diff --git "a/checkpoint/trainer_state.json" "b/checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint/trainer_state.json"
@@ -0,0 +1,21016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.843216896831844,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4096181965881397e-05,
+      "loss": 1.7364,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4092363931762796e-05,
+      "loss": 1.4598,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4088545897644193e-05,
+      "loss": 1.5923,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.408472786352559e-05,
+      "loss": 1.5136,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4080909829406987e-05,
+      "loss": 1.7184,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4077091795288384e-05,
+      "loss": 1.5414,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.407327376116978e-05,
+      "loss": 1.5266,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4069455727051177e-05,
+      "loss": 1.3991,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4065637692932576e-05,
+      "loss": 1.5951,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4061819658813973e-05,
+      "loss": 1.626,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.405800162469537e-05,
+      "loss": 1.3523,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4054183590576768e-05,
+      "loss": 1.5133,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4050365556458165e-05,
+      "loss": 1.4678,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.404654752233956e-05,
+      "loss": 1.4675,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.404272948822096e-05,
+      "loss": 1.6265,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4038911454102356e-05,
+      "loss": 1.4609,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4035093419983753e-05,
+      "loss": 1.6426,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4031275385865152e-05,
+      "loss": 1.5437,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4027457351746549e-05,
+      "loss": 1.397,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4023639317627945e-05,
+      "loss": 2.0914,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4019821283509342e-05,
+      "loss": 1.7421,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4016003249390741e-05,
+      "loss": 1.4436,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4012185215272136e-05,
+      "loss": 1.2738,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4008367181153533e-05,
+      "loss": 1.6343,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4004549147034932e-05,
+      "loss": 1.3553,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4000731112916329e-05,
+      "loss": 1.4495,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3996913078797725e-05,
+      "loss": 1.6103,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3993095044679124e-05,
+      "loss": 1.2189,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3989277010560521e-05,
+      "loss": 1.3933,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3985458976441918e-05,
+      "loss": 1.427,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3981640942323316e-05,
+      "loss": 1.623,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3977822908204712e-05,
+      "loss": 1.6029,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3974004874086109e-05,
+      "loss": 1.5448,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3970186839967505e-05,
+      "loss": 1.283,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3966368805848904e-05,
+      "loss": 1.5284,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3962550771730301e-05,
+      "loss": 1.4035,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3958732737611698e-05,
+      "loss": 1.3968,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3954914703493096e-05,
+      "loss": 1.4202,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3951096669374493e-05,
+      "loss": 1.8046,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.394727863525589e-05,
+      "loss": 1.6019,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3943460601137287e-05,
+      "loss": 1.7246,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3939642567018684e-05,
+      "loss": 1.3567,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3935824532900081e-05,
+      "loss": 1.6781,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3932006498781478e-05,
+      "loss": 1.2438,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3928188464662877e-05,
+      "loss": 1.7724,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3924370430544273e-05,
+      "loss": 1.3301,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.392055239642567e-05,
+      "loss": 1.1059,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3916734362307069e-05,
+      "loss": 1.6276,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3912916328188466e-05,
+      "loss": 1.4095,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3909098294069863e-05,
+      "loss": 1.3497,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.390528025995126e-05,
+      "loss": 1.4345,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3901462225832657e-05,
+      "loss": 1.6274,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3897644191714053e-05,
+      "loss": 1.4522,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3893826157595452e-05,
+      "loss": 1.2801,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3890008123476849e-05,
+      "loss": 1.2884,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3886190089358246e-05,
+      "loss": 1.259,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3882372055239643e-05,
+      "loss": 1.6678,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3878554021121041e-05,
+      "loss": 1.2951,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3874735987002437e-05,
+      "loss": 1.3352,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3870917952883834e-05,
+      "loss": 1.8426,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3867099918765232e-05,
+      "loss": 1.2748,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3863281884646629e-05,
+      "loss": 1.5492,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3859463850528026e-05,
+      "loss": 1.4666,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3855645816409425e-05,
+      "loss": 1.4496,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3851827782290821e-05,
+      "loss": 1.3837,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3848009748172218e-05,
+      "loss": 1.4447,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3844191714053617e-05,
+      "loss": 1.5778,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3840373679935012e-05,
+      "loss": 1.6288,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3836555645816409e-05,
+      "loss": 1.3052,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3832737611697806e-05,
+      "loss": 1.2759,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3828919577579205e-05,
+      "loss": 1.3967,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3825101543460601e-05,
+      "loss": 1.5585,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3821283509341998e-05,
+      "loss": 1.4811,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3817465475223397e-05,
+      "loss": 1.6148,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3813647441104794e-05,
+      "loss": 1.4459,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.380982940698619e-05,
+      "loss": 1.7072,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3806011372867588e-05,
+      "loss": 1.5611,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3802193338748985e-05,
+      "loss": 1.9849,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3798375304630381e-05,
+      "loss": 1.1936,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.379455727051178e-05,
+      "loss": 1.5539,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3790739236393177e-05,
+      "loss": 1.4975,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3786921202274574e-05,
+      "loss": 1.1984,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.378310316815597e-05,
+      "loss": 1.4066,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.377928513403737e-05,
+      "loss": 1.4101,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3775467099918766e-05,
+      "loss": 1.3219,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3771649065800163e-05,
+      "loss": 1.2763,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.376783103168156e-05,
+      "loss": 1.4774,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3764012997562957e-05,
+      "loss": 1.4323,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3760194963444354e-05,
+      "loss": 1.3116,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3756376929325753e-05,
+      "loss": 1.342,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.375255889520715e-05,
+      "loss": 1.3991,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3748740861088546e-05,
+      "loss": 1.3024,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3744922826969945e-05,
+      "loss": 1.3091,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3741104792851342e-05,
+      "loss": 1.3673,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3737286758732739e-05,
+      "loss": 1.4539,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3733468724614134e-05,
+      "loss": 1.2119,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3729650690495533e-05,
+      "loss": 1.27,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.372583265637693e-05,
+      "loss": 1.6691,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3722014622258326e-05,
+      "loss": 1.5645,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3718196588139725e-05,
+      "loss": 1.6277,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3714378554021122e-05,
+      "loss": 1.2884,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3710560519902519e-05,
+      "loss": 1.3435,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3706742485783917e-05,
+      "loss": 1.3577,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3702924451665314e-05,
+      "loss": 1.5759,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.369910641754671e-05,
+      "loss": 1.34,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3695288383428106e-05,
+      "loss": 1.1277,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3691470349309505e-05,
+      "loss": 1.0814,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3687652315190902e-05,
+      "loss": 1.2223,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3683834281072299e-05,
+      "loss": 1.2427,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3680016246953697e-05,
+      "loss": 1.5602,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3676198212835094e-05,
+      "loss": 1.4551,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3672380178716491e-05,
+      "loss": 1.3893,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3668562144597888e-05,
+      "loss": 1.4563,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3664744110479285e-05,
+      "loss": 1.5714,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3660926076360682e-05,
+      "loss": 1.5505,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.365710804224208e-05,
+      "loss": 1.2814,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3653290008123477e-05,
+      "loss": 1.7689,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3649471974004874e-05,
+      "loss": 1.2512,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3645653939886271e-05,
+      "loss": 1.381,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.364183590576767e-05,
+      "loss": 1.4703,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3638017871649067e-05,
+      "loss": 1.3237,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3634199837530464e-05,
+      "loss": 1.3075,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.363038180341186e-05,
+      "loss": 1.6228,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3626563769293257e-05,
+      "loss": 1.4757,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3622745735174654e-05,
+      "loss": 1.0566,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3618927701056053e-05,
+      "loss": 1.4489,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.361510966693745e-05,
+      "loss": 1.2202,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3611291632818847e-05,
+      "loss": 1.3654,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3607473598700245e-05,
+      "loss": 1.7373,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3603655564581642e-05,
+      "loss": 1.3072,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.359983753046304e-05,
+      "loss": 1.7585,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3596019496344434e-05,
+      "loss": 1.1749,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3592201462225833e-05,
+      "loss": 1.2398,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.358838342810723e-05,
+      "loss": 1.1244,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3584565393988627e-05,
+      "loss": 1.4318,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3580747359870025e-05,
+      "loss": 1.5061,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3576929325751422e-05,
+      "loss": 1.2021,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.357311129163282e-05,
+      "loss": 1.2036,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3569293257514218e-05,
+      "loss": 1.2787,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3565475223395615e-05,
+      "loss": 1.2214,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.356165718927701e-05,
+      "loss": 1.5319,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3557839155158409e-05,
+      "loss": 1.3172,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3554021121039805e-05,
+      "loss": 1.1769,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3550203086921202e-05,
+      "loss": 1.643,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.35463850528026e-05,
+      "loss": 1.5852,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3542567018683998e-05,
+      "loss": 1.2812,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3538748984565395e-05,
+      "loss": 1.5434,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3534930950446792e-05,
+      "loss": 1.4349,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.353111291632819e-05,
+      "loss": 1.9226,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3527294882209586e-05,
+      "loss": 1.1818,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3523476848090982e-05,
+      "loss": 1.2534,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3519658813972381e-05,
+      "loss": 1.2356,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3515840779853778e-05,
+      "loss": 1.53,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3512022745735175e-05,
+      "loss": 1.4172,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3508204711616573e-05,
+      "loss": 1.2815,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.350438667749797e-05,
+      "loss": 1.2669,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3500568643379367e-05,
+      "loss": 1.371,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3496750609260764e-05,
+      "loss": 1.1214,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3492932575142161e-05,
+      "loss": 1.2766,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3489114541023558e-05,
+      "loss": 1.5737,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3485296506904955e-05,
+      "loss": 1.0817,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3481478472786353e-05,
+      "loss": 1.2881,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.347766043866775e-05,
+      "loss": 1.5085,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3473842404549147e-05,
+      "loss": 1.3642,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3470024370430546e-05,
+      "loss": 1.1971,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3466206336311943e-05,
+      "loss": 1.4039,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.346238830219334e-05,
+      "loss": 1.352,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3458570268074735e-05,
+      "loss": 1.0812,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3454752233956134e-05,
+      "loss": 1.7978,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.345093419983753e-05,
+      "loss": 1.4957,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3447116165718927e-05,
+      "loss": 1.3394,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3443298131600326e-05,
+      "loss": 1.1684,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3439480097481723e-05,
+      "loss": 1.4979,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.343566206336312e-05,
+      "loss": 1.2344,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3431844029244518e-05,
+      "loss": 1.2271,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3428025995125915e-05,
+      "loss": 1.1061,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.342420796100731e-05,
+      "loss": 1.331,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3420389926888709e-05,
+      "loss": 1.1172,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3416571892770106e-05,
+      "loss": 1.2676,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3412753858651503e-05,
+      "loss": 1.2083,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.34089358245329e-05,
+      "loss": 1.1811,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3405117790414298e-05,
+      "loss": 1.2359,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3401299756295695e-05,
+      "loss": 1.3174,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3397481722177092e-05,
+      "loss": 1.3029,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.339366368805849e-05,
+      "loss": 1.4456,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3389845653939886e-05,
+      "loss": 1.6496,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3386027619821283e-05,
+      "loss": 1.1957,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3382209585702681e-05,
+      "loss": 1.2209,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3378391551584078e-05,
+      "loss": 1.4883,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3374573517465475e-05,
+      "loss": 1.1816,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3370755483346874e-05,
+      "loss": 1.4626,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.336693744922827e-05,
+      "loss": 1.3695,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3363119415109668e-05,
+      "loss": 1.3418,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3359301380991065e-05,
+      "loss": 1.4083,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3355483346872462e-05,
+      "loss": 0.9517,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3351665312753858e-05,
+      "loss": 1.3114,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3347847278635255e-05,
+      "loss": 1.3146,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3344029244516654e-05,
+      "loss": 1.5107,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.334021121039805e-05,
+      "loss": 1.3794,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3336393176279448e-05,
+      "loss": 1.4163,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3332575142160846e-05,
+      "loss": 1.4267,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3328757108042243e-05,
+      "loss": 1.5498,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.332493907392364e-05,
+      "loss": 1.2369,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3321121039805037e-05,
+      "loss": 1.026,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3317303005686434e-05,
+      "loss": 1.4414,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3313484971567831e-05,
+      "loss": 1.3372,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3309666937449228e-05,
+      "loss": 1.7031,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3305848903330626e-05,
+      "loss": 1.3151,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3302030869212023e-05,
+      "loss": 1.3416,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.329821283509342e-05,
+      "loss": 1.3431,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3294394800974819e-05,
+      "loss": 1.8496,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3290576766856216e-05,
+      "loss": 1.2586,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3286758732737611e-05,
+      "loss": 1.2754,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.328294069861901e-05,
+      "loss": 1.082,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3279122664500406e-05,
+      "loss": 1.4809,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3275304630381803e-05,
+      "loss": 1.0506,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3271486596263202e-05,
+      "loss": 1.2994,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3267668562144599e-05,
+      "loss": 1.1595,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3263850528025996e-05,
+      "loss": 1.2154,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3260032493907393e-05,
+      "loss": 1.2641,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3256214459788791e-05,
+      "loss": 1.4224,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3252396425670186e-05,
+      "loss": 1.4111,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3248578391551583e-05,
+      "loss": 2.1452,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3244760357432982e-05,
+      "loss": 1.1907,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3240942323314379e-05,
+      "loss": 1.3457,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3237124289195776e-05,
+      "loss": 1.1665,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3233306255077174e-05,
+      "loss": 1.2745,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3229488220958571e-05,
+      "loss": 1.2863,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3225670186839968e-05,
+      "loss": 1.2205,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3221852152721367e-05,
+      "loss": 1.2055,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3218034118602762e-05,
+      "loss": 1.205,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3214216084484159e-05,
+      "loss": 1.047,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3210398050365556e-05,
+      "loss": 1.4517,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3206580016246954e-05,
+      "loss": 1.2015,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3202761982128351e-05,
+      "loss": 1.2577,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3198943948009748e-05,
+      "loss": 1.3627,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3195125913891147e-05,
+      "loss": 1.2385,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3191307879772544e-05,
+      "loss": 1.1771,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.318748984565394e-05,
+      "loss": 1.3341,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3183671811535338e-05,
+      "loss": 1.5681,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3179853777416734e-05,
+      "loss": 1.2609,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3176035743298131e-05,
+      "loss": 1.3366,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3172217709179528e-05,
+      "loss": 1.3648,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3168399675060927e-05,
+      "loss": 1.4856,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3164581640942324e-05,
+      "loss": 1.2875,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.316076360682372e-05,
+      "loss": 1.323,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.315694557270512e-05,
+      "loss": 1.0661,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3153127538586516e-05,
+      "loss": 1.0919,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3149309504467913e-05,
+      "loss": 1.1812,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.314549147034931e-05,
+      "loss": 1.4574,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3141673436230707e-05,
+      "loss": 1.1595,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3137855402112104e-05,
+      "loss": 1.3059,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3134037367993502e-05,
+      "loss": 1.1999,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.31302193338749e-05,
+      "loss": 1.2358,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3126401299756296e-05,
+      "loss": 1.1647,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3122583265637693e-05,
+      "loss": 1.3368,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3118765231519092e-05,
+      "loss": 1.2559,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3114947197400489e-05,
+      "loss": 1.1328,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3111129163281884e-05,
+      "loss": 1.4725,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3107311129163282e-05,
+      "loss": 1.1961,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.310349309504468e-05,
+      "loss": 1.394,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3099675060926076e-05,
+      "loss": 1.3679,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3095857026807475e-05,
+      "loss": 1.3838,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3092038992688872e-05,
+      "loss": 1.4826,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3088220958570269e-05,
+      "loss": 1.2843,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3084402924451667e-05,
+      "loss": 1.2408,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3080584890333062e-05,
+      "loss": 1.4491,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.307676685621446e-05,
+      "loss": 1.2547,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3072948822095856e-05,
+      "loss": 1.3169,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3069130787977255e-05,
+      "loss": 1.557,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3065312753858652e-05,
+      "loss": 1.1323,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3061494719740049e-05,
+      "loss": 1.0853,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3057676685621447e-05,
+      "loss": 1.2355,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3053858651502844e-05,
+      "loss": 1.3763,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3050040617384241e-05,
+      "loss": 1.603,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3046222583265638e-05,
+      "loss": 1.3193,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3042404549147035e-05,
+      "loss": 1.2491,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3038586515028432e-05,
+      "loss": 1.4878,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.303476848090983e-05,
+      "loss": 1.2079,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3030950446791227e-05,
+      "loss": 1.5355,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3027132412672624e-05,
+      "loss": 1.3766,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3023314378554021e-05,
+      "loss": 1.2499,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.301949634443542e-05,
+      "loss": 1.2008,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3015678310316817e-05,
+      "loss": 1.4684,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3011860276198214e-05,
+      "loss": 1.0759,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.300804224207961e-05,
+      "loss": 1.4697,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3004224207961007e-05,
+      "loss": 1.1966,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3000406173842404e-05,
+      "loss": 1.2562,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.2996588139723803e-05,
+      "loss": 1.1989,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.29927701056052e-05,
+      "loss": 1.1619,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2988952071486597e-05,
+      "loss": 1.4786,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2985134037367995e-05,
+      "loss": 1.4511,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2981316003249392e-05,
+      "loss": 1.2313,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2977497969130789e-05,
+      "loss": 1.2406,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2973679935012184e-05,
+      "loss": 1.3208,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2969861900893583e-05,
+      "loss": 1.1513,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.296604386677498e-05,
+      "loss": 1.1818,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2962225832656377e-05,
+      "loss": 1.2264,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2958407798537775e-05,
+      "loss": 1.2968,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2954589764419172e-05,
+      "loss": 1.5004,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.2950771730300569e-05,
+      "loss": 1.2143,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2946953696181968e-05,
+      "loss": 1.2487,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2943135662063365e-05,
+      "loss": 1.1975,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.293931762794476e-05,
+      "loss": 1.2309,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2935499593826157e-05,
+      "loss": 1.4145,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2931681559707555e-05,
+      "loss": 1.398,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2927863525588952e-05,
+      "loss": 1.0731,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2924045491470349e-05,
+      "loss": 1.3804,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2920227457351748e-05,
+      "loss": 1.2296,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2916409423233145e-05,
+      "loss": 1.2221,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2912591389114542e-05,
+      "loss": 1.3532,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.290877335499594e-05,
+      "loss": 1.3318,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.2904955320877335e-05,
+      "loss": 1.2845,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2901137286758732e-05,
+      "loss": 1.1968,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2897319252640131e-05,
+      "loss": 1.3599,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2893501218521528e-05,
+      "loss": 1.5475,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2889683184402925e-05,
+      "loss": 1.342,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2885865150284322e-05,
+      "loss": 1.2008,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.288204711616572e-05,
+      "loss": 1.2056,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2878229082047117e-05,
+      "loss": 1.5116,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2874411047928514e-05,
+      "loss": 1.2637,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2870593013809911e-05,
+      "loss": 1.2431,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2866774979691308e-05,
+      "loss": 1.6368,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2862956945572705e-05,
+      "loss": 1.3055,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2859138911454103e-05,
+      "loss": 1.214,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.28553208773355e-05,
+      "loss": 1.2848,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2851502843216897e-05,
+      "loss": 1.1676,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2847684809098296e-05,
+      "loss": 0.835,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2843866774979693e-05,
+      "loss": 1.3964,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.284004874086109e-05,
+      "loss": 1.1669,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2836230706742485e-05,
+      "loss": 1.2763,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2832412672623883e-05,
+      "loss": 1.2741,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.282859463850528e-05,
+      "loss": 1.9296,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2824776604386677e-05,
+      "loss": 1.4038,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2820958570268076e-05,
+      "loss": 1.5115,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2817140536149473e-05,
+      "loss": 1.1778,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.281332250203087e-05,
+      "loss": 1.0735,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2809504467912268e-05,
+      "loss": 1.0437,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2805686433793665e-05,
+      "loss": 1.3501,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.280186839967506e-05,
+      "loss": 1.3226,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2798050365556459e-05,
+      "loss": 1.3203,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2794232331437856e-05,
+      "loss": 1.2979,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2790414297319253e-05,
+      "loss": 1.3188,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.278659626320065e-05,
+      "loss": 1.4391,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2782778229082048e-05,
+      "loss": 1.2842,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2778960194963445e-05,
+      "loss": 1.0727,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2775142160844842e-05,
+      "loss": 1.2095,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.277132412672624e-05,
+      "loss": 1.1623,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2767506092607636e-05,
+      "loss": 1.3381,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2763688058489033e-05,
+      "loss": 1.6069,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2759870024370431e-05,
+      "loss": 1.1111,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2756051990251828e-05,
+      "loss": 1.3486,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2752233956133225e-05,
+      "loss": 1.4335,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2748415922014624e-05,
+      "loss": 1.2564,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.274459788789602e-05,
+      "loss": 1.3395,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2740779853777418e-05,
+      "loss": 1.0812,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2736961819658814e-05,
+      "loss": 1.1838,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2733143785540211e-05,
+      "loss": 1.1919,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2729325751421608e-05,
+      "loss": 1.1085,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2725507717303005e-05,
+      "loss": 1.5125,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2721689683184404e-05,
+      "loss": 1.4167,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.27178716490658e-05,
+      "loss": 1.3451,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2714053614947198e-05,
+      "loss": 1.1706,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2710235580828596e-05,
+      "loss": 1.4633,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2706417546709993e-05,
+      "loss": 1.3078,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.270259951259139e-05,
+      "loss": 1.2378,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2698781478472785e-05,
+      "loss": 0.9722,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2694963444354184e-05,
+      "loss": 1.4604,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.269114541023558e-05,
+      "loss": 1.1942,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2687327376116978e-05,
+      "loss": 1.2488,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2683509341998376e-05,
+      "loss": 1.0365,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2679691307879773e-05,
+      "loss": 1.2516,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.267587327376117e-05,
+      "loss": 1.2657,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2672055239642569e-05,
+      "loss": 1.5658,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2668237205523966e-05,
+      "loss": 1.6101,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.266441917140536e-05,
+      "loss": 1.6651,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.266060113728676e-05,
+      "loss": 1.0557,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2656783103168156e-05,
+      "loss": 1.3023,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2652965069049553e-05,
+      "loss": 1.4082,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.264914703493095e-05,
+      "loss": 1.1101,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2645329000812349e-05,
+      "loss": 1.3318,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2641510966693746e-05,
+      "loss": 1.3711,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2637692932575142e-05,
+      "loss": 1.2684,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2633874898456541e-05,
+      "loss": 1.2863,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2630056864337936e-05,
+      "loss": 1.2028,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2626238830219333e-05,
+      "loss": 1.1381,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2622420796100732e-05,
+      "loss": 1.4131,
+      "step": 387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2618602761982129e-05,
+      "loss": 1.7112,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2614784727863526e-05,
+      "loss": 1.1919,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2610966693744924e-05,
+      "loss": 1.1407,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2607148659626321e-05,
+      "loss": 1.3704,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2603330625507718e-05,
+      "loss": 1.1797,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2599512591389115e-05,
+      "loss": 1.3275,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2595694557270512e-05,
+      "loss": 1.2145,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2591876523151909e-05,
+      "loss": 1.2464,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2588058489033306e-05,
+      "loss": 1.2583,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2584240454914704e-05,
+      "loss": 1.7095,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2580422420796101e-05,
+      "loss": 1.1147,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2576604386677498e-05,
+      "loss": 1.2907,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2572786352558897e-05,
+      "loss": 1.3906,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2568968318440294e-05,
+      "loss": 1.2231,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.256515028432169e-05,
+      "loss": 1.4364,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2561332250203087e-05,
+      "loss": 1.26,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2557514216084484e-05,
+      "loss": 1.5812,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2553696181965881e-05,
+      "loss": 1.4986,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2549878147847278e-05,
+      "loss": 1.7229,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2546060113728677e-05,
+      "loss": 1.3166,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2542242079610074e-05,
+      "loss": 1.5326,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.253842404549147e-05,
+      "loss": 1.5269,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2534606011372869e-05,
+      "loss": 1.2414,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2530787977254266e-05,
+      "loss": 1.3807,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2526969943135663e-05,
+      "loss": 1.2941,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.252315190901706e-05,
+      "loss": 1.2205,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2519333874898457e-05,
+      "loss": 1.4878,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2515515840779854e-05,
+      "loss": 1.3011,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2511697806661252e-05,
+      "loss": 1.1604,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2507879772542649e-05,
+      "loss": 1.1375,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2504061738424046e-05,
+      "loss": 1.5259,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2500243704305443e-05,
+      "loss": 1.2929,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2496425670186842e-05,
+      "loss": 1.4951,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2492607636068237e-05,
+      "loss": 1.248,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2488789601949634e-05,
+      "loss": 1.1547,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2484971567831032e-05,
+      "loss": 1.1854,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.248115353371243e-05,
+      "loss": 1.2538,
+      "step": 424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2477335499593826e-05,
+      "loss": 0.9611,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2473517465475225e-05,
+      "loss": 1.2368,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2469699431356622e-05,
+      "loss": 1.0084,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2465881397238018e-05,
+      "loss": 1.0193,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2462063363119415e-05,
+      "loss": 1.307,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2458245329000812e-05,
+      "loss": 1.2318,
+      "step": 430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.245442729488221e-05,
+      "loss": 1.1886,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2450609260763606e-05,
+      "loss": 1.2178,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2446791226645005e-05,
+      "loss": 1.4579,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2442973192526402e-05,
+      "loss": 1.3895,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2439155158407799e-05,
+      "loss": 1.2385,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2435337124289197e-05,
+      "loss": 1.2793,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2431519090170594e-05,
+      "loss": 1.1442,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2427701056051991e-05,
+      "loss": 1.2889,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2423883021933388e-05,
+      "loss": 1.4712,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2420064987814785e-05,
+      "loss": 1.4199,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2416246953696182e-05,
+      "loss": 1.2222,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2412428919577579e-05,
+      "loss": 1.1913,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2408610885458977e-05,
+      "loss": 1.1754,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2404792851340374e-05,
+      "loss": 1.1702,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2400974817221771e-05,
+      "loss": 1.2166,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.239715678310317e-05,
+      "loss": 1.2796,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2393338748984566e-05,
+      "loss": 1.0887,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2389520714865963e-05,
+      "loss": 1.1644,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.238570268074736e-05,
+      "loss": 1.2355,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2381884646628757e-05,
+      "loss": 1.5222,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2378066612510154e-05,
+      "loss": 0.9849,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2374248578391553e-05,
+      "loss": 1.5012,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.237043054427295e-05,
+      "loss": 1.2993,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2366612510154347e-05,
+      "loss": 1.3072,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2362794476035743e-05,
+      "loss": 1.3487,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2358976441917142e-05,
+      "loss": 1.1573,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2355158407798539e-05,
+      "loss": 1.2227,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2351340373679934e-05,
+      "loss": 1.047,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2347522339561333e-05,
+      "loss": 1.0145,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.234370430544273e-05,
+      "loss": 1.3946,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2339886271324127e-05,
+      "loss": 1.169,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2336068237205525e-05,
+      "loss": 1.0491,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2332250203086922e-05,
+      "loss": 1.1048,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2328432168968319e-05,
+      "loss": 1.4523,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2324614134849718e-05,
+      "loss": 1.3573,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2320796100731114e-05,
+      "loss": 1.5579,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.231697806661251e-05,
+      "loss": 1.0969,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2313160032493907e-05,
+      "loss": 1.1757,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2309341998375305e-05,
+      "loss": 1.298,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2305523964256702e-05,
+      "loss": 1.2109,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2301705930138099e-05,
+      "loss": 1.1973,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2297887896019498e-05,
+      "loss": 1.307,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2294069861900894e-05,
+      "loss": 1.208,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2290251827782291e-05,
+      "loss": 1.363,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2286433793663688e-05,
+      "loss": 1.3982,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2282615759545085e-05,
+      "loss": 1.3655,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2278797725426482e-05,
+      "loss": 1.1725,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.227497969130788e-05,
+      "loss": 1.722,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2271161657189278e-05,
+      "loss": 1.2896,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2267343623070675e-05,
+      "loss": 1.2675,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2263525588952071e-05,
+      "loss": 1.199,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.225970755483347e-05,
+      "loss": 1.3848,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2255889520714867e-05,
+      "loss": 1.0419,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2252071486596264e-05,
+      "loss": 1.3636,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.224825345247766e-05,
+      "loss": 1.3982,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2244435418359058e-05,
+      "loss": 1.5918,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2240617384240455e-05,
+      "loss": 1.2972,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2236799350121853e-05,
+      "loss": 1.2789,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.223298131600325e-05,
+      "loss": 1.2725,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2229163281884647e-05,
+      "loss": 1.3092,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2225345247766046e-05,
+      "loss": 1.2579,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2221527213647442e-05,
+      "loss": 1.1162,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.221770917952884e-05,
+      "loss": 1.3029,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2213891145410235e-05,
+      "loss": 1.3149,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2210073111291633e-05,
+      "loss": 1.4344,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.220625507717303e-05,
+      "loss": 1.2729,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2202437043054427e-05,
+      "loss": 1.2587,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2198619008935826e-05,
+      "loss": 1.2244,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2194800974817223e-05,
+      "loss": 1.1055,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.219098294069862e-05,
+      "loss": 1.2137,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2187164906580018e-05,
+      "loss": 1.252,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2183346872461415e-05,
+      "loss": 1.2903,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.217952883834281e-05,
+      "loss": 1.2722,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2175710804224207e-05,
+      "loss": 1.5099,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2171892770105606e-05,
+      "loss": 1.1612,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2168074735987003e-05,
