 0.3271,
+      "step": 145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9835867895971015e-05,
+      "loss": 0.3206,
+      "step": 146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9832404031898554e-05,
+      "loss": 0.3202,
+      "step": 147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9828904306402564e-05,
+      "loss": 0.3649,
+      "step": 148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.982536873224748e-05,
+      "loss": 0.3175,
+      "step": 149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9821797322328497e-05,
+      "loss": 0.3293,
+      "step": 150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.981819008967151e-05,
+      "loss": 0.3196,
+      "step": 151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.981454704743306e-05,
+      "loss": 0.348,
+      "step": 152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9810868208900306e-05,
+      "loss": 0.3163,
+      "step": 153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9807153587490963e-05,
+      "loss": 0.3719,
+      "step": 154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.980340319675326e-05,
+      "loss": 0.3303,
+      "step": 155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.979961705036587e-05,
+      "loss": 0.3235,
+      "step": 156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9795795162137896e-05,
+      "loss": 0.333,
+      "step": 157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.979193754600879e-05,
+      "loss": 0.3124,
+      "step": 158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9788044216048314e-05,
+      "loss": 0.3646,
+      "step": 159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9784115186456496e-05,
+      "loss": 0.3462,
+      "step": 160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.978015047156356e-05,
+      "loss": 0.3336,
+      "step": 161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.977615008582989e-05,
+      "loss": 0.2962,
+      "step": 162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9772114043845968e-05,
+      "loss": 0.3366,
+      "step": 163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9768042360332328e-05,
+      "loss": 0.3216,
+      "step": 164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.976393505013949e-05,
+      "loss": 0.3072,
+      "step": 165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9759792128247922e-05,
+      "loss": 0.3363,
+      "step": 166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.975561360976798e-05,
+      "loss": 0.3214,
+      "step": 167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9751399509939837e-05,
+      "loss": 0.3367,
+      "step": 168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974714984413346e-05,
+      "loss": 0.2889,
+      "step": 169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974286462784851e-05,
+      "loss": 0.3585,
+      "step": 170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9738543876714335e-05,
+      "loss": 0.3525,
+      "step": 171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973418760648987e-05,
+      "loss": 0.3219,
+      "step": 172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9729795833063614e-05,
+      "loss": 0.3435,
+      "step": 173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725368572453538e-05,
+      "loss": 0.3429,
+      "step": 174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9720905840807055e-05,
+      "loss": 0.3196,
+      "step": 175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9716407654400954e-05,
+      "loss": 0.3261,
+      "step": 176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971187402964132e-05,
+      "loss": 0.3091,
+      "step": 177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9707304983063512e-05,
+      "loss": 0.3021,
+      "step": 178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.970270053133207e-05,
+      "loss": 0.3251,
+      "step": 179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9698060691240665e-05,
+      "loss": 0.3633,
+      "step": 180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9693385479712047e-05,
+      "loss": 0.2938,
+      "step": 181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968867491379797e-05,
+      "loss": 0.3004,
+      "step": 182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9683929010679136e-05,
+      "loss": 0.3164,
+      "step": 183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9679147787665128e-05,
+      "loss": 0.284,
+      "step": 184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9674331262194353e-05,
+      "loss": 0.325,
+      "step": 185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9669479451833976e-05,
+      "loss": 0.3009,
+      "step": 186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.966459237427986e-05,
+      "loss": 0.3441,
+      "step": 187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9659670047356486e-05,
+      "loss": 0.3349,
+      "step": 188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.965471248901691e-05,
+      "loss": 0.3594,
+      "step": 189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9649719717342673e-05,
+      "loss": 0.3445,
+      "step": 190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.964469175054377e-05,
+      "loss": 0.3003,
+      "step": 191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 0.3294,
+      "step": 192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9634530305053622e-05,
+      "loss": 0.2857,
+      "step": 193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9629396863423914e-05,
+      "loss": 0.3287,
+      "step": 194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9624228300792446e-05,
+      "loss": 0.3331,
+      "step": 195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.961902463601036e-05,
+      "loss": 0.3767,
+      "step": 196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9613785888056827e-05,
+      "loss": 0.3213,
+      "step": 197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9608512076038964e-05,
+      "loss": 0.3122,
+      "step": 198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9603203219191785e-05,
+      "loss": 0.3432,
+      "step": 199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9597859336878117e-05,
+      "loss": 0.2853,
+      "step": 200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.959248044858854e-05,
+      "loss": 0.3485,
+      "step": 201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.958706657394131e-05,
+      "loss": 0.3494,
+      "step": 202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.958161773268228e-05,
+      "loss": 0.3063,
+      "step": 203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9576133944684842e-05,
+      "loss": 0.4,
+      "step": 204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9570615229949844e-05,
+      "loss": 0.3508,
+      "step": 205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9565061608605526e-05,
+      "loss": 0.3159,
+      "step": 206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.955947310090744e-05,
+      "loss": 0.3262,
+      "step": 207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9553849727238375e-05,
+      "loss": 0.285,
+      "step": 208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9548191508108294e-05,
+      "loss": 0.3195,
+      "step": 209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.954249846415424e-05,
+      "loss": 0.2972,
+      "step": 210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9536770616140277e-05,
+      "loss": 0.3283,
+      "step": 211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9531007984957408e-05,
+      "loss": 0.3577,
+      "step": 212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9525210591623502e-05,
+      "loss": 0.2111,
+      "step": 213
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.951937845728321e-05,
+      "loss": 0.2268,
+      "step": 214
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.95135116032079e-05,
+      "loss": 0.2056,
+      "step": 215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.950761005079556e-05,
+      "loss": 0.2218,
+      "step": 216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9501673821570743e-05,
+      "loss": 0.2358,
+      "step": 217
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.949570293718448e-05,
+      "loss": 0.2399,
+      "step": 218
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.948969741941418e-05,
+      "loss": 0.2189,
+      "step": 219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9483657290163598e-05,
+      "loss": 0.219,
+      "step": 220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9477582571462706e-05,
+      "loss": 0.2076,
+      "step": 221
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.947147328546763e-05,
+      "loss": 0.2543,
+      "step": 222
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9465329454460587e-05,
+      "loss": 0.2072,
+      "step": 223
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9459151100849783e-05,
+      "loss": 0.2369,
+      "step": 224
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9452938247169336e-05,
+      "loss": 0.2108,
+      "step": 225
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.944669091607919e-05,
+      "loss": 0.2048,
+      "step": 226
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9440409130365044e-05,
+      "loss": 0.2157,
+      "step": 227