+      "loss": 1.3153,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.21642567018684e-05,
+      "loss": 1.5121,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2160438667749798e-05,
+      "loss": 1.5355,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2156620633631195e-05,
+      "loss": 1.4545,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2152802599512592e-05,
+      "loss": 1.2684,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.214898456539399e-05,
+      "loss": 1.4268,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2145166531275386e-05,
+      "loss": 1.0874,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2141348497156783e-05,
+      "loss": 1.4574,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2137530463038181e-05,
+      "loss": 1.3059,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2133712428919578e-05,
+      "loss": 1.2899,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2129894394800975e-05,
+      "loss": 1.3041,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2126076360682372e-05,
+      "loss": 1.2746,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.212225832656377e-05,
+      "loss": 1.3411,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2118440292445167e-05,
+      "loss": 1.242,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2114622258326564e-05,
+      "loss": 1.4562,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2110804224207961e-05,
+      "loss": 1.0046,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2106986190089358e-05,
+      "loss": 1.5939,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2103168155970755e-05,
+      "loss": 1.2245,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2099350121852154e-05,
+      "loss": 1.9058,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.209553208773355e-05,
+      "loss": 1.3161,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2091714053614947e-05,
+      "loss": 1.2841,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2087896019496346e-05,
+      "loss": 1.3196,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2084077985377743e-05,
+      "loss": 1.2283,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.208025995125914e-05,
+      "loss": 1.3551,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2076441917140535e-05,
+      "loss": 1.1194,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2072623883021934e-05,
+      "loss": 1.1296,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.206880584890333e-05,
+      "loss": 1.5611,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2064987814784727e-05,
+      "loss": 1.517,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2061169780666126e-05,
+      "loss": 1.173,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2057351746547523e-05,
+      "loss": 1.1013,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.205353371242892e-05,
+      "loss": 1.2922,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2049715678310318e-05,
+      "loss": 1.1695,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2045897644191715e-05,
+      "loss": 1.4589,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.204207961007311e-05,
+      "loss": 1.287,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.203826157595451e-05,
+      "loss": 1.2812,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2034443541835906e-05,
+      "loss": 1.3639,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2030625507717303e-05,
+      "loss": 1.223,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.20268074735987e-05,
+      "loss": 1.0352,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2022989439480099e-05,
+      "loss": 1.1497,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2019171405361495e-05,
+      "loss": 1.4294,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2015353371242892e-05,
+      "loss": 1.1222,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2011535337124291e-05,
+      "loss": 0.9926,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007717303005686e-05,
+      "loss": 1.1344,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003899268887083e-05,
+      "loss": 1.0363,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000081234768482e-05,
+      "loss": 1.2383,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996263200649879e-05,
+      "loss": 1.475,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992445166531275e-05,
+      "loss": 1.7158,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988627132412674e-05,
+      "loss": 1.4556,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984809098294071e-05,
+      "loss": 1.2886,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980991064175468e-05,
+      "loss": 1.2778,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977173030056865e-05,
+      "loss": 1.561,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973354995938262e-05,
+      "loss": 1.1873,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969536961819659e-05,
+      "loss": 1.3695,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965718927701055e-05,
+      "loss": 1.1422,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961900893582454e-05,
+      "loss": 1.193,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1958082859463851e-05,
+      "loss": 1.1546,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1954264825345248e-05,
+      "loss": 1.132,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1950446791226647e-05,
+      "loss": 1.0971,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1946628757108043e-05,
+      "loss": 1.1648,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.194281072298944e-05,
+      "loss": 1.3712,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1938992688870837e-05,
+      "loss": 1.0937,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1935174654752234e-05,
+      "loss": 1.5984,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931356620633631e-05,
+      "loss": 1.0894,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927538586515028e-05,
+      "loss": 1.346,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923720552396427e-05,
+      "loss": 1.438,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1919902518277823e-05,
+      "loss": 1.2988,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191608448415922e-05,
+      "loss": 1.0288,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1912266450040619e-05,
+      "loss": 1.3138,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1908448415922016e-05,
+      "loss": 1.2912,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1904630381803411e-05,
+      "loss": 1.08,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.190081234768481e-05,
+      "loss": 1.1938,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1896994313566207e-05,
+      "loss": 1.0847,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1893176279447603e-05,
+      "loss": 1.8922,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1889358245329e-05,
+      "loss": 1.1215,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1885540211210399e-05,
+      "loss": 1.2131,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1881722177091796e-05,
+      "loss": 1.0385,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1877904142973193e-05,
+      "loss": 1.0671,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1874086108854591e-05,
+      "loss": 1.3873,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1870268074735987e-05,
+      "loss": 1.4788,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1866450040617384e-05,
+      "loss": 1.206,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1862632006498782e-05,
+      "loss": 1.2943,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1858813972380179e-05,
+      "loss": 1.3638,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1854995938261576e-05,
+      "loss": 1.5215,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1851177904142975e-05,
+      "loss": 1.3175,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1847359870024371e-05,
+      "loss": 1.1919,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1843541835905768e-05,
+      "loss": 1.0603,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1839723801787165e-05,
+      "loss": 1.4277,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1835905767668562e-05,
+      "loss": 1.3256,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1832087733549959e-05,
+      "loss": 1.3303,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1828269699431356e-05,
+      "loss": 1.4598,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1824451665312755e-05,
+      "loss": 1.0705,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1820633631194151e-05,
+      "loss": 1.1575,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1816815597075548e-05,
+      "loss": 1.3328,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1812997562956947e-05,
+      "loss": 1.0676,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1809179528838344e-05,
+      "loss": 1.1909,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.180536149471974e-05,
+      "loss": 1.7348,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1801543460601138e-05,
+      "loss": 1.2883,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1797725426482535e-05,
+      "loss": 1.4185,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1793907392363932e-05,
+      "loss": 1.3655,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1790089358245328e-05,
+      "loss": 1.2338,
+      "step": 605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1786271324126727e-05,
+      "loss": 1.0758,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1782453290008124e-05,
+      "loss": 1.2588,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.177863525588952e-05,
+      "loss": 1.1753,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.177481722177092e-05,
+      "loss": 1.3152,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1770999187652316e-05,
+      "loss": 1.4114,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1767181153533713e-05,
+      "loss": 1.2669,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.176336311941511e-05,
+      "loss": 1.2817,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1759545085296507e-05,
+      "loss": 1.3967,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1755727051177904e-05,
+      "loss": 1.3292,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1751909017059303e-05,
+      "loss": 1.2903,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.17480909829407e-05,
+      "loss": 1.1566,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1744272948822096e-05,
+      "loss": 1.1358,
+      "step": 617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1740454914703493e-05,
+      "loss": 1.0327,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1736636880584892e-05,
+      "loss": 1.4046,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1732818846466289e-05,
+      "loss": 1.2574,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1729000812347684e-05,
+      "loss": 1.2465,
+      "step": 621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1725182778229083e-05,
+      "loss": 1.1982,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.172136474411048e-05,
+      "loss": 1.0592,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1717546709991876e-05,
+      "loss": 1.1432,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1713728675873275e-05,
+      "loss": 1.4513,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1709910641754672e-05,
+      "loss": 1.1525,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1706092607636069e-05,
+      "loss": 1.2067,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1702274573517466e-05,
+      "loss": 1.1721,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1698456539398863e-05,
+      "loss": 1.5066,
+      "step": 629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.169463850528026e-05,
+      "loss": 1.039,
+      "step": 630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1690820471161656e-05,
+      "loss": 1.3772,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1687002437043055e-05,
+      "loss": 1.0785,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1683184402924452e-05,
+      "loss": 1.3153,
+      "step": 633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1679366368805849e-05,
+      "loss": 1.5971,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1675548334687247e-05,
+      "loss": 1.3179,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1671730300568644e-05,
+      "loss": 1.3033,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1667912266450041e-05,
+      "loss": 1.2759,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1664094232331438e-05,
+      "loss": 1.322,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1660276198212835e-05,
+      "loss": 1.2389,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1656458164094232e-05,
+      "loss": 1.6611,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1652640129975629e-05,
+      "loss": 1.2118,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1648822095857027e-05,
+      "loss": 1.1312,
+      "step": 642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1645004061738424e-05,
+      "loss": 1.2191,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1641186027619821e-05,
+      "loss": 1.1598,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.163736799350122e-05,
+      "loss": 1.4002,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1633549959382617e-05,
+      "loss": 1.049,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1629731925264014e-05,
+      "loss": 1.2497,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.162591389114541e-05,
+      "loss": 1.1847,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1622095857026808e-05,
+      "loss": 1.7279,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1618277822908204e-05,
+      "loss": 0.8989,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1614459788789603e-05,
+      "loss": 1.2363,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1610641754671e-05,
+      "loss": 1.1373,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1606823720552397e-05,
+      "loss": 1.0867,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1603005686433794e-05,
+      "loss": 1.3068,
+      "step": 654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1599187652315192e-05,
+      "loss": 1.1635,
+      "step": 655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.159536961819659e-05,
+      "loss": 1.0955,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1591551584077984e-05,
+      "loss": 1.4084,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1587733549959383e-05,
+      "loss": 1.3179,
+      "step": 658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.158391551584078e-05,
+      "loss": 1.3404,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1580097481722177e-05,
+      "loss": 1.7859,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1576279447603575e-05,
+      "loss": 1.1056,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1572461413484972e-05,
+      "loss": 1.4442,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.156864337936637e-05,
+      "loss": 1.0448,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1564825345247768e-05,
+      "loss": 1.1269,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1561007311129165e-05,
+      "loss": 1.2428,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.155718927701056e-05,
+      "loss": 1.3183,
+      "step": 666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1553371242891957e-05,
+      "loss": 0.9205,
+      "step": 667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1549553208773355e-05,
+      "loss": 1.1075,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1545735174654752e-05,
+      "loss": 1.6057,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.154191714053615e-05,
+      "loss": 1.2492,
+      "step": 670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1538099106417548e-05,
+      "loss": 1.3823,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1534281072298945e-05,
+      "loss": 0.997,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1530463038180342e-05,
+      "loss": 1.3166,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.152664500406174e-05,
+      "loss": 1.1456,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1522826969943136e-05,
+      "loss": 1.1964,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1519008935824532e-05,
+      "loss": 1.1157,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1515190901705931e-05,
+      "loss": 1.3513,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1511372867587328e-05,
+      "loss": 1.2138,
+      "step": 678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1507554833468725e-05,
+      "loss": 1.5413,
+      "step": 679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1503736799350122e-05,
+      "loss": 1.3288,
+      "step": 680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.149991876523152e-05,
+      "loss": 1.4602,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1496100731112917e-05,
+      "loss": 1.2841,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1492282696994314e-05,
+      "loss": 1.2338,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1488464662875711e-05,
+      "loss": 1.191,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1484646628757108e-05,
+      "loss": 1.3097,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1480828594638505e-05,
+      "loss": 1.1927,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1477010560519903e-05,
+      "loss": 1.4275,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.14731925264013e-05,
+      "loss": 1.1329,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1469374492282697e-05,
+      "loss": 1.2345,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1465556458164096e-05,
+      "loss": 1.4304,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1461738424045493e-05,
+      "loss": 1.2694,
+      "step": 691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.145792038992689e-05,
+      "loss": 1.0006,
+      "step": 692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1454102355808285e-05,
+      "loss": 1.3425,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.1450284321689684e-05,
+      "loss": 1.5635,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.144646628757108e-05,
+      "loss": 1.2046,
+      "step": 695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1442648253452477e-05,
+      "loss": 1.1436,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1438830219333876e-05,
+      "loss": 1.1949,
+      "step": 697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1435012185215273e-05,
+      "loss": 1.2221,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.143119415109667e-05,
+      "loss": 1.2027,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1427376116978068e-05,
+      "loss": 1.1946,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1423558082859465e-05,
+      "loss": 1.1909,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.141974004874086e-05,
+      "loss": 1.6419,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1415922014622257e-05,
+      "loss": 1.2764,
+      "step": 703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1412103980503656e-05,
+      "loss": 1.519,
+      "step": 704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1408285946385053e-05,
+      "loss": 1.3435,
+      "step": 705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.140446791226645e-05,
+      "loss": 1.3154,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1400649878147848e-05,
+      "loss": 1.1236,
+      "step": 707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1396831844029245e-05,
+      "loss": 1.1232,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1393013809910642e-05,
+      "loss": 1.1325,
+      "step": 709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.138919577579204e-05,
+      "loss": 1.042,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1385377741673436e-05,
+      "loss": 1.3185,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1381559707554833e-05,
+      "loss": 1.3925,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1377741673436232e-05,
+      "loss": 1.1978,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1373923639317628e-05,
+      "loss": 0.9484,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1370105605199025e-05,
+      "loss": 1.3317,
+      "step": 715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1366287571080422e-05,
+      "loss": 1.5639,
+      "step": 716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.136246953696182e-05,
+      "loss": 1.1367,
+      "step": 717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1358651502843218e-05,
+      "loss": 1.5438,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1354833468724615e-05,
+      "loss": 1.1721,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.1351015434606012e-05,
+      "loss": 1.3215,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1347197400487408e-05,
+      "loss": 1.1008,
+      "step": 721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1343379366368805e-05,
+      "loss": 1.4077,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1339561332250204e-05,
+      "loss": 1.445,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1335743298131601e-05,
+      "loss": 1.1155,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1331925264012998e-05,
+      "loss": 1.1268,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1328107229894396e-05,
+      "loss": 1.2215,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1324289195775793e-05,
+      "loss": 1.3477,
+      "step": 727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.132047116165719e-05,
+      "loss": 1.4048,
+      "step": 728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1316653127538585e-05,
+      "loss": 1.4026,
+      "step": 729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1312835093419984e-05,
+      "loss": 1.4353,
+      "step": 730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1309017059301381e-05,
+      "loss": 1.2021,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.1305199025182778e-05,
+      "loss": 1.121,
+      "step": 732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1301380991064176e-05,
+      "loss": 0.9781,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1297562956945573e-05,
+      "loss": 1.3155,
+      "step": 734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.129374492282697e-05,
+      "loss": 1.4046,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1289926888708369e-05,
+      "loss": 1.1548,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1286108854589766e-05,
+      "loss": 1.1915,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1282290820471161e-05,
+      "loss": 1.0436,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.127847278635256e-05,
+      "loss": 1.1255,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1274654752233956e-05,
+      "loss": 1.0866,
+      "step": 740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1270836718115353e-05,
+      "loss": 1.0312,
+      "step": 741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.126701868399675e-05,
+      "loss": 1.1865,
+      "step": 742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1263200649878149e-05,
+      "loss": 1.2614,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1259382615759546e-05,
+      "loss": 1.129,
+      "step": 744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1255564581640943e-05,
+      "loss": 1.3916,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1251746547522341e-05,
+      "loss": 1.0696,
+      "step": 746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1247928513403736e-05,
+      "loss": 1.3455,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1244110479285133e-05,
+      "loss": 1.5139,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1240292445166532e-05,
+      "loss": 1.3222,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1236474411047929e-05,
+      "loss": 1.349,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1232656376929326e-05,
+      "loss": 1.198,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1228838342810724e-05,
+      "loss": 1.43,
+      "step": 752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1225020308692121e-05,
+      "loss": 1.0844,
+      "step": 753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1221202274573518e-05,
+      "loss": 1.2448,
+      "step": 754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1217384240454915e-05,
+      "loss": 1.2499,
+      "step": 755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1213566206336312e-05,
+      "loss": 1.3281,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1209748172217709e-05,
+      "loss": 1.2027,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1205930138099106e-05,
+      "loss": 1.5015,
+      "step": 758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1202112103980504e-05,
+      "loss": 1.1981,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1198294069861901e-05,
+      "loss": 1.4424,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1194476035743298e-05,
+      "loss": 1.2847,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1190658001624697e-05,
+      "loss": 1.2458,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1186839967506094e-05,
+      "loss": 1.1548,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.118302193338749e-05,
+      "loss": 1.1573,
+      "step": 764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1179203899268888e-05,
+      "loss": 1.0995,
+      "step": 765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1175385865150284e-05,
+      "loss": 1.2283,
+      "step": 766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1171567831031681e-05,
+      "loss": 1.3006,
+      "step": 767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1167749796913078e-05,
+      "loss": 1.2423,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1163931762794477e-05,
+      "loss": 1.2606,
+      "step": 769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1160113728675874e-05,
+      "loss": 1.2804,
+      "step": 770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.115629569455727e-05,
+      "loss": 1.3783,
+      "step": 771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.115247766043867e-05,
+      "loss": 1.2538,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1148659626320066e-05,
+      "loss": 1.1269,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1144841592201463e-05,
+      "loss": 1.4891,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.114102355808286e-05,
+      "loss": 1.402,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1137205523964257e-05,
+      "loss": 1.1069,
+      "step": 776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1133387489845654e-05,
+      "loss": 1.2073,
+      "step": 777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.112956945572705e-05,
+      "loss": 1.5322,
+      "step": 778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.112575142160845e-05,
+      "loss": 1.2538,
+      "step": 779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1121933387489846e-05,
+      "loss": 1.3175,
+      "step": 780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1118115353371243e-05,
+      "loss": 1.594,
+      "step": 781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1114297319252642e-05,
+      "loss": 1.164,
+      "step": 782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1110479285134037e-05,
+      "loss": 1.2087,
+      "step": 783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1106661251015434e-05,
+      "loss": 0.9543,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1102843216896832e-05,
+      "loss": 0.8164,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.109902518277823e-05,
+      "loss": 1.4995,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1095207148659626e-05,
+      "loss": 1.407,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1091389114541025e-05,
+      "loss": 1.1888,
+      "step": 788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1087571080422422e-05,
+      "loss": 1.175,
+      "step": 789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1083753046303819e-05,
+      "loss": 1.2168,
+      "step": 790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1079935012185216e-05,
+      "loss": 1.0654,
+      "step": 791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1076116978066612e-05,
+      "loss": 1.2199,
+      "step": 792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.107229894394801e-05,
+      "loss": 1.1629,
+      "step": 793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1068480909829406e-05,
+      "loss": 1.2226,
+      "step": 794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1064662875710805e-05,
+      "loss": 1.1859,
+      "step": 795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1060844841592202e-05,
+      "loss": 1.0681,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1057026807473599e-05,
+      "loss": 1.2967,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1053208773354997e-05,
+      "loss": 1.3267,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1049390739236394e-05,
+      "loss": 1.2851,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1045572705117791e-05,
+      "loss": 1.4181,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1041754670999188e-05,
+      "loss": 1.2417,
+      "step": 801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1037936636880585e-05,
+      "loss": 1.4157,
+      "step": 802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1034118602761982e-05,
+      "loss": 1.0419,
+      "step": 803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1030300568643379e-05,
+      "loss": 1.018,
+      "step": 804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1026482534524777e-05,
+      "loss": 1.4254,
+      "step": 805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1022664500406174e-05,
+      "loss": 1.4498,
+      "step": 806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1018846466287571e-05,
+      "loss": 1.6383,
+      "step": 807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.101502843216897e-05,
+      "loss": 1.0662,
+      "step": 808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1011210398050367e-05,
+      "loss": 1.2264,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1007392363931764e-05,
+      "loss": 1.2401,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.100357432981316e-05,
+      "loss": 1.3032,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0999756295694557e-05,
+      "loss": 1.853,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0995938261575954e-05,
+      "loss": 1.2458,
+      "step": 813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0992120227457353e-05,
+      "loss": 1.5764,
+      "step": 814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.098830219333875e-05,
+      "loss": 1.44,
+      "step": 815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0984484159220147e-05,
+      "loss": 1.2445,
+      "step": 816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0980666125101544e-05,
+      "loss": 1.0822,
+      "step": 817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0976848090982942e-05,
+      "loss": 1.4951,
+      "step": 818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0973030056864339e-05,
+      "loss": 1.3974,
+      "step": 819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0969212022745734e-05,
+      "loss": 1.2482,
+      "step": 820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0965393988627133e-05,
+      "loss": 1.4313,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.096157595450853e-05,
+      "loss": 1.3009,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0957757920389927e-05,
+      "loss": 1.0535,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0953939886271325e-05,
+      "loss": 1.5214,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0950121852152722e-05,
+      "loss": 1.1056,
+      "step": 825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0946303818034119e-05,
+      "loss": 1.296,
+      "step": 826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0942485783915516e-05,
+      "loss": 1.1873,
+      "step": 827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0938667749796915e-05,
+      "loss": 1.226,
+      "step": 828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.093484971567831e-05,
+      "loss": 1.406,
+      "step": 829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0931031681559707e-05,
+      "loss": 1.5169,
+      "step": 830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0927213647441105e-05,
+      "loss": 1.1998,
+      "step": 831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0923395613322502e-05,
+      "loss": 1.2791,
+      "step": 832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.09195775792039e-05,
+      "loss": 1.303,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0915759545085298e-05,
+      "loss": 1.236,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0911941510966695e-05,
+      "loss": 1.2135,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0908123476848092e-05,
+      "loss": 1.1651,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0904305442729488e-05,
+      "loss": 1.1581,
+      "step": 837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0900487408610885e-05,
+      "loss": 1.5056,
+      "step": 838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0896669374492282e-05,
+      "loss": 1.3057,
+      "step": 839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.089285134037368e-05,
+      "loss": 1.2618,
+      "step": 840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0889033306255078e-05,
+      "loss": 1.4792,
+      "step": 841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0885215272136475e-05,
+      "loss": 1.2261,
+      "step": 842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0881397238017872e-05,
+      "loss": 1.5709,
+      "step": 843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.087757920389927e-05,
+      "loss": 1.3903,
+      "step": 844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0873761169780667e-05,
+      "loss": 1.3522,
+      "step": 845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0869943135662064e-05,
+      "loss": 1.1985,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0866125101543461e-05,
+      "loss": 1.1378,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0862307067424858e-05,
+      "loss": 1.3062,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0858489033306255e-05,
+      "loss": 1.1787,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0854670999187653e-05,
+      "loss": 1.061,
+      "step": 850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.085085296506905e-05,
+      "loss": 1.4303,
+      "step": 851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0847034930950447e-05,
+      "loss": 1.3004,
+      "step": 852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0843216896831844e-05,
+      "loss": 1.248,
+      "step": 853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0839398862713243e-05,
+      "loss": 1.2064,
+      "step": 854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.083558082859464e-05,
+      "loss": 1.3966,
+      "step": 855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0831762794476035e-05,
+      "loss": 1.5002,
+      "step": 856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0827944760357433e-05,
+      "loss": 1.2859,
+      "step": 857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.082412672623883e-05,
+      "loss": 1.5141,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0820308692120227e-05,
+      "loss": 1.0866,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0816490658001626e-05,
+      "loss": 1.4934,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0812672623883023e-05,
+      "loss": 1.2185,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.080885458976442e-05,
+      "loss": 1.2412,
+      "step": 862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0805036555645818e-05,
+      "loss": 1.3071,
+      "step": 863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0801218521527215e-05,
+      "loss": 1.3472,
+      "step": 864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.079740048740861e-05,
+      "loss": 1.2904,
+      "step": 865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0793582453290007e-05,
+      "loss": 1.1016,
+      "step": 866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0789764419171406e-05,
+      "loss": 1.2245,
+      "step": 867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0785946385052803e-05,
+      "loss": 1.8545,
+      "step": 868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.07821283509342e-05,
+      "loss": 1.3424,
+      "step": 869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0778310316815598e-05,
+      "loss": 1.1084,
+      "step": 870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0774492282696995e-05,
+      "loss": 1.4077,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0770674248578392e-05,
+      "loss": 1.2488,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.076685621445979e-05,
+      "loss": 1.2305,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0763038180341186e-05,
+      "loss": 1.2281,
+      "step": 874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0759220146222583e-05,
+      "loss": 1.3109,
+      "step": 875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0755402112103981e-05,
+      "loss": 1.3624,
+      "step": 876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0751584077985378e-05,
+      "loss": 1.4247,
+      "step": 877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0747766043866775e-05,
+      "loss": 1.2562,
+      "step": 878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0743948009748172e-05,
+      "loss": 1.3077,
+      "step": 879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.074012997562957e-05,
+      "loss": 1.229,
+      "step": 880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0736311941510968e-05,
+      "loss": 1.3954,
+      "step": 881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0732493907392364e-05,
+      "loss": 1.1284,
+      "step": 882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0728675873273761e-05,
+      "loss": 1.2577,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0724857839155158e-05,
+      "loss": 1.1091,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0721039805036555e-05,
+      "loss": 0.9297,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0717221770917954e-05,
+      "loss": 1.048,
+      "step": 886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.071340373679935e-05,
+      "loss": 1.3362,
+      "step": 887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0709585702680748e-05,
+      "loss": 1.1442,
+      "step": 888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0705767668562146e-05,
+      "loss": 1.3094,
+      "step": 889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0701949634443543e-05,
+      "loss": 1.1601,
+      "step": 890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.069813160032494e-05,
+      "loss": 1.2726,
+      "step": 891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0694313566206335e-05,
+      "loss": 1.1209,
+      "step": 892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0690495532087734e-05,
+      "loss": 1.3876,
+      "step": 893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.068667749796913e-05,
+      "loss": 1.3868,
+      "step": 894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0682859463850528e-05,
+      "loss": 0.8962,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0679041429731926e-05,
+      "loss": 1.2099,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0675223395613323e-05,
+      "loss": 1.2756,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.067140536149472e-05,
+      "loss": 1.3977,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0667587327376119e-05,
+      "loss": 1.4919,
+      "step": 899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0663769293257516e-05,
+      "loss": 1.2793,
+      "step": 900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.065995125913891e-05,
+      "loss": 1.0765,
+      "step": 901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0656133225020308e-05,
+      "loss": 1.1568,
+      "step": 902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0652315190901706e-05,
+      "loss": 1.1012,
+      "step": 903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0648497156783103e-05,
+      "loss": 1.2104,
+      "step": 904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.06446791226645e-05,
+      "loss": 1.4515,
+      "step": 905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0640861088545899e-05,
+      "loss": 1.3073,
+      "step": 906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0637043054427296e-05,
+      "loss": 1.4104,
+      "step": 907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0633225020308692e-05,
+      "loss": 1.0889,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0629406986190091e-05,
+      "loss": 1.2455,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0625588952071486e-05,
+      "loss": 1.1995,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0621770917952883e-05,
+      "loss": 1.0894,
+      "step": 911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0617952883834282e-05,
+      "loss": 1.1828,
+      "step": 912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0614134849715679e-05,
+      "loss": 1.1741,
+      "step": 913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0610316815597076e-05,
+      "loss": 1.0733,
+      "step": 914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0606498781478473e-05,
+      "loss": 1.1024,
+      "step": 915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0602680747359871e-05,
+      "loss": 1.2032,
+      "step": 916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.0598862713241268e-05,
+      "loss": 1.3693,
+      "step": 917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0595044679122665e-05,
+      "loss": 1.0772,
+      "step": 918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0591226645004062e-05,
+      "loss": 1.1194,
+      "step": 919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0587408610885459e-05,
+      "loss": 1.2266,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0583590576766856e-05,
+      "loss": 1.1499,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0579772542648254e-05,
+      "loss": 1.0707,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0575954508529651e-05,
+      "loss": 1.3278,
+      "step": 923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0572136474411048e-05,
+      "loss": 1.046,
+      "step": 924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0568318440292447e-05,
+      "loss": 1.2883,
+      "step": 925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0564500406173844e-05,
+      "loss": 1.3172,
+      "step": 926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.056068237205524e-05,
+      "loss": 1.1092,
+      "step": 927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0556864337936637e-05,
+      "loss": 1.261,
+      "step": 928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.0553046303818034e-05,
+      "loss": 1.0017,
+      "step": 929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0549228269699431e-05,
+      "loss": 1.4889,
+      "step": 930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0545410235580828e-05,
+      "loss": 1.0641,
+      "step": 931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0541592201462227e-05,
+      "loss": 1.0892,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0537774167343624e-05,
+      "loss": 1.3155,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.053395613322502e-05,
+      "loss": 1.1426,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0530138099106419e-05,
+      "loss": 1.4453,
+      "step": 935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0526320064987816e-05,
+      "loss": 1.4714,
+      "step": 936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0522502030869211e-05,
+      "loss": 2.0258,
+      "step": 937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.051868399675061e-05,
+      "loss": 1.3093,
+      "step": 938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0514865962632007e-05,
+      "loss": 1.1843,
+      "step": 939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.0511047928513404e-05,
+      "loss": 1.1837,
+      "step": 940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.05072298943948e-05,
+      "loss": 1.0802,
+      "step": 941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.05034118602762e-05,
+      "loss": 1.3281,
+      "step": 942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0499593826157596e-05,
+      "loss": 1.2504,
+      "step": 943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0495775792038993e-05,
+      "loss": 1.3957,
+      "step": 944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0491957757920392e-05,
+      "loss": 1.0257,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0488139723801787e-05,
+      "loss": 1.0866,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0484321689683184e-05,
+      "loss": 1.1458,
+      "step": 947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0480503655564582e-05,
+      "loss": 1.07,
+      "step": 948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.047668562144598e-05,
+      "loss": 0.9955,
+      "step": 949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0472867587327376e-05,
+      "loss": 1.193,
+      "step": 950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0469049553208775e-05,
+      "loss": 1.1742,
+      "step": 951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0465231519090172e-05,
+      "loss": 1.3066,
+      "step": 952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0461413484971569e-05,
+      "loss": 1.3953,
+      "step": 953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0457595450852965e-05,
+      "loss": 1.231,
+      "step": 954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0453777416734362e-05,
+      "loss": 1.1173,
+      "step": 955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.044995938261576e-05,
+      "loss": 1.1899,
+      "step": 956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0446141348497156e-05,
+      "loss": 1.1624,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0442323314378555e-05,
+      "loss": 1.1283,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0438505280259952e-05,
+      "loss": 0.9126,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0434687246141349e-05,
+      "loss": 1.2147,
+      "step": 960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0430869212022747e-05,
+      "loss": 1.311,
+      "step": 961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0427051177904144e-05,
+      "loss": 1.3655,
+      "step": 962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0423233143785541e-05,
+      "loss": 1.3857,
+      "step": 963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0419415109666938e-05,
+      "loss": 1.1637,
+      "step": 964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0415597075548335e-05,
+      "loss": 1.6959,
+      "step": 965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0411779041429732e-05,
+      "loss": 1.2575,
+      "step": 966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0407961007311129e-05,
+      "loss": 1.0416,
+      "step": 967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0404142973192527e-05,
+      "loss": 1.3079,
+      "step": 968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0400324939073924e-05,
+      "loss": 1.2461,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0396506904955321e-05,
+      "loss": 1.3727,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.039268887083672e-05,
+      "loss": 1.2977,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0388870836718116e-05,
+      "loss": 0.8774,
+      "step": 972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0385052802599513e-05,
+      "loss": 1.1124,
+      "step": 973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.038123476848091e-05,
+      "loss": 1.1715,
+      "step": 974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0377416734362307e-05,
+      "loss": 1.2862,
+      "step": 975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0373598700243704e-05,
+      "loss": 1.0927,
+      "step": 976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0369780666125101e-05,