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.943409291293827e-05,
+      "loss": 0.2374,
+      "step": 228
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9427742286835804e-05,
+      "loss": 0.1925,
+      "step": 229
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.942135727522009e-05,
+      "loss": 0.2104,
+      "step": 230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.941493790137898e-05,
+      "loss": 0.2189,
+      "step": 231
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9408484188725668e-05,
+      "loss": 0.238,
+      "step": 232
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9401996160798574e-05,
+      "loss": 0.238,
+      "step": 233
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9395473841261282e-05,
+      "loss": 0.2015,
+      "step": 234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9388917253902446e-05,
+      "loss": 0.2094,
+      "step": 235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9382326422635705e-05,
+      "loss": 0.2431,
+      "step": 236
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.93757013714996e-05,
+      "loss": 0.2381,
+      "step": 237
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9369042124657468e-05,
+      "loss": 0.2203,
+      "step": 238
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 0.2187,
+      "step": 239
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9355621141132022e-05,
+      "loss": 0.2582,
+      "step": 240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.934885945339865e-05,
+      "loss": 0.2083,
+      "step": 241
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9342063667858948e-05,
+      "loss": 0.2332,
+      "step": 242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9335233809298982e-05,
+      "loss": 0.2277,
+      "step": 243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9328369902629084e-05,
+      "loss": 0.2256,
+      "step": 244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9321471972883763e-05,
+      "loss": 0.2238,
+      "step": 245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9314540045221628e-05,
+      "loss": 0.2429,
+      "step": 246
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9307574144925288e-05,
+      "loss": 0.2295,
+      "step": 247
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9300574297401252e-05,
+      "loss": 0.2142,
+      "step": 248
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9293540528179843e-05,
+      "loss": 0.2261,
+      "step": 249
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9286472862915122e-05,
+      "loss": 0.2322,
+      "step": 250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.927937132738476e-05,
+      "loss": 0.2294,
+      "step": 251
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9272235947489964e-05,
+      "loss": 0.2164,
+      "step": 252
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.926506674925539e-05,
+      "loss": 0.2535,
+      "step": 253
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9257863758829038e-05,
+      "loss": 0.2234,
+      "step": 254
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9250627002482137e-05,
+      "loss": 0.2296,
+      "step": 255
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.92433565066091e-05,
+      "loss": 0.2456,
+      "step": 256
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9236052297727377e-05,
+      "loss": 0.2181,
+      "step": 257
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.922871440247738e-05,
+      "loss": 0.242,
+      "step": 258
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9221342847622394e-05,
+      "loss": 0.2445,
+      "step": 259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.921393766004846e-05,
+      "loss": 0.1934,
+      "step": 260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.920649886676429e-05,
+      "loss": 0.2257,
+      "step": 261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9199026494901163e-05,
+      "loss": 0.236,
+      "step": 262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9191520571712835e-05,
+      "loss": 0.2088,
+      "step": 263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.918398112457542e-05,
+      "loss": 0.2225,
+      "step": 264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.917640818098732e-05,
+      "loss": 0.2413,
+      "step": 265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.916880176856909e-05,
+      "loss": 0.2263,
+      "step": 266
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.916116191506337e-05,
+      "loss": 0.2007,
+      "step": 267
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.915348864833476e-05,
+      "loss": 0.2203,
+      "step": 268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9145781996369725e-05,
+      "loss": 0.2045,
+      "step": 269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9138041987276498e-05,
+      "loss": 0.2151,
+      "step": 270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9130268649284982e-05,
+      "loss": 0.2558,
+      "step": 271
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.912246201074663e-05,
+      "loss": 0.2253,
+      "step": 272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.911462210013435e-05,
+      "loss": 0.2194,
+      "step": 273
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9106748946042407e-05,
+      "loss": 0.2107,
+      "step": 274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9098842577186315e-05,
+      "loss": 0.2248,
+      "step": 275
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.909090302240273e-05,
+      "loss": 0.2365,
+      "step": 276
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.908293031064934e-05,
+      "loss": 0.2205,
+      "step": 277
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.907492447100477e-05,
+      "loss": 0.245,
+      "step": 278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.906688553266848e-05,
+      "loss": 0.2432,
+      "step": 279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9058813524960636e-05,
+      "loss": 0.2309,
+      "step": 280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.905070847732202e-05,
+      "loss": 0.226,
+      "step": 281
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9042570419313927e-05,
+      "loss": 0.2369,
+      "step": 282
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.903439938061804e-05,
+      "loss": 0.242,
+      "step": 283
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.902619539103634e-05,
+      "loss": 0.2395,
+      "step": 284
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9017958480490978e-05,
+      "loss": 0.2257,
+      "step": 285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.226,
+      "step": 286
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.900138601679817e-05,
+      "loss": 0.2349,
+      "step": 287
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8993050524094957e-05,
+      "loss": 0.2352,
+      "step": 288
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8984682231316335e-05,
+      "loss": 0.2537,
+      "step": 289
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8976281168983718e-05,
+      "loss": 0.2213,
+      "step": 290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.896784736773805e-05,
+      "loss": 0.2244,
+      "step": 291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8959380858339664e-05,
+      "loss": 0.23,
+      "step": 292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8950881671668206e-05,
+      "loss": 0.2525,
+      "step": 293
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8942349838722487e-05,
+      "loss": 0.2213,
+      "step": 294
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8933785390620408e-05,
+      "loss": 0.2371,
+      "step": 295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8925188358598815e-05,
+      "loss": 0.2246,
+      "step": 296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.891655877401339e-05,
+      "loss": 0.2095,
+      "step": 297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8907896668338558e-05,
+      "loss": 0.2103,
+      "step": 298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8899202073167345e-05,
+      "loss": 0.2301,
+      "step": 299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.889047502021128e-05,
+      "loss": 0.2336,
+      "step": 300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8881715541300278e-05,
+      "loss": 0.2233,
+      "step": 301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8872923668382507e-05,
+      "loss": 0.2094,
+      "step": 302
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8864099433524302e-05,
+      "loss": 0.2458,
+      "step": 303
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.885524286891002e-05,
+      "loss": 0.2351,
+      "step": 304
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8846354006841933e-05,
+      "loss": 0.2417,
+      "step": 305
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8837432879740113e-05,
+      "loss": 0.2541,
+      "step": 306
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.8828479520142316e-05,
+      "loss": 0.2499,
+      "step": 307
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.8819493960703853e-05,
+      "loss": 0.2506,
+      "step": 308
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.881047623419747e-05,
+      "loss": 0.2256,
+      "step": 309
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.880142637351325e-05,