+      "loss": 1.1174,
+      "step": 977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03659626320065e-05,
+      "loss": 1.1281,
+      "step": 978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0362144597887897e-05,
+      "loss": 1.1329,
+      "step": 979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0358326563769293e-05,
+      "loss": 1.2884,
+      "step": 980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0354508529650692e-05,
+      "loss": 1.1835,
+      "step": 981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0350690495532089e-05,
+      "loss": 0.9296,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0346872461413484e-05,
+      "loss": 1.3209,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0343054427294883e-05,
+      "loss": 1.2061,
+      "step": 984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033923639317628e-05,
+      "loss": 1.0518,
+      "step": 985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0335418359057677e-05,
+      "loss": 1.2051,
+      "step": 986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0331600324939075e-05,
+      "loss": 1.3484,
+      "step": 987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0327782290820472e-05,
+      "loss": 1.2829,
+      "step": 988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0323964256701869e-05,
+      "loss": 1.3154,
+      "step": 989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0320146222583266e-05,
+      "loss": 1.2984,
+      "step": 990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0316328188464663e-05,
+      "loss": 1.4039,
+      "step": 991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.031251015434606e-05,
+      "loss": 1.0201,
+      "step": 992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0308692120227457e-05,
+      "loss": 1.246,
+      "step": 993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0304874086108855e-05,
+      "loss": 0.8834,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0301056051990252e-05,
+      "loss": 1.3792,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0297238017871649e-05,
+      "loss": 1.2863,
+      "step": 996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0293419983753048e-05,
+      "loss": 1.1189,
+      "step": 997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0289601949634445e-05,
+      "loss": 1.2408,
+      "step": 998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0285783915515841e-05,
+      "loss": 1.0703,
+      "step": 999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0281965881397238e-05,
+      "loss": 1.1974,
+      "step": 1000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0278147847278635e-05,
+      "loss": 0.9156,
+      "step": 1001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0274329813160032e-05,
+      "loss": 1.4194,
+      "step": 1002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0270511779041429e-05,
+      "loss": 1.2589,
+      "step": 1003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0266693744922828e-05,
+      "loss": 1.1948,
+      "step": 1004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0262875710804225e-05,
+      "loss": 1.2959,
+      "step": 1005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0259057676685621e-05,
+      "loss": 1.4115,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.025523964256702e-05,
+      "loss": 1.2279,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0251421608448417e-05,
+      "loss": 1.182,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0247603574329814e-05,
+      "loss": 1.1,
+      "step": 1009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.024378554021121e-05,
+      "loss": 1.1616,
+      "step": 1010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0239967506092608e-05,
+      "loss": 1.3482,
+      "step": 1011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0236149471974005e-05,
+      "loss": 1.2281,
+      "step": 1012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0232331437855403e-05,
+      "loss": 1.1083,
+      "step": 1013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.02285134037368e-05,
+      "loss": 1.2744,
+      "step": 1014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0224695369618197e-05,
+      "loss": 1.308,
+      "step": 1015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0220877335499594e-05,
+      "loss": 1.2527,
+      "step": 1016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0217059301380992e-05,
+      "loss": 1.7981,
+      "step": 1017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.021324126726239e-05,
+      "loss": 1.0463,
+      "step": 1018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0209423233143785e-05,
+      "loss": 1.0348,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0205605199025183e-05,
+      "loss": 1.7496,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.020178716490658e-05,
+      "loss": 1.2111,
+      "step": 1021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0197969130787977e-05,
+      "loss": 1.2565,
+      "step": 1022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0194151096669376e-05,
+      "loss": 1.1222,
+      "step": 1023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0190333062550773e-05,
+      "loss": 1.1116,
+      "step": 1024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.018651502843217e-05,
+      "loss": 1.0817,
+      "step": 1025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0182696994313566e-05,
+      "loss": 1.5623,
+      "step": 1026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0178878960194965e-05,
+      "loss": 1.2752,
+      "step": 1027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.017506092607636e-05,
+      "loss": 1.096,
+      "step": 1028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0171242891957757e-05,
+      "loss": 1.2503,
+      "step": 1029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0167424857839156e-05,
+      "loss": 1.1224,
+      "step": 1030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0163606823720553e-05,
+      "loss": 1.125,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.015978878960195e-05,
+      "loss": 1.1948,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0155970755483348e-05,
+      "loss": 1.3924,
+      "step": 1033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0152152721364745e-05,
+      "loss": 1.2164,
+      "step": 1034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0148334687246142e-05,
+      "loss": 1.3653,
+      "step": 1035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.014451665312754e-05,
+      "loss": 0.9916,
+      "step": 1036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0140698619008936e-05,
+      "loss": 1.4422,
+      "step": 1037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0136880584890333e-05,
+      "loss": 1.1385,
+      "step": 1038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.013306255077173e-05,
+      "loss": 1.0138,
+      "step": 1039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0129244516653128e-05,
+      "loss": 1.1856,
+      "step": 1040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0125426482534525e-05,
+      "loss": 1.1525,
+      "step": 1041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0121608448415922e-05,
+      "loss": 1.2385,
+      "step": 1042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.011779041429732e-05,
+      "loss": 1.2826,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0113972380178717e-05,
+      "loss": 1.3187,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0110154346060114e-05,
+      "loss": 1.4194,
+      "step": 1045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0106336311941511e-05,
+      "loss": 1.2473,
+      "step": 1046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0102518277822908e-05,
+      "loss": 1.4013,
+      "step": 1047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0098700243704305e-05,
+      "loss": 1.4582,
+      "step": 1048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0094882209585704e-05,
+      "loss": 1.3714,
+      "step": 1049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.00910641754671e-05,
+      "loss": 1.1471,
+      "step": 1050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0087246141348497e-05,
+      "loss": 1.0234,
+      "step": 1051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0083428107229894e-05,
+      "loss": 1.4046,
+      "step": 1052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0079610073111293e-05,
+      "loss": 1.2267,
+      "step": 1053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007579203899269e-05,
+      "loss": 1.2125,
+      "step": 1054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0071974004874085e-05,
+      "loss": 1.3672,
+      "step": 1055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0068155970755484e-05,
+      "loss": 1.4148,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006433793663688e-05,
+      "loss": 1.0114,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0060519902518277e-05,
+      "loss": 1.3814,
+      "step": 1058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0056701868399676e-05,
+      "loss": 1.0602,
+      "step": 1059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0052883834281073e-05,
+      "loss": 1.2034,
+      "step": 1060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004906580016247e-05,
+      "loss": 1.2566,
+      "step": 1061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0045247766043869e-05,
+      "loss": 1.0687,
+      "step": 1062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0041429731925265e-05,
+      "loss": 1.3685,
+      "step": 1063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003761169780666e-05,
+      "loss": 1.2772,
+      "step": 1064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0033793663688058e-05,
+      "loss": 1.3747,
+      "step": 1065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0029975629569456e-05,
+      "loss": 1.3674,
+      "step": 1066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0026157595450853e-05,
+      "loss": 1.7534,
+      "step": 1067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.002233956133225e-05,
+      "loss": 1.1444,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0018521527213649e-05,
+      "loss": 1.108,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0014703493095045e-05,
+      "loss": 1.0534,
+      "step": 1070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0010885458976442e-05,
+      "loss": 1.3135,
+      "step": 1071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0007067424857841e-05,
+      "loss": 1.5357,
+      "step": 1072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0003249390739236e-05,
+      "loss": 1.0313,
+      "step": 1073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.999431356620633e-06,
+      "loss": 1.2297,
+      "step": 1074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.995613322502032e-06,
+      "loss": 1.0966,
+      "step": 1075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.991795288383429e-06,
+      "loss": 1.5094,
+      "step": 1076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.987977254264825e-06,
+      "loss": 1.2864,
+      "step": 1077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.984159220146222e-06,
+      "loss": 1.3172,
+      "step": 1078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.980341186027621e-06,
+      "loss": 1.264,
+      "step": 1079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.976523151909018e-06,
+      "loss": 1.1057,
+      "step": 1080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.972705117790415e-06,
+      "loss": 1.0534,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.968887083671812e-06,
+      "loss": 1.1644,
+      "step": 1082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.965069049553209e-06,
+      "loss": 1.1936,
+      "step": 1083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.961251015434606e-06,
+      "loss": 1.1576,
+      "step": 1084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.957432981316004e-06,
+      "loss": 1.0996,
+      "step": 1085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.953614947197401e-06,
+      "loss": 1.4836,
+      "step": 1086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.949796913078798e-06,
+      "loss": 1.6307,
+      "step": 1087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.945978878960197e-06,
+      "loss": 1.1495,
+      "step": 1088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.942160844841593e-06,
+      "loss": 1.4318,
+      "step": 1089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.93834281072299e-06,
+      "loss": 1.1842,
+      "step": 1090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.934524776604386e-06,
+      "loss": 1.1181,
+      "step": 1091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.930706742485784e-06,
+      "loss": 1.1626,
+      "step": 1092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.926888708367181e-06,
+      "loss": 1.0731,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.923070674248578e-06,
+      "loss": 1.6103,
+      "step": 1094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.919252640129977e-06,
+      "loss": 1.0093,
+      "step": 1095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.915434606011373e-06,
+      "loss": 1.3279,
+      "step": 1096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.91161657189277e-06,
+      "loss": 1.4665,
+      "step": 1097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.907798537774169e-06,
+      "loss": 1.0119,
+      "step": 1098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.903980503655566e-06,
+      "loss": 1.1317,
+      "step": 1099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.900162469536961e-06,
+      "loss": 1.1193,
+      "step": 1100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.896344435418358e-06,
+      "loss": 1.1096,
+      "step": 1101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.892526401299757e-06,
+      "loss": 1.358,
+      "step": 1102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.888708367181153e-06,
+      "loss": 1.2791,
+      "step": 1103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.88489033306255e-06,
+      "loss": 1.082,
+      "step": 1104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.881072298943949e-06,
+      "loss": 1.1266,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.877254264825346e-06,
+      "loss": 1.4393,
+      "step": 1106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.873436230706743e-06,
+      "loss": 1.2296,
+      "step": 1107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.869618196588141e-06,
+      "loss": 1.0641,
+      "step": 1108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.865800162469537e-06,
+      "loss": 1.0335,
+      "step": 1109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.861982128350934e-06,
+      "loss": 1.2318,
+      "step": 1110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.858164094232332e-06,
+      "loss": 1.4526,
+      "step": 1111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.854346060113729e-06,
+      "loss": 1.1435,
+      "step": 1112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.850528025995126e-06,
+      "loss": 1.4337,
+      "step": 1113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.846709991876523e-06,
+      "loss": 1.0889,
+      "step": 1114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.842891957757921e-06,
+      "loss": 1.1064,
+      "step": 1115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.839073923639318e-06,
+      "loss": 1.0351,
+      "step": 1116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.835255889520715e-06,
+      "loss": 1.1312,
+      "step": 1117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.831437855402112e-06,
+      "loss": 0.9053,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.827619821283509e-06,
+      "loss": 1.5388,
+      "step": 1119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.823801787164906e-06,
+      "loss": 1.1709,
+      "step": 1120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.819983753046305e-06,
+      "loss": 1.2214,
+      "step": 1121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.816165718927701e-06,
+      "loss": 1.3566,
+      "step": 1122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.812347684809098e-06,
+      "loss": 1.1095,
+      "step": 1123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.808529650690497e-06,
+      "loss": 1.4132,
+      "step": 1124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.804711616571894e-06,
+      "loss": 1.0921,
+      "step": 1125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.80089358245329e-06,
+      "loss": 1.2282,
+      "step": 1126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.797075548334688e-06,
+      "loss": 1.2642,
+      "step": 1127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.793257514216085e-06,
+      "loss": 1.0976,
+      "step": 1128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.789439480097482e-06,
+      "loss": 1.388,
+      "step": 1129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.785621445978878e-06,
+      "loss": 1.5094,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.781803411860277e-06,
+      "loss": 1.3931,
+      "step": 1131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.777985377741674e-06,
+      "loss": 1.2129,
+      "step": 1132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.77416734362307e-06,
+      "loss": 0.9761,
+      "step": 1133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.77034930950447e-06,
+      "loss": 1.5877,
+      "step": 1134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.766531275385866e-06,
+      "loss": 1.1997,
+      "step": 1135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.762713241267263e-06,
+      "loss": 1.2313,
+      "step": 1136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.75889520714866e-06,
+      "loss": 1.1567,
+      "step": 1137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.755077173030057e-06,
+      "loss": 1.3108,
+      "step": 1138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.751259138911454e-06,
+      "loss": 1.0625,
+      "step": 1139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.747441104792851e-06,
+      "loss": 1.327,
+      "step": 1140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.74362307067425e-06,
+      "loss": 1.1247,
+      "step": 1141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.739805036555646e-06,
+      "loss": 1.2402,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.735987002437043e-06,
+      "loss": 1.3179,
+      "step": 1143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.732168968318442e-06,
+      "loss": 1.0806,
+      "step": 1144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.728350934199837e-06,
+      "loss": 0.8872,
+      "step": 1145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.724532900081234e-06,
+      "loss": 1.0064,
+      "step": 1146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.720714865962633e-06,
+      "loss": 1.3048,
+      "step": 1147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.71689683184403e-06,
+      "loss": 1.3137,
+      "step": 1148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.713078797725426e-06,
+      "loss": 1.3628,
+      "step": 1149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.709260763606825e-06,
+      "loss": 1.2837,
+      "step": 1150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.705442729488222e-06,
+      "loss": 1.0633,
+      "step": 1151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.701624695369619e-06,
+      "loss": 1.1037,
+      "step": 1152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.697806661251016e-06,
+      "loss": 1.442,
+      "step": 1153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.693988627132413e-06,
+      "loss": 1.4552,
+      "step": 1154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.69017059301381e-06,
+      "loss": 1.1276,
+      "step": 1155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.686352558895206e-06,
+      "loss": 1.3445,
+      "step": 1156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.682534524776605e-06,
+      "loss": 1.3988,
+      "step": 1157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.678716490658002e-06,
+      "loss": 1.0994,
+      "step": 1158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.674898456539399e-06,
+      "loss": 1.1683,
+      "step": 1159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.671080422420797e-06,
+      "loss": 1.1745,
+      "step": 1160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.667262388302194e-06,
+      "loss": 1.2423,
+      "step": 1161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.663444354183591e-06,
+      "loss": 1.2724,
+      "step": 1162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.659626320064988e-06,
+      "loss": 1.2282,
+      "step": 1163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.655808285946385e-06,
+      "loss": 1.1862,
+      "step": 1164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.651990251827782e-06,
+      "loss": 1.0723,
+      "step": 1165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.648172217709179e-06,
+      "loss": 1.1658,
+      "step": 1166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.644354183590577e-06,
+      "loss": 1.219,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.640536149471974e-06,
+      "loss": 1.1063,
+      "step": 1168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.636718115353371e-06,
+      "loss": 1.126,
+      "step": 1169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.63290008123477e-06,
+      "loss": 1.2756,
+      "step": 1170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.629082047116167e-06,
+      "loss": 1.4547,
+      "step": 1171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.625264012997564e-06,
+      "loss": 1.3717,
+      "step": 1172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.62144597887896e-06,
+      "loss": 1.0853,
+      "step": 1173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.617627944760358e-06,
+      "loss": 1.1388,
+      "step": 1174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.613809910641754e-06,
+      "loss": 1.1957,
+      "step": 1175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.609991876523151e-06,
+      "loss": 1.0138,
+      "step": 1176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.60617384240455e-06,
+      "loss": 1.1351,
+      "step": 1177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.602355808285947e-06,
+      "loss": 1.1172,
+      "step": 1178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.598537774167344e-06,
+      "loss": 1.309,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.594719740048742e-06,
+      "loss": 1.2848,
+      "step": 1180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.59090170593014e-06,
+      "loss": 1.2518,
+      "step": 1181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.587083671811534e-06,
+      "loss": 1.1677,
+      "step": 1182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.583265637692933e-06,
+      "loss": 1.2421,
+      "step": 1183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.57944760357433e-06,
+      "loss": 1.1893,
+      "step": 1184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.575629569455727e-06,
+      "loss": 1.3374,
+      "step": 1185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.571811535337125e-06,
+      "loss": 1.5525,
+      "step": 1186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.567993501218522e-06,
+      "loss": 1.3313,
+      "step": 1187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.56417546709992e-06,
+      "loss": 1.2026,
+      "step": 1188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.560357432981316e-06,
+      "loss": 1.2826,
+      "step": 1189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.556539398862715e-06,
+      "loss": 1.169,
+      "step": 1190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.55272136474411e-06,
+      "loss": 1.0221,
+      "step": 1191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.548903330625507e-06,
+      "loss": 1.1002,
+      "step": 1192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.545085296506906e-06,
+      "loss": 1.5505,
+      "step": 1193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.541267262388302e-06,
+      "loss": 1.5081,
+      "step": 1194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.5374492282697e-06,
+      "loss": 1.3284,
+      "step": 1195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.533631194151098e-06,
+      "loss": 1.163,
+      "step": 1196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.529813160032495e-06,
+      "loss": 1.1169,
+      "step": 1197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.525995125913892e-06,
+      "loss": 1.3503,
+      "step": 1198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.522177091795289e-06,
+      "loss": 1.0617,
+      "step": 1199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.518359057676686e-06,
+      "loss": 1.3156,
+      "step": 1200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.514541023558082e-06,
+      "loss": 1.1821,
+      "step": 1201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.51072298943948e-06,
+      "loss": 1.3296,
+      "step": 1202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.506904955320878e-06,
+      "loss": 1.4198,
+      "step": 1203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.503086921202275e-06,
+      "loss": 1.3094,
+      "step": 1204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.499268887083672e-06,
+      "loss": 0.9187,
+      "step": 1205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.49545085296507e-06,
+      "loss": 0.9563,
+      "step": 1206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.491632818846467e-06,
+      "loss": 1.2326,
+      "step": 1207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.487814784727864e-06,
+      "loss": 1.2181,
+      "step": 1208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.483996750609261e-06,
+      "loss": 1.1023,
+      "step": 1209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.480178716490658e-06,
+      "loss": 1.3816,
+      "step": 1210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.476360682372055e-06,
+      "loss": 1.526,
+      "step": 1211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.472542648253453e-06,
+      "loss": 0.991,
+      "step": 1212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.46872461413485e-06,
+      "loss": 1.3934,
+      "step": 1213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.464906580016247e-06,
+      "loss": 1.1254,
+      "step": 1214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.461088545897644e-06,
+      "loss": 1.2407,
+      "step": 1215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.457270511779043e-06,
+      "loss": 1.0393,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.45345247766044e-06,
+      "loss": 1.1736,
+      "step": 1217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.449634443541835e-06,
+      "loss": 0.9456,
+      "step": 1218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.445816409423234e-06,
+      "loss": 1.3603,
+      "step": 1219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.44199837530463e-06,
+      "loss": 1.2319,
+      "step": 1220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.438180341186027e-06,
+      "loss": 1.2531,
+      "step": 1221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.434362307067426e-06,
+      "loss": 1.1609,
+      "step": 1222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.430544272948823e-06,
+      "loss": 1.2971,
+      "step": 1223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.42672623883022e-06,
+      "loss": 0.9689,
+      "step": 1224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.422908204711617e-06,
+      "loss": 1.2177,
+      "step": 1225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.419090170593015e-06,
+      "loss": 1.4716,
+      "step": 1226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.41527213647441e-06,
+      "loss": 1.2168,
+      "step": 1227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.411454102355807e-06,
+      "loss": 0.9961,
+      "step": 1228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.407636068237206e-06,
+      "loss": 1.4192,
+      "step": 1229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.403818034118603e-06,
+      "loss": 1.4503,
+      "step": 1230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.4e-06,
+      "loss": 1.2995,
+      "step": 1231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.396181965881398e-06,
+      "loss": 1.4374,
+      "step": 1232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.392363931762795e-06,
+      "loss": 1.1279,
+      "step": 1233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.388545897644192e-06,
+      "loss": 1.421,
+      "step": 1234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.38472786352559e-06,
+      "loss": 1.0035,
+      "step": 1235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.380909829406986e-06,
+      "loss": 1.5774,
+      "step": 1236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.377091795288383e-06,
+      "loss": 1.3653,
+      "step": 1237
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.37327376116978e-06,
+      "loss": 1.4478,
+      "step": 1238
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.369455727051178e-06,
+      "loss": 1.3553,
+      "step": 1239
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.365637692932575e-06,
+      "loss": 1.0728,
+      "step": 1240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.361819658813972e-06,
+      "loss": 1.4196,
+      "step": 1241
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.35800162469537e-06,
+      "loss": 1.5148,
+      "step": 1242
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.354183590576768e-06,
+      "loss": 1.1548,
+      "step": 1243
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.350365556458165e-06,
+      "loss": 1.4328,
+      "step": 1244
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.346547522339562e-06,
+      "loss": 1.0754,
+      "step": 1245
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.342729488220958e-06,
+      "loss": 1.4977,
+      "step": 1246
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.338911454102355e-06,
+      "loss": 1.3043,
+      "step": 1247
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.335093419983754e-06,
+      "loss": 1.0641,
+      "step": 1248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.331275385865151e-06,
+      "loss": 1.1847,
+      "step": 1249
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.327457351746548e-06,
+      "loss": 1.0785,
+      "step": 1250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.323639317627945e-06,
+      "loss": 1.6394,
+      "step": 1251
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.319821283509343e-06,
+      "loss": 1.0891,
+      "step": 1252
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.31600324939074e-06,
+      "loss": 0.9547,
+      "step": 1253
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.312185215272135e-06,
+      "loss": 1.314,
+      "step": 1254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.308367181153534e-06,
+      "loss": 1.307,
+      "step": 1255
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.304549147034931e-06,
+      "loss": 1.3619,
+      "step": 1256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.300731112916328e-06,
+      "loss": 1.1189,
+      "step": 1257
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.296913078797726e-06,
+      "loss": 1.1426,
+      "step": 1258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.293095044679123e-06,
+      "loss": 1.3617,
+      "step": 1259
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.28927701056052e-06,
+      "loss": 1.2545,
+      "step": 1260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.285458976441919e-06,
+      "loss": 1.2997,
+      "step": 1261
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.281640942323316e-06,
+      "loss": 1.7025,
+      "step": 1262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.277822908204711e-06,
+      "loss": 1.1386,
+      "step": 1263
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.274004874086108e-06,
+      "loss": 1.2196,
+      "step": 1264
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.270186839967506e-06,
+      "loss": 1.2754,
+      "step": 1265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.266368805848903e-06,
+      "loss": 1.2758,
+      "step": 1266
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.2625507717303e-06,
+      "loss": 1.4984,
+      "step": 1267
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.258732737611699e-06,
+      "loss": 1.1959,
+      "step": 1268
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.254914703493096e-06,
+      "loss": 1.3381,
+      "step": 1269
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.251096669374493e-06,
+      "loss": 1.224,
+      "step": 1270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.247278635255891e-06,
+      "loss": 1.1716,
+      "step": 1271
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.243460601137286e-06,
+      "loss": 1.2526,
+      "step": 1272
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.239642567018683e-06,
+      "loss": 1.1345,
+      "step": 1273
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.235824532900082e-06,
+      "loss": 0.9847,
+      "step": 1274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.232006498781479e-06,
+      "loss": 1.323,
+      "step": 1275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.228188464662876e-06,
+      "loss": 1.2271,
+      "step": 1276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.224370430544273e-06,
+      "loss": 0.9824,
+      "step": 1277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.220552396425671e-06,
+      "loss": 1.3585,
+      "step": 1278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.216734362307068e-06,
+      "loss": 1.1128,
+      "step": 1279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.212916328188465e-06,
+      "loss": 1.1019,
+      "step": 1280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.209098294069862e-06,
+      "loss": 1.2307,
+      "step": 1281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.205280259951259e-06,
+      "loss": 1.1246,
+      "step": 1282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.201462225832656e-06,
+      "loss": 0.9906,
+      "step": 1283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.197644191714054e-06,
+      "loss": 1.169,
+      "step": 1284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.193826157595451e-06,
+      "loss": 1.1756,
+      "step": 1285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.190008123476848e-06,
+      "loss": 1.4146,
+      "step": 1286
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.186190089358245e-06,
+      "loss": 1.2447,
+      "step": 1287
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.182372055239644e-06,
+      "loss": 1.3621,
+      "step": 1288
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.17855402112104e-06,
+      "loss": 1.4531,
+      "step": 1289
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.174735987002438e-06,
+      "loss": 1.139,
+      "step": 1290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.170917952883834e-06,
+      "loss": 1.0144,
+      "step": 1291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.167099918765231e-06,
+      "loss": 1.0287,
+      "step": 1292
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.163281884646628e-06,
+      "loss": 1.1961,
+      "step": 1293
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.159463850528027e-06,
+      "loss": 1.5825,
+      "step": 1294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.155645816409424e-06,
+      "loss": 1.4047,
+      "step": 1295
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.15182778229082e-06,
+      "loss": 1.0546,
+      "step": 1296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.14800974817222e-06,
+      "loss": 1.3601,
+      "step": 1297
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.144191714053616e-06,
+      "loss": 1.2665,
+      "step": 1298
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.140373679935013e-06,
+      "loss": 1.1563,
+      "step": 1299
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.136555645816408e-06,
+      "loss": 1.1204,
+      "step": 1300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.132737611697807e-06,
+      "loss": 1.2561,
+      "step": 1301
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.128919577579204e-06,
+      "loss": 1.3249,
+      "step": 1302
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.1251015434606e-06,
+      "loss": 1.1101,
+      "step": 1303
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.121283509342e-06,
+      "loss": 1.3898,
+      "step": 1304
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.117465475223396e-06,
+      "loss": 1.1611,
+      "step": 1305
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.113647441104793e-06,
+      "loss": 1.1039,
+      "step": 1306
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.109829406986192e-06,
+      "loss": 1.3518,
+      "step": 1307
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.106011372867587e-06,
+      "loss": 1.2205,
+      "step": 1308
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.102193338748984e-06,
+      "loss": 1.0865,
+      "step": 1309
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.098375304630382e-06,
+      "loss": 1.2775,
+      "step": 1310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.09455727051178e-06,
+      "loss": 0.9097,
+      "step": 1311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.090739236393176e-06,
+      "loss": 1.2786,
+      "step": 1312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.086921202274573e-06,
+      "loss": 1.3156,
+      "step": 1313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.083103168155972e-06,
+      "loss": 1.6502,
+      "step": 1314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.079285134037369e-06,
+      "loss": 1.1631,
+      "step": 1315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.075467099918766e-06,
+      "loss": 1.0827,
+      "step": 1316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.071649065800162e-06,
+      "loss": 1.0456,
+      "step": 1317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.06783103168156e-06,
+      "loss": 1.5506,
+      "step": 1318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.064012997562956e-06,
+      "loss": 1.1369,
+      "step": 1319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.060194963444355e-06,
+      "loss": 1.1464,
+      "step": 1320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.056376929325752e-06,
+      "loss": 1.3474,
+      "step": 1321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.052558895207149e-06,
+      "loss": 1.1399,
+      "step": 1322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.048740861088547e-06,
+      "loss": 1.1841,
+      "step": 1323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.044922826969944e-06,
+      "loss": 1.415,
+      "step": 1324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.041104792851341e-06,
+      "loss": 1.5333,
+      "step": 1325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.037286758732738e-06,
+      "loss": 1.3441,
+      "step": 1326
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.033468724614135e-06,
+      "loss": 1.3183,
+      "step": 1327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.029650690495532e-06,
+      "loss": 1.6283,
+      "step": 1328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.025832656376929e-06,
+      "loss": 1.138,
+      "step": 1329
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.022014622258327e-06,
+      "loss": 1.2685,
+      "step": 1330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.018196588139724e-06,
+      "loss": 1.2918,
+      "step": 1331
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.014378554021121e-06,
+      "loss": 1.215,
+      "step": 1332
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.01056051990252e-06,
+      "loss": 1.0487,
+      "step": 1333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.006742485783917e-06,
+      "loss": 1.2569,
+      "step": 1334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.002924451665314e-06,
+      "loss": 1.133,
+      "step": 1335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.99910641754671e-06,
+      "loss": 1.1822,
+      "step": 1336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.995288383428107e-06,
+      "loss": 1.1522,
+      "step": 1337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.991470349309504e-06,
+      "loss": 1.0733,
+      "step": 1338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.987652315190901e-06,
+      "loss": 1.0347,
+      "step": 1339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.9838342810723e-06,
+      "loss": 1.5016,
+      "step": 1340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.980016246953697e-06,
+      "loss": 1.5628,
+      "step": 1341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.976198212835094e-06,
+      "loss": 1.3989,
+      "step": 1342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.972380178716492e-06,
+      "loss": 1.1647,
+      "step": 1343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.968562144597889e-06,
+      "loss": 1.3605,
+      "step": 1344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.964744110479284e-06,
+      "loss": 1.3449,
+      "step": 1345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.960926076360683e-06,
+      "loss": 1.3798,
+      "step": 1346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.95710804224208e-06,
+      "loss": 1.5248,
+      "step": 1347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.953290008123477e-06,
+      "loss": 1.2141,
+      "step": 1348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.949471974004875e-06,
+      "loss": 1.2362,
+      "step": 1349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.945653939886272e-06,
+      "loss": 1.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.941835905767669e-06,
+      "loss": 0.9853,
+      "step": 1351
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.938017871649066e-06,
+      "loss": 1.3012,
+      "step": 1352
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.934199837530465e-06,
+      "loss": 1.0875,
+      "step": 1353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.93038180341186e-06,
+      "loss": 1.07,
+      "step": 1354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.926563769293257e-06,
+      "loss": 1.1205,
+      "step": 1355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.922745735174655e-06,
+      "loss": 1.4332,
+      "step": 1356
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.918927701056052e-06,
+      "loss": 1.229,
+      "step": 1357
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.91510966693745e-06,
+      "loss": 1.1664,
+      "step": 1358
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.911291632818848e-06,
+      "loss": 1.1774,
+      "step": 1359
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.907473598700245e-06,
+      "loss": 1.4849,
+      "step": 1360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.903655564581642e-06,
+      "loss": 1.3488,
+      "step": 1361
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.899837530463038e-06,
+      "loss": 1.4134,
+      "step": 1362
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.896019496344435e-06,
+      "loss": 1.248,
+      "step": 1363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.892201462225832e-06,
+      "loss": 1.7011,
+      "step": 1364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.88838342810723e-06,
+      "loss": 0.97,
+      "step": 1365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.884565393988628e-06,
+      "loss": 1.1684,
+      "step": 1366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.880747359870025e-06,
+      "loss": 1.2153,
+      "step": 1367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.876929325751422e-06,
+      "loss": 1.2356,
+      "step": 1368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.87311129163282e-06,
+      "loss": 1.008,
+      "step": 1369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.869293257514217e-06,
+      "loss": 1.1836,
+      "step": 1370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.865475223395614e-06,
+      "loss": 1.3309,
+      "step": 1371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.861657189277011e-06,
+      "loss": 1.2051,
+      "step": 1372
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.857839155158408e-06,
+      "loss": 1.3267,
+      "step": 1373
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.854021121039805e-06,
+      "loss": 1.2279,
+      "step": 1374
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.850203086921202e-06,
+      "loss": 1.2553,
+      "step": 1375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.8463850528026e-06,
+      "loss": 1.1256,
+      "step": 1376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.842567018683997e-06,
+      "loss": 1.2699,
+      "step": 1377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.838748984565394e-06,
+      "loss": 1.2695,
+      "step": 1378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.834930950446793e-06,
+      "loss": 0.9318,
+      "step": 1379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.83111291632819e-06,
+      "loss": 1.1653,
+      "step": 1380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.827294882209585e-06,
+      "loss": 1.1585,
+      "step": 1381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.823476848090983e-06,
+      "loss": 1.2721,
+      "step": 1382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.81965881397238e-06,
+      "loss": 1.3655,
+      "step": 1383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.815840779853777e-06,
+      "loss": 1.1801,
+      "step": 1384
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.812022745735176e-06,
+      "loss": 0.9812,
+      "step": 1385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.808204711616573e-06,
+      "loss": 1.2466,
+      "step": 1386
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.80438667749797e-06,
+      "loss": 0.9495,
+      "step": 1387
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.800568643379367e-06,
+      "loss": 1.3519,
+      "step": 1388
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.796750609260765e-06,
+      "loss": 1.1982,
+      "step": 1389
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.79293257514216e-06,
+      "loss": 1.0683,
+      "step": 1390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.789114541023557e-06,
+      "loss": 1.1893,
+      "step": 1391
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.785296506904956e-06,
+      "loss": 1.5005,
+      "step": 1392
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.781478472786353e-06,
+      "loss": 1.4345,
+      "step": 1393
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.77766043866775e-06,
+      "loss": 1.1088,
+      "step": 1394
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.773842404549148e-06,
+      "loss": 1.0415,
+      "step": 1395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.770024370430545e-06,
+      "loss": 0.8846,
+      "step": 1396
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.766206336311942e-06,
+      "loss": 0.9664,
+      "step": 1397
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.76238830219334e-06,
+      "loss": 2.087,
+      "step": 1398
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.758570268074736e-06,
+      "loss": 1.2369,
+      "step": 1399
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.754752233956133e-06,
+      "loss": 1.596,
+      "step": 1400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.75093419983753e-06,
+      "loss": 1.2559,
+      "step": 1401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.747116165718928e-06,
+      "loss": 1.0841,
+      "step": 1402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.743298131600325e-06,
+      "loss": 1.1794,