+      "loss": 0.2433,
+      "step": 310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.879234441165847e-05,
+      "loss": 0.2094,
+      "step": 311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.8783230381757484e-05,
+      "loss": 0.2043,
+      "step": 312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.877408431705162e-05,
+      "loss": 0.2177,
+      "step": 313
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.8764906250899026e-05,
+      "loss": 0.2452,
+      "step": 314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.875569621677459e-05,
+      "loss": 0.215,
+      "step": 315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8746454248269777e-05,
+      "loss": 0.2394,
+      "step": 316
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8737180379092536e-05,
+      "loss": 0.212,
+      "step": 317
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8727874643067166e-05,
+      "loss": 0.2091,
+      "step": 318
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.871853707413419e-05,
+      "loss": 0.224,
+      "step": 319
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8709167706350234e-05,
+      "loss": 0.2242,
+      "step": 320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8699766573887902e-05,
+      "loss": 0.2492,
+      "step": 321
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8690333711035655e-05,
+      "loss": 0.2367,
+      "step": 322
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.868086915219768e-05,
+      "loss": 0.2216,
+      "step": 323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.8671372931893775e-05,
+      "loss": 0.2063,
+      "step": 324
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.866184508475921e-05,
+      "loss": 0.2057,
+      "step": 325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.8652285645544602e-05,
+      "loss": 0.2325,
+      "step": 326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.864269464911581e-05,
+      "loss": 0.2343,
+      "step": 327
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.8633072130453766e-05,
+      "loss": 0.2165,
+      "step": 328
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.8623418124654394e-05,
+      "loss": 0.2566,
+      "step": 329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8613732666928445e-05,
+      "loss": 0.2134,
+      "step": 330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8604015792601395e-05,
+      "loss": 0.2218,
+      "step": 331
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8594267537113298e-05,
+      "loss": 0.2182,
+      "step": 332
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 0.2302,
+      "step": 333
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.857467702498633e-05,
+      "loss": 0.2286,
+      "step": 334
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.856483483979933e-05,
+      "loss": 0.2336,
+      "step": 335
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.855496141635476e-05,
+      "loss": 0.2323,
+      "step": 336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8545056790663653e-05,
+      "loss": 0.202,
+      "step": 337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.853512099885085e-05,
+      "loss": 0.2361,
+      "step": 338
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8525154077154846e-05,
+      "loss": 0.2552,
+      "step": 339
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.85151560619277e-05,
+      "loss": 0.208,
+      "step": 340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.850512698963485e-05,
+      "loss": 0.2265,
+      "step": 341
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8495066896855038e-05,
+      "loss": 0.2333,
+      "step": 342
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8484975820280115e-05,
+      "loss": 0.2265,
+      "step": 343
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.847485379671496e-05,
+      "loss": 0.2165,
+      "step": 344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8464700863077313e-05,
+      "loss": 0.2263,
+      "step": 345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8454517056397663e-05,
+      "loss": 0.2532,
+      "step": 346
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8444302413819082e-05,
+      "loss": 0.2121,
+      "step": 347
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8434056972597135e-05,
+      "loss": 0.2077,
+      "step": 348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.842378077009969e-05,
+      "loss": 0.2428,
+      "step": 349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8413473843806828e-05,
+      "loss": 0.2311,
+      "step": 350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8403136231310686e-05,
+      "loss": 0.2524,
+      "step": 351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8392767970315314e-05,
+      "loss": 0.2196,
+      "step": 352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.838236909863655e-05,
+      "loss": 0.2101,
+      "step": 353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.837193965420188e-05,
+      "loss": 0.231,
+      "step": 354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8361479675050296e-05,
+      "loss": 0.2714,
+      "step": 355
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8350989199332156e-05,
+      "loss": 0.2244,
+      "step": 356
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8340468265309043e-05,
+      "loss": 0.2206,
+      "step": 357
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8329916911353637e-05,
+      "loss": 0.2426,
+      "step": 358
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.831933517594957e-05,
+      "loss": 0.2282,
+      "step": 359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8308723097691277e-05,
+      "loss": 0.2375,
+      "step": 360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.829808071528386e-05,
+      "loss": 0.1973,
+      "step": 361
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.828740806754295e-05,
+      "loss": 0.2196,
+      "step": 362
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8276705193394574e-05,
+      "loss": 0.2295,
+      "step": 363
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.826597213187499e-05,
+      "loss": 0.2284,
+      "step": 364
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.825520892213056e-05,
+      "loss": 0.2778,
+      "step": 365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8244415603417603e-05,
+      "loss": 0.2451,
+      "step": 366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8233592215102268e-05,
+      "loss": 0.2392,
+      "step": 367
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8222738796660353e-05,
+      "loss": 0.2538,
+      "step": 368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8211855387677203e-05,
+      "loss": 0.2121,
+      "step": 369
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.820094202784754e-05,
+      "loss": 0.2303,
+      "step": 370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8189998756975318e-05,
+      "loss": 0.2321,
+      "step": 371
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.81790256149736e-05,
+      "loss": 0.2229,
+      "step": 372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.816802264186438e-05,
+      "loss": 0.2069,
+      "step": 373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.815698987777846e-05,
+      "loss": 0.2379,
+      "step": 374
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8145927362955308e-05,
+      "loss": 0.2345,
+      "step": 375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.813483513774289e-05,
+      "loss": 0.2444,
+      "step": 376
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8123713242597534e-05,
+      "loss": 0.2299,
+      "step": 377
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8112561718083785e-05,
+      "loss": 0.2404,
+      "step": 378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8101380604874255e-05,
+      "loss": 0.2382,
+      "step": 379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.2186,
+      "step": 380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.807892977559774e-05,
+      "loss": 0.2211,
+      "step": 381
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.806766014141497e-05,
+      "loss": 0.2102,
+      "step": 382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.805636108230455e-05,
+      "loss": 0.2324,
+      "step": 383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8045032639477195e-05,
+      "loss": 0.2245,
+      "step": 384
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8033674854250778e-05,
+      "loss": 0.2355,
+      "step": 385
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.80222877680502e-05,
+      "loss": 0.2272,
+      "step": 386
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8010871422407238e-05,
+      "loss": 0.2094,
+      "step": 387
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.799942585896036e-05,
+      "loss": 0.2633,
+      "step": 388
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.798795111945463e-05,
+      "loss": 0.2334,
+      "step": 389
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.797644724574151e-05,
+      "loss": 0.219,
+      "step": 390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7964914279778716e-05,
+      "loss": 0.198,
+      "step": 391
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.7953352263630082e-05,
+      "loss": 0.2126,
+      "step": 392
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.794176123946539e-05,
+      "loss": 0.2207,
+      "step": 393
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.7930141249560235e-05,
+      "loss": 0.2157,
+      "step": 394
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.791849233629583e-05,
+      "loss": 0.2041,
+      "step": 395
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7906814542158913e-05,
+      "loss": 0.2261,
+      "step": 396
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.789510790974153e-05,
+      "loss": 0.237,
+      "step": 397
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7883372481740928e-05,
+      "loss": 0.2284,
+      "step": 398
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7871608300959367e-05,
+      "loss": 0.2277,
+      "step": 399
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.785981541030398e-05,
+      "loss": 0.2555,
+      "step": 400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7847993852786612e-05,
+      "loss": 0.2609,
+      "step": 401
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7836143671523666e-05,
+      "loss": 0.2217,
+      "step": 402
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7824264909735943e-05,
+      "loss": 0.2311,
+      "step": 403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7812357610748487e-05,
+      "loss": 0.2513,
+      "step": 404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7800421817990414e-05,
+      "loss": 0.2524,
+      "step": 405
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7788457574994777e-05,
+      "loss": 0.2391,
+      "step": 406
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7776464925398392e-05,
+      "loss": 0.2117,
+      "step": 407
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7764443912941675e-05,
+      "loss": 0.2212,
+      "step": 408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7752394581468494e-05,
+      "loss": 0.1823,
+      "step": 409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7740316974926005e-05,
+      "loss": 0.2349,
+      "step": 410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.772821113736449e-05,
+      "loss": 0.2244,
+      "step": 411
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7716077112937195e-05,
+      "loss": 0.2297,
+      "step": 412
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.770391494590017e-05,
+      "loss": 0.2274,
+      "step": 413
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.769172468061212e-05,
+      "loss": 0.2409,
+      "step": 414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7679506361534216e-05,
+      "loss": 0.2377,
+      "step": 415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7667260033229953e-05,
+      "loss": 0.2362,
+      "step": 416
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7654985740364992e-05,
+      "loss": 0.2181,
+      "step": 417
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.764268352770698e-05,
+      "loss": 0.228,
+      "step": 418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7630353440125402e-05,
+      "loss": 0.2063,
+      "step": 419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7617995522591396e-05,
+      "loss": 0.2322,
+      "step": 420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.760560982017762e-05,
+      "loss": 0.2367,
+      "step": 421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.759319637805806e-05,
+      "loss": 0.2334,
+      "step": 422
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.758075524150788e-05,
+      "loss": 0.2381,
+      "step": 423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7568286455903257e-05,
+      "loss": 0.2215,
+      "step": 424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7555790066721207e-05,
+      "loss": 0.1579,
+      "step": 425
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.7543266119539424e-05,
+      "loss": 0.1218,
+      "step": 426
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.1333,
+      "step": 427
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.751813573398982e-05,
+      "loss": 0.1151,
+      "step": 428
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.750552938727928e-05,
+      "loss": 0.1205,
+      "step": 429
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7492895665883228e-05,
+      "loss": 0.1436,
+      "step": 430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7480234615880247e-05,
+      "loss": 0.1176,
+      "step": 431
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7467546283448596e-05,
+      "loss": 0.125,
+      "step": 432
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7454830714866036e-05,
+      "loss": 0.114,
+      "step": 433
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7442087956509667e-05,
+      "loss": 0.1135,
+      "step": 434
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7429318054855756e-05,
+      "loss": 0.1215,
+      "step": 435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7416521056479577e-05,
+      "loss": 0.1286,
+      "step": 436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.740369700805522e-05,
+      "loss": 0.1095,
+      "step": 437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7390845956355447e-05,
+      "loss": 0.1156,
+      "step": 438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.73779679482515e-05,
+      "loss": 0.1184,
+      "step": 439
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7365063030712933e-05,
+      "loss": 0.1175,
+      "step": 440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7352131250807466e-05,
+      "loss": 0.1267,
+      "step": 441
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7339172655700773e-05,
+      "loss": 0.1257,
+      "step": 442
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7326187292656332e-05,
+      "loss": 0.115,
+      "step": 443
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.731317520903527e-05,
+      "loss": 0.127,
+      "step": 444
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.730013645229615e-05,
+      "loss": 0.1437,
+      "step": 445
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.728707106999482e-05,
+      "loss": 0.1218,
+      "step": 446
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.727397910978426e-05,
+      "loss": 0.1252,
+      "step": 447
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.726086061941436e-05,
+      "loss": 0.1235,
+      "step": 448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.7247715646731792e-05,
+      "loss": 0.1114,
+      "step": 449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.7234544239679807e-05,
+      "loss": 0.1208,
+      "step": 450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.722134644629807e-05,
+      "loss": 0.1335,
+      "step": 451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7208122314722487e-05,
+      "loss": 0.1217,
+      "step": 452
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.719487189318502e-05,
+      "loss": 0.1178,
+      "step": 453
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.7181595230013526e-05,
+      "loss": 0.1468,
+      "step": 454
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.716829237363157e-05,
+      "loss": 0.1258,
+      "step": 455
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7154963372558246e-05,
+      "loss": 0.1115,
+      "step": 456
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.714160827540801e-05,
+      "loss": 0.1228,
+      "step": 457
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7128227130890494e-05,
+      "loss": 0.1127,
+      "step": 458
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.711481998781033e-05,
+      "loss": 0.1278,
+      "step": 459
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7101386895066987e-05,
+      "loss": 0.1273,
+      "step": 460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7087927901654558e-05,
+      "loss": 0.1365,
+      "step": 461
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.707444305666162e-05,
+      "loss": 0.1257,
+      "step": 462
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7060932409271028e-05,
+      "loss": 0.1296,
+      "step": 463
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7047396008759755e-05,
+      "loss": 0.1162,
+      "step": 464
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.703383390449869e-05,
+      "loss": 0.126,
+      "step": 465
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.702024614595248e-05,
+      "loss": 0.1207,
+      "step": 466
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7006632782679336e-05,
+      "loss": 0.1238,
+      "step": 467
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.699299386433086e-05,
+      "loss": 0.1122,
+      "step": 468
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.6979329440651857e-05,
+      "loss": 0.1346,
+      "step": 469
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6965639561480152e-05,
+      "loss": 0.1142,
+      "step": 470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6951924276746425e-05,
+      "loss": 0.1245,
+      "step": 471
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.6938183636474007e-05,
+      "loss": 0.1257,
+      "step": 472
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.6924417690778713e-05,
+      "loss": 0.1521,
+      "step": 473
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.1147,