+      "step": 1403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.739480097481722e-06,
+      "loss": 1.1964,
+      "step": 1404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.73566206336312e-06,
+      "loss": 1.029,
+      "step": 1405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.731844029244518e-06,
+      "loss": 1.1683,
+      "step": 1406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.728025995125914e-06,
+      "loss": 1.0366,
+      "step": 1407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.724207961007311e-06,
+      "loss": 0.9699,
+      "step": 1408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.720389926888708e-06,
+      "loss": 1.1729,
+      "step": 1409
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.716571892770105e-06,
+      "loss": 1.4872,
+      "step": 1410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.712753858651504e-06,
+      "loss": 1.1294,
+      "step": 1411
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.7089358245329e-06,
+      "loss": 1.0249,
+      "step": 1412
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.705117790414298e-06,
+      "loss": 1.106,
+      "step": 1413
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.701299756295695e-06,
+      "loss": 1.0279,
+      "step": 1414
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.697481722177093e-06,
+      "loss": 1.2714,
+      "step": 1415
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.69366368805849e-06,
+      "loss": 1.1467,
+      "step": 1416
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.689845653939885e-06,
+      "loss": 1.1949,
+      "step": 1417
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.686027619821284e-06,
+      "loss": 1.3733,
+      "step": 1418
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.68220958570268e-06,
+      "loss": 1.465,
+      "step": 1419
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.678391551584078e-06,
+      "loss": 1.3019,
+      "step": 1420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.674573517465476e-06,
+      "loss": 1.223,
+      "step": 1421
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.670755483346873e-06,
+      "loss": 1.5889,
+      "step": 1422
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.66693744922827e-06,
+      "loss": 1.478,
+      "step": 1423
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.663119415109667e-06,
+      "loss": 1.3251,
+      "step": 1424
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.659301380991066e-06,
+      "loss": 1.4523,
+      "step": 1425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.65548334687246e-06,
+      "loss": 1.368,
+      "step": 1426
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.651665312753858e-06,
+      "loss": 1.4663,
+      "step": 1427
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.647847278635256e-06,
+      "loss": 1.4585,
+      "step": 1428
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.644029244516653e-06,
+      "loss": 1.5435,
+      "step": 1429
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.64021121039805e-06,
+      "loss": 1.3867,
+      "step": 1430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.636393176279449e-06,
+      "loss": 1.4274,
+      "step": 1431
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.632575142160846e-06,
+      "loss": 0.9814,
+      "step": 1432
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.628757108042243e-06,
+      "loss": 1.1046,
+      "step": 1433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.624939073923641e-06,
+      "loss": 1.0657,
+      "step": 1434
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.621121039805036e-06,
+      "loss": 0.9423,
+      "step": 1435
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.617303005686433e-06,
+      "loss": 1.3726,
+      "step": 1436
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.61348497156783e-06,
+      "loss": 1.1973,
+      "step": 1437
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.609666937449229e-06,
+      "loss": 1.1481,
+      "step": 1438
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.605848903330626e-06,
+      "loss": 1.2406,
+      "step": 1439
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.602030869212023e-06,
+      "loss": 1.1608,
+      "step": 1440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.598212835093421e-06,
+      "loss": 1.3295,
+      "step": 1441
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.594394800974818e-06,
+      "loss": 1.0101,
+      "step": 1442
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.590576766856215e-06,
+      "loss": 1.4845,
+      "step": 1443
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.586758732737612e-06,
+      "loss": 0.9978,
+      "step": 1444
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.582940698619009e-06,
+      "loss": 1.3235,
+      "step": 1445
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.579122664500406e-06,
+      "loss": 1.3332,
+      "step": 1446
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.575304630381804e-06,
+      "loss": 1.4228,
+      "step": 1447
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.571486596263201e-06,
+      "loss": 1.4412,
+      "step": 1448
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.567668562144598e-06,
+      "loss": 1.1972,
+      "step": 1449
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.563850528025995e-06,
+      "loss": 1.2094,
+      "step": 1450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.560032493907394e-06,
+      "loss": 1.1787,
+      "step": 1451
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.55621445978879e-06,
+      "loss": 1.1603,
+      "step": 1452
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.552396425670187e-06,
+      "loss": 0.9799,
+      "step": 1453
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.548578391551584e-06,
+      "loss": 1.0531,
+      "step": 1454
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.544760357432981e-06,
+      "loss": 1.1855,
+      "step": 1455
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.540942323314378e-06,
+      "loss": 1.169,
+      "step": 1456
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.537124289195777e-06,
+      "loss": 1.4206,
+      "step": 1457
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.533306255077174e-06,
+      "loss": 1.2123,
+      "step": 1458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.52948822095857e-06,
+      "loss": 1.4194,
+      "step": 1459
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.525670186839969e-06,
+      "loss": 1.1016,
+      "step": 1460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.521852152721366e-06,
+      "loss": 1.0128,
+      "step": 1461
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.518034118602761e-06,
+      "loss": 1.2266,
+      "step": 1462
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.514216084484158e-06,
+      "loss": 1.1276,
+      "step": 1463
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.510398050365557e-06,
+      "loss": 1.0834,
+      "step": 1464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.506580016246954e-06,
+      "loss": 1.0554,
+      "step": 1465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.50276198212835e-06,
+      "loss": 1.4328,
+      "step": 1466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.49894394800975e-06,
+      "loss": 1.5663,
+      "step": 1467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.495125913891146e-06,
+      "loss": 1.2021,
+      "step": 1468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.491307879772543e-06,
+      "loss": 1.1385,
+      "step": 1469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.487489845653942e-06,
+      "loss": 1.0667,
+      "step": 1470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.483671811535337e-06,
+      "loss": 1.4587,
+      "step": 1471
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.479853777416734e-06,
+      "loss": 1.459,
+      "step": 1472
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.476035743298132e-06,
+      "loss": 1.0915,
+      "step": 1473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.47221770917953e-06,
+      "loss": 1.4452,
+      "step": 1474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.468399675060926e-06,
+      "loss": 1.3287,
+      "step": 1475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.464581640942323e-06,
+      "loss": 1.1695,
+      "step": 1476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.460763606823722e-06,
+      "loss": 1.563,
+      "step": 1477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.456945572705119e-06,
+      "loss": 1.2846,
+      "step": 1478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.453127538586515e-06,
+      "loss": 1.2418,
+      "step": 1479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.449309504467912e-06,
+      "loss": 1.087,
+      "step": 1480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.44549147034931e-06,
+      "loss": 1.2187,
+      "step": 1481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.441673436230706e-06,
+      "loss": 1.147,
+      "step": 1482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.437855402112105e-06,
+      "loss": 1.3729,
+      "step": 1483
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.434037367993502e-06,
+      "loss": 1.0043,
+      "step": 1484
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.430219333874899e-06,
+      "loss": 1.3682,
+      "step": 1485
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.426401299756295e-06,
+      "loss": 0.8864,
+      "step": 1486
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.422583265637694e-06,
+      "loss": 1.3209,
+      "step": 1487
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.418765231519091e-06,
+      "loss": 1.0166,
+      "step": 1488
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.414947197400488e-06,
+      "loss": 1.2842,
+      "step": 1489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.411129163281885e-06,
+      "loss": 1.2182,
+      "step": 1490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.407311129163282e-06,
+      "loss": 1.1595,
+      "step": 1491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.403493095044679e-06,
+      "loss": 1.0129,
+      "step": 1492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.399675060926077e-06,
+      "loss": 1.1567,
+      "step": 1493
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.395857026807474e-06,
+      "loss": 1.0968,
+      "step": 1494
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.392038992688871e-06,
+      "loss": 1.1369,
+      "step": 1495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.38822095857027e-06,
+      "loss": 0.8953,
+      "step": 1496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.384402924451667e-06,
+      "loss": 1.5419,
+      "step": 1497
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.380584890333063e-06,
+      "loss": 1.3553,
+      "step": 1498
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.376766856214459e-06,
+      "loss": 1.276,
+      "step": 1499
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.372948822095857e-06,
+      "loss": 1.1088,
+      "step": 1500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.369130787977254e-06,
+      "loss": 1.144,
+      "step": 1501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.365312753858651e-06,
+      "loss": 1.0651,
+      "step": 1502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.36149471974005e-06,
+      "loss": 1.1147,
+      "step": 1503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.357676685621447e-06,
+      "loss": 1.2024,
+      "step": 1504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.353858651502843e-06,
+      "loss": 1.033,
+      "step": 1505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.350040617384242e-06,
+      "loss": 1.37,
+      "step": 1506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.346222583265639e-06,
+      "loss": 1.2436,
+      "step": 1507
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.342404549147034e-06,
+      "loss": 1.3604,
+      "step": 1508
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.338586515028433e-06,
+      "loss": 1.0774,
+      "step": 1509
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.33476848090983e-06,
+      "loss": 1.5044,
+      "step": 1510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.330950446791227e-06,
+      "loss": 1.2274,
+      "step": 1511
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.327132412672623e-06,
+      "loss": 1.0815,
+      "step": 1512
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.323314378554022e-06,
+      "loss": 1.3802,
+      "step": 1513
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.319496344435419e-06,
+      "loss": 1.2911,
+      "step": 1514
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.315678310316816e-06,
+      "loss": 1.5323,
+      "step": 1515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.311860276198213e-06,
+      "loss": 1.3911,
+      "step": 1516
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.30804224207961e-06,
+      "loss": 1.3154,
+      "step": 1517
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.304224207961007e-06,
+      "loss": 1.1335,
+      "step": 1518
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.300406173842405e-06,
+      "loss": 1.3069,
+      "step": 1519
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.296588139723802e-06,
+      "loss": 1.2644,
+      "step": 1520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.292770105605199e-06,
+      "loss": 1.5374,
+      "step": 1521
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.288952071486598e-06,
+      "loss": 1.4099,
+      "step": 1522
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.285134037367995e-06,
+      "loss": 1.2048,
+      "step": 1523
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.281316003249391e-06,
+      "loss": 1.1628,
+      "step": 1524
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.277497969130788e-06,
+      "loss": 1.3232,
+      "step": 1525
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.273679935012185e-06,
+      "loss": 1.1057,
+      "step": 1526
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.269861900893582e-06,
+      "loss": 1.0896,
+      "step": 1527
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.266043866774979e-06,
+      "loss": 1.6067,
+      "step": 1528
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.262225832656378e-06,
+      "loss": 1.2691,
+      "step": 1529
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.258407798537775e-06,
+      "loss": 1.3431,
+      "step": 1530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.254589764419171e-06,
+      "loss": 1.1946,
+      "step": 1531
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.25077173030057e-06,
+      "loss": 1.6415,
+      "step": 1532
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.246953696181967e-06,
+      "loss": 1.1838,
+      "step": 1533
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.243135662063364e-06,
+      "loss": 1.227,
+      "step": 1534
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.23931762794476e-06,
+      "loss": 1.0083,
+      "step": 1535
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.235499593826158e-06,
+      "loss": 1.3868,
+      "step": 1536
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.231681559707555e-06,
+      "loss": 0.931,
+      "step": 1537
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.227863525588951e-06,
+      "loss": 1.3326,
+      "step": 1538
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.22404549147035e-06,
+      "loss": 1.299,
+      "step": 1539
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.220227457351747e-06,
+      "loss": 1.4396,
+      "step": 1540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.216409423233144e-06,
+      "loss": 1.1092,
+      "step": 1541
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.212591389114543e-06,
+      "loss": 1.1191,
+      "step": 1542
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.20877335499594e-06,
+      "loss": 1.3849,
+      "step": 1543
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.204955320877335e-06,
+      "loss": 1.1427,
+      "step": 1544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.201137286758733e-06,
+      "loss": 1.1167,
+      "step": 1545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.19731925264013e-06,
+      "loss": 1.471,
+      "step": 1546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.193501218521527e-06,
+      "loss": 1.2282,
+      "step": 1547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.189683184402926e-06,
+      "loss": 1.1606,
+      "step": 1548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.185865150284323e-06,
+      "loss": 1.3297,
+      "step": 1549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.18204711616572e-06,
+      "loss": 1.2887,
+      "step": 1550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.178229082047116e-06,
+      "loss": 1.0192,
+      "step": 1551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.174411047928515e-06,
+      "loss": 1.2355,
+      "step": 1552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.17059301380991e-06,
+      "loss": 1.202,
+      "step": 1553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.166774979691307e-06,
+      "loss": 1.1864,
+      "step": 1554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.162956945572706e-06,
+      "loss": 1.2496,
+      "step": 1555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.159138911454103e-06,
+      "loss": 0.9648,
+      "step": 1556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.1553208773355e-06,
+      "loss": 0.9881,
+      "step": 1557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.151502843216898e-06,
+      "loss": 1.2554,
+      "step": 1558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.147684809098295e-06,
+      "loss": 1.2249,
+      "step": 1559
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.143866774979692e-06,
+      "loss": 1.332,
+      "step": 1560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.140048740861089e-06,
+      "loss": 1.0104,
+      "step": 1561
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.136230706742486e-06,
+      "loss": 1.1077,
+      "step": 1562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.132412672623883e-06,
+      "loss": 1.1455,
+      "step": 1563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.12859463850528e-06,
+      "loss": 1.6546,
+      "step": 1564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.124776604386678e-06,
+      "loss": 1.2246,
+      "step": 1565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.120958570268075e-06,
+      "loss": 1.1715,
+      "step": 1566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.117140536149472e-06,
+      "loss": 0.962,
+      "step": 1567
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.11332250203087e-06,
+      "loss": 1.1199,
+      "step": 1568
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.109504467912267e-06,
+      "loss": 1.4921,
+      "step": 1569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.105686433793664e-06,
+      "loss": 1.3232,
+      "step": 1570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.101868399675061e-06,
+      "loss": 1.5596,
+      "step": 1571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.098050365556458e-06,
+      "loss": 1.4387,
+      "step": 1572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.094232331437855e-06,
+      "loss": 1.1671,
+      "step": 1573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.090414297319252e-06,
+      "loss": 0.8515,
+      "step": 1574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.08659626320065e-06,
+      "loss": 1.3172,
+      "step": 1575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.082778229082047e-06,
+      "loss": 1.2373,
+      "step": 1576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.078960194963444e-06,
+      "loss": 1.2374,
+      "step": 1577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.075142160844843e-06,
+      "loss": 1.0602,
+      "step": 1578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.07132412672624e-06,
+      "loss": 1.0684,
+      "step": 1579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.067506092607635e-06,
+      "loss": 1.2363,
+      "step": 1580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.063688058489034e-06,
+      "loss": 1.0497,
+      "step": 1581
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.05987002437043e-06,
+      "loss": 1.6015,
+      "step": 1582
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.056051990251828e-06,
+      "loss": 1.3373,
+      "step": 1583
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.052233956133226e-06,
+      "loss": 1.3406,
+      "step": 1584
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.048415922014623e-06,
+      "loss": 1.4491,
+      "step": 1585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.04459788789602e-06,
+      "loss": 1.3498,
+      "step": 1586
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.040779853777417e-06,
+      "loss": 1.1876,
+      "step": 1587
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.036961819658815e-06,
+      "loss": 1.3227,
+      "step": 1588
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.03314378554021e-06,
+      "loss": 1.0699,
+      "step": 1589
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.029325751421608e-06,
+      "loss": 1.0671,
+      "step": 1590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.025507717303006e-06,
+      "loss": 1.4481,
+      "step": 1591
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.021689683184403e-06,
+      "loss": 0.9665,
+      "step": 1592
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.0178716490658e-06,
+      "loss": 0.9403,
+      "step": 1593
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.014053614947199e-06,
+      "loss": 1.326,
+      "step": 1594
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.010235580828595e-06,
+      "loss": 1.2875,
+      "step": 1595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.006417546709992e-06,
+      "loss": 1.1379,
+      "step": 1596
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.002599512591391e-06,
+      "loss": 1.2929,
+      "step": 1597
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.998781478472786e-06,
+      "loss": 1.2793,
+      "step": 1598
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.994963444354183e-06,
+      "loss": 1.1433,
+      "step": 1599
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.99114541023558e-06,
+      "loss": 1.1176,
+      "step": 1600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.987327376116979e-06,
+      "loss": 1.1404,
+      "step": 1601
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.983509341998375e-06,
+      "loss": 1.3162,
+      "step": 1602
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.979691307879772e-06,
+      "loss": 1.0017,
+      "step": 1603
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.975873273761171e-06,
+      "loss": 1.4147,
+      "step": 1604
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.972055239642568e-06,
+      "loss": 1.5927,
+      "step": 1605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.968237205523965e-06,
+      "loss": 1.3603,
+      "step": 1606
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.964419171405362e-06,
+      "loss": 1.0814,
+      "step": 1607
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.960601137286759e-06,
+      "loss": 1.2822,
+      "step": 1608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.956783103168156e-06,
+      "loss": 1.1045,
+      "step": 1609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.952965069049554e-06,
+      "loss": 1.3176,
+      "step": 1610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.949147034930951e-06,
+      "loss": 1.1113,
+      "step": 1611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.945329000812348e-06,
+      "loss": 1.0633,
+      "step": 1612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.941510966693745e-06,
+      "loss": 1.5352,
+      "step": 1613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.937692932575143e-06,
+      "loss": 1.2789,
+      "step": 1614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.93387489845654e-06,
+      "loss": 1.1051,
+      "step": 1615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.930056864337936e-06,
+      "loss": 1.6237,
+      "step": 1616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.926238830219334e-06,
+      "loss": 1.4639,
+      "step": 1617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.922420796100731e-06,
+      "loss": 1.3349,
+      "step": 1618
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.918602761982128e-06,
+      "loss": 1.1651,
+      "step": 1619
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.914784727863527e-06,
+      "loss": 1.2457,
+      "step": 1620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.910966693744923e-06,
+      "loss": 1.2838,
+      "step": 1621
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.90714865962632e-06,
+      "loss": 1.1615,
+      "step": 1622
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.903330625507717e-06,
+      "loss": 1.0631,
+      "step": 1623
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.899512591389116e-06,
+      "loss": 1.0957,
+      "step": 1624
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.895694557270511e-06,
+      "loss": 1.2597,
+      "step": 1625
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.891876523151908e-06,
+      "loss": 1.4333,
+      "step": 1626
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.888058489033307e-06,
+      "loss": 1.1186,
+      "step": 1627
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.884240454914704e-06,
+      "loss": 1.0334,
+      "step": 1628
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.8804224207961e-06,
+      "loss": 1.2422,
+      "step": 1629
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.876604386677499e-06,
+      "loss": 1.1046,
+      "step": 1630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.872786352558896e-06,
+      "loss": 1.0343,
+      "step": 1631
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.868968318440293e-06,
+      "loss": 1.0609,
+      "step": 1632
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.865150284321691e-06,
+      "loss": 1.2536,
+      "step": 1633
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.861332250203087e-06,
+      "loss": 1.3595,
+      "step": 1634
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.857514216084484e-06,
+      "loss": 1.2147,
+      "step": 1635
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.85369618196588e-06,
+      "loss": 1.2373,
+      "step": 1636
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.849878147847279e-06,
+      "loss": 1.1152,
+      "step": 1637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.846060113728676e-06,
+      "loss": 1.1062,
+      "step": 1638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.842242079610073e-06,
+      "loss": 1.365,
+      "step": 1639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.838424045491471e-06,
+      "loss": 1.2896,
+      "step": 1640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.834606011372868e-06,
+      "loss": 1.2065,
+      "step": 1641
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.830787977254265e-06,
+      "loss": 1.3802,
+      "step": 1642
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.826969943135662e-06,
+      "loss": 1.1649,
+      "step": 1643
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.823151909017059e-06,
+      "loss": 1.338,
+      "step": 1644
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.819333874898456e-06,
+      "loss": 1.0807,
+      "step": 1645
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.815515840779855e-06,
+      "loss": 1.2104,
+      "step": 1646
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.811697806661251e-06,
+      "loss": 1.112,
+      "step": 1647
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.807879772542648e-06,
+      "loss": 1.3352,
+      "step": 1648
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.804061738424045e-06,
+      "loss": 1.0243,
+      "step": 1649
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.800243704305444e-06,
+      "loss": 1.0938,
+      "step": 1650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.79642567018684e-06,
+      "loss": 1.1042,
+      "step": 1651
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.792607636068238e-06,
+      "loss": 1.1017,
+      "step": 1652
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.788789601949635e-06,
+      "loss": 1.1551,
+      "step": 1653
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.784971567831032e-06,
+      "loss": 1.3674,
+      "step": 1654
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.781153533712428e-06,
+      "loss": 1.1834,
+      "step": 1655
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.777335499593827e-06,
+      "loss": 1.0032,
+      "step": 1656
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.773517465475224e-06,
+      "loss": 1.3947,
+      "step": 1657
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.769699431356621e-06,
+      "loss": 1.0302,
+      "step": 1658
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.76588139723802e-06,
+      "loss": 1.1447,
+      "step": 1659
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.762063363119416e-06,
+      "loss": 1.18,
+      "step": 1660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.758245329000813e-06,
+      "loss": 1.2411,
+      "step": 1661
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.754427294882208e-06,
+      "loss": 1.2681,
+      "step": 1662
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.750609260763607e-06,
+      "loss": 1.2489,
+      "step": 1663
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.746791226645004e-06,
+      "loss": 1.1793,
+      "step": 1664
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.742973192526401e-06,
+      "loss": 1.1595,
+      "step": 1665
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.7391551584078e-06,
+      "loss": 1.2826,
+      "step": 1666
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.735337124289196e-06,
+      "loss": 1.0433,
+      "step": 1667
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.731519090170593e-06,
+      "loss": 1.3056,
+      "step": 1668
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.727701056051992e-06,
+      "loss": 1.1778,
+      "step": 1669
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.723883021933387e-06,
+      "loss": 1.4959,
+      "step": 1670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.720064987814784e-06,
+      "loss": 1.1014,
+      "step": 1671
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.716246953696183e-06,
+      "loss": 1.6187,
+      "step": 1672
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.71242891957758e-06,
+      "loss": 1.4133,
+      "step": 1673
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.708610885458976e-06,
+      "loss": 1.2647,
+      "step": 1674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.704792851340373e-06,
+      "loss": 1.3591,
+      "step": 1675
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.700974817221772e-06,
+      "loss": 1.0935,
+      "step": 1676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.697156783103169e-06,
+      "loss": 1.111,
+      "step": 1677
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.693338748984566e-06,
+      "loss": 1.3023,
+      "step": 1678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.689520714865963e-06,
+      "loss": 1.1549,
+      "step": 1679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.68570268074736e-06,
+      "loss": 1.3954,
+      "step": 1680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.681884646628756e-06,
+      "loss": 1.1491,
+      "step": 1681
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.678066612510155e-06,
+      "loss": 1.3943,
+      "step": 1682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.674248578391552e-06,
+      "loss": 1.3747,
+      "step": 1683
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.670430544272949e-06,
+      "loss": 1.342,
+      "step": 1684
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.666612510154346e-06,
+      "loss": 1.3102,
+      "step": 1685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.662794476035744e-06,
+      "loss": 1.076,
+      "step": 1686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.658976441917141e-06,
+      "loss": 1.1587,
+      "step": 1687
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.655158407798538e-06,
+      "loss": 1.3968,
+      "step": 1688
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.651340373679935e-06,
+      "loss": 1.1951,
+      "step": 1689
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.647522339561332e-06,
+      "loss": 1.2069,
+      "step": 1690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.643704305442729e-06,
+      "loss": 1.0847,
+      "step": 1691
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.639886271324128e-06,
+      "loss": 0.9607,
+      "step": 1692
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.636068237205524e-06,
+      "loss": 1.1307,
+      "step": 1693
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.632250203086921e-06,
+      "loss": 1.1246,
+      "step": 1694
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.628432168968319e-06,
+      "loss": 1.29,
+      "step": 1695
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.624614134849716e-06,
+      "loss": 1.2926,
+      "step": 1696
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.620796100731113e-06,
+      "loss": 0.9386,
+      "step": 1697
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.61697806661251e-06,
+      "loss": 1.3189,
+      "step": 1698
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.613160032493908e-06,
+      "loss": 1.1799,
+      "step": 1699
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.6093419983753044e-06,
+      "loss": 1.2992,
+      "step": 1700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.605523964256701e-06,
+      "loss": 1.3748,
+      "step": 1701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.6017059301381e-06,
+      "loss": 1.243,
+      "step": 1702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.597887896019497e-06,
+      "loss": 1.086,
+      "step": 1703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.594069861900894e-06,
+      "loss": 1.2256,
+      "step": 1704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.5902518277822915e-06,
+      "loss": 1.1325,
+      "step": 1705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.586433793663688e-06,
+      "loss": 1.2503,
+      "step": 1706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.582615759545085e-06,
+      "loss": 1.3043,
+      "step": 1707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.578797725426484e-06,
+      "loss": 1.4964,
+      "step": 1708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.57497969130788e-06,
+      "loss": 1.1571,
+      "step": 1709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.571161657189277e-06,
+      "loss": 1.234,
+      "step": 1710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.567343623070674e-06,
+      "loss": 1.0626,
+      "step": 1711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.563525588952072e-06,
+      "loss": 1.7373,
+      "step": 1712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.559707554833469e-06,
+      "loss": 1.1637,
+      "step": 1713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.555889520714865e-06,
+      "loss": 1.411,
+      "step": 1714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.552071486596264e-06,
+      "loss": 1.375,
+      "step": 1715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.548253452477661e-06,
+      "loss": 1.104,
+      "step": 1716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.544435418359058e-06,
+      "loss": 1.0757,
+      "step": 1717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.5406173842404555e-06,
+      "loss": 1.1771,
+      "step": 1718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.5367993501218524e-06,
+      "loss": 1.186,
+      "step": 1719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.532981316003249e-06,
+      "loss": 1.1974,
+      "step": 1720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.529163281884648e-06,
+      "loss": 1.1644,
+      "step": 1721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.525345247766044e-06,
+      "loss": 1.2898,
+      "step": 1722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.521527213647441e-06,
+      "loss": 1.0907,
+      "step": 1723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.517709179528838e-06,
+      "loss": 1.3536,
+      "step": 1724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.513891145410236e-06,
+      "loss": 1.5711,
+      "step": 1725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.510073111291633e-06,
+      "loss": 1.0278,
+      "step": 1726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.506255077173029e-06,
+      "loss": 1.2048,
+      "step": 1727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.502437043054428e-06,
+      "loss": 1.4462,
+      "step": 1728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.498619008935825e-06,
+      "loss": 1.3657,
+      "step": 1729
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.494800974817222e-06,
+      "loss": 1.0311,
+      "step": 1730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.4909829406986195e-06,
+      "loss": 1.0734,
+      "step": 1731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.4871649065800164e-06,
+      "loss": 1.097,
+      "step": 1732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.483346872461413e-06,
+      "loss": 1.1399,
+      "step": 1733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.479528838342812e-06,
+      "loss": 1.0247,
+      "step": 1734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.475710804224209e-06,
+      "loss": 1.0235,
+      "step": 1735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.471892770105605e-06,
+      "loss": 1.3829,
+      "step": 1736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.468074735987002e-06,
+      "loss": 1.2384,
+      "step": 1737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.4642567018684e-06,
+      "loss": 1.1507,
+      "step": 1738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.460438667749797e-06,
+      "loss": 1.2788,
+      "step": 1739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.456620633631194e-06,
+      "loss": 1.3733,
+      "step": 1740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.452802599512592e-06,
+      "loss": 1.1111,
+      "step": 1741
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.448984565393989e-06,
+      "loss": 1.3768,
+      "step": 1742
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.445166531275386e-06,
+      "loss": 1.1837,
+      "step": 1743
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.441348497156784e-06,
+      "loss": 1.1143,
+      "step": 1744
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.4375304630381804e-06,
+      "loss": 1.3183,
+      "step": 1745
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.433712428919577e-06,
+      "loss": 1.2333,
+      "step": 1746
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.429894394800976e-06,
+      "loss": 0.9954,
+      "step": 1747
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.426076360682373e-06,
+      "loss": 1.1551,
+      "step": 1748
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.42225832656377e-06,
+      "loss": 1.4931,
+      "step": 1749
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.418440292445166e-06,
+      "loss": 1.1393,
+      "step": 1750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.414622258326564e-06,
+      "loss": 1.363,
+      "step": 1751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.410804224207961e-06,
+      "loss": 1.2215,
+      "step": 1752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.406986190089358e-06,
+      "loss": 1.4448,
+      "step": 1753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.403168155970756e-06,
+      "loss": 1.1486,
+      "step": 1754
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.399350121852153e-06,
+      "loss": 1.0916,
+      "step": 1755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.39553208773355e-06,
+      "loss": 1.2512,
+      "step": 1756
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.391714053614948e-06,
+      "loss": 1.1467,
+      "step": 1757
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.387896019496345e-06,
+      "loss": 1.0408,
+      "step": 1758
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.384077985377741e-06,
+      "loss": 1.2074,
+      "step": 1759
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.380259951259138e-06,
+      "loss": 1.0239,
+      "step": 1760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.376441917140537e-06,
+      "loss": 1.17,
+      "step": 1761
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.372623883021934e-06,
+      "loss": 1.1235,
+      "step": 1762
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.368805848903331e-06,
+      "loss": 1.3512,
+      "step": 1763
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.3649878147847284e-06,
+      "loss": 1.0877,
+      "step": 1764
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.361169780666125e-06,
+      "loss": 1.1451,
+      "step": 1765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.357351746547522e-06,
+      "loss": 1.105,
+      "step": 1766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.353533712428921e-06,
+      "loss": 1.0802,
+      "step": 1767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.349715678310317e-06,
+      "loss": 1.3312,
+      "step": 1768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.345897644191714e-06,
+      "loss": 1.0935,
+      "step": 1769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.342079610073112e-06,
+      "loss": 1.0295,
+      "step": 1770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.338261575954509e-06,
+      "loss": 1.0997,
+      "step": 1771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.334443541835905e-06,
+      "loss": 1.1746,
+      "step": 1772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.330625507717302e-06,
+      "loss": 1.1772,
+      "step": 1773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.326807473598701e-06,
+      "loss": 1.1659,
+      "step": 1774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.322989439480098e-06,
+      "loss": 1.5071,
+      "step": 1775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.319171405361495e-06,
+      "loss": 1.4304,
+      "step": 1776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.3153533712428924e-06,
+      "loss": 1.237,
+      "step": 1777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.311535337124289e-06,
+      "loss": 1.3959,
+      "step": 1778
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.307717303005686e-06,
+      "loss": 1.339,
+      "step": 1779
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.303899268887085e-06,
+      "loss": 1.3102,
+      "step": 1780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.300081234768481e-06,
+      "loss": 1.4347,
+      "step": 1781
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.296263200649878e-06,
+      "loss": 1.5785,
+      "step": 1782
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.292445166531276e-06,
+      "loss": 1.3871,
+      "step": 1783
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.288627132412673e-06,
+      "loss": 1.3808,
+      "step": 1784
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.28480909829407e-06,
+      "loss": 1.2581,
+      "step": 1785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.280991064175466e-06,
+      "loss": 1.2657,
+      "step": 1786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.277173030056865e-06,
+      "loss": 1.4247,
+      "step": 1787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.273354995938262e-06,
+      "loss": 1.2386,
+      "step": 1788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.269536961819659e-06,
+      "loss": 1.202,
+      "step": 1789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.2657189277010565e-06,
+      "loss": 1.4012,
+      "step": 1790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.261900893582453e-06,
+      "loss": 1.4356,
+      "step": 1791
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.25808285946385e-06,
+      "loss": 0.983,
+      "step": 1792
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.254264825345249e-06,
+      "loss": 1.6904,
+      "step": 1793
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.250446791226646e-06,
+      "loss": 1.2722,
+      "step": 1794
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.246628757108042e-06,
+      "loss": 0.8669,
+      "step": 1795
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.2428107229894404e-06,
+      "loss": 1.3188,
+      "step": 1796
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.238992688870837e-06,
+      "loss": 1.1627,
+      "step": 1797
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.235174654752234e-06,
+      "loss": 1.2463,
+      "step": 1798
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.231356620633631e-06,
+      "loss": 1.16,
+      "step": 1799
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.227538586515029e-06,
+      "loss": 1.2327,
+      "step": 1800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.223720552396426e-06,
+      "loss": 1.0085,
+      "step": 1801
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.219902518277823e-06,
+      "loss": 1.1172,
+      "step": 1802
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.216084484159221e-06,
+      "loss": 1.3963,
+      "step": 1803
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.212266450040617e-06,
+      "loss": 1.218,
+      "step": 1804
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.208448415922014e-06,
+      "loss": 1.054,
+      "step": 1805
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.204630381803413e-06,
+      "loss": 1.1046,
+      "step": 1806
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.20081234768481e-06,
+      "loss": 1.1692,
+      "step": 1807
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.196994313566207e-06,
+      "loss": 1.4657,
+      "step": 1808
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.1931762794476044e-06,
+      "loss": 1.3604,
+      "step": 1809
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.189358245329001e-06,
+      "loss": 1.0994,
+      "step": 1810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.185540211210398e-06,
+      "loss": 1.23,
+      "step": 1811
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.181722177091795e-06,
+      "loss": 1.0138,
+      "step": 1812
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.177904142973193e-06,
+      "loss": 1.2348,
+      "step": 1813
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.17408610885459e-06,
+      "loss": 1.4747,
+      "step": 1814
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.170268074735987e-06,
+      "loss": 1.3148,
+      "step": 1815
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.166450040617385e-06,
+      "loss": 1.4759,
+      "step": 1816
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.162632006498782e-06,
+      "loss": 1.3385,
+      "step": 1817
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.158813972380178e-06,
+      "loss": 1.4195,
+      "step": 1818
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.154995938261577e-06,
+      "loss": 1.562,
+      "step": 1819
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.151177904142974e-06,
+      "loss": 1.2365,
+      "step": 1820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.147359870024371e-06,
+      "loss": 1.2196,
+      "step": 1821
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.143541835905767e-06,
+      "loss": 1.325,
+      "step": 1822
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.139723801787165e-06,
+      "loss": 1.155,
+      "step": 1823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135905767668562e-06,
+      "loss": 1.0286,
+      "step": 1824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.132087733549959e-06,
+      "loss": 1.5711,
+      "step": 1825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.128269699431357e-06,
+      "loss": 1.3596,
+      "step": 1826
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.124451665312754e-06,
+      "loss": 1.1808,
+      "step": 1827
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.120633631194151e-06,
+      "loss": 1.1352,
+      "step": 1828
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.116815597075549e-06,
+      "loss": 1.1891,
+      "step": 1829
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.112997562956946e-06,
+      "loss": 1.148,
+      "step": 1830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.109179528838342e-06,
+      "loss": 1.1082,
+      "step": 1831
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.105361494719741e-06,
+      "loss": 1.3764,
+      "step": 1832
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.101543460601138e-06,
+      "loss": 0.9558,
+      "step": 1833
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.097725426482535e-06,
+      "loss": 1.523,
+      "step": 1834
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.093907392363932e-06,
+      "loss": 0.9497,
+      "step": 1835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.090089358245329e-06,
+      "loss": 1.5993,
+      "step": 1836
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.086271324126726e-06,
+      "loss": 0.999,
+      "step": 1837
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.082453290008123e-06,
+      "loss": 1.1727,
+      "step": 1838
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.078635255889522e-06,
+      "loss": 1.2332,
+      "step": 1839
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.074817221770918e-06,
+      "loss": 1.3204,
+      "step": 1840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.070999187652315e-06,
+      "loss": 1.3731,
+      "step": 1841
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.067181153533713e-06,
+      "loss": 1.0962,
+      "step": 1842
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.06336311941511e-06,
+      "loss": 1.206,
+      "step": 1843
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.059545085296507e-06,
+      "loss": 1.5335,
+      "step": 1844
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.055727051177905e-06,
+      "loss": 1.0838,
+      "step": 1845
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.051909017059302e-06,
+      "loss": 1.0687,
+      "step": 1846
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.048090982940699e-06,
+      "loss": 1.6538,
+      "step": 1847
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.0442729488220965e-06,
+      "loss": 1.3462,
+      "step": 1848
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.040454914703493e-06,
+      "loss": 1.2744,
+      "step": 1849
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.03663688058489e-06,
+      "loss": 1.1659,
+      "step": 1850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.032818846466288e-06,
+      "loss": 1.0555,
+      "step": 1851
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.029000812347685e-06,
+      "loss": 1.0847,
+      "step": 1852
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.025182778229083e-06,
+      "loss": 0.8951,
+      "step": 1853
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.02136474411048e-06,
+      "loss": 1.3694,
+      "step": 1854
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.0175467099918765e-06,
+      "loss": 1.1248,
+      "step": 1855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.013728675873274e-06,
+      "loss": 0.9653,
+      "step": 1856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.009910641754671e-06,
+      "loss": 1.3381,
+      "step": 1857
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.006092607636068e-06,
+      "loss": 1.1565,
+      "step": 1858
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.002274573517466e-06,
+      "loss": 1.0872,
+      "step": 1859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.998456539398863e-06,
+      "loss": 1.7592,
+      "step": 1860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.9946385052802605e-06,
+      "loss": 1.0652,
+      "step": 1861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.990820471161658e-06,
+      "loss": 1.5282,
+      "step": 1862
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.987002437043054e-06,
+      "loss": 1.4475,
+      "step": 1863
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.983184402924452e-06,
+      "loss": 1.64,
+      "step": 1864
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.979366368805849e-06,
+      "loss": 1.1714,
+      "step": 1865
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.975548334687247e-06,
+      "loss": 1.5052,
+      "step": 1866
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.971730300568644e-06,
+      "loss": 1.1267,
+      "step": 1867
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.9679122664500405e-06,
+      "loss": 1.0476,
+      "step": 1868
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.964094232331438e-06,
+      "loss": 1.3266,
+      "step": 1869
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.960276198212835e-06,
+      "loss": 1.143,
+      "step": 1870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.956458164094233e-06,
+      "loss": 1.0894,
+      "step": 1871
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.95264012997563e-06,
+      "loss": 1.1978,
+      "step": 1872
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.948822095857027e-06,
+      "loss": 1.0552,
+      "step": 1873
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.9450040617384245e-06,
+      "loss": 1.3441,
+      "step": 1874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.941186027619821e-06,
+      "loss": 1.3093,
+      "step": 1875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.937367993501218e-06,
+      "loss": 1.0242,
+      "step": 1876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.933549959382616e-06,
+      "loss": 0.987,
+      "step": 1877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.929731925264013e-06,
+      "loss": 1.0968,
+      "step": 1878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.925913891145411e-06,
+      "loss": 1.1627,
+      "step": 1879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.9220958570268085e-06,
+      "loss": 1.0352,
+      "step": 1880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.9182778229082045e-06,
+      "loss": 1.3504,
+      "step": 1881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.914459788789602e-06,
+      "loss": 1.0819,
+      "step": 1882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.910641754670999e-06,
+      "loss": 1.0706,
+      "step": 1883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.906823720552397e-06,
+      "loss": 1.1186,
+      "step": 1884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.903005686433794e-06,
+      "loss": 1.2957,
+      "step": 1885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.899187652315191e-06,
+      "loss": 1.4243,
+      "step": 1886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.8953696181965885e-06,
+      "loss": 1.277,
+      "step": 1887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.891551584077985e-06,
+      "loss": 1.3897,
+      "step": 1888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.887733549959383e-06,
+      "loss": 1.1421,
+      "step": 1889
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.88391551584078e-06,
+      "loss": 1.1176,
+      "step": 1890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.880097481722177e-06,
+      "loss": 1.2201,
+      "step": 1891
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.876279447603575e-06,
+      "loss": 1.1159,
+      "step": 1892
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.8724614134849725e-06,
+      "loss": 1.588,
+      "step": 1893
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.868643379366369e-06,
+      "loss": 1.1244,
+      "step": 1894
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.864825345247766e-06,
+      "loss": 1.1802,
+      "step": 1895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.861007311129163e-06,
+      "loss": 1.4733,
+      "step": 1896
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.857189277010561e-06,
+      "loss": 1.4586,
+      "step": 1897
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.853371242891959e-06,
+      "loss": 0.9617,
+      "step": 1898
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.849553208773355e-06,
+      "loss": 1.2491,
+      "step": 1899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.8457351746547525e-06,
+      "loss": 1.197,
+      "step": 1900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.841917140536149e-06,
+      "loss": 1.182,
+      "step": 1901
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.838099106417547e-06,
+      "loss": 1.1269,
+      "step": 1902
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.834281072298944e-06,
+      "loss": 1.2899,
+      "step": 1903
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.830463038180341e-06,
+      "loss": 1.1255,
+      "step": 1904
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.826645004061739e-06,
+      "loss": 1.2052,
+      "step": 1905
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.822826969943136e-06,
+      "loss": 1.3266,
+      "step": 1906
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.819008935824533e-06,
+      "loss": 1.2019,
+      "step": 1907
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.81519090170593e-06,
+      "loss": 1.048,
+      "step": 1908
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.811372867587327e-06,
+      "loss": 1.0599,
+      "step": 1909
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.807554833468725e-06,
+      "loss": 1.3649,
+      "step": 1910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.803736799350123e-06,
+      "loss": 1.6464,
+      "step": 1911
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.79991876523152e-06,
+      "loss": 1.4144,
+      "step": 1912
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.7961007311129165e-06,
+      "loss": 1.3006,
+      "step": 1913
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.7922826969943134e-06,
+      "loss": 1.0216,
+      "step": 1914
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.788464662875711e-06,
+      "loss": 1.3803,
+      "step": 1915
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.784646628757109e-06,
+      "loss": 1.1939,
+      "step": 1916
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.780828594638505e-06,
+      "loss": 1.1428,
+      "step": 1917
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.777010560519903e-06,
+      "loss": 1.3374,
+      "step": 1918
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.7731925264013e-06,
+      "loss": 1.2343,
+      "step": 1919
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.769374492282697e-06,
+      "loss": 1.1663,
+      "step": 1920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.765556458164095e-06,
+      "loss": 0.9419,
+      "step": 1921
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.761738424045491e-06,
+      "loss": 1.4383,
+      "step": 1922
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.757920389926889e-06,
+      "loss": 1.3896,
+      "step": 1923
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.754102355808287e-06,
+      "loss": 1.236,
+      "step": 1924
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.750284321689684e-06,
+      "loss": 1.1307,
+      "step": 1925
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.7464662875710805e-06,
+      "loss": 1.455,
+      "step": 1926
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.7426482534524774e-06,
+      "loss": 1.2693,
+      "step": 1927
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.738830219333875e-06,
+      "loss": 1.3357,
+      "step": 1928
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.735012185215273e-06,
+      "loss": 1.0256,
+      "step": 1929
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.73119415109667e-06,
+      "loss": 1.2048,
+      "step": 1930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.727376116978067e-06,
+      "loss": 1.0859,
+      "step": 1931
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.723558082859464e-06,
+      "loss": 1.3393,
+      "step": 1932
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.719740048740861e-06,
+      "loss": 1.5182,
+      "step": 1933
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.715922014622259e-06,
+      "loss": 1.4252,
+      "step": 1934
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.712103980503655e-06,
+      "loss": 1.3535,
+      "step": 1935
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.708285946385053e-06,
+      "loss": 1.2248,
+      "step": 1936
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.70446791226645e-06,
+      "loss": 1.1534,
+      "step": 1937
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.700649878147848e-06,
+      "loss": 1.2394,
+      "step": 1938
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.696831844029245e-06,
+      "loss": 1.0808,
+      "step": 1939
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.6930138099106414e-06,
+      "loss": 1.2574,
+      "step": 1940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.689195775792039e-06,
+      "loss": 1.4682,
+      "step": 1941
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.685377741673437e-06,
+      "loss": 1.0615,
+      "step": 1942
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.681559707554834e-06,
+      "loss": 1.3404,
+      "step": 1943
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.677741673436231e-06,
+      "loss": 1.2337,
+      "step": 1944
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.673923639317628e-06,
+      "loss": 0.8663,
+      "step": 1945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.670105605199025e-06,
+      "loss": 1.2897,
+      "step": 1946
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.666287571080423e-06,
+      "loss": 1.2999,
+      "step": 1947
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.66246953696182e-06,
+      "loss": 1.1909,
+      "step": 1948
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.658651502843217e-06,
+      "loss": 1.1098,
+      "step": 1949
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.654833468724614e-06,
+      "loss": 1.1777,
+      "step": 1950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.651015434606012e-06,
+      "loss": 1.2781,
+      "step": 1951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.647197400487409e-06,
+      "loss": 1.3208,
+      "step": 1952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.6433793663688055e-06,
+      "loss": 1.4315,
+      "step": 1953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.639561332250203e-06,
+      "loss": 1.1699,
+      "step": 1954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.635743298131601e-06,
+      "loss": 1.2692,
+      "step": 1955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.631925264012998e-06,
+      "loss": 1.2452,
+      "step": 1956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.628107229894396e-06,
+      "loss": 0.9517,
+      "step": 1957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.624289195775792e-06,
+      "loss": 1.273,
+      "step": 1958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.6204711616571894e-06,
+      "loss": 1.2345,
+      "step": 1959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.616653127538587e-06,
+      "loss": 1.1952,
+      "step": 1960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.612835093419984e-06,
+      "loss": 1.4269,
+      "step": 1961
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.609017059301381e-06,
+      "loss": 1.2022,
+      "step": 1962
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.605199025182778e-06,
+      "loss": 1.1655,
+      "step": 1963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.601380991064176e-06,
+      "loss": 1.0362,
+      "step": 1964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.597562956945573e-06,
+      "loss": 1.272,
+      "step": 1965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.59374492282697e-06,
+      "loss": 1.1468,
+      "step": 1966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.589926888708367e-06,
+      "loss": 1.1554,
+      "step": 1967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.586108854589764e-06,
+      "loss": 1.234,
+      "step": 1968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.582290820471162e-06,
+      "loss": 1.3968,
+      "step": 1969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.57847278635256e-06,
+      "loss": 1.2292,
+      "step": 1970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.5746547522339565e-06,
+      "loss": 1.2218,
+      "step": 1971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.5708367181153534e-06,
+      "loss": 1.3323,
+      "step": 1972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.567018683996751e-06,
+      "loss": 1.6455,
+      "step": 1973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.563200649878148e-06,
+      "loss": 1.4074,
+      "step": 1974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.559382615759546e-06,
+      "loss": 1.1893,
+      "step": 1975
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.555564581640942e-06,
+      "loss": 1.1222,
+      "step": 1976
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.55174654752234e-06,
+      "loss": 1.368,
+      "step": 1977
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.547928513403737e-06,
+      "loss": 1.4091,
+      "step": 1978
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.544110479285134e-06,
+      "loss": 1.4678,
+      "step": 1979
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.540292445166531e-06,
+      "loss": 1.1297,
+      "step": 1980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.536474411047928e-06,
+      "loss": 1.2021,
+      "step": 1981
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.532656376929326e-06,
+      "loss": 1.4884,
+      "step": 1982
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.528838342810724e-06,
+      "loss": 1.3075,
+      "step": 1983
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.5250203086921205e-06,
+      "loss": 1.1899,
+      "step": 1984
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.5212022745735175e-06,
+      "loss": 1.2655,
+      "step": 1985
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.517384240454915e-06,
+      "loss": 1.6249,
+      "step": 1986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.513566206336312e-06,
+      "loss": 0.9382,
+      "step": 1987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.50974817221771e-06,
+      "loss": 1.2127,
+      "step": 1988
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.505930138099107e-06,
+      "loss": 1.1303,
+      "step": 1989
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.502112103980504e-06,
+      "loss": 1.0056,
+      "step": 1990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.4982940698619014e-06,
+      "loss": 1.5219,
+      "step": 1991
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.494476035743298e-06,
+      "loss": 1.3994,
+      "step": 1992
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.490658001624696e-06,
+      "loss": 1.1469,
+      "step": 1993
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.486839967506092e-06,
+      "loss": 0.9173,
+      "step": 1994
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.48302193338749e-06,
+      "loss": 1.1785,
+      "step": 1995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.479203899268888e-06,
+      "loss": 1.3641,
+      "step": 1996
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.4753858651502846e-06,
+      "loss": 1.1117,
+      "step": 1997
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.471567831031682e-06,
+      "loss": 1.2437,
+      "step": 1998
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.467749796913078e-06,
+      "loss": 1.3381,
+      "step": 1999
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.463931762794476e-06,
+      "loss": 1.3646,
+      "step": 2000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.460113728675874e-06,
+      "loss": 1.3008,
+      "step": 2001
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.456295694557271e-06,
+      "loss": 0.9979,
+      "step": 2002
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.452477660438668e-06,
+      "loss": 1.1968,
+      "step": 2003
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.4486596263200654e-06,
+      "loss": 1.6118,
+      "step": 2004
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.444841592201462e-06,
+      "loss": 1.1227,
+      "step": 2005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.44102355808286e-06,
+      "loss": 1.2585,
+      "step": 2006
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.437205523964257e-06,
+      "loss": 1.3078,
+      "step": 2007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.433387489845654e-06,
+      "loss": 1.5037,
+      "step": 2008
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.429569455727052e-06,
+      "loss": 1.1215,
+      "step": 2009
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.4257514216084486e-06,
+      "loss": 1.1486,
+      "step": 2010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.421933387489846e-06,
+      "loss": 1.2492,
+      "step": 2011
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.418115353371242e-06,
+      "loss": 1.2498,
+      "step": 2012
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.41429731925264e-06,
+      "loss": 1.2495,
+      "step": 2013
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.410479285134038e-06,
+      "loss": 1.0423,
+      "step": 2014
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.406661251015435e-06,
+      "loss": 1.1878,
+      "step": 2015
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.4028432168968325e-06,
+      "loss": 1.3244,
+      "step": 2016
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.3990251827782294e-06,
+      "loss": 1.1861,
+      "step": 2017
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.395207148659626e-06,
+      "loss": 1.0773,
+      "step": 2018
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.391389114541024e-06,
+      "loss": 1.0632,
+      "step": 2019
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.387571080422421e-06,
+      "loss": 1.4665,
+      "step": 2020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.383753046303818e-06,
+      "loss": 1.4674,
+      "step": 2021
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.379935012185216e-06,
+      "loss": 1.391,
+      "step": 2022
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.3761169780666126e-06,
+      "loss": 1.0341,
+      "step": 2023
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.37229894394801e-06,
+      "loss": 0.9712,
+      "step": 2024
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.368480909829407e-06,
+      "loss": 1.2903,
+      "step": 2025
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.364662875710804e-06,
+      "loss": 1.337,
+      "step": 2026
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.360844841592202e-06,
+      "loss": 1.0495,
+      "step": 2027
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.357026807473599e-06,
+      "loss": 1.0629,
+      "step": 2028
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.3532087733549966e-06,
+      "loss": 1.0291,
+      "step": 2029
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.349390739236393e-06,
+      "loss": 1.1371,
+      "step": 2030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.34557270511779e-06,
+      "loss": 1.5099,
+      "step": 2031
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.341754670999188e-06,
+      "loss": 1.3115,
+      "step": 2032
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.337936636880585e-06,
+      "loss": 1.0546,
+      "step": 2033
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.334118602761983e-06,
+      "loss": 1.167,
+      "step": 2034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.33030056864338e-06,
+      "loss": 1.1969,
+      "step": 2035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.326482534524777e-06,
+      "loss": 0.9972,
+      "step": 2036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.322664500406174e-06,
+      "loss": 1.1717,
+      "step": 2037
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.318846466287571e-06,
+      "loss": 0.9703,
+      "step": 2038
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.315028432168968e-06,
+      "loss": 1.1472,
+      "step": 2039
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.311210398050366e-06,
+      "loss": 1.3269,
+      "step": 2040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.307392363931763e-06,
+      "loss": 1.055,
+      "step": 2041
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.3035743298131606e-06,
+      "loss": 1.5715,
+      "step": 2042
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.2997562956945575e-06,
+      "loss": 1.3874,
+      "step": 2043
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.295938261575954e-06,
+      "loss": 1.735,
+      "step": 2044
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.292120227457352e-06,
+      "loss": 1.5117,
+      "step": 2045
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.288302193338749e-06,
+      "loss": 1.1773,
+      "step": 2046
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.284484159220147e-06,
+      "loss": 1.356,
+      "step": 2047
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.280666125101544e-06,
+      "loss": 1.2444,
+      "step": 2048
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.276848090982941e-06,
+      "loss": 1.31,
+      "step": 2049
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.273030056864338e-06,
+      "loss": 1.1003,
+      "step": 2050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.269212022745735e-06,
+      "loss": 1.3896,
+      "step": 2051
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.265393988627133e-06,
+      "loss": 0.981,
+      "step": 2052
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.26157595450853e-06,
+      "loss": 1.0206,
+      "step": 2053
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.257757920389927e-06,
+      "loss": 0.966,
+      "step": 2054
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.2539398862713246e-06,
+      "loss": 1.2336,
+      "step": 2055
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.2501218521527215e-06,
+      "loss": 1.3558,
+      "step": 2056
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.246303818034118e-06,
+      "loss": 1.0742,
+      "step": 2057
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.242485783915516e-06,
+      "loss": 1.1767,
+      "step": 2058
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.238667749796913e-06,
+      "loss": 1.1974,
+      "step": 2059
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.234849715678311e-06,
+      "loss": 1.2148,
+      "step": 2060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.231031681559708e-06,
+      "loss": 1.2676,
+      "step": 2061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.227213647441105e-06,
+      "loss": 1.1113,
+      "step": 2062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.223395613322502e-06,
+      "loss": 1.3153,
+      "step": 2063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.219577579203899e-06,
+      "loss": 1.4191,
+      "step": 2064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.215759545085297e-06,
+      "loss": 1.0299,
+      "step": 2065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.211941510966694e-06,
+      "loss": 1.0096,
+      "step": 2066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.208123476848091e-06,
+      "loss": 1.2997,
+      "step": 2067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.204305442729489e-06,
+      "loss": 1.0947,
+      "step": 2068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.2004874086108855e-06,
+      "loss": 1.1126,
+      "step": 2069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.196669374492283e-06,
+      "loss": 1.2182,
+      "step": 2070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.19285134037368e-06,
+      "loss": 1.3949,
+      "step": 2071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.189033306255077e-06,
+      "loss": 1.237,
+      "step": 2072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.185215272136475e-06,
+      "loss": 1.081,
+      "step": 2073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.181397238017872e-06,
+      "loss": 1.4528,
+      "step": 2074
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.1775792038992695e-06,
+      "loss": 1.1563,
+      "step": 2075
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.173761169780666e-06,
+      "loss": 1.449,
+      "step": 2076
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.169943135662063e-06,
+      "loss": 1.2814,
+      "step": 2077
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.166125101543461e-06,
+      "loss": 1.1492,
+      "step": 2078
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.162307067424859e-06,
+      "loss": 1.2484,
+      "step": 2079
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.158489033306255e-06,
+      "loss": 1.213,
+      "step": 2080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.154670999187653e-06,
+      "loss": 1.6631,
+      "step": 2081
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.1508529650690495e-06,
+      "loss": 1.2634,
+      "step": 2082
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.147034930950447e-06,
+      "loss": 1.2207,
+      "step": 2083
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.143216896831844e-06,
+      "loss": 1.3858,
+      "step": 2084
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.139398862713241e-06,
+      "loss": 1.3427,
+      "step": 2085
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.135580828594639e-06,
+      "loss": 1.2068,
+      "step": 2086
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.131762794476036e-06,
+      "loss": 1.4787,
+      "step": 2087
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.1279447603574335e-06,
+      "loss": 1.1405,
+      "step": 2088
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.12412672623883e-06,
+      "loss": 1.0543,
+      "step": 2089
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.120308692120227e-06,
+      "loss": 1.3046,
+      "step": 2090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.116490658001625e-06,
+      "loss": 1.3706,
+      "step": 2091
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.112672623883023e-06,
+      "loss": 1.648,
+      "step": 2092
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.10885458976442e-06,
+      "loss": 1.2489,
+      "step": 2093
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.105036555645817e-06,
+      "loss": 1.1104,
+      "step": 2094
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.1012185215272135e-06,
+      "loss": 1.3661,
+      "step": 2095
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.097400487408611e-06,
+      "loss": 1.29,
+      "step": 2096
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.093582453290009e-06,
+      "loss": 1.0587,
+      "step": 2097
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.089764419171405e-06,
+      "loss": 1.1282,
+      "step": 2098
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.085946385052803e-06,
+      "loss": 1.0545,
+      "step": 2099
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.0821283509342e-06,
+      "loss": 1.4685,
+      "step": 2100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.0783103168155975e-06,
+      "loss": 1.2286,
+      "step": 2101
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.074492282696995e-06,
+      "loss": 1.6816,
+      "step": 2102
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.070674248578391e-06,
+      "loss": 1.0954,
+      "step": 2103
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.066856214459789e-06,
+      "loss": 1.3571,
+      "step": 2104
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.063038180341186e-06,
+      "loss": 1.6313,
+      "step": 2105
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.059220146222584e-06,
+      "loss": 1.1799,
+      "step": 2106
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.055402112103981e-06,
+      "loss": 1.2489,
+      "step": 2107
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.0515840779853775e-06,
+      "loss": 1.7904,
+      "step": 2108
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.047766043866775e-06,
+      "loss": 1.0585,
+      "step": 2109
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.043948009748173e-06,
+      "loss": 1.0259,
+      "step": 2110
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.04012997562957e-06,
+      "loss": 1.1644,
+      "step": 2111
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.036311941510967e-06,
+      "loss": 1.2564,
+      "step": 2112
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.032493907392364e-06,
+      "loss": 1.2841,
+      "step": 2113
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.0286758732737615e-06,
+      "loss": 1.5045,
+      "step": 2114
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.024857839155159e-06,
+      "loss": 1.1467,
+      "step": 2115
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.021039805036555e-06,
+      "loss": 1.383,
+      "step": 2116
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.017221770917953e-06,
+      "loss": 1.0672,
+      "step": 2117
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.01340373679935e-06,
+      "loss": 1.2015,
+      "step": 2118
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.009585702680748e-06,
+      "loss": 1.3291,
+      "step": 2119
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.0057676685621455e-06,
+      "loss": 1.2678,
+      "step": 2120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.0019496344435415e-06,
+      "loss": 1.1776,
+      "step": 2121
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.998131600324939e-06,
+      "loss": 1.1811,
+      "step": 2122
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.994313566206337e-06,
+      "loss": 1.1229,
+      "step": 2123
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.990495532087734e-06,
+      "loss": 1.2695,
+      "step": 2124
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.986677497969131e-06,
+      "loss": 1.2114,
+      "step": 2125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.982859463850528e-06,
+      "loss": 1.0532,
+      "step": 2126
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9790414297319255e-06,
+      "loss": 1.1981,
+      "step": 2127
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.975223395613323e-06,
+      "loss": 1.1747,
+      "step": 2128
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.97140536149472e-06,
+      "loss": 1.1748,
+      "step": 2129
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.967587327376117e-06,
+      "loss": 1.3916,
+      "step": 2130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.963769293257514e-06,
+      "loss": 1.3435,
+      "step": 2131
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.959951259138912e-06,
+      "loss": 1.0759,
+      "step": 2132
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9561332250203095e-06,
+      "loss": 1.1504,
+      "step": 2133
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9523151909017055e-06,
+      "loss": 1.063,
+      "step": 2134
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.948497156783103e-06,
+      "loss": 1.1958,
+      "step": 2135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.9446791226645e-06,
+      "loss": 1.1892,
+      "step": 2136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.940861088545898e-06,
+      "loss": 1.0063,
+      "step": 2137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.937043054427296e-06,
+      "loss": 1.1311,
+      "step": 2138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.933225020308692e-06,
+      "loss": 1.0651,
+      "step": 2139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.9294069861900895e-06,
+      "loss": 1.4284,
+      "step": 2140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.925588952071487e-06,
+      "loss": 1.679,
+      "step": 2141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.921770917952884e-06,
+      "loss": 1.2247,
+      "step": 2142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.917952883834281e-06,
+      "loss": 1.3588,
+      "step": 2143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.914134849715678e-06,
+      "loss": 1.1114,
+      "step": 2144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.910316815597076e-06,
+      "loss": 0.8954,
+      "step": 2145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.9064987814784735e-06,
+      "loss": 1.1338,
+      "step": 2146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.90268074735987e-06,
+      "loss": 1.307,
+      "step": 2147
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.898862713241267e-06,
+      "loss": 1.086,
+      "step": 2148
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.895044679122664e-06,
+      "loss": 1.1822,
+      "step": 2149
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.891226645004062e-06,
+      "loss": 1.2563,
+      "step": 2150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.88740861088546e-06,
+      "loss": 1.3857,
+      "step": 2151
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.883590576766857e-06,
+      "loss": 1.1154,
+      "step": 2152
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.8797725426482535e-06,
+      "loss": 1.1183,
+      "step": 2153
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.875954508529651e-06,
+      "loss": 1.1581,
+      "step": 2154
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.872136474411048e-06,
+      "loss": 1.0355,
+      "step": 2155
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.868318440292446e-06,
+      "loss": 1.1809,
+      "step": 2156
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.864500406173842e-06,
+      "loss": 1.4289,
+      "step": 2157
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.86068237205524e-06,
+      "loss": 1.0345,
+      "step": 2158
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.8568643379366375e-06,
+      "loss": 1.0098,
+      "step": 2159
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.853046303818034e-06,
+      "loss": 1.3494,
+      "step": 2160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.849228269699431e-06,
+      "loss": 1.246,
+      "step": 2161
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.845410235580828e-06,
+      "loss": 1.3511,
+      "step": 2162
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.841592201462226e-06,
+      "loss": 1.339,
+      "step": 2163
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.837774167343624e-06,
+      "loss": 1.4885,
+      "step": 2164
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.833956133225021e-06,
+      "loss": 1.2059,
+      "step": 2165
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.8301380991064175e-06,
+      "loss": 1.4041,
+      "step": 2166
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.8263200649878144e-06,
+      "loss": 1.0019,
+      "step": 2167
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.822502030869212e-06,
+      "loss": 1.4921,
+      "step": 2168
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.81868399675061e-06,
+      "loss": 1.3274,
+      "step": 2169
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.814865962632007e-06,
+      "loss": 1.2524,
+      "step": 2170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.811047928513404e-06,
+      "loss": 1.4745,
+      "step": 2171
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.8072298943948015e-06,
+      "loss": 1.0844,
+      "step": 2172
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.803411860276198e-06,
+      "loss": 1.2402,
+      "step": 2173
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.799593826157596e-06,
+      "loss": 1.3559,
+      "step": 2174
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.795775792038992e-06,
+      "loss": 1.2847,
+      "step": 2175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.79195775792039e-06,
+      "loss": 1.2833,
+      "step": 2176
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.788139723801788e-06,
+      "loss": 1.014,
+      "step": 2177
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.784321689683185e-06,
+      "loss": 1.0939,
+      "step": 2178
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.780503655564582e-06,
+      "loss": 0.963,
+      "step": 2179
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7766856214459784e-06,
+      "loss": 1.3121,
+      "step": 2180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.772867587327376e-06,
+      "loss": 1.3053,
+      "step": 2181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.769049553208774e-06,
+      "loss": 1.1604,
+      "step": 2182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.765231519090171e-06,
+      "loss": 1.4455,
+      "step": 2183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.761413484971568e-06,
+      "loss": 1.2919,
+      "step": 2184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7575954508529655e-06,
+      "loss": 1.3992,
+      "step": 2185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7537774167343624e-06,
+      "loss": 1.248,
+      "step": 2186
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.74995938261576e-06,
+      "loss": 1.0907,
+      "step": 2187
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.746141348497157e-06,
+      "loss": 1.1301,
+      "step": 2188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.742323314378554e-06,
+      "loss": 1.2215,
+      "step": 2189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.738505280259952e-06,
+      "loss": 1.3561,
+      "step": 2190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.734687246141349e-06,
+      "loss": 1.2558,
+      "step": 2191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.730869212022746e-06,
+      "loss": 1.1807,
+      "step": 2192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7270511779041425e-06,
+      "loss": 1.2392,
+      "step": 2193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.72323314378554e-06,
+      "loss": 1.2216,
+      "step": 2194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.719415109666938e-06,
+      "loss": 1.4111,
+      "step": 2195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.715597075548335e-06,
+      "loss": 1.2646,
+      "step": 2196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.711779041429733e-06,
+      "loss": 1.1363,
+      "step": 2197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.707961007311129e-06,
+      "loss": 1.4286,
+      "step": 2198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7041429731925264e-06,
+      "loss": 1.8406,
+      "step": 2199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.700324939073924e-06,
+      "loss": 1.2716,
+      "step": 2200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.696506904955321e-06,
+      "loss": 1.3237,
+      "step": 2201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.692688870836718e-06,
+      "loss": 1.2159,
+      "step": 2202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.688870836718116e-06,
+      "loss": 1.0427,
+      "step": 2203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.685052802599513e-06,
+      "loss": 1.4347,
+      "step": 2204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.68123476848091e-06,
+      "loss": 1.0977,
+      "step": 2205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.677416734362307e-06,
+      "loss": 1.1926,
+      "step": 2206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.673598700243704e-06,
+      "loss": 1.02,
+      "step": 2207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.669780666125102e-06,
+      "loss": 1.2844,
+      "step": 2208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.665962632006499e-06,
+      "loss": 1.2606,
+      "step": 2209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.662144597887897e-06,
+      "loss": 1.2628,
+      "step": 2210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.658326563769293e-06,
+      "loss": 1.2383,
+      "step": 2211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.6545085296506904e-06,
+      "loss": 1.4839,
+      "step": 2212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.650690495532088e-06,
+      "loss": 1.4329,
+      "step": 2213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.646872461413485e-06,
+      "loss": 1.1473,
+      "step": 2214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.643054427294883e-06,
+      "loss": 1.042,
+      "step": 2215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.63923639317628e-06,
+      "loss": 0.9663,
+      "step": 2216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.635418359057677e-06,
+      "loss": 1.3064,
+      "step": 2217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.631600324939074e-06,
+      "loss": 1.1662,
+      "step": 2218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.627782290820471e-06,
+      "loss": 1.2608,
+      "step": 2219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.623964256701868e-06,
+      "loss": 1.3376,
+      "step": 2220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.620146222583266e-06,
+      "loss": 1.1031,
+      "step": 2221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.616328188464663e-06,
+      "loss": 1.2083,
+      "step": 2222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.612510154346061e-06,
+      "loss": 1.1778,
+      "step": 2223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.6086921202274575e-06,
+      "loss": 1.0896,
+      "step": 2224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.6048740861088545e-06,
+      "loss": 1.2797,
+      "step": 2225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.601056051990252e-06,
+      "loss": 0.8819,
+      "step": 2226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.597238017871649e-06,
+      "loss": 1.2927,
+      "step": 2227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.593419983753047e-06,
+      "loss": 1.223,
+      "step": 2228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.589601949634444e-06,
+      "loss": 1.1527,
+      "step": 2229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.585783915515841e-06,
+      "loss": 1.376,
+      "step": 2230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.5819658813972384e-06,
+      "loss": 1.2449,
+      "step": 2231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.578147847278635e-06,
+      "loss": 1.0579,
+      "step": 2232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.574329813160033e-06,
+      "loss": 1.1291,
+      "step": 2233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.57051177904143e-06,
+      "loss": 1.3718,
+      "step": 2234
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.566693744922827e-06,
+      "loss": 1.2012,
+      "step": 2235
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.562875710804225e-06,
+      "loss": 1.3176,
+      "step": 2236
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.5590576766856216e-06,
+      "loss": 1.2965,
+      "step": 2237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.5552396425670185e-06,
+      "loss": 1.0887,
+      "step": 2238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.551421608448416e-06,
+      "loss": 1.3362,
+      "step": 2239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.547603574329813e-06,
+      "loss": 1.1482,
+      "step": 2240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.543785540211211e-06,
+      "loss": 1.2583,
+      "step": 2241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.539967506092608e-06,
+      "loss": 1.1239,
+      "step": 2242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.536149471974005e-06,
+      "loss": 1.2504,