+      "step": 474
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6896810084044037e-05,
+      "loss": 0.1339,
+      "step": 475
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.688296852369703e-05,
+      "loss": 0.1199,
+      "step": 476
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.686910185931152e-05,
+      "loss": 0.0992,
+      "step": 477
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.6855210141462964e-05,
+      "loss": 0.1223,
+      "step": 478
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.6841293420818198e-05,
+      "loss": 0.1216,
+      "step": 479
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.6827351748135244e-05,
+      "loss": 0.1216,
+      "step": 480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6813385174263137e-05,
+      "loss": 0.1201,
+      "step": 481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6799393750141727e-05,
+      "loss": 0.1251,
+      "step": 482
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.678537752680151e-05,
+      "loss": 0.1417,
+      "step": 483
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6771336555363418e-05,
+      "loss": 0.1214,
+      "step": 484
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6757270887038653e-05,
+      "loss": 0.1355,
+      "step": 485
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6743180573128494e-05,
+      "loss": 0.1295,
+      "step": 486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.672906566502411e-05,
+      "loss": 0.131,
+      "step": 487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.6714926214206365e-05,
+      "loss": 0.1107,
+      "step": 488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6700762272245646e-05,
+      "loss": 0.1189,
+      "step": 489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6686573890801654e-05,
+      "loss": 0.1228,
+      "step": 490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6672361121623238e-05,
+      "loss": 0.1209,
+      "step": 491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.66581240165482e-05,
+      "loss": 0.1277,
+      "step": 492
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.664386262750308e-05,
+      "loss": 0.132,
+      "step": 493
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.6629577006503012e-05,
+      "loss": 0.1363,
+      "step": 494
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.661526720565149e-05,
+      "loss": 0.1253,
+      "step": 495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.660093327714022e-05,
+      "loss": 0.1279,
+      "step": 496
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.6586575273248895e-05,
+      "loss": 0.1075,
+      "step": 497
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.6572193246345012e-05,
+      "loss": 0.1318,
+      "step": 498
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.6557787248883698e-05,
+      "loss": 0.1435,
+      "step": 499
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.65433573334075e-05,
+      "loss": 0.1263,
+      "step": 500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.6528903552546207e-05,
+      "loss": 0.1351,
+      "step": 501
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.6514425959016643e-05,
+      "loss": 0.1313,
+      "step": 502
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.6499924605622493e-05,
+      "loss": 0.1212,
+      "step": 503
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.648539954525409e-05,
+      "loss": 0.1343,
+      "step": 504
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6470850830888247e-05,
+      "loss": 0.1066,
+      "step": 505
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6456278515588023e-05,
+      "loss": 0.1404,
+      "step": 506
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6441682652502594e-05,
+      "loss": 0.1141,
+      "step": 507
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6427063294866988e-05,
+      "loss": 0.1236,
+      "step": 508
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.641242049600194e-05,
+      "loss": 0.1293,
+      "step": 509
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6397754309313676e-05,
+      "loss": 0.1239,
+      "step": 510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.638306478829373e-05,
+      "loss": 0.1335,
+      "step": 511
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6368351986518735e-05,
+      "loss": 0.1141,
+      "step": 512
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.635361595765024e-05,
+      "loss": 0.118,
+      "step": 513
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6338856755434506e-05,
+      "loss": 0.1274,
+      "step": 514
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6324074433702317e-05,
+      "loss": 0.1208,
+      "step": 515
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6309269046368777e-05,
+      "loss": 0.1212,
+      "step": 516
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.629444064743312e-05,
+      "loss": 0.1293,
+      "step": 517
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6279589290978503e-05,
+      "loss": 0.136,
+      "step": 518
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6264715031171827e-05,
+      "loss": 0.135,
+      "step": 519
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6249817922263518e-05,
+      "loss": 0.1243,
+      "step": 520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.1328,
+      "step": 521
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.621995537456019e-05,
+      "loss": 0.1218,
+      "step": 522
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6204990044681924e-05,
+      "loss": 0.1124,
+      "step": 523
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6190002083535122e-05,
+      "loss": 0.1368,
+      "step": 524
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.617499154578491e-05,
+      "loss": 0.1222,
+      "step": 525
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.615995848617876e-05,
+      "loss": 0.1195,
+      "step": 526
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6144902959546286e-05,
+      "loss": 0.1324,
+      "step": 527
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6129825020799048e-05,
+      "loss": 0.1388,
+      "step": 528
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.611472472493034e-05,
+      "loss": 0.1488,
+      "step": 529
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.609960212701501e-05,
+      "loss": 0.1138,
+      "step": 530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6084457282209244e-05,
+      "loss": 0.1286,
+      "step": 531
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6069290245750365e-05,
+      "loss": 0.1502,
+      "step": 532
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.605410107295663e-05,
+      "loss": 0.1035,
+      "step": 533
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6038889819227047e-05,
+      "loss": 0.1446,
+      "step": 534
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.602365654004115e-05,
+      "loss": 0.1302,
+      "step": 535
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6008401290958806e-05,
+      "loss": 0.139,
+      "step": 536
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5993124127620018e-05,
+      "loss": 0.1267,
+      "step": 537
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5977825105744708e-05,
+      "loss": 0.1303,
+      "step": 538
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5962504281132528e-05,
+      "loss": 0.1332,
+      "step": 539
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5947161709662647e-05,
+      "loss": 0.1207,
+      "step": 540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5931797447293553e-05,
+      "loss": 0.1212,
+      "step": 541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5916411550062846e-05,
+      "loss": 0.1236,
+      "step": 542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.590100407408703e-05,
+      "loss": 0.1413,
+      "step": 543
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5885575075561327e-05,
+      "loss": 0.1133,
+      "step": 544
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.587012461075943e-05,
+      "loss": 0.1149,
+      "step": 545
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5854652736033353e-05,
+      "loss": 0.1305,
+      "step": 546
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.583915950781319e-05,
+      "loss": 0.1227,
+      "step": 547
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5823644982606905e-05,
+      "loss": 0.1362,
+      "step": 548
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.580810921700015e-05,
+      "loss": 0.1266,
+      "step": 549
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5792552267656046e-05,
+      "loss": 0.1247,
+      "step": 550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.577697419131497e-05,
+      "loss": 0.1186,
+      "step": 551
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5761375044794354e-05,
+      "loss": 0.1315,
+      "step": 552
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5745754884988495e-05,
+      "loss": 0.132,
+      "step": 553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5730113768868312e-05,
+      "loss": 0.1251,
+      "step": 554
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.571445175348117e-05,
+      "loss": 0.1328,
+      "step": 555
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5698768895950644e-05,
+      "loss": 0.1255,
+      "step": 556
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5683065253476347e-05,
+      "loss": 0.1341,
+      "step": 557
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.566734088333368e-05,
+      "loss": 0.1118,
+      "step": 558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5651595842873657e-05,
+      "loss": 0.131,
+      "step": 559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.563583018952268e-05,
+      "loss": 0.1131,
+      "step": 560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5620043980782327e-05,
+      "loss": 0.1278,
+      "step": 561
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.560423727422915e-05,
+      "loss": 0.1348,
+      "step": 562
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.558841012751446e-05,
+      "loss": 0.1213,
+      "step": 563
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.557256259836412e-05,
+      "loss": 0.1206,
+      "step": 564
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5556694744578336e-05,
+      "loss": 0.1357,
+      "step": 565
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.554080662403144e-05,
+      "loss": 0.1115,
+      "step": 566
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5524898294671685e-05,
+      "loss": 0.134,
+      "step": 567
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.1325,
+      "step": 568
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.549302124167492e-05,
+      "loss": 0.1082,
+      "step": 569
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.54770526343021e-05,
+      "loss": 0.1461,
+      "step": 570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.546106405064438e-05,
+      "loss": 0.1235,
+      "step": 571
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5445055549016425e-05,
+      "loss": 0.1178,
+      "step": 572
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5429027187805542e-05,
+      "loss": 0.1215,
+      "step": 573
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.541297902547149e-05,
+      "loss": 0.1172,
+      "step": 574
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5396911120546224e-05,
+      "loss": 0.1328,
+      "step": 575
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5380823531633727e-05,
+      "loss": 0.1421,
+      "step": 576
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5364716317409763e-05,
+      "loss": 0.1512,
+      "step": 577
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.534858953662167e-05,
+      "loss": 0.1286,
+      "step": 578
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.533244324808816e-05,
+      "loss": 0.1197,
+      "step": 579
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5316277510699103e-05,
+      "loss": 0.1093,
+      "step": 580
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5300092383415282e-05,
+      "loss": 0.1318,
+      "step": 581
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5283887925268213e-05,
+      "loss": 0.1392,
+      "step": 582
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5267664195359917e-05,
+      "loss": 0.1413,
+      "step": 583
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5251421252862708e-05,
+      "loss": 0.1231,
+      "step": 584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5235159157018962e-05,
+      "loss": 0.1246,
+      "step": 585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5218877967140921e-05,
+      "loss": 0.1304,
+      "step": 586
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.520257774261047e-05,
+      "loss": 0.1587,
+      "step": 587
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5186258542878907e-05,
+      "loss": 0.1508,
+      "step": 588
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5169920427466749e-05,
+      "loss": 0.1297,
+      "step": 589
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5153563455963501e-05,
+      "loss": 0.1117,
+      "step": 590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5137187688027437e-05,
+      "loss": 0.1396,
+      "step": 591
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5120793183385391e-05,
+      "loss": 0.1271,
+      "step": 592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5104380001832535e-05,
+      "loss": 0.1111,
+      "step": 593
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5087948203232157e-05,
+      "loss": 0.1424,
+      "step": 594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5071497847515449e-05,
+      "loss": 0.1324,
+      "step": 595
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5055028994681284e-05,
+      "loss": 0.1424,
+      "step": 596
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5038541704796004e-05,
+      "loss": 0.1421,
+      "step": 597
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5022036037993193e-05,
+      "loss": 0.1343,
+      "step": 598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.500551205447345e-05,
+      "loss": 0.142,
+      "step": 599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4988969814504199e-05,
+      "loss": 0.1256,
+      "step": 600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4972409378419439e-05,
+      "loss": 0.117,
+      "step": 601
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4955830806619535e-05,
+      "loss": 0.1428,
+      "step": 602
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4939234159571001e-05,
+      "loss": 0.1197,
+      "step": 603
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4922619497806276e-05,
+      "loss": 0.1293,
+      "step": 604
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4905986881923502e-05,
+      "loss": 0.131,
+      "step": 605
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4889336372586305e-05,
+      "loss": 0.129,
+      "step": 606
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4872668030523574e-05,
+      "loss": 0.1396,
+      "step": 607
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4855981916529239e-05,
+      "loss": 0.1282,
+      "step": 608
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4839278091462044e-05,
+      "loss": 0.1122,
+      "step": 609
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4822556616245338e-05,
+      "loss": 0.1286,
+      "step": 610
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4805817551866839e-05,
+      "loss": 0.1182,
+      "step": 611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4789060959378421e-05,
+      "loss": 0.1373,
+      "step": 612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.477228689989589e-05,
+      "loss": 0.1263,
+      "step": 613
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4755495434598745e-05,
+      "loss": 0.1351,
+      "step": 614
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.1125,
+      "step": 615
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4721860531595868e-05,
+      "loss": 0.1414,
+      "step": 616
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4705017216565676e-05,
+      "loss": 0.1414,
+      "step": 617
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4688156741071513e-05,
+      "loss": 0.1162,
+      "step": 618
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4671279166608074e-05,
+      "loss": 0.1312,
+      "step": 619
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4654384554732412e-05,
+      "loss": 0.127,
+      "step": 620
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4637472967063721e-05,
+      "loss": 0.1209,
+      "step": 621
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4620544465283118e-05,
+      "loss": 0.1154,
+      "step": 622
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4603599111133402e-05,
+      "loss": 0.1301,
+      "step": 623
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4586636966418838e-05,
+      "loss": 0.1252,
+      "step": 624
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4569658093004935e-05,
+      "loss": 0.1245,
+      "step": 625
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4552662552818211e-05,
+      "loss": 0.1274,
+      "step": 626
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.453565040784598e-05,
+      "loss": 0.1327,
+      "step": 627
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4518621720136105e-05,
+      "loss": 0.1294,
+      "step": 628
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4501576551796803e-05,
+      "loss": 0.1251,
+      "step": 629
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4484514964996381e-05,
+      "loss": 0.1198,
+      "step": 630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.446743702196304e-05,
+      "loss": 0.1256,
+      "step": 631
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4450342784984632e-05,
+      "loss": 0.1397,
+      "step": 632
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4433232316408446e-05,
+      "loss": 0.1201,
+      "step": 633
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4416105678640962e-05,
+      "loss": 0.1249,
+      "step": 634
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4398962934147638e-05,
+      "loss": 0.1317,
+      "step": 635
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4381804145452672e-05,
+      "loss": 0.1272,
+      "step": 636
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.436462937513879e-05,
+      "loss": 0.0599,
+      "step": 637
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4347438685846997e-05,