+      "step": 2243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.5323314378554024e-06,
+      "loss": 1.1769,
+      "step": 2244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.528513403736799e-06,
+      "loss": 0.8937,
+      "step": 2245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.524695369618197e-06,
+      "loss": 1.0974,
+      "step": 2246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.520877335499594e-06,
+      "loss": 1.2524,
+      "step": 2247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.517059301380991e-06,
+      "loss": 1.293,
+      "step": 2248
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.513241267262389e-06,
+      "loss": 1.4927,
+      "step": 2249
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.5094232331437856e-06,
+      "loss": 1.3283,
+      "step": 2250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.505605199025183e-06,
+      "loss": 1.1354,
+      "step": 2251
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.50178716490658e-06,
+      "loss": 1.0972,
+      "step": 2252
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.497969130787977e-06,
+      "loss": 1.2314,
+      "step": 2253
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.494151096669375e-06,
+      "loss": 1.1937,
+      "step": 2254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.490333062550772e-06,
+      "loss": 1.3577,
+      "step": 2255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.4865150284321695e-06,
+      "loss": 1.1895,
+      "step": 2256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.4826969943135664e-06,
+      "loss": 1.2721,
+      "step": 2257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.478878960194963e-06,
+      "loss": 1.2671,
+      "step": 2258
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.475060926076361e-06,
+      "loss": 1.1962,
+      "step": 2259
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.471242891957758e-06,
+      "loss": 1.4573,
+      "step": 2260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.467424857839155e-06,
+      "loss": 1.421,
+      "step": 2261
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.463606823720553e-06,
+      "loss": 1.4239,
+      "step": 2262
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.45978878960195e-06,
+      "loss": 1.0979,
+      "step": 2263
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.455970755483347e-06,
+      "loss": 0.9238,
+      "step": 2264
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.452152721364744e-06,
+      "loss": 0.9691,
+      "step": 2265
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.448334687246141e-06,
+      "loss": 1.4013,
+      "step": 2266
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.444516653127539e-06,
+      "loss": 1.1962,
+      "step": 2267
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.440698619008936e-06,
+      "loss": 1.6593,
+      "step": 2268
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.4368805848903336e-06,
+      "loss": 1.494,
+      "step": 2269
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.4330625507717305e-06,
+      "loss": 1.4977,
+      "step": 2270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.429244516653127e-06,
+      "loss": 0.9586,
+      "step": 2271
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.425426482534525e-06,
+      "loss": 1.1901,
+      "step": 2272
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.421608448415922e-06,
+      "loss": 1.2554,
+      "step": 2273
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.41779041429732e-06,
+      "loss": 1.2884,
+      "step": 2274
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.413972380178717e-06,
+      "loss": 1.2315,
+      "step": 2275
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.410154346060114e-06,
+      "loss": 1.1399,
+      "step": 2276
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.406336311941511e-06,
+      "loss": 1.5347,
+      "step": 2277
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.402518277822909e-06,
+      "loss": 1.3201,
+      "step": 2278
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.398700243704305e-06,
+      "loss": 1.1041,
+      "step": 2279
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.394882209585703e-06,
+      "loss": 1.2514,
+      "step": 2280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.3910641754671e-06,
+      "loss": 1.2981,
+      "step": 2281
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.3872461413484976e-06,
+      "loss": 1.3586,
+      "step": 2282
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.383428107229895e-06,
+      "loss": 1.2278,
+      "step": 2283
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.379610073111291e-06,
+      "loss": 1.3011,
+      "step": 2284
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.375792038992689e-06,
+      "loss": 1.5073,
+      "step": 2285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.371974004874086e-06,
+      "loss": 1.4135,
+      "step": 2286
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.368155970755484e-06,
+      "loss": 1.2017,
+      "step": 2287
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.364337936636881e-06,
+      "loss": 1.1916,
+      "step": 2288
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.360519902518278e-06,
+      "loss": 1.3492,
+      "step": 2289
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.356701868399675e-06,
+      "loss": 1.2316,
+      "step": 2290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.352883834281073e-06,
+      "loss": 0.9778,
+      "step": 2291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.34906580016247e-06,
+      "loss": 1.0325,
+      "step": 2292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.345247766043867e-06,
+      "loss": 1.1908,
+      "step": 2293
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.341429731925264e-06,
+      "loss": 1.2091,
+      "step": 2294
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.3376116978066616e-06,
+      "loss": 1.4295,
+      "step": 2295
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.333793663688059e-06,
+      "loss": 1.1021,
+      "step": 2296
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.329975629569455e-06,
+      "loss": 1.3304,
+      "step": 2297
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.326157595450853e-06,
+      "loss": 1.417,
+      "step": 2298
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.32233956133225e-06,
+      "loss": 1.4284,
+      "step": 2299
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.318521527213648e-06,
+      "loss": 1.1143,
+      "step": 2300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.3147034930950455e-06,
+      "loss": 1.1568,
+      "step": 2301
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.310885458976442e-06,
+      "loss": 1.2586,
+      "step": 2302
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.307067424857839e-06,
+      "loss": 1.3278,
+      "step": 2303
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.303249390739236e-06,
+      "loss": 1.3805,
+      "step": 2304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.299431356620634e-06,
+      "loss": 1.3916,
+      "step": 2305
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.295613322502031e-06,
+      "loss": 1.0098,
+      "step": 2306
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.291795288383428e-06,
+      "loss": 1.1417,
+      "step": 2307
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.287977254264826e-06,
+      "loss": 1.0168,
+      "step": 2308
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.284159220146223e-06,
+      "loss": 1.1633,
+      "step": 2309
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.28034118602762e-06,
+      "loss": 1.0257,
+      "step": 2310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.276523151909017e-06,
+      "loss": 1.1512,
+      "step": 2311
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.272705117790414e-06,
+      "loss": 1.1662,
+      "step": 2312
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.268887083671812e-06,
+      "loss": 1.6144,
+      "step": 2313
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.2650690495532096e-06,
+      "loss": 1.2113,
+      "step": 2314
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.261251015434606e-06,
+      "loss": 1.2639,
+      "step": 2315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.257432981316003e-06,
+      "loss": 1.2432,
+      "step": 2316
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.2536149471974e-06,
+      "loss": 1.0308,
+      "step": 2317
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.249796913078798e-06,
+      "loss": 1.2035,
+      "step": 2318
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.245978878960196e-06,
+      "loss": 1.1117,
+      "step": 2319
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.242160844841592e-06,
+      "loss": 1.2287,
+      "step": 2320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.23834281072299e-06,
+      "loss": 1.3698,
+      "step": 2321
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.234524776604387e-06,
+      "loss": 1.3018,
+      "step": 2322
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.230706742485784e-06,
+      "loss": 1.0734,
+      "step": 2323
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.226888708367181e-06,
+      "loss": 1.3161,
+      "step": 2324
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.223070674248578e-06,
+      "loss": 1.376,
+      "step": 2325
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.219252640129976e-06,
+      "loss": 1.3306,
+      "step": 2326
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.2154346060113736e-06,
+      "loss": 1.3877,
+      "step": 2327
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.2116165718927705e-06,
+      "loss": 1.072,
+      "step": 2328
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.207798537774167e-06,
+      "loss": 1.2107,
+      "step": 2329
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.203980503655564e-06,
+      "loss": 1.1447,
+      "step": 2330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.200162469536962e-06,
+      "loss": 1.3814,
+      "step": 2331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.19634443541836e-06,
+      "loss": 1.0621,
+      "step": 2332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.192526401299757e-06,
+      "loss": 1.0961,
+      "step": 2333
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.188708367181154e-06,
+      "loss": 1.2629,
+      "step": 2334
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.1848903330625505e-06,
+      "loss": 1.5022,
+      "step": 2335
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.181072298943948e-06,
+      "loss": 1.0854,
+      "step": 2336
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.177254264825346e-06,
+      "loss": 1.3665,
+      "step": 2337
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.173436230706742e-06,
+      "loss": 1.0776,
+      "step": 2338
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.16961819658814e-06,
+      "loss": 1.0421,
+      "step": 2339
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.165800162469538e-06,
+      "loss": 1.2965,
+      "step": 2340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.1619821283509345e-06,
+      "loss": 1.2035,
+      "step": 2341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.158164094232331e-06,
+      "loss": 1.183,
+      "step": 2342
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.154346060113728e-06,
+      "loss": 1.0379,
+      "step": 2343
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.150528025995126e-06,
+      "loss": 1.6185,
+      "step": 2344
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.146709991876524e-06,
+      "loss": 1.2167,
+      "step": 2345
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.142891957757921e-06,
+      "loss": 1.327,
+      "step": 2346
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.139073923639318e-06,
+      "loss": 1.3264,
+      "step": 2347
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.1352558895207145e-06,
+      "loss": 1.3237,
+      "step": 2348
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.131437855402112e-06,
+      "loss": 1.1189,
+      "step": 2349
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.12761982128351e-06,
+      "loss": 1.3293,
+      "step": 2350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.123801787164907e-06,
+      "loss": 1.1389,
+      "step": 2351
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.119983753046304e-06,
+      "loss": 1.1961,
+      "step": 2352
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.116165718927702e-06,
+      "loss": 1.2185,
+      "step": 2353
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.1123476848090985e-06,
+      "loss": 1.3853,
+      "step": 2354
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.108529650690496e-06,
+      "loss": 1.2397,
+      "step": 2355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.104711616571892e-06,
+      "loss": 0.9276,
+      "step": 2356
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.10089358245329e-06,
+      "loss": 1.1678,
+      "step": 2357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.097075548334688e-06,
+      "loss": 1.1346,
+      "step": 2358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.093257514216085e-06,
+      "loss": 1.1283,
+      "step": 2359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.0894394800974825e-06,
+      "loss": 1.3182,
+      "step": 2360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.0856214459788785e-06,
+      "loss": 1.4638,
+      "step": 2361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.081803411860276e-06,
+      "loss": 1.3557,
+      "step": 2362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.077985377741674e-06,
+      "loss": 1.2971,
+      "step": 2363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.074167343623071e-06,
+      "loss": 1.0147,
+      "step": 2364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.070349309504468e-06,
+      "loss": 0.9299,
+      "step": 2365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.066531275385865e-06,
+      "loss": 1.1347,
+      "step": 2366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.0627132412672625e-06,
+      "loss": 1.4784,
+      "step": 2367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.05889520714866e-06,
+      "loss": 1.275,
+      "step": 2368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.055077173030057e-06,
+      "loss": 0.9635,
+      "step": 2369
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.051259138911454e-06,
+      "loss": 1.819,
+      "step": 2370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.047441104792852e-06,
+      "loss": 1.0772,
+      "step": 2371
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.043623070674249e-06,
+      "loss": 1.2096,
+      "step": 2372
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.0398050365556465e-06,
+      "loss": 1.1423,
+      "step": 2373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.0359870024370425e-06,
+      "loss": 1.3658,
+      "step": 2374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.03216896831844e-06,
+      "loss": 1.1531,
+      "step": 2375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.028350934199838e-06,
+      "loss": 1.3394,
+      "step": 2376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.024532900081235e-06,
+      "loss": 0.9667,
+      "step": 2377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.020714865962633e-06,
+      "loss": 1.3733,
+      "step": 2378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.016896831844029e-06,
+      "loss": 0.9694,
+      "step": 2379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.0130787977254265e-06,
+      "loss": 1.2456,
+      "step": 2380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.009260763606824e-06,
+      "loss": 1.0933,
+      "step": 2381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.005442729488221e-06,
+      "loss": 1.1659,
+      "step": 2382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.001624695369618e-06,
+      "loss": 1.032,
+      "step": 2383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.997806661251016e-06,
+      "loss": 1.0966,
+      "step": 2384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.993988627132413e-06,
+      "loss": 1.3161,
+      "step": 2385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9901705930138105e-06,
+      "loss": 1.1847,
+      "step": 2386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.986352558895207e-06,
+      "loss": 1.1887,
+      "step": 2387
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.982534524776604e-06,
+      "loss": 1.3708,
+      "step": 2388
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.978716490658002e-06,
+      "loss": 1.3687,
+      "step": 2389
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.974898456539399e-06,
+      "loss": 1.0117,
+      "step": 2390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.971080422420797e-06,
+      "loss": 1.3718,
+      "step": 2391
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.967262388302193e-06,
+      "loss": 1.0394,
+      "step": 2392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9634443541835905e-06,
+      "loss": 1.1015,
+      "step": 2393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.959626320064988e-06,
+      "loss": 1.3893,
+      "step": 2394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.955808285946385e-06,
+      "loss": 1.7484,
+      "step": 2395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.951990251827783e-06,
+      "loss": 1.5269,
+      "step": 2396
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.948172217709179e-06,
+      "loss": 1.2671,
+      "step": 2397
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.944354183590577e-06,
+      "loss": 1.0196,
+      "step": 2398
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9405361494719745e-06,
+      "loss": 1.7034,
+      "step": 2399
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.936718115353371e-06,
+      "loss": 1.2574,
+      "step": 2400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.932900081234768e-06,
+      "loss": 1.0915,
+      "step": 2401
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.929082047116166e-06,
+      "loss": 1.0365,
+      "step": 2402
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.925264012997563e-06,
+      "loss": 1.1794,
+      "step": 2403
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.921445978878961e-06,
+      "loss": 1.2588,
+      "step": 2404
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.917627944760358e-06,
+      "loss": 1.1885,
+      "step": 2405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9138099106417545e-06,
+      "loss": 1.0398,
+      "step": 2406
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.909991876523152e-06,
+      "loss": 0.9649,
+      "step": 2407
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.906173842404549e-06,
+      "loss": 1.1877,
+      "step": 2408
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.902355808285947e-06,
+      "loss": 1.2083,
+      "step": 2409
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.898537774167344e-06,
+      "loss": 1.206,
+      "step": 2410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.894719740048741e-06,
+      "loss": 1.0593,
+      "step": 2411
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8909017059301385e-06,
+      "loss": 1.2591,
+      "step": 2412
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.887083671811535e-06,
+      "loss": 1.3338,
+      "step": 2413
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.883265637692933e-06,
+      "loss": 1.5039,
+      "step": 2414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.87944760357433e-06,
+      "loss": 1.1721,
+      "step": 2415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.875629569455727e-06,
+      "loss": 1.1124,
+      "step": 2416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.871811535337125e-06,
+      "loss": 1.4731,
+      "step": 2417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.867993501218522e-06,
+      "loss": 0.9747,
+      "step": 2418
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.8641754670999185e-06,
+      "loss": 1.294,
+      "step": 2419
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.860357432981316e-06,
+      "loss": 1.2516,
+      "step": 2420
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.856539398862713e-06,
+      "loss": 1.0385,
+      "step": 2421
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.852721364744111e-06,
+      "loss": 1.2086,
+      "step": 2422
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.848903330625508e-06,
+      "loss": 1.5578,
+      "step": 2423
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.845085296506905e-06,
+      "loss": 1.0729,
+      "step": 2424
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.8412672623883025e-06,
+      "loss": 1.3062,
+      "step": 2425
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.8374492282696994e-06,
+      "loss": 1.1885,
+      "step": 2426
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.833631194151097e-06,
+      "loss": 1.1589,
+      "step": 2427
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.829813160032494e-06,
+      "loss": 0.9997,
+      "step": 2428
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.825995125913891e-06,
+      "loss": 1.2706,
+      "step": 2429
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.822177091795289e-06,
+      "loss": 1.1866,
+      "step": 2430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.818359057676686e-06,
+      "loss": 1.1348,
+      "step": 2431
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.814541023558083e-06,
+      "loss": 1.2893,
+      "step": 2432
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.81072298943948e-06,
+      "loss": 1.1461,
+      "step": 2433
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.806904955320877e-06,
+      "loss": 0.9759,
+      "step": 2434
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.803086921202275e-06,
+      "loss": 1.3508,
+      "step": 2435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.799268887083672e-06,
+      "loss": 0.8831,
+      "step": 2436
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.79545085296507e-06,
+      "loss": 1.1778,
+      "step": 2437
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.7916328188464665e-06,
+      "loss": 0.9644,
+      "step": 2438
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.7878147847278634e-06,
+      "loss": 0.958,
+      "step": 2439
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.783996750609261e-06,
+      "loss": 1.4355,
+      "step": 2440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.780178716490658e-06,
+      "loss": 1.5726,
+      "step": 2441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.776360682372055e-06,
+      "loss": 1.2028,
+      "step": 2442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.772542648253453e-06,
+      "loss": 1.4899,
+      "step": 2443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.76872461413485e-06,
+      "loss": 1.2124,
+      "step": 2444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.764906580016247e-06,
+      "loss": 1.097,
+      "step": 2445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.761088545897644e-06,
+      "loss": 1.294,
+      "step": 2446
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.757270511779041e-06,
+      "loss": 1.2114,
+      "step": 2447
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.753452477660439e-06,
+      "loss": 1.0377,
+      "step": 2448
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.749634443541836e-06,
+      "loss": 1.2182,
+      "step": 2449
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.745816409423234e-06,
+      "loss": 1.359,
+      "step": 2450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.7419983753046305e-06,
+      "loss": 1.4016,
+      "step": 2451
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.7381803411860274e-06,
+      "loss": 1.1123,
+      "step": 2452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.734362307067425e-06,
+      "loss": 1.1442,
+      "step": 2453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.730544272948822e-06,
+      "loss": 1.6332,
+      "step": 2454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.72672623883022e-06,
+      "loss": 1.2632,
+      "step": 2455
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.722908204711617e-06,
+      "loss": 1.419,
+      "step": 2456
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.719090170593014e-06,
+      "loss": 1.0885,
+      "step": 2457
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.715272136474411e-06,
+      "loss": 1.1614,
+      "step": 2458
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.711454102355808e-06,
+      "loss": 1.497,
+      "step": 2459
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.707636068237205e-06,
+      "loss": 1.5183,
+      "step": 2460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.703818034118603e-06,
+      "loss": 1.1857,
+      "step": 2461
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7e-06,
+      "loss": 1.1046,
+      "step": 2462
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.696181965881398e-06,
+      "loss": 1.2982,
+      "step": 2463
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.692363931762795e-06,
+      "loss": 1.2905,
+      "step": 2464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6885458976441915e-06,
+      "loss": 1.3208,
+      "step": 2465
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.684727863525589e-06,
+      "loss": 1.0439,
+      "step": 2466
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.680909829406986e-06,
+      "loss": 1.1429,
+      "step": 2467
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.677091795288384e-06,
+      "loss": 1.1934,
+      "step": 2468
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.673273761169781e-06,
+      "loss": 1.2124,
+      "step": 2469
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.669455727051178e-06,
+      "loss": 1.4119,
+      "step": 2470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6656376929325754e-06,
+      "loss": 1.2111,
+      "step": 2471
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.661819658813972e-06,
+      "loss": 1.4178,
+      "step": 2472
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.65800162469537e-06,
+      "loss": 1.1002,
+      "step": 2473
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.654183590576767e-06,
+      "loss": 1.2924,
+      "step": 2474
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.650365556458164e-06,
+      "loss": 1.1639,
+      "step": 2475
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.646547522339562e-06,
+      "loss": 1.3091,
+      "step": 2476
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.642729488220959e-06,
+      "loss": 1.3236,
+      "step": 2477
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6389114541023555e-06,
+      "loss": 1.2179,
+      "step": 2478
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.635093419983753e-06,
+      "loss": 1.4322,
+      "step": 2479
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.63127538586515e-06,
+      "loss": 1.2464,
+      "step": 2480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.627457351746548e-06,
+      "loss": 0.957,
+      "step": 2481
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.623639317627946e-06,
+      "loss": 1.0464,
+      "step": 2482
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.619821283509342e-06,
+      "loss": 1.2555,
+      "step": 2483
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6160032493907394e-06,
+      "loss": 1.3757,
+      "step": 2484
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.612185215272136e-06,
+      "loss": 1.718,
+      "step": 2485
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.608367181153534e-06,
+      "loss": 1.0189,
+      "step": 2486
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.604549147034931e-06,
+      "loss": 1.3384,
+      "step": 2487
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.600731112916328e-06,
+      "loss": 1.2529,
+      "step": 2488
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.596913078797726e-06,
+      "loss": 1.3485,
+      "step": 2489
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.5930950446791226e-06,
+      "loss": 1.3047,
+      "step": 2490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.58927701056052e-06,
+      "loss": 1.0708,
+      "step": 2491
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.585458976441917e-06,
+      "loss": 1.2799,
+      "step": 2492
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.581640942323314e-06,
+      "loss": 1.2583,
+      "step": 2493
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.577822908204712e-06,
+      "loss": 1.1616,
+      "step": 2494
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.57400487408611e-06,
+      "loss": 1.2377,
+      "step": 2495
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.5701868399675065e-06,
+      "loss": 1.3221,
+      "step": 2496
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.5663688058489035e-06,
+      "loss": 1.1865,
+      "step": 2497
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.5625507717303e-06,
+      "loss": 1.0754,
+      "step": 2498
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.558732737611698e-06,
+      "loss": 1.4962,
+      "step": 2499
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.554914703493096e-06,
+      "loss": 1.6034,
+      "step": 2500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.551096669374492e-06,
+      "loss": 1.3732,
+      "step": 2501
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.54727863525589e-06,
+      "loss": 1.043,
+      "step": 2502
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.543460601137287e-06,
+      "loss": 1.1798,
+      "step": 2503
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.539642567018684e-06,
+      "loss": 1.3348,
+      "step": 2504
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.535824532900081e-06,
+      "loss": 1.3239,
+      "step": 2505
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.532006498781478e-06,
+      "loss": 1.1349,
+      "step": 2506
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.528188464662876e-06,
+      "loss": 1.3621,
+      "step": 2507
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.524370430544274e-06,
+      "loss": 1.1938,
+      "step": 2508
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5205523964256706e-06,
+      "loss": 1.1612,
+      "step": 2509
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5167343623070675e-06,
+      "loss": 1.2509,
+      "step": 2510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.512916328188464e-06,
+      "loss": 1.0236,
+      "step": 2511
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.509098294069862e-06,
+      "loss": 1.0134,
+      "step": 2512
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.50528025995126e-06,
+      "loss": 1.1535,
+      "step": 2513
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.501462225832657e-06,
+      "loss": 0.973,
+      "step": 2514
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.497644191714054e-06,
+      "loss": 1.1757,
+      "step": 2515
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.493826157595451e-06,
+      "loss": 1.0982,
+      "step": 2516
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.490008123476848e-06,
+      "loss": 1.0905,
+      "step": 2517
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.486190089358246e-06,
+      "loss": 1.1503,
+      "step": 2518
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.482372055239642e-06,
+      "loss": 1.3331,
+      "step": 2519
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.47855402112104e-06,
+      "loss": 0.96,
+      "step": 2520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.474735987002438e-06,
+      "loss": 1.4992,
+      "step": 2521
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.4709179528838346e-06,
+      "loss": 1.3731,
+      "step": 2522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.467099918765232e-06,
+      "loss": 1.2652,
+      "step": 2523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.463281884646628e-06,
+      "loss": 1.2439,
+      "step": 2524
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.459463850528026e-06,
+      "loss": 1.1659,
+      "step": 2525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.455645816409424e-06,
+      "loss": 1.2688,
+      "step": 2526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.451827782290821e-06,
+      "loss": 1.3544,
+      "step": 2527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.448009748172218e-06,
+      "loss": 1.3829,
+      "step": 2528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.444191714053615e-06,
+      "loss": 1.2052,
+      "step": 2529
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.440373679935012e-06,
+      "loss": 1.1235,
+      "step": 2530
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.43655564581641e-06,
+      "loss": 1.4654,
+      "step": 2531
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.432737611697807e-06,
+      "loss": 1.0127,
+      "step": 2532
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.428919577579204e-06,
+      "loss": 1.014,
+      "step": 2533
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.425101543460601e-06,
+      "loss": 1.1028,
+      "step": 2534
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.421283509341999e-06,
+      "loss": 1.1839,
+      "step": 2535
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.417465475223396e-06,
+      "loss": 1.1125,
+      "step": 2536
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.413647441104792e-06,
+      "loss": 1.4104,
+      "step": 2537
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.40982940698619e-06,
+      "loss": 1.2946,
+      "step": 2538
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.406011372867588e-06,
+      "loss": 1.0772,
+      "step": 2539
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.402193338748985e-06,
+      "loss": 1.1739,
+      "step": 2540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.3983753046303826e-06,
+      "loss": 1.3382,
+      "step": 2541
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.394557270511779e-06,
+      "loss": 1.0707,
+      "step": 2542
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.390739236393176e-06,
+      "loss": 1.2706,
+      "step": 2543
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.386921202274574e-06,
+      "loss": 1.2741,
+      "step": 2544
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.383103168155971e-06,
+      "loss": 1.203,
+      "step": 2545
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.379285134037368e-06,
+      "loss": 1.1616,
+      "step": 2546
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.375467099918765e-06,
+      "loss": 0.9967,
+      "step": 2547
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.371649065800163e-06,
+      "loss": 1.0492,
+      "step": 2548
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.36783103168156e-06,
+      "loss": 1.429,
+      "step": 2549
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.364012997562957e-06,
+      "loss": 1.8079,
+      "step": 2550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.360194963444354e-06,
+      "loss": 1.0287,
+      "step": 2551
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.356376929325752e-06,
+      "loss": 1.1025,
+      "step": 2552
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.352558895207149e-06,
+      "loss": 1.456,
+      "step": 2553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.3487408610885466e-06,
+      "loss": 1.3382,
+      "step": 2554
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.344922826969943e-06,
+      "loss": 1.4377,
+      "step": 2555
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.34110479285134e-06,
+      "loss": 1.1077,
+      "step": 2556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.337286758732738e-06,
+      "loss": 1.3626,
+      "step": 2557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.333468724614135e-06,
+      "loss": 0.979,
+      "step": 2558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.329650690495533e-06,
+      "loss": 1.1074,
+      "step": 2559
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.325832656376929e-06,
+      "loss": 1.0535,
+      "step": 2560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.322014622258327e-06,
+      "loss": 1.0996,
+      "step": 2561
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.318196588139724e-06,
+      "loss": 1.275,
+      "step": 2562
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.314378554021121e-06,
+      "loss": 1.1306,
+      "step": 2563
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.310560519902518e-06,
+      "loss": 1.3867,
+      "step": 2564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.306742485783915e-06,
+      "loss": 1.2062,
+      "step": 2565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.302924451665313e-06,
+      "loss": 1.3904,
+      "step": 2566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.2991064175467106e-06,
+      "loss": 1.3346,
+      "step": 2567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.2952883834281075e-06,
+      "loss": 1.2204,
+      "step": 2568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.291470349309504e-06,
+      "loss": 1.1196,
+      "step": 2569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.287652315190902e-06,
+      "loss": 1.3332,
+      "step": 2570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.283834281072299e-06,
+      "loss": 1.2179,
+      "step": 2571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.280016246953697e-06,
+      "loss": 1.3152,
+      "step": 2572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.276198212835094e-06,
+      "loss": 1.3106,
+      "step": 2573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.272380178716491e-06,
+      "loss": 1.1769,
+      "step": 2574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.268562144597888e-06,
+      "loss": 1.2831,
+      "step": 2575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.264744110479285e-06,
+      "loss": 1.3894,
+      "step": 2576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.260926076360683e-06,
+      "loss": 1.0413,
+      "step": 2577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.257108042242079e-06,
+      "loss": 1.2491,
+      "step": 2578
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.253290008123477e-06,
+      "loss": 1.3427,
+      "step": 2579
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.249471974004875e-06,
+      "loss": 1.3655,
+      "step": 2580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2456539398862715e-06,
+      "loss": 1.2606,
+      "step": 2581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.241835905767668e-06,
+      "loss": 1.2398,
+      "step": 2582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.238017871649066e-06,
+      "loss": 1.0044,
+      "step": 2583
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.234199837530463e-06,
+      "loss": 1.2496,
+      "step": 2584
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.230381803411861e-06,
+      "loss": 1.2933,
+      "step": 2585
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.226563769293258e-06,
+      "loss": 1.3395,
+      "step": 2586
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.222745735174655e-06,
+      "loss": 1.2368,
+      "step": 2587
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.218927701056052e-06,
+      "loss": 1.1795,
+      "step": 2588
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.215109666937449e-06,
+      "loss": 1.1273,
+      "step": 2589
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.211291632818847e-06,
+      "loss": 1.272,
+      "step": 2590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.207473598700244e-06,
+      "loss": 1.3743,
+      "step": 2591
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.203655564581641e-06,
+      "loss": 1.0504,
+      "step": 2592
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.199837530463039e-06,
+      "loss": 0.9864,
+      "step": 2593
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1960194963444355e-06,
+      "loss": 1.3541,
+      "step": 2594
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.192201462225833e-06,
+      "loss": 1.0984,
+      "step": 2595
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.188383428107229e-06,
+      "loss": 1.4958,
+      "step": 2596
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.184565393988627e-06,
+      "loss": 1.3144,
+      "step": 2597
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.180747359870025e-06,
+      "loss": 1.1359,
+      "step": 2598
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.176929325751422e-06,
+      "loss": 1.1191,
+      "step": 2599
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1731112916328195e-06,
+      "loss": 1.2221,
+      "step": 2600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.169293257514216e-06,
+      "loss": 1.0044,
+      "step": 2601
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.165475223395613e-06,
+      "loss": 1.0461,
+      "step": 2602
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.161657189277011e-06,
+      "loss": 1.4078,
+      "step": 2603
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157839155158408e-06,
+      "loss": 1.61,
+      "step": 2604
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154021121039805e-06,
+      "loss": 1.1901,
+      "step": 2605
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.150203086921203e-06,
+      "loss": 1.6118,
+      "step": 2606
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1463850528025995e-06,
+      "loss": 1.0597,
+      "step": 2607
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142567018683997e-06,
+      "loss": 1.1739,
+      "step": 2608
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138748984565394e-06,
+      "loss": 1.321,
+      "step": 2609
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.134930950446791e-06,
+      "loss": 1.38,
+      "step": 2610
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.131112916328189e-06,
+      "loss": 1.5087,
+      "step": 2611
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.127294882209586e-06,
+      "loss": 1.0078,
+      "step": 2612
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1234768480909835e-06,
+      "loss": 1.1351,
+      "step": 2613
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.11965881397238e-06,
+      "loss": 1.2573,
+      "step": 2614
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.115840779853777e-06,
+      "loss": 1.2799,
+      "step": 2615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112022745735175e-06,
+      "loss": 1.0016,
+      "step": 2616
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108204711616572e-06,
+      "loss": 1.2017,
+      "step": 2617
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10438667749797e-06,
+      "loss": 1.0752,
+      "step": 2618
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100568643379367e-06,
+      "loss": 1.0247,
+      "step": 2619
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0967506092607635e-06,
+      "loss": 1.098,
+      "step": 2620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092932575142161e-06,
+      "loss": 1.3535,
+      "step": 2621
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.089114541023558e-06,
+      "loss": 1.1574,
+      "step": 2622
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085296506904955e-06,
+      "loss": 1.2926,
+      "step": 2623
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081478472786353e-06,
+      "loss": 1.2264,
+      "step": 2624
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07766043866775e-06,
+      "loss": 1.4944,
+      "step": 2625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0738424045491475e-06,
+      "loss": 1.1803,
+      "step": 2626
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070024370430544e-06,
+      "loss": 1.1232,
+      "step": 2627
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066206336311941e-06,
+      "loss": 1.3007,
+      "step": 2628
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.062388302193339e-06,
+      "loss": 1.4422,
+      "step": 2629
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.058570268074736e-06,
+      "loss": 1.1745,
+      "step": 2630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.054752233956134e-06,
+      "loss": 1.3986,
+      "step": 2631
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050934199837531e-06,
+      "loss": 1.2824,
+      "step": 2632
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0471161657189275e-06,
+      "loss": 1.2932,
+      "step": 2633
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043298131600325e-06,
+      "loss": 1.2496,
+      "step": 2634
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.039480097481722e-06,
+      "loss": 0.9785,
+      "step": 2635
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.03566206336312e-06,
+      "loss": 1.2204,
+      "step": 2636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031844029244517e-06,
+      "loss": 1.1285,
+      "step": 2637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028025995125914e-06,
+      "loss": 1.161,
+      "step": 2638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0242079610073115e-06,
+      "loss": 1.316,
+      "step": 2639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020389926888708e-06,
+      "loss": 1.2626,
+      "step": 2640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.016571892770105e-06,
+      "loss": 1.4363,
+      "step": 2641
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.012753858651503e-06,
+      "loss": 0.994,
+      "step": 2642
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0089358245329e-06,
+      "loss": 1.1917,
+      "step": 2643
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.005117790414298e-06,
+      "loss": 1.1333,
+      "step": 2644
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0012997562956955e-06,
+      "loss": 1.3813,
+      "step": 2645
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9974817221770915e-06,
+      "loss": 1.265,
+      "step": 2646
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.993663688058489e-06,
+      "loss": 1.2585,
+      "step": 2647
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.989845653939886e-06,
+      "loss": 1.4186,
+      "step": 2648
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.986027619821284e-06,
+      "loss": 1.4532,
+      "step": 2649
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.982209585702681e-06,
+      "loss": 1.1907,
+      "step": 2650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.978391551584078e-06,
+      "loss": 1.3656,
+      "step": 2651
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9745735174654755e-06,
+      "loss": 1.1755,
+      "step": 2652
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.970755483346872e-06,
+      "loss": 1.0387,
+      "step": 2653
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.96693744922827e-06,
+      "loss": 1.1671,
+      "step": 2654
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.963119415109667e-06,
+      "loss": 1.2362,
+      "step": 2655
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.959301380991064e-06,
+      "loss": 1.4558,
+      "step": 2656
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.955483346872462e-06,
+      "loss": 1.0336,
+      "step": 2657
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.951665312753859e-06,
+      "loss": 1.2656,
+      "step": 2658
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9478472786352555e-06,
+      "loss": 1.2486,
+      "step": 2659
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.944029244516653e-06,
+      "loss": 1.0736,
+      "step": 2660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.94021121039805e-06,
+      "loss": 1.129,
+      "step": 2661
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.936393176279448e-06,
+      "loss": 1.1074,
+      "step": 2662
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.932575142160846e-06,
+      "loss": 1.2459,
+      "step": 2663
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928757108042242e-06,
+      "loss": 1.2108,
+      "step": 2664
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9249390739236395e-06,
+      "loss": 1.2379,
+      "step": 2665
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9211210398050364e-06,
+      "loss": 1.1492,
+      "step": 2666
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917303005686434e-06,
+      "loss": 1.5371,
+      "step": 2667
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.913484971567831e-06,
+      "loss": 1.0087,
+      "step": 2668
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.909666937449228e-06,
+      "loss": 1.0896,
+      "step": 2669
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.905848903330626e-06,
+      "loss": 1.5909,
+      "step": 2670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.902030869212023e-06,
+      "loss": 1.1947,
+      "step": 2671
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.89821283509342e-06,
+      "loss": 1.1605,
+      "step": 2672
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.894394800974817e-06,
+      "loss": 1.2766,
+      "step": 2673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.890576766856214e-06,
+      "loss": 1.0039,
+      "step": 2674
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.886758732737612e-06,
+      "loss": 1.1453,
+      "step": 2675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.88294069861901e-06,
+      "loss": 1.4472,
+      "step": 2676
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.879122664500407e-06,
+      "loss": 1.1324,
+      "step": 2677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8753046303818035e-06,
+      "loss": 1.4442,
+      "step": 2678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8714865962632004e-06,
+      "loss": 1.1516,
+      "step": 2679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.867668562144598e-06,
+      "loss": 1.2522,
+      "step": 2680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.863850528025996e-06,
+      "loss": 1.2547,
+      "step": 2681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.860032493907392e-06,
+      "loss": 1.208,
+      "step": 2682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.85621445978879e-06,
+      "loss": 1.2998,
+      "step": 2683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.852396425670187e-06,
+      "loss": 1.3195,
+      "step": 2684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.848578391551584e-06,
+      "loss": 1.3385,
+      "step": 2685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.844760357432981e-06,
+      "loss": 1.1229,
+      "step": 2686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.840942323314378e-06,
+      "loss": 1.0619,
+      "step": 2687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.837124289195776e-06,
+      "loss": 1.3792,
+      "step": 2688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.833306255077173e-06,
+      "loss": 1.3146,
+      "step": 2689
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.829488220958571e-06,
+      "loss": 1.0396,
+      "step": 2690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8256701868399675e-06,
+      "loss": 1.613,
+      "step": 2691
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8218521527213645e-06,
+      "loss": 1.1738,
+      "step": 2692
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.818034118602762e-06,
+      "loss": 1.639,
+      "step": 2693
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8142160844841595e-06,
+      "loss": 1.2657,
+      "step": 2694
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8103980503655564e-06,
+      "loss": 1.3639,
+      "step": 2695
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.806580016246954e-06,
+      "loss": 1.2069,
+      "step": 2696
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8027619821283507e-06,
+      "loss": 1.0169,
+      "step": 2697
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7989439480097484e-06,
+      "loss": 1.3674,
+      "step": 2698
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7951259138911458e-06,
+      "loss": 1.034,
+      "step": 2699
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7913078797725427e-06,
+      "loss": 1.2138,
+      "step": 2700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.78748984565394e-06,
+      "loss": 1.2788,
+      "step": 2701
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.783671811535337e-06,
+      "loss": 1.1479,
+      "step": 2702
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7798537774167346e-06,
+      "loss": 1.0876,
+      "step": 2703
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.776035743298132e-06,
+      "loss": 1.4059,
+      "step": 2704
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.772217709179529e-06,
+      "loss": 1.1834,
+      "step": 2705
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7683996750609262e-06,
+      "loss": 1.2754,
+      "step": 2706
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.764581640942324e-06,
+      "loss": 1.3408,
+      "step": 2707
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7607636068237204e-06,
+      "loss": 1.2457,
+      "step": 2708
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.756945572705118e-06,
+      "loss": 1.2171,
+      "step": 2709
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7531275385865147e-06,
+      "loss": 1.071,
+      "step": 2710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7493095044679124e-06,
+      "loss": 1.4609,
+      "step": 2711
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7454914703493098e-06,
+      "loss": 1.3085,
+      "step": 2712
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7416734362307067e-06,
+      "loss": 1.2867,
+      "step": 2713
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.7378554021121044e-06,
+      "loss": 1.21,
+      "step": 2714
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.734037367993501e-06,
+      "loss": 1.5404,
+      "step": 2715
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7302193338748987e-06,
+      "loss": 1.1934,
+      "step": 2716
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.726401299756296e-06,
+      "loss": 1.1393,
+      "step": 2717
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.722583265637693e-06,
+      "loss": 1.2781,
+      "step": 2718
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7187652315190902e-06,
+      "loss": 1.0681,
+      "step": 2719
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.714947197400488e-06,
+      "loss": 1.0527,
+      "step": 2720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.711129163281885e-06,
+      "loss": 1.171,
+      "step": 2721
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.707311129163282e-06,
+      "loss": 1.2593,
+      "step": 2722
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.703493095044679e-06,
+      "loss": 1.0716,
+      "step": 2723
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6996750609260764e-06,
+      "loss": 1.3597,
+      "step": 2724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.695857026807474e-06,
+      "loss": 1.2401,
+      "step": 2725
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6920389926888707e-06,
+      "loss": 1.3368,
+      "step": 2726
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6882209585702684e-06,
+      "loss": 1.2246,
+      "step": 2727
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6844029244516653e-06,
+      "loss": 1.1718,
+      "step": 2728
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6805848903330627e-06,
+      "loss": 1.124,
+      "step": 2729
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6767668562144604e-06,
+      "loss": 1.1193,
+      "step": 2730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.672948822095857e-06,
+      "loss": 1.1004,
+      "step": 2731
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6691307879772547e-06,
+      "loss": 0.9524,
+      "step": 2732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.665312753858651e-06,
+      "loss": 1.2145,
+      "step": 2733
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.661494719740049e-06,
+      "loss": 1.4569,
+      "step": 2734
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6576766856214462e-06,
+      "loss": 1.3217,
+      "step": 2735
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.653858651502843e-06,
+      "loss": 1.4451,
+      "step": 2736
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6500406173842405e-06,
+      "loss": 1.3742,
+      "step": 2737
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.646222583265638e-06,
+      "loss": 1.2175,
+      "step": 2738
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.642404549147035e-06,
+      "loss": 1.049,
+      "step": 2739
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6385865150284324e-06,
+      "loss": 1.2291,
+      "step": 2740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6347684809098294e-06,
+      "loss": 1.3836,
+      "step": 2741
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6309504467912267e-06,
+      "loss": 1.5459,
+      "step": 2742
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6271324126726244e-06,
+      "loss": 1.1501,
+      "step": 2743
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623314378554021e-06,
+      "loss": 1.1965,
+      "step": 2744
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6194963444354187e-06,
+      "loss": 1.1924,
+      "step": 2745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6156783103168156e-06,
+      "loss": 1.2319,
+      "step": 2746
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.611860276198213e-06,
+      "loss": 1.1957,
+      "step": 2747
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6080422420796107e-06,
+      "loss": 1.2754,
+      "step": 2748
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.604224207961007e-06,
+      "loss": 1.1421,
+      "step": 2749
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.600406173842405e-06,
+      "loss": 1.3871,
+      "step": 2750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.5965881397238022e-06,
+      "loss": 1.4682,
+      "step": 2751
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.592770105605199e-06,
+      "loss": 1.1667,
+      "step": 2752
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5889520714865965e-06,
+      "loss": 1.4151,
+      "step": 2753
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5851340373679934e-06,
+      "loss": 1.0841,
+      "step": 2754
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.581316003249391e-06,
+      "loss": 1.2595,
+      "step": 2755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5774979691307884e-06,
+      "loss": 1.1497,
+      "step": 2756
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5736799350121853e-06,
+      "loss": 1.0655,
+      "step": 2757
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5698619008935827e-06,
+      "loss": 1.0764,
+      "step": 2758
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5660438667749796e-06,
+      "loss": 1.3389,
+      "step": 2759
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.562225832656377e-06,
+      "loss": 1.049,
+      "step": 2760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.5584077985377747e-06,
+      "loss": 1.1197,
+      "step": 2761
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.554589764419171e-06,
+      "loss": 1.1994,
+      "step": 2762
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.550771730300569e-06,
+      "loss": 1.2619,
+      "step": 2763
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.546953696181966e-06,
+      "loss": 1.0832,
+      "step": 2764
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.543135662063363e-06,
+      "loss": 1.2195,
+      "step": 2765
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.539317627944761e-06,
+      "loss": 1.3226,
+      "step": 2766
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5354995938261574e-06,
+      "loss": 1.3226,
+      "step": 2767
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.531681559707555e-06,
+      "loss": 1.1889,
+      "step": 2768
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5278635255889525e-06,
+      "loss": 1.1914,
+      "step": 2769
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5240454914703494e-06,
+      "loss": 1.0473,
+      "step": 2770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5202274573517467e-06,
+      "loss": 1.1167,
+      "step": 2771
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.516409423233144e-06,
+      "loss": 1.4741,
+      "step": 2772
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5125913891145413e-06,
+      "loss": 1.0644,
+      "step": 2773
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5087733549959383e-06,
+      "loss": 1.0161,
+      "step": 2774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5049553208773356e-06,
+      "loss": 1.1959,
+      "step": 2775
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.501137286758733e-06,
+      "loss": 1.2554,
+      "step": 2776
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4973192526401302e-06,
+      "loss": 1.0731,
+      "step": 2777
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.493501218521527e-06,
+      "loss": 1.1188,
+      "step": 2778
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4896831844029245e-06,
+      "loss": 1.188,
+      "step": 2779
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.485865150284322e-06,
+      "loss": 0.9127,
+      "step": 2780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.482047116165719e-06,
+      "loss": 1.0495,
+      "step": 2781
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4782290820471165e-06,
+      "loss": 1.0693,
+      "step": 2782
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4744110479285134e-06,
+      "loss": 1.1227,
+      "step": 2783
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4705930138099107e-06,
+      "loss": 1.2625,
+      "step": 2784
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.466774979691308e-06,
+      "loss": 1.196,
+      "step": 2785
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4629569455727054e-06,
+      "loss": 1.2925,
+      "step": 2786
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4591389114541023e-06,
+      "loss": 1.2381,
+      "step": 2787
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4553208773354996e-06,
+      "loss": 1.1737,
+      "step": 2788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.451502843216897e-06,
+      "loss": 1.3673,
+      "step": 2789
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4476848090982942e-06,
+      "loss": 1.1858,
+      "step": 2790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4438667749796916e-06,
+      "loss": 1.3298,
+      "step": 2791
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4400487408610885e-06,
+      "loss": 1.0504,
+      "step": 2792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4362307067424862e-06,
+      "loss": 1.0938,
+      "step": 2793
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.432412672623883e-06,
+      "loss": 1.1415,
+      "step": 2794
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4285946385052805e-06,
+      "loss": 1.4619,
+      "step": 2795
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4247766043866774e-06,
+      "loss": 1.514,
+      "step": 2796
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4209585702680747e-06,
+      "loss": 1.0813,
+      "step": 2797
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.417140536149472e-06,
+      "loss": 1.8013,
+      "step": 2798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4133225020308694e-06,
+      "loss": 1.301,
+      "step": 2799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4095044679122667e-06,
+      "loss": 1.2865,
+      "step": 2800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.4056864337936636e-06,
+      "loss": 1.2078,
+      "step": 2801
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.4018683996750614e-06,
+      "loss": 1.2368,
+      "step": 2802
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3980503655564583e-06,
+      "loss": 1.0792,
+      "step": 2803
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3942323314378556e-06,
+      "loss": 1.0086,
+      "step": 2804
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3904142973192525e-06,
+      "loss": 1.2497,
+      "step": 2805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.38659626320065e-06,
+      "loss": 1.0812,
+      "step": 2806
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3827782290820476e-06,
+      "loss": 1.3046,
+      "step": 2807
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3789601949634445e-06,
+      "loss": 1.0834,
+      "step": 2808
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.375142160844842e-06,
+      "loss": 1.1275,
+      "step": 2809
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3713241267262387e-06,
+      "loss": 1.6953,
+      "step": 2810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3675060926076365e-06,
+      "loss": 1.3275,
+      "step": 2811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3636880584890334e-06,
+      "loss": 1.0934,
+      "step": 2812
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3598700243704307e-06,
+      "loss": 1.4386,
+      "step": 2813
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3560519902518276e-06,
+      "loss": 1.0478,
+      "step": 2814
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.352233956133225e-06,
+      "loss": 1.2628,
+      "step": 2815
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3484159220146227e-06,
+      "loss": 1.2602,
+      "step": 2816
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3445978878960196e-06,
+      "loss": 1.3285,
+      "step": 2817
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.340779853777417e-06,
+      "loss": 1.2311,
+      "step": 2818
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.336961819658814e-06,
+      "loss": 1.0928,
+      "step": 2819
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3331437855402116e-06,
+      "loss": 1.0455,
+      "step": 2820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3293257514216085e-06,
+      "loss": 1.0765,
+      "step": 2821
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.325507717303006e-06,
+      "loss": 1.0833,
+      "step": 2822
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3216896831844027e-06,
+      "loss": 1.2094,
+      "step": 2823
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.3178716490658005e-06,
+      "loss": 1.2618,
+      "step": 2824
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.314053614947198e-06,
+      "loss": 1.055,
+      "step": 2825
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.3102355808285947e-06,
+      "loss": 1.3527,
+      "step": 2826
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.306417546709992e-06,
+      "loss": 1.2093,
+      "step": 2827
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.302599512591389e-06,
+      "loss": 1.0981,
+      "step": 2828
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.2987814784727867e-06,
+      "loss": 1.6305,
+      "step": 2829
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.2949634443541836e-06,
+      "loss": 1.4024,
+      "step": 2830
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.291145410235581e-06,
+      "loss": 1.5047,
+      "step": 2831
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.2873273761169783e-06,
+      "loss": 1.4443,
+      "step": 2832
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.2835093419983756e-06,
+      "loss": 1.2602,
+      "step": 2833
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.279691307879773e-06,
+      "loss": 1.0726,
+      "step": 2834
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.27587327376117e-06,
+      "loss": 1.5362,
+      "step": 2835
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.272055239642567e-06,
+      "loss": 1.207,
+      "step": 2836
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.268237205523964e-06,
+      "loss": 1.191,
+      "step": 2837
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.264419171405362e-06,
+      "loss": 1.0208,
+      "step": 2838
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.2606011372867587e-06,
+      "loss": 1.3693,
+      "step": 2839
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.256783103168156e-06,
+      "loss": 1.2076,
+      "step": 2840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.2529650690495534e-06,
+      "loss": 1.119,
+      "step": 2841
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.2491470349309507e-06,
+      "loss": 1.2194,
+      "step": 2842
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.245329000812348e-06,
+      "loss": 1.2884,
+      "step": 2843
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.241510966693745e-06,
+      "loss": 1.1694,
+      "step": 2844
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.2376929325751423e-06,
+      "loss": 1.2258,
+      "step": 2845
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.233874898456539e-06,
+      "loss": 1.2361,
+      "step": 2846
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.230056864337937e-06,
+      "loss": 1.2387,
+      "step": 2847
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.226238830219334e-06,
+      "loss": 1.3695,
+      "step": 2848
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.222420796100731e-06,
+      "loss": 1.3181,
+      "step": 2849
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.2186027619821285e-06,
+      "loss": 1.3082,
+      "step": 2850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.214784727863526e-06,
+      "loss": 1.3427,
+      "step": 2851
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.210966693744923e-06,
+      "loss": 0.9888,
+      "step": 2852
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.20714865962632e-06,
+      "loss": 0.7959,
+      "step": 2853
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.2033306255077174e-06,
+      "loss": 1.0005,
+      "step": 2854
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.1995125913891147e-06,
+      "loss": 1.4307,
+      "step": 2855
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.195694557270512e-06,
+      "loss": 1.3072,
+      "step": 2856
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.191876523151909e-06,
+      "loss": 1.3744,
+      "step": 2857
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.1880584890333063e-06,
+      "loss": 1.1169,
+      "step": 2858
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.1842404549147036e-06,
+      "loss": 1.4182,
+      "step": 2859
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.180422420796101e-06,
+      "loss": 1.0297,
+      "step": 2860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.1766043866774983e-06,
+      "loss": 1.0172,
+      "step": 2861
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.172786352558895e-06,
+      "loss": 1.0397,
+      "step": 2862
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1689683184402925e-06,
+      "loss": 1.0614,
+      "step": 2863
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.16515028432169e-06,
+      "loss": 1.1979,
+      "step": 2864
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.161332250203087e-06,
+      "loss": 1.2538,
+      "step": 2865
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.157514216084484e-06,
+      "loss": 1.0802,
+      "step": 2866
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1536961819658814e-06,
+      "loss": 0.9884,
+      "step": 2867
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1498781478472787e-06,
+      "loss": 1.247,
+      "step": 2868
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.146060113728676e-06,
+      "loss": 1.047,
+      "step": 2869
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1422420796100734e-06,
+      "loss": 1.2294,
+      "step": 2870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1384240454914703e-06,
+      "loss": 0.9381,
+      "step": 2871
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1346060113728676e-06,
+      "loss": 1.1256,
+      "step": 2872
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.130787977254265e-06,
+      "loss": 1.6428,
+      "step": 2873
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.1269699431356623e-06,
+      "loss": 1.2112,
+      "step": 2874
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.123151909017059e-06,
+      "loss": 1.215,
+      "step": 2875
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.1193338748984565e-06,
+      "loss": 1.2376,
+      "step": 2876
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.115515840779854e-06,
+      "loss": 1.0358,
+      "step": 2877
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.111697806661251e-06,
+      "loss": 1.0281,
+      "step": 2878
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.1078797725426485e-06,
+      "loss": 1.1889,
+      "step": 2879
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.1040617384240454e-06,
+      "loss": 1.195,
+      "step": 2880
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.1002437043054427e-06,
+      "loss": 1.0753,
+      "step": 2881
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.09642567018684e-06,
+      "loss": 1.1976,
+      "step": 2882
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0926076360682374e-06,
+      "loss": 1.1829,
+      "step": 2883
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0887896019496347e-06,
+      "loss": 1.3467,
+      "step": 2884
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0849715678310316e-06,
+      "loss": 1.56,
+      "step": 2885
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0811535337124294e-06,
+      "loss": 1.1941,
+      "step": 2886
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0773354995938263e-06,
+      "loss": 1.5097,
+      "step": 2887
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0735174654752236e-06,
+      "loss": 1.3461,
+      "step": 2888
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0696994313566205e-06,
+      "loss": 1.1204,
+      "step": 2889
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.065881397238018e-06,
+      "loss": 1.5029,
+      "step": 2890
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.062063363119415e-06,
+      "loss": 1.1294,
+      "step": 2891
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0582453290008125e-06,
+      "loss": 1.1849,
+      "step": 2892
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.05442729488221e-06,
+      "loss": 0.9588,
+      "step": 2893
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0506092607636068e-06,
+      "loss": 1.2241,
+      "step": 2894
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0467912266450045e-06,
+      "loss": 1.2013,
+      "step": 2895
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0429731925264014e-06,
+      "loss": 1.3803,
+      "step": 2896
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0391551584077987e-06,
+      "loss": 1.2171,
+      "step": 2897
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0353371242891956e-06,
+      "loss": 1.1938,
+      "step": 2898
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.031519090170593e-06,
+      "loss": 1.4031,
+      "step": 2899
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0277010560519903e-06,
+      "loss": 1.8172,
+      "step": 2900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0238830219333876e-06,
+      "loss": 1.1701,
+      "step": 2901
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.020064987814785e-06,
+      "loss": 1.5422,
+      "step": 2902
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.016246953696182e-06,
+      "loss": 1.0069,
+      "step": 2903
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0124289195775796e-06,
+      "loss": 1.1549,
+      "step": 2904
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0086108854589765e-06,
+      "loss": 1.1101,
+      "step": 2905
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.004792851340374e-06,
+      "loss": 1.0022,
+      "step": 2906
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0009748172217708e-06,
+      "loss": 1.0385,
+      "step": 2907
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.9971567831031685e-06,
+      "loss": 1.1097,
+      "step": 2908
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.9933387489845654e-06,
+      "loss": 1.4117,
+      "step": 2909
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.9895207148659628e-06,
+      "loss": 1.1758,
+      "step": 2910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.98570268074736e-06,
+      "loss": 1.2225,
+      "step": 2911
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.981884646628757e-06,
+      "loss": 1.1111,
+      "step": 2912
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.9780666125101547e-06,
+      "loss": 1.1619,
+      "step": 2913
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.9742485783915516e-06,
+      "loss": 1.1312,
+      "step": 2914
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.970430544272949e-06,
+      "loss": 1.0345,
+      "step": 2915
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.966612510154346e-06,
+      "loss": 1.4091,
+      "step": 2916
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.9627944760357436e-06,
+      "loss": 1.1814,
+      "step": 2917
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.9589764419171405e-06,
+      "loss": 1.5463,
+      "step": 2918
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.955158407798538e-06,
+      "loss": 1.3325,
+      "step": 2919
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.951340373679935e-06,
+      "loss": 1.4518,
+      "step": 2920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.947522339561332e-06,
+      "loss": 1.3925,
+      "step": 2921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.94370430544273e-06,
+      "loss": 1.0626,
+      "step": 2922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.9398862713241268e-06,
+      "loss": 1.0669,
+      "step": 2923
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.936068237205524e-06,
+      "loss": 1.3293,
+      "step": 2924
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.932250203086921e-06,
+      "loss": 0.7243,
+      "step": 2925
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.9284321689683187e-06,
+      "loss": 1.2547,
+      "step": 2926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.9246141348497157e-06,
+      "loss": 1.3172,
+      "step": 2927
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.920796100731113e-06,
+      "loss": 1.2501,
+      "step": 2928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.9169780666125103e-06,
+      "loss": 1.1075,
+      "step": 2929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.9131600324939072e-06,
+      "loss": 1.4591,
+      "step": 2930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.909341998375305e-06,
+      "loss": 1.6315,
+      "step": 2931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.905523964256702e-06,
+      "loss": 1.1993,
+      "step": 2932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.901705930138099e-06,
+      "loss": 0.9308,
+      "step": 2933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.897887896019496e-06,
+      "loss": 1.251,
+      "step": 2934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.894069861900894e-06,
+      "loss": 1.1851,
+      "step": 2935
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.890251827782291e-06,
+      "loss": 1.2112,
+      "step": 2936
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.886433793663688e-06,
+      "loss": 1.2096,
+      "step": 2937
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8826157595450854e-06,
+      "loss": 1.3684,
+      "step": 2938
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8787977254264828e-06,
+      "loss": 1.1459,
+      "step": 2939
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.87497969130788e-06,
+      "loss": 1.1667,
+      "step": 2940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.871161657189277e-06,
+      "loss": 1.2389,
+      "step": 2941
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8673436230706743e-06,
+      "loss": 1.1586,
+      "step": 2942
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8635255889520712e-06,
+      "loss": 1.0839,
+      "step": 2943
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.859707554833469e-06,
+      "loss": 1.3129,
+      "step": 2944
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8558895207148663e-06,
+      "loss": 1.3999,
+      "step": 2945
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8520714865962632e-06,
+      "loss": 1.1941,
+      "step": 2946
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8482534524776605e-06,
+      "loss": 1.0675,
+      "step": 2947
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.844435418359058e-06,
+      "loss": 1.5633,
+      "step": 2948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.840617384240455e-06,
+      "loss": 1.0541,
+      "step": 2949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.836799350121852e-06,
+      "loss": 1.2487,
+      "step": 2950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8329813160032494e-06,
+      "loss": 1.282,
+      "step": 2951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8291632818846463e-06,
+      "loss": 1.2276,
+      "step": 2952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.825345247766044e-06,
+      "loss": 1.3582,
+      "step": 2953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8215272136474414e-06,
+      "loss": 1.42,
+      "step": 2954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8177091795288383e-06,
+      "loss": 1.2196,
+      "step": 2955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8138911454102357e-06,
+      "loss": 1.1703,
+      "step": 2956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.810073111291633e-06,
+      "loss": 1.1551,
+      "step": 2957
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8062550771730303e-06,
+      "loss": 1.1537,
+      "step": 2958
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.8024370430544272e-06,
+      "loss": 1.4934,
+      "step": 2959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7986190089358246e-06,
+      "loss": 1.4635,
+      "step": 2960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.794800974817222e-06,
+      "loss": 1.3167,
+      "step": 2961
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7909829406986192e-06,
+      "loss": 1.3083,
+      "step": 2962
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7871649065800165e-06,
+      "loss": 1.2118,
+      "step": 2963
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7833468724614134e-06,
+      "loss": 1.0067,
+      "step": 2964
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7795288383428108e-06,
+      "loss": 1.6314,
+      "step": 2965
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.775710804224208e-06,
+      "loss": 1.3792,
+      "step": 2966
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7718927701056054e-06,
+      "loss": 0.8231,
+      "step": 2967
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7680747359870023e-06,
+      "loss": 1.1228,
+      "step": 2968
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7642567018683997e-06,
+      "loss": 1.0727,
+      "step": 2969
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.760438667749797e-06,
+      "loss": 1.2436,
+      "step": 2970
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7566206336311943e-06,
+      "loss": 1.0953,
+      "step": 2971
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7528025995125917e-06,
+      "loss": 1.1383,
+      "step": 2972
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7489845653939886e-06,
+      "loss": 1.1838,
+      "step": 2973
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.745166531275386e-06,
+      "loss": 1.2835,
+      "step": 2974
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7413484971567832e-06,
+      "loss": 1.0012,
+      "step": 2975
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7375304630381806e-06,
+      "loss": 0.9537,
+      "step": 2976
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7337124289195775e-06,
+      "loss": 1.248,
+      "step": 2977
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.729894394800975e-06,
+      "loss": 1.1701,
+      "step": 2978
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.726076360682372e-06,
+      "loss": 1.3665,
+      "step": 2979
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7222583265637694e-06,
+      "loss": 1.2099,
+      "step": 2980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7184402924451668e-06,
+      "loss": 1.3204,
+      "step": 2981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7146222583265637e-06,
+      "loss": 1.0174,
+      "step": 2982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.710804224207961e-06,
+      "loss": 1.3942,
+      "step": 2983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7069861900893583e-06,
+      "loss": 1.0234,
+      "step": 2984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7031681559707557e-06,
+      "loss": 1.2999,
+      "step": 2985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6993501218521526e-06,
+      "loss": 1.2503,
+      "step": 2986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.69553208773355e-06,
+      "loss": 1.1402,
+      "step": 2987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6917140536149477e-06,
+      "loss": 1.2125,
+      "step": 2988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6878960194963446e-06,
+      "loss": 1.3393,
+      "step": 2989
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.684077985377742e-06,
+      "loss": 1.3247,
+      "step": 2990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.680259951259139e-06,
+      "loss": 1.033,
+      "step": 2991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6764419171405366e-06,
+      "loss": 1.1087,
+      "step": 2992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6726238830219335e-06,
+      "loss": 1.0575,
+      "step": 2993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6688058489033308e-06,
+      "loss": 1.184,
+      "step": 2994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6649878147847277e-06,
+      "loss": 1.0611,
+      "step": 2995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.661169780666125e-06,
+      "loss": 1.0816,
+      "step": 2996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.6573517465475228e-06,
+      "loss": 1.199,
+      "step": 2997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6535337124289197e-06,
+      "loss": 1.2008,
+      "step": 2998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.649715678310317e-06,
+      "loss": 1.0125,
+      "step": 2999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.645897644191714e-06,
+      "loss": 1.1655,
+      "step": 3000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6420796100731117e-06,
+      "loss": 1.1978,
+      "step": 3001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6382615759545086e-06,
+      "loss": 1.3752,
+      "step": 3002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.634443541835906e-06,
+      "loss": 1.1532,
+      "step": 3003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.630625507717303e-06,
+      "loss": 1.476,
+      "step": 3004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6268074735987e-06,
+      "loss": 1.0439,
+      "step": 3005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.622989439480098e-06,
+      "loss": 1.3042,
+      "step": 3006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.619171405361495e-06,
+      "loss": 1.1535,
+      "step": 3007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.615353371242892e-06,
+      "loss": 0.9431,
+      "step": 3008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.611535337124289e-06,
+      "loss": 1.4894,
+      "step": 3009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6077173030056868e-06,
+      "loss": 1.2985,
+      "step": 3010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6038992688870837e-06,
+      "loss": 1.0841,
+      "step": 3011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.600081234768481e-06,
+      "loss": 1.335,
+      "step": 3012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5962632006498783e-06,
+      "loss": 1.1589,
+      "step": 3013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5924451665312753e-06,
+      "loss": 1.3579,
+      "step": 3014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.588627132412673e-06,
+      "loss": 1.6086,
+      "step": 3015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.58480909829407e-06,
+      "loss": 1.241,
+      "step": 3016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5809910641754672e-06,
+      "loss": 1.1687,
+      "step": 3017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.577173030056864e-06,
+      "loss": 1.2164,
+      "step": 3018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.573354995938262e-06,
+      "loss": 1.1976,
+      "step": 3019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.569536961819659e-06,
+      "loss": 1.2904,
+      "step": 3020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.565718927701056e-06,
+      "loss": 1.2326,
+      "step": 3021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5619008935824535e-06,
+      "loss": 0.9622,
+      "step": 3022
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.558082859463851e-06,
+      "loss": 1.3732,
+      "step": 3023
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.554264825345248e-06,
+      "loss": 1.1738,
+      "step": 3024
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.550446791226645e-06,
+      "loss": 1.5599,
+      "step": 3025
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5466287571080424e-06,
+      "loss": 1.1004,
+      "step": 3026
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5428107229894393e-06,
+      "loss": 1.1501,
+      "step": 3027
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.538992688870837e-06,
+      "loss": 1.2334,
+      "step": 3028
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.535174654752234e-06,
+      "loss": 1.4726,
+      "step": 3029
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5313566206336313e-06,
+      "loss": 1.3294,
+      "step": 3030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5275385865150286e-06,
+      "loss": 1.0619,
+      "step": 3031
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.523720552396426e-06,
+      "loss": 1.2333,
+      "step": 3032
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5199025182778232e-06,
+      "loss": 1.4564,
+      "step": 3033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.51608448415922e-06,
+      "loss": 1.1402,
+      "step": 3034
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5122664500406175e-06,
+      "loss": 1.1277,
+      "step": 3035
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5084484159220144e-06,
+      "loss": 1.1976,
+      "step": 3036
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.504630381803412e-06,
+      "loss": 1.0698,
+      "step": 3037
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.500812347684809e-06,
+      "loss": 1.14,
+      "step": 3038
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4969943135662064e-06,
+      "loss": 1.1146,
+      "step": 3039
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4931762794476037e-06,
+      "loss": 1.2389,
+      "step": 3040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.489358245329001e-06,
+      "loss": 1.1424,
+      "step": 3041
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4855402112103984e-06,
+      "loss": 1.278,
+      "step": 3042
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4817221770917953e-06,
+      "loss": 0.9951,
+      "step": 3043
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4779041429731926e-06,
+      "loss": 1.1655,
+      "step": 3044
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4740861088545895e-06,
+      "loss": 1.136,
+      "step": 3045
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4702680747359873e-06,
+      "loss": 1.2592,
+      "step": 3046
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.466450040617384e-06,
+      "loss": 1.199,
+      "step": 3047
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4626320064987815e-06,
+      "loss": 0.9867,
+      "step": 3048
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.458813972380179e-06,
+      "loss": 1.0769,
+      "step": 3049
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.454995938261576e-06,
+      "loss": 1.1118,
+      "step": 3050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4511779041429735e-06,
+      "loss": 1.2503,
+      "step": 3051
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4473598700243704e-06,
+      "loss": 1.0892,
+      "step": 3052
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4435418359057677e-06,
+      "loss": 1.2784,
+      "step": 3053
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.439723801787165e-06,
+      "loss": 1.0496,
+      "step": 3054
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4359057676685624e-06,
+      "loss": 1.2067,
+      "step": 3055
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4320877335499593e-06,
+      "loss": 1.1668,
+      "step": 3056
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4282696994313566e-06,
+      "loss": 1.1345,
+      "step": 3057
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.424451665312754e-06,
+      "loss": 1.3488,
+      "step": 3058
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4206336311941513e-06,
+      "loss": 1.1156,
+      "step": 3059
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4168155970755486e-06,
+      "loss": 1.2203,
+      "step": 3060
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4129975629569455e-06,
+      "loss": 1.2238,
+      "step": 3061
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.409179528838343e-06,
+      "loss": 1.2093,
+      "step": 3062
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.40536149471974e-06,
+      "loss": 1.2481,
+      "step": 3063
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4015434606011375e-06,
+      "loss": 1.0602,
+      "step": 3064
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.397725426482535e-06,
+      "loss": 1.1429,
+      "step": 3065
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.3939073923639317e-06,
+      "loss": 1.0385,
+      "step": 3066
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.390089358245329e-06,
+      "loss": 1.1408,
+      "step": 3067
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.3862713241267264e-06,
+      "loss": 1.4718,
+      "step": 3068
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.3824532900081237e-06,
+      "loss": 1.2822,
+      "step": 3069
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.3786352558895206e-06,
+      "loss": 1.1756,
+      "step": 3070
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.374817221770918e-06,
+      "loss": 1.0474,
+      "step": 3071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3709991876523153e-06,
+      "loss": 0.9565,
+      "step": 3072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3671811535337126e-06,
+      "loss": 1.5245,
+      "step": 3073
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.36336311941511e-06,
+      "loss": 1.1534,
+      "step": 3074
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.359545085296507e-06,
+      "loss": 1.2895,
+      "step": 3075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.355727051177904e-06,
+      "loss": 1.0628,
+      "step": 3076
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3519090170593015e-06,
+      "loss": 1.3489,
+      "step": 3077
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.348090982940699e-06,
+      "loss": 1.2919,
+      "step": 3078
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3442729488220957e-06,
+      "loss": 1.3919,
+      "step": 3079
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.340454914703493e-06,
+      "loss": 1.2686,
+      "step": 3080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3366368805848904e-06,
+      "loss": 1.3054,
+      "step": 3081
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.3328188464662877e-06,
+      "loss": 1.2298,
+      "step": 3082