+      "loss": 0.0728,
+      "step": 638
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4330232140276365e-05,
+      "loss": 0.062,
+      "step": 639
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4313009801183799e-05,
+      "loss": 0.0591,
+      "step": 640
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4295771731383799e-05,
+      "loss": 0.0542,
+      "step": 641
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4278517993748244e-05,
+      "loss": 0.0605,
+      "step": 642
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4261248651206163e-05,
+      "loss": 0.0566,
+      "step": 643
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4243963766743495e-05,
+      "loss": 0.0598,
+      "step": 644
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4226663403402866e-05,
+      "loss": 0.0552,
+      "step": 645
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4209347624283352e-05,
+      "loss": 0.0662,
+      "step": 646
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4192016492540263e-05,
+      "loss": 0.0597,
+      "step": 647
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4174670071384899e-05,
+      "loss": 0.0577,
+      "step": 648
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4157308424084329e-05,
+      "loss": 0.0596,
+      "step": 649
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4139931613961149e-05,
+      "loss": 0.0728,
+      "step": 650
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4122539704393265e-05,
+      "loss": 0.0641,
+      "step": 651
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4105132758813648e-05,
+      "loss": 0.0514,
+      "step": 652
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.408771084071012e-05,
+      "loss": 0.0558,
+      "step": 653
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4070274013625096e-05,
+      "loss": 0.0606,
+      "step": 654
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4052822341155381e-05,
+      "loss": 0.0579,
+      "step": 655
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4035355886951924e-05,
+      "loss": 0.0509,
+      "step": 656
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.401787471471958e-05,
+      "loss": 0.0655,
+      "step": 657
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4000378888216886e-05,
+      "loss": 0.0544,
+      "step": 658
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3982868471255835e-05,
+      "loss": 0.0544,
+      "step": 659
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3965343527701629e-05,
+      "loss": 0.0621,
+      "step": 660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3947804121472453e-05,
+      "loss": 0.0557,
+      "step": 661
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.0542,
+      "step": 662
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3912682176925439e-05,
+      "loss": 0.0536,
+      "step": 663
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3895099766706786e-05,
+      "loss": 0.0671,
+      "step": 664
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3877503150011062e-05,
+      "loss": 0.0653,
+      "step": 665
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3859892391017867e-05,
+      "loss": 0.0572,
+      "step": 666
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3842267553958373e-05,
+      "loss": 0.0708,
+      "step": 667
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3824628703115104e-05,
+      "loss": 0.0573,
+      "step": 668
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.38069759028217e-05,
+      "loss": 0.0653,
+      "step": 669
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.378930921746267e-05,
+      "loss": 0.0536,
+      "step": 670
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3771628711473173e-05,
+      "loss": 0.0572,
+      "step": 671
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3753934449338773e-05,
+      "loss": 0.0584,
+      "step": 672
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3736226495595202e-05,
+      "loss": 0.0589,
+      "step": 673
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3718504914828135e-05,
+      "loss": 0.0515,
+      "step": 674
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3700769771672944e-05,
+      "loss": 0.0609,
+      "step": 675
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.368302113081447e-05,
+      "loss": 0.0514,
+      "step": 676
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.366525905698679e-05,
+      "loss": 0.0575,
+      "step": 677
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3647483614972953e-05,
+      "loss": 0.0546,
+      "step": 678
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3629694869604791e-05,
+      "loss": 0.0508,
+      "step": 679
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3611892885762643e-05,
+      "loss": 0.0623,
+      "step": 680
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3594077728375129e-05,
+      "loss": 0.0537,
+      "step": 681
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3576249462418931e-05,
+      "loss": 0.0624,
+      "step": 682
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3558408152918528e-05,
+      "loss": 0.0497,
+      "step": 683
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3540553864945975e-05,
+      "loss": 0.0554,
+      "step": 684
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3522686663620668e-05,
+      "loss": 0.0615,
+      "step": 685
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3504806614109098e-05,
+      "loss": 0.048,
+      "step": 686
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.348691378162462e-05,
+      "loss": 0.0627,
+      "step": 687
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3469008231427207e-05,
+      "loss": 0.0587,
+      "step": 688
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3451090028823217e-05,
+      "loss": 0.0588,
+      "step": 689
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3433159239165162e-05,
+      "loss": 0.0563,
+      "step": 690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.341521592785145e-05,
+      "loss": 0.0591,
+      "step": 691
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3397260160326173e-05,
+      "loss": 0.0594,
+      "step": 692
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3379292002078845e-05,
+      "loss": 0.0528,
+      "step": 693
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3361311518644173e-05,
+      "loss": 0.0612,
+      "step": 694
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.334331877560182e-05,
+      "loss": 0.0565,
+      "step": 695
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.332531383857616e-05,
+      "loss": 0.0567,
+      "step": 696
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.330729677323605e-05,
+      "loss": 0.0554,
+      "step": 697
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.328926764529457e-05,
+      "loss": 0.0622,
+      "step": 698
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3271226520508802e-05,
+      "loss": 0.0533,
+      "step": 699
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3253173464679588e-05,
+      "loss": 0.0568,
+      "step": 700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3235108543651272e-05,
+      "loss": 0.0509,
+      "step": 701
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3217031823311488e-05,
+      "loss": 0.0633,
+      "step": 702
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3198943369590905e-05,
+      "loss": 0.0615,
+      "step": 703
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3180843248462973e-05,
+      "loss": 0.0569,
+      "step": 704
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3162731525943706e-05,
+      "loss": 0.0591,
+      "step": 705
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3144608268091435e-05,
+      "loss": 0.0524,
+      "step": 706
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3126473541006552e-05,
+      "loss": 0.0605,
+      "step": 707
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3108327410831291e-05,
+      "loss": 0.058,
+      "step": 708
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.0597,
+      "step": 709
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3072001205986269e-05,
+      "loss": 0.0587,
+      "step": 710
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3053821263807947e-05,
+      "loss": 0.0564,
+      "step": 711
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3035630183521653e-05,
+      "loss": 0.0532,
+      "step": 712
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3017428031475155e-05,
+      "loss": 0.0566,
+      "step": 713
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2999214874056596e-05,
+      "loss": 0.0535,
+      "step": 714
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2980990777694262e-05,
+      "loss": 0.0535,
+      "step": 715
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2962755808856341e-05,
+      "loss": 0.0585,
+      "step": 716
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2944510034050669e-05,
+      "loss": 0.0622,
+      "step": 717
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2926253519824496e-05,
+      "loss": 0.059,
+      "step": 718
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2907986332764245e-05,
+      "loss": 0.0567,
+      "step": 719
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.288970853949526e-05,
+      "loss": 0.0562,
+      "step": 720
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2871420206681573e-05,
+      "loss": 0.0601,
+      "step": 721
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2853121401025656e-05,
+      "loss": 0.052,
+      "step": 722
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.283481218926818e-05,
+      "loss": 0.0558,
+      "step": 723
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2816492638187763e-05,
+      "loss": 0.062,
+      "step": 724
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2798162814600743e-05,
+      "loss": 0.0559,
+      "step": 725
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2779822785360913e-05,
+      "loss": 0.0628,
+      "step": 726
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2761472617359298e-05,
+      "loss": 0.057,
+      "step": 727
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.27431123775239e-05,
+      "loss": 0.0532,
+      "step": 728
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2724742132819454e-05,
+      "loss": 0.0797,
+      "step": 729
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.270636195024719e-05,
+      "loss": 0.0568,
+      "step": 730
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2687971896844575e-05,
+      "loss": 0.053,
+      "step": 731
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.266957203968508e-05,
+      "loss": 0.0599,
+      "step": 732
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2651162445877945e-05,
+      "loss": 0.0689,
+      "step": 733
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2632743182567905e-05,
+      "loss": 0.0537,
+      "step": 734
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2614314316934974e-05,
+      "loss": 0.0643,
+      "step": 735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2595875916194188e-05,
+      "loss": 0.0589,
+      "step": 736
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2577428047595343e-05,
+      "loss": 0.0619,
+      "step": 737
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2558970778422791e-05,
+      "loss": 0.0549,
+      "step": 738
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2540504175995161e-05,
+      "loss": 0.0542,
+      "step": 739
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2522028307665117e-05,
+      "loss": 0.058,
+      "step": 740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2503543240819127e-05,
+      "loss": 0.058,
+      "step": 741
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.24850490428772e-05,
+      "loss": 0.0509,
+      "step": 742
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2466545781292653e-05,
+      "loss": 0.0677,
+      "step": 743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2448033523551866e-05,
+      "loss": 0.0714,
+      "step": 744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2429512337174017e-05,
+      "loss": 0.0656,
+      "step": 745
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2410982289710865e-05,
+      "loss": 0.0504,
+      "step": 746
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2392443448746474e-05,
+      "loss": 0.0667,
+      "step": 747
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2373895881896988e-05,
+      "loss": 0.0635,
+      "step": 748
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2355339656810372e-05,
+      "loss": 0.0736,
+      "step": 749
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2336774841166181e-05,
+      "loss": 0.0659,
+      "step": 750
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2318201502675285e-05,
+      "loss": 0.0525,
+      "step": 751
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2299619709079651e-05,
+      "loss": 0.0612,
+      "step": 752
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2281029528152079e-05,
+      "loss": 0.0531,
+      "step": 753
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2262431027695964e-05,
+      "loss": 0.0518,
+      "step": 754
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2243824275545042e-05,
+      "loss": 0.0554,
+      "step": 755
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.0637,
+      "step": 756
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2206586287643955e-05,
+      "loss": 0.0586,
+      "step": 757
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2187955187710752e-05,
+      "loss": 0.0572,
+      "step": 758
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2169316107716171e-05,
+      "loss": 0.0697,
+      "step": 759
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2150669115641955e-05,
+      "loss": 0.0526,
+      "step": 760
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2132014279498702e-05,
+      "loss": 0.06,
+      "step": 761
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2113351667325622e-05,
+      "loss": 0.0555,
+      "step": 762
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2094681347190278e-05,
+      "loss": 0.0542,
+      "step": 763
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2076003387188353e-05,
+      "loss": 0.0598,
+      "step": 764
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2057317855443395e-05,
+      "loss": 0.0677,
+      "step": 765
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2038624820106572e-05,
+      "loss": 0.0623,
+      "step": 766
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2019924349356407e-05,
+      "loss": 0.0622,
+      "step": 767
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2001216511398554e-05,
+      "loss": 0.0552,
+      "step": 768
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1982501374465528e-05,
+      "loss": 0.0657,
+      "step": 769
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1963779006816475e-05,
+      "loss": 0.0531,
+      "step": 770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1945049476736905e-05,
+      "loss": 0.0759,
+      "step": 771
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1926312852538456e-05,
+      "loss": 0.0654,
+      "step": 772
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1907569202558637e-05,
+      "loss": 0.0672,
+      "step": 773
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1888818595160585e-05,
+      "loss": 0.0563,
+      "step": 774
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.187006109873281e-05,
+      "loss": 0.0549,
+      "step": 775
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1851296781688952e-05,
+      "loss": 0.0625,
+      "step": 776
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1832525712467522e-05,
+      "loss": 0.0553,
+      "step": 777
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1813747959531662e-05,
+      "loss": 0.057,
+      "step": 778
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1794963591368893e-05,
+      "loss": 0.0663,
+      "step": 779
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1776172676490855e-05,
+      "loss": 0.0625,
+      "step": 780
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1757375283433077e-05,
+      "loss": 0.0591,
+      "step": 781
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1738571480754705e-05,
+      "loss": 0.0652,
+      "step": 782
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1719761337038273e-05,
+      "loss": 0.0608,
+      "step": 783
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1700944920889437e-05,
+      "loss": 0.0539,
+      "step": 784
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1682122300936725e-05,
+      "loss": 0.0615,
+      "step": 785
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1663293545831302e-05,
+      "loss": 0.0622,
+      "step": 786
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1644458724246704e-05,
+      "loss": 0.0646,
+      "step": 787
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.162561790487859e-05,
+      "loss": 0.0627,
+      "step": 788
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1606771156444508e-05,
+      "loss": 0.063,
+      "step": 789
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1587918547683612e-05,
+      "loss": 0.0647,
+      "step": 790
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1569060147356441e-05,
+      "loss": 0.0492,
+      "step": 791
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.155019602424466e-05,
+      "loss": 0.0604,
+      "step": 792
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1531326247150802e-05,
+      "loss": 0.0527,
+      "step": 793
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1512450884898024e-05,
+      "loss": 0.0609,
+      "step": 794
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1493570006329842e-05,
+      "loss": 0.0548,
+      "step": 795
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1474683680309913e-05,
+      "loss": 0.0651,
+      "step": 796
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1455791975721741e-05,
+      "loss": 0.0617,
+      "step": 797
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1436894961468463e-05,
+      "loss": 0.063,
+      "step": 798
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1417992706472576e-05,
+      "loss": 0.0636,
+      "step": 799
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1399085279675688e-05,
+      "loss": 0.061,
+      "step": 800
+    }
+  ],
+  "max_steps": 1696,
+  "num_train_epochs": 8,
+  "total_flos": 1.9738787363527066e+17,
+  "trial_name": null,
+  "trial_params": null
+}