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.329000812347685e-06,
+      "loss": 1.1383,
+      "step": 3083
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.325182778229082e-06,
+      "loss": 1.1229,
+      "step": 3084
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.3213647441104797e-06,
+      "loss": 1.0848,
+      "step": 3085
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.3175467099918766e-06,
+      "loss": 0.9466,
+      "step": 3086
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.313728675873274e-06,
+      "loss": 1.0105,
+      "step": 3087
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.309910641754671e-06,
+      "loss": 1.1853,
+      "step": 3088
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.306092607636068e-06,
+      "loss": 1.2703,
+      "step": 3089
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.3022745735174655e-06,
+      "loss": 1.3668,
+      "step": 3090
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.298456539398863e-06,
+      "loss": 1.3945,
+      "step": 3091
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.29463850528026e-06,
+      "loss": 1.1371,
+      "step": 3092
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.290820471161657e-06,
+      "loss": 1.3344,
+      "step": 3093
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.287002437043055e-06,
+      "loss": 1.2944,
+      "step": 3094
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2831844029244517e-06,
+      "loss": 1.2205,
+      "step": 3095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.279366368805849e-06,
+      "loss": 1.2022,
+      "step": 3096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.275548334687246e-06,
+      "loss": 1.0183,
+      "step": 3097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2717303005686433e-06,
+      "loss": 1.4014,
+      "step": 3098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2679122664500406e-06,
+      "loss": 1.1218,
+      "step": 3099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.264094232331438e-06,
+      "loss": 1.1105,
+      "step": 3100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2602761982128353e-06,
+      "loss": 1.107,
+      "step": 3101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.256458164094232e-06,
+      "loss": 1.1325,
+      "step": 3102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.25264012997563e-06,
+      "loss": 1.3981,
+      "step": 3103
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.248822095857027e-06,
+      "loss": 1.2952,
+      "step": 3104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.245004061738424e-06,
+      "loss": 1.1213,
+      "step": 3105
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.241186027619821e-06,
+      "loss": 1.3107,
+      "step": 3106
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.237367993501219e-06,
+      "loss": 1.2215,
+      "step": 3107
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.233549959382616e-06,
+      "loss": 1.0505,
+      "step": 3108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.229731925264013e-06,
+      "loss": 1.5156,
+      "step": 3109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2259138911454104e-06,
+      "loss": 1.3903,
+      "step": 3110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2220958570268073e-06,
+      "loss": 1.1184,
+      "step": 3111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.218277822908205e-06,
+      "loss": 1.1483,
+      "step": 3112
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.214459788789602e-06,
+      "loss": 1.1275,
+      "step": 3113
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2106417546709993e-06,
+      "loss": 2.0806,
+      "step": 3114
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.206823720552396e-06,
+      "loss": 1.0525,
+      "step": 3115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.203005686433794e-06,
+      "loss": 1.0446,
+      "step": 3116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.1991876523151913e-06,
+      "loss": 1.2429,
+      "step": 3117
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.195369618196588e-06,
+      "loss": 1.5203,
+      "step": 3118
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.1915515840779855e-06,
+      "loss": 1.2606,
+      "step": 3119
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.1877335499593824e-06,
+      "loss": 1.2527,
+      "step": 3120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.18391551584078e-06,
+      "loss": 1.1833,
+      "step": 3121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.180097481722177e-06,
+      "loss": 0.9683,
+      "step": 3122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1762794476035744e-06,
+      "loss": 1.161,
+      "step": 3123
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1724614134849713e-06,
+      "loss": 1.3915,
+      "step": 3124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.168643379366369e-06,
+      "loss": 1.2608,
+      "step": 3125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1648253452477664e-06,
+      "loss": 1.2476,
+      "step": 3126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1610073111291633e-06,
+      "loss": 1.3402,
+      "step": 3127
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1571892770105606e-06,
+      "loss": 1.1644,
+      "step": 3128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1533712428919575e-06,
+      "loss": 1.2949,
+      "step": 3129
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1495532087733553e-06,
+      "loss": 1.0578,
+      "step": 3130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.145735174654752e-06,
+      "loss": 1.3696,
+      "step": 3131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.1419171405361495e-06,
+      "loss": 1.2927,
+      "step": 3132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.138099106417547e-06,
+      "loss": 1.3467,
+      "step": 3133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.134281072298944e-06,
+      "loss": 1.836,
+      "step": 3134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1304630381803415e-06,
+      "loss": 1.0599,
+      "step": 3135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1266450040617384e-06,
+      "loss": 1.2043,
+      "step": 3136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1228269699431357e-06,
+      "loss": 1.1144,
+      "step": 3137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.119008935824533e-06,
+      "loss": 1.3649,
+      "step": 3138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1151909017059304e-06,
+      "loss": 1.1816,
+      "step": 3139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1113728675873273e-06,
+      "loss": 1.3498,
+      "step": 3140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.1075548334687246e-06,
+      "loss": 1.1218,
+      "step": 3141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.103736799350122e-06,
+      "loss": 1.0275,
+      "step": 3142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0999187652315193e-06,
+      "loss": 1.4742,
+      "step": 3143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0961007311129166e-06,
+      "loss": 1.301,
+      "step": 3144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0922826969943135e-06,
+      "loss": 1.124,
+      "step": 3145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.088464662875711e-06,
+      "loss": 1.1061,
+      "step": 3146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.084646628757108e-06,
+      "loss": 1.2084,
+      "step": 3147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0808285946385055e-06,
+      "loss": 1.216,
+      "step": 3148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0770105605199024e-06,
+      "loss": 1.3301,
+      "step": 3149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0731925264012998e-06,
+      "loss": 1.3432,
+      "step": 3150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.069374492282697e-06,
+      "loss": 1.2623,
+      "step": 3151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0655564581640944e-06,
+      "loss": 1.2284,
+      "step": 3152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0617384240454917e-06,
+      "loss": 0.8887,
+      "step": 3153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0579203899268886e-06,
+      "loss": 1.0692,
+      "step": 3154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.054102355808286e-06,
+      "loss": 1.394,
+      "step": 3155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0502843216896833e-06,
+      "loss": 1.3857,
+      "step": 3156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0464662875710806e-06,
+      "loss": 1.0436,
+      "step": 3157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0426482534524775e-06,
+      "loss": 1.1933,
+      "step": 3158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.038830219333875e-06,
+      "loss": 0.9878,
+      "step": 3159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.035012185215272e-06,
+      "loss": 1.2211,
+      "step": 3160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0311941510966695e-06,
+      "loss": 1.1407,
+      "step": 3161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.027376116978067e-06,
+      "loss": 1.1739,
+      "step": 3162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0235580828594638e-06,
+      "loss": 1.4513,
+      "step": 3163
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.019740048740861e-06,
+      "loss": 1.2399,
+      "step": 3164
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0159220146222584e-06,
+      "loss": 1.3686,
+      "step": 3165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0121039805036558e-06,
+      "loss": 1.0723,
+      "step": 3166
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0082859463850527e-06,
+      "loss": 1.196,
+      "step": 3167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.00446791226645e-06,
+      "loss": 1.1182,
+      "step": 3168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0006498781478477e-06,
+      "loss": 1.2196,
+      "step": 3169
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9968318440292446e-06,
+      "loss": 1.2977,
+      "step": 3170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.993013809910642e-06,
+      "loss": 1.3639,
+      "step": 3171
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.989195775792039e-06,
+      "loss": 1.1181,
+      "step": 3172
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.985377741673436e-06,
+      "loss": 1.0577,
+      "step": 3173
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9815597075548335e-06,
+      "loss": 1.212,
+      "step": 3174
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.977741673436231e-06,
+      "loss": 1.1133,
+      "step": 3175
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9739236393176278e-06,
+      "loss": 1.145,
+      "step": 3176
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.970105605199025e-06,
+      "loss": 1.1922,
+      "step": 3177
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.966287571080423e-06,
+      "loss": 1.0362,
+      "step": 3178
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9624695369618198e-06,
+      "loss": 1.2696,
+      "step": 3179
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.958651502843217e-06,
+      "loss": 1.1105,
+      "step": 3180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.954833468724614e-06,
+      "loss": 1.1314,
+      "step": 3181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9510154346060113e-06,
+      "loss": 1.062,
+      "step": 3182
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9471974004874087e-06,
+      "loss": 1.1974,
+      "step": 3183
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.943379366368806e-06,
+      "loss": 1.2126,
+      "step": 3184
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9395613322502033e-06,
+      "loss": 1.31,
+      "step": 3185
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9357432981316002e-06,
+      "loss": 1.3816,
+      "step": 3186
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.931925264012998e-06,
+      "loss": 1.234,
+      "step": 3187
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.928107229894395e-06,
+      "loss": 0.9614,
+      "step": 3188
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.924289195775792e-06,
+      "loss": 1.3264,
+      "step": 3189
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.920471161657189e-06,
+      "loss": 1.2931,
+      "step": 3190
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9166531275385864e-06,
+      "loss": 1.3964,
+      "step": 3191
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9128350934199838e-06,
+      "loss": 1.264,
+      "step": 3192
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.909017059301381e-06,
+      "loss": 1.4421,
+      "step": 3193
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9051990251827782e-06,
+      "loss": 1.1666,
+      "step": 3194
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9013809910641753e-06,
+      "loss": 1.1212,
+      "step": 3195
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8975629569455729e-06,
+      "loss": 1.1616,
+      "step": 3196
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.89374492282697e-06,
+      "loss": 1.1454,
+      "step": 3197
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8899268887083673e-06,
+      "loss": 1.1419,
+      "step": 3198
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8861088545897644e-06,
+      "loss": 1.0538,
+      "step": 3199
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.882290820471162e-06,
+      "loss": 1.4296,
+      "step": 3200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.878472786352559e-06,
+      "loss": 1.3963,
+      "step": 3201
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8746547522339562e-06,
+      "loss": 1.0642,
+      "step": 3202
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8708367181153533e-06,
+      "loss": 1.1886,
+      "step": 3203
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8670186839967505e-06,
+      "loss": 1.0287,
+      "step": 3204
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.863200649878148e-06,
+      "loss": 1.0349,
+      "step": 3205
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8593826157595451e-06,
+      "loss": 1.0506,
+      "step": 3206
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8555645816409424e-06,
+      "loss": 1.3189,
+      "step": 3207
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8517465475223396e-06,
+      "loss": 1.5045,
+      "step": 3208
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.847928513403737e-06,
+      "loss": 1.2563,
+      "step": 3209
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8441104792851342e-06,
+      "loss": 1.1368,
+      "step": 3210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8402924451665313e-06,
+      "loss": 1.0951,
+      "step": 3211
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8364744110479285e-06,
+      "loss": 1.3589,
+      "step": 3212
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8326563769293256e-06,
+      "loss": 1.3027,
+      "step": 3213
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8288383428107231e-06,
+      "loss": 1.322,
+      "step": 3214
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8250203086921202e-06,
+      "loss": 1.2266,
+      "step": 3215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8212022745735176e-06,
+      "loss": 1.0776,
+      "step": 3216
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8173842404549147e-06,
+      "loss": 1.1845,
+      "step": 3217
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8135662063363122e-06,
+      "loss": 1.0582,
+      "step": 3218
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8097481722177093e-06,
+      "loss": 1.2485,
+      "step": 3219
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8059301380991065e-06,
+      "loss": 1.3605,
+      "step": 3220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8021121039805036e-06,
+      "loss": 1.1476,
+      "step": 3221
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7982940698619011e-06,
+      "loss": 1.2707,
+      "step": 3222
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7944760357432982e-06,
+      "loss": 1.1589,
+      "step": 3223
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7906580016246956e-06,
+      "loss": 1.2133,
+      "step": 3224
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7868399675060927e-06,
+      "loss": 1.2212,
+      "step": 3225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7830219333874898e-06,
+      "loss": 1.123,
+      "step": 3226
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7792038992688873e-06,
+      "loss": 1.096,
+      "step": 3227
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7753858651502845e-06,
+      "loss": 1.2871,
+      "step": 3228
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7715678310316816e-06,
+      "loss": 0.891,
+      "step": 3229
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7677497969130787e-06,
+      "loss": 1.4492,
+      "step": 3230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7639317627944762e-06,
+      "loss": 0.8361,
+      "step": 3231
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7601137286758733e-06,
+      "loss": 0.9317,
+      "step": 3232
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7562956945572707e-06,
+      "loss": 1.5525,
+      "step": 3233
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7524776604386678e-06,
+      "loss": 1.4447,
+      "step": 3234
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7486596263200651e-06,
+      "loss": 0.9761,
+      "step": 3235
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7448415922014622e-06,
+      "loss": 1.3543,
+      "step": 3236
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7410235580828596e-06,
+      "loss": 1.1355,
+      "step": 3237
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7372055239642567e-06,
+      "loss": 1.3614,
+      "step": 3238
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.733387489845654e-06,
+      "loss": 1.1474,
+      "step": 3239
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7295694557270511e-06,
+      "loss": 1.2726,
+      "step": 3240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7257514216084485e-06,
+      "loss": 1.2977,
+      "step": 3241
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7219333874898458e-06,
+      "loss": 1.5058,
+      "step": 3242
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7181153533712431e-06,
+      "loss": 1.1108,
+      "step": 3243
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7142973192526402e-06,
+      "loss": 1.2318,
+      "step": 3244
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7104792851340374e-06,
+      "loss": 0.9476,
+      "step": 3245
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7066612510154347e-06,
+      "loss": 1.2315,
+      "step": 3246
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7028432168968318e-06,
+      "loss": 1.4864,
+      "step": 3247
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6990251827782291e-06,
+      "loss": 1.389,
+      "step": 3248
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6952071486596262e-06,
+      "loss": 1.6087,
+      "step": 3249
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6913891145410238e-06,
+      "loss": 1.3744,
+      "step": 3250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.687571080422421e-06,
+      "loss": 1.0956,
+      "step": 3251
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6837530463038182e-06,
+      "loss": 1.1493,
+      "step": 3252
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6799350121852154e-06,
+      "loss": 1.4383,
+      "step": 3253
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6761169780666125e-06,
+      "loss": 1.2556,
+      "step": 3254
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6722989439480098e-06,
+      "loss": 1.1855,
+      "step": 3255
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.668480909829407e-06,
+      "loss": 1.3203,
+      "step": 3256
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6646628757108042e-06,
+      "loss": 1.5263,
+      "step": 3257
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6608448415922014e-06,
+      "loss": 1.1525,
+      "step": 3258
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.657026807473599e-06,
+      "loss": 1.2163,
+      "step": 3259
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.653208773354996e-06,
+      "loss": 1.1309,
+      "step": 3260
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6493907392363934e-06,
+      "loss": 1.0838,
+      "step": 3261
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6455727051177905e-06,
+      "loss": 1.1012,
+      "step": 3262
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6417546709991878e-06,
+      "loss": 0.9711,
+      "step": 3263
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.637936636880585e-06,
+      "loss": 1.4923,
+      "step": 3264
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.634118602761982e-06,
+      "loss": 1.2209,
+      "step": 3265
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6303005686433794e-06,
+      "loss": 1.0753,
+      "step": 3266
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6264825345247767e-06,
+      "loss": 1.1783,
+      "step": 3267
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.622664500406174e-06,
+      "loss": 1.2448,
+      "step": 3268
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6188464662875711e-06,
+      "loss": 1.1788,
+      "step": 3269
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6150284321689685e-06,
+      "loss": 1.2925,
+      "step": 3270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6112103980503656e-06,
+      "loss": 1.0713,
+      "step": 3271
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.607392363931763e-06,
+      "loss": 1.1665,
+      "step": 3272
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.60357432981316e-06,
+      "loss": 1.3812,
+      "step": 3273
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5997562956945574e-06,
+      "loss": 0.9873,
+      "step": 3274
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5959382615759545e-06,
+      "loss": 1.2582,
+      "step": 3275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5921202274573518e-06,
+      "loss": 1.2071,
+      "step": 3276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5883021933387491e-06,
+      "loss": 0.9225,
+      "step": 3277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5844841592201463e-06,
+      "loss": 1.3651,
+      "step": 3278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5806661251015436e-06,
+      "loss": 1.3147,
+      "step": 3279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5768480909829407e-06,
+      "loss": 1.5237,
+      "step": 3280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.573030056864338e-06,
+      "loss": 1.2697,
+      "step": 3281
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5692120227457351e-06,
+      "loss": 1.2763,
+      "step": 3282
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5653939886271325e-06,
+      "loss": 1.0197,
+      "step": 3283
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5615759545085296e-06,
+      "loss": 1.2777,
+      "step": 3284
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.557757920389927e-06,
+      "loss": 1.3561,
+      "step": 3285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5539398862713243e-06,
+      "loss": 1.4432,
+      "step": 3286
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5501218521527214e-06,
+      "loss": 1.2751,
+      "step": 3287
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5463038180341187e-06,
+      "loss": 1.3156,
+      "step": 3288
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5424857839155158e-06,
+      "loss": 1.2555,
+      "step": 3289
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5386677497969131e-06,
+      "loss": 0.9841,
+      "step": 3290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5348497156783103e-06,
+      "loss": 1.6377,
+      "step": 3291
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5310316815597076e-06,
+      "loss": 1.1211,
+      "step": 3292
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.527213647441105e-06,
+      "loss": 1.4438,
+      "step": 3293
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5233956133225023e-06,
+      "loss": 1.3199,
+      "step": 3294
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5195775792038994e-06,
+      "loss": 1.6366,
+      "step": 3295
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5157595450852965e-06,
+      "loss": 1.5036,
+      "step": 3296
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5119415109666938e-06,
+      "loss": 1.0814,
+      "step": 3297
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.508123476848091e-06,
+      "loss": 1.1469,
+      "step": 3298
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5043054427294883e-06,
+      "loss": 1.2886,
+      "step": 3299
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5004874086108854e-06,
+      "loss": 1.0358,
+      "step": 3300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4966693744922827e-06,
+      "loss": 1.4884,
+      "step": 3301
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.49285134037368e-06,
+      "loss": 1.1064,
+      "step": 3302
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4890333062550774e-06,
+      "loss": 1.3744,
+      "step": 3303
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4852152721364745e-06,
+      "loss": 1.2089,
+      "step": 3304
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4813972380178718e-06,
+      "loss": 1.205,
+      "step": 3305
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.477579203899269e-06,
+      "loss": 1.1613,
+      "step": 3306
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.473761169780666e-06,
+      "loss": 1.0183,
+      "step": 3307
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4699431356620634e-06,
+      "loss": 1.157,
+      "step": 3308
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4661251015434605e-06,
+      "loss": 1.0012,
+      "step": 3309
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4623070674248578e-06,
+      "loss": 1.4991,
+      "step": 3310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4584890333062552e-06,
+      "loss": 1.3983,
+      "step": 3311
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4546709991876525e-06,
+      "loss": 1.2051,
+      "step": 3312
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4508529650690496e-06,
+      "loss": 1.1688,
+      "step": 3313
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.447034930950447e-06,
+      "loss": 1.639,
+      "step": 3314
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.443216896831844e-06,
+      "loss": 1.0653,
+      "step": 3315
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4393988627132414e-06,
+      "loss": 1.0457,
+      "step": 3316
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4355808285946385e-06,
+      "loss": 1.1505,
+      "step": 3317
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4317627944760356e-06,
+      "loss": 1.132,
+      "step": 3318
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4279447603574332e-06,
+      "loss": 1.4484,
+      "step": 3319
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4241267262388303e-06,
+      "loss": 1.1038,
+      "step": 3320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4203086921202276e-06,
+      "loss": 1.1195,
+      "step": 3321
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4164906580016247e-06,
+      "loss": 1.0623,
+      "step": 3322
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.412672623883022e-06,
+      "loss": 1.5046,
+      "step": 3323
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4088545897644192e-06,
+      "loss": 1.3341,
+      "step": 3324
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4050365556458165e-06,
+      "loss": 1.1228,
+      "step": 3325
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4012185215272136e-06,
+      "loss": 1.3111,
+      "step": 3326
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.397400487408611e-06,
+      "loss": 1.1569,
+      "step": 3327
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3935824532900083e-06,
+      "loss": 1.1326,
+      "step": 3328
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3897644191714054e-06,
+      "loss": 1.0593,
+      "step": 3329
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3859463850528027e-06,
+      "loss": 1.2356,
+      "step": 3330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3821283509341998e-06,
+      "loss": 1.1251,
+      "step": 3331
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3783103168155972e-06,
+      "loss": 1.2622,
+      "step": 3332
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3744922826969943e-06,
+      "loss": 1.5379,
+      "step": 3333
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3706742485783916e-06,
+      "loss": 1.4408,
+      "step": 3334
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3668562144597887e-06,
+      "loss": 1.0707,
+      "step": 3335
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.363038180341186e-06,
+      "loss": 1.3076,
+      "step": 3336
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3592201462225834e-06,
+      "loss": 1.0354,
+      "step": 3337
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3554021121039805e-06,
+      "loss": 1.447,
+      "step": 3338
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3515840779853778e-06,
+      "loss": 1.2181,
+      "step": 3339
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.347766043866775e-06,
+      "loss": 1.4855,
+      "step": 3340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3439480097481723e-06,
+      "loss": 1.0498,
+      "step": 3341
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.3401299756295694e-06,
+      "loss": 0.9977,
+      "step": 3342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3363119415109667e-06,
+      "loss": 1.1386,
+      "step": 3343
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3324939073923638e-06,
+      "loss": 1.0431,
+      "step": 3344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3286758732737614e-06,
+      "loss": 1.325,
+      "step": 3345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3248578391551585e-06,
+      "loss": 1.2642,
+      "step": 3346
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3210398050365558e-06,
+      "loss": 1.2505,
+      "step": 3347
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.317221770917953e-06,
+      "loss": 1.2228,
+      "step": 3348
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.31340373679935e-06,
+      "loss": 1.1311,
+      "step": 3349
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3095857026807474e-06,
+      "loss": 1.1873,
+      "step": 3350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3057676685621445e-06,
+      "loss": 1.8201,
+      "step": 3351
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3019496344435418e-06,
+      "loss": 1.3626,
+      "step": 3352
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2981316003249392e-06,
+      "loss": 1.2572,
+      "step": 3353
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2943135662063365e-06,
+      "loss": 1.2038,
+      "step": 3354
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2904955320877336e-06,
+      "loss": 1.2798,
+      "step": 3355
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.286677497969131e-06,
+      "loss": 1.17,
+      "step": 3356
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.282859463850528e-06,
+      "loss": 1.1103,
+      "step": 3357
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2790414297319254e-06,
+      "loss": 1.105,
+      "step": 3358
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2752233956133225e-06,
+      "loss": 1.1038,
+      "step": 3359
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2714053614947196e-06,
+      "loss": 1.2435,
+      "step": 3360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.267587327376117e-06,
+      "loss": 1.2881,
+      "step": 3361
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2637692932575143e-06,
+      "loss": 1.1212,
+      "step": 3362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2599512591389116e-06,
+      "loss": 1.4228,
+      "step": 3363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2561332250203087e-06,
+      "loss": 0.9671,
+      "step": 3364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.252315190901706e-06,
+      "loss": 1.167,
+      "step": 3365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2484971567831032e-06,
+      "loss": 1.4042,
+      "step": 3366
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2446791226645005e-06,
+      "loss": 1.2318,
+      "step": 3367
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2408610885458976e-06,
+      "loss": 1.2757,
+      "step": 3368
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2370430544272947e-06,
+      "loss": 1.1491,
+      "step": 3369
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.233225020308692e-06,
+      "loss": 1.3187,
+      "step": 3370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2294069861900894e-06,
+      "loss": 1.2255,
+      "step": 3371
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2255889520714867e-06,
+      "loss": 1.0096,
+      "step": 3372
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2217709179528839e-06,
+      "loss": 1.0453,
+      "step": 3373
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2179528838342812e-06,
+      "loss": 0.9742,
+      "step": 3374
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2141348497156783e-06,
+      "loss": 1.2587,
+      "step": 3375
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2103168155970756e-06,
+      "loss": 1.0972,
+      "step": 3376
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2064987814784727e-06,
+      "loss": 1.2045,
+      "step": 3377
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.20268074735987e-06,
+      "loss": 1.4295,
+      "step": 3378
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1988627132412674e-06,
+      "loss": 1.2411,
+      "step": 3379
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1950446791226645e-06,
+      "loss": 1.3769,
+      "step": 3380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1912266450040619e-06,
+      "loss": 1.3411,
+      "step": 3381
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.187408610885459e-06,
+      "loss": 1.466,
+      "step": 3382
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1835905767668563e-06,
+      "loss": 1.0963,
+      "step": 3383
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1797725426482534e-06,
+      "loss": 1.3445,
+      "step": 3384
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1759545085296507e-06,
+      "loss": 1.0876,
+      "step": 3385
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1721364744110479e-06,
+      "loss": 1.113,
+      "step": 3386
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1683184402924452e-06,
+      "loss": 1.0518,
+      "step": 3387
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1645004061738425e-06,
+      "loss": 1.2615,
+      "step": 3388
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1606823720552399e-06,
+      "loss": 1.2972,
+      "step": 3389
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.156864337936637e-06,
+      "loss": 1.0919,
+      "step": 3390
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.153046303818034e-06,
+      "loss": 0.9377,
+      "step": 3391
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1492282696994314e-06,
+      "loss": 1.4153,
+      "step": 3392
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1454102355808285e-06,
+      "loss": 1.2336,
+      "step": 3393
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1415922014622259e-06,
+      "loss": 1.2874,
+      "step": 3394
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.137774167343623e-06,
+      "loss": 1.3572,
+      "step": 3395
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1339561332250203e-06,
+      "loss": 1.3536,
+      "step": 3396
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1301380991064176e-06,
+      "loss": 1.7276,
+      "step": 3397
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.126320064987815e-06,
+      "loss": 1.1715,
+      "step": 3398
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.122502030869212e-06,
+      "loss": 1.0169,
+      "step": 3399
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1186839967506094e-06,
+      "loss": 1.3878,
+      "step": 3400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1148659626320065e-06,
+      "loss": 1.409,
+      "step": 3401
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1110479285134037e-06,
+      "loss": 1.103,
+      "step": 3402
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.107229894394801e-06,
+      "loss": 1.0928,
+      "step": 3403
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.103411860276198e-06,
+      "loss": 1.2855,
+      "step": 3404
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0995938261575956e-06,
+      "loss": 1.3271,
+      "step": 3405
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0957757920389928e-06,
+      "loss": 1.3762,
+      "step": 3406
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.09195775792039e-06,
+      "loss": 1.058,
+      "step": 3407
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0881397238017872e-06,
+      "loss": 0.9654,
+      "step": 3408
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0843216896831845e-06,
+      "loss": 1.2775,
+      "step": 3409
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0805036555645816e-06,
+      "loss": 1.3722,
+      "step": 3410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0766856214459788e-06,
+      "loss": 1.2211,
+      "step": 3411
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.072867587327376e-06,
+      "loss": 1.2196,
+      "step": 3412
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0690495532087734e-06,
+      "loss": 1.1152,
+      "step": 3413
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0652315190901708e-06,
+      "loss": 1.2757,
+      "step": 3414
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0614134849715679e-06,
+      "loss": 1.1371,
+      "step": 3415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0575954508529652e-06,
+      "loss": 1.2603,
+      "step": 3416
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0537774167343623e-06,
+      "loss": 1.0924,
+      "step": 3417
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0499593826157596e-06,
+      "loss": 1.5531,
+      "step": 3418
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0461413484971568e-06,
+      "loss": 1.0239,
+      "step": 3419
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.042323314378554e-06,
+      "loss": 1.2806,
+      "step": 3420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0385052802599512e-06,
+      "loss": 1.5476,
+      "step": 3421
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0346872461413485e-06,
+      "loss": 1.2718,
+      "step": 3422
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0308692120227459e-06,
+      "loss": 0.9536,
+      "step": 3423
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.027051177904143e-06,
+      "loss": 1.4222,
+      "step": 3424
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0232331437855403e-06,
+      "loss": 1.1241,
+      "step": 3425
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0194151096669374e-06,
+      "loss": 1.1793,
+      "step": 3426
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0155970755483348e-06,
+      "loss": 1.1683,
+      "step": 3427
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0117790414297319e-06,
+      "loss": 1.1788,
+      "step": 3428
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0079610073111292e-06,
+      "loss": 1.2959,
+      "step": 3429
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0041429731925263e-06,
+      "loss": 1.2121,
+      "step": 3430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0003249390739239e-06,
+      "loss": 1.1723,
+      "step": 3431
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.96506904955321e-07,
+      "loss": 1.2296,
+      "step": 3432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.92688870836718e-07,
+      "loss": 1.5024,
+      "step": 3433
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.888708367181154e-07,
+      "loss": 1.2,
+      "step": 3434
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.850528025995126e-07,
+      "loss": 1.2566,
+      "step": 3435
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.812347684809099e-07,
+      "loss": 1.1409,
+      "step": 3436
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.77416734362307e-07,
+      "loss": 1.1141,
+      "step": 3437
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.735987002437043e-07,
+      "loss": 1.0752,
+      "step": 3438
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.697806661251017e-07,
+      "loss": 1.2015,
+      "step": 3439
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.65962632006499e-07,
+      "loss": 1.2634,
+      "step": 3440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.62144597887896e-07,
+      "loss": 1.105,
+      "step": 3441
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.583265637692932e-07,
+      "loss": 1.5504,
+      "step": 3442
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.545085296506906e-07,
+      "loss": 1.1307,
+      "step": 3443
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.506904955320877e-07,
+      "loss": 1.2526,
+      "step": 3444
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.46872461413485e-07,
+      "loss": 1.2637,
+      "step": 3445
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.430544272948822e-07,
+      "loss": 1.1722,
+      "step": 3446
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.392363931762796e-07,
+      "loss": 1.2645,
+      "step": 3447
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.354183590576767e-07,
+      "loss": 1.1859,
+      "step": 3448
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.31600324939074e-07,
+      "loss": 1.4728,
+      "step": 3449
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.277822908204712e-07,
+      "loss": 1.0398,
+      "step": 3450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.239642567018685e-07,
+      "loss": 1.4692,
+      "step": 3451
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.201462225832657e-07,
+      "loss": 1.2681,
+      "step": 3452
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.163281884646628e-07,
+      "loss": 1.2052,
+      "step": 3453
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.125101543460601e-07,
+      "loss": 1.1499,
+      "step": 3454
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.086921202274573e-07,
+      "loss": 1.1194,
+      "step": 3455
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.048740861088547e-07,
+      "loss": 1.3785,
+      "step": 3456
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.010560519902518e-07,
+      "loss": 1.2781,
+      "step": 3457
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.972380178716491e-07,
+      "loss": 1.2799,
+      "step": 3458
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.934199837530463e-07,
+      "loss": 1.1471,
+      "step": 3459
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.896019496344437e-07,
+      "loss": 1.5076,
+      "step": 3460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.857839155158408e-07,
+      "loss": 1.2437,
+      "step": 3461
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.819658813972381e-07,
+      "loss": 1.0072,
+      "step": 3462
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.781478472786353e-07,
+      "loss": 1.1333,
+      "step": 3463
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.743298131600326e-07,
+      "loss": 1.1842,
+      "step": 3464
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.705117790414298e-07,
+      "loss": 1.2243,
+      "step": 3465
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.66693744922827e-07,
+      "loss": 1.1168,
+      "step": 3466
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.628757108042242e-07,
+      "loss": 1.2461,
+      "step": 3467
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.590576766856216e-07,
+      "loss": 1.0863,
+      "step": 3468
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.552396425670187e-07,
+      "loss": 1.1641,
+      "step": 3469
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.514216084484159e-07,
+      "loss": 1.5065,
+      "step": 3470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.476035743298131e-07,
+      "loss": 1.361,
+      "step": 3471
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.437855402112105e-07,
+      "loss": 1.4547,
+      "step": 3472
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.399675060926077e-07,
+      "loss": 1.1074,
+      "step": 3473
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.361494719740049e-07,
+      "loss": 1.137,
+      "step": 3474
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.323314378554021e-07,
+      "loss": 1.1055,
+      "step": 3475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.285134037367995e-07,
+      "loss": 1.2929,
+      "step": 3476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.246953696181967e-07,
+      "loss": 1.1365,
+      "step": 3477
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.208773354995939e-07,
+      "loss": 1.2249,
+      "step": 3478
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.17059301380991e-07,
+      "loss": 1.0702,
+      "step": 3479
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.132412672623883e-07,
+      "loss": 1.0664,
+      "step": 3480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.094232331437856e-07,
+      "loss": 1.5471,
+      "step": 3481
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.056051990251828e-07,
+      "loss": 1.2446,
+      "step": 3482
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.0178716490658e-07,
+      "loss": 1.2417,
+      "step": 3483
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.979691307879772e-07,
+      "loss": 1.1043,
+      "step": 3484
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.941510966693746e-07,
+      "loss": 1.1666,
+      "step": 3485
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.903330625507718e-07,
+      "loss": 1.2683,
+      "step": 3486
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.86515028432169e-07,
+      "loss": 1.0575,
+      "step": 3487
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.826969943135662e-07,
+      "loss": 1.276,
+      "step": 3488
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.788789601949635e-07,
+      "loss": 1.3447,
+      "step": 3489
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.750609260763607e-07,
+      "loss": 0.9672,
+      "step": 3490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.712428919577579e-07,
+      "loss": 1.3142,
+      "step": 3491
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.674248578391551e-07,
+      "loss": 1.0675,
+      "step": 3492
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.636068237205525e-07,
+      "loss": 1.1678,
+      "step": 3493
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.597887896019497e-07,
+      "loss": 0.9339,
+      "step": 3494
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.559707554833469e-07,
+      "loss": 1.373,
+      "step": 3495
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.521527213647441e-07,
+      "loss": 1.1817,
+      "step": 3496
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.483346872461414e-07,
+      "loss": 1.2574,
+      "step": 3497
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.445166531275387e-07,
+      "loss": 1.178,
+      "step": 3498
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.406986190089359e-07,
+      "loss": 1.2327,
+      "step": 3499
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.36880584890333e-07,
+      "loss": 1.2716,
+      "step": 3500
+    }
+  ],
+  "max_steps": 3693,
+  "num_train_epochs": 3,
+  "total_flos": 2.7881354984005632e+17,
+  "trial_name": null,
+  "trial_params": null
+}