diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30176 @@
+{
+  "best_metric": 1.4090298414230347,
+  "best_model_checkpoint": "/home/yuchenl/models/reverse-instruct/prefixgen_flan-small/checkpoint-5000",
+  "epoch": 14.792899408284024,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99802761341223e-05,
+      "loss": 3.8917,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996055226824458e-05,
+      "loss": 3.674,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994082840236687e-05,
+      "loss": 3.6946,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992110453648915e-05,
+      "loss": 3.5351,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990138067061144e-05,
+      "loss": 3.3956,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988165680473373e-05,
+      "loss": 3.4182,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986193293885602e-05,
+      "loss": 3.3563,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.984220907297831e-05,
+      "loss": 3.3066,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982248520710059e-05,
+      "loss": 3.2331,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980276134122289e-05,
+      "loss": 3.2169,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978303747534518e-05,
+      "loss": 3.1005,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.976331360946746e-05,
+      "loss": 3.0535,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974358974358975e-05,
+      "loss": 3.1198,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972386587771204e-05,
+      "loss": 3.0663,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970414201183432e-05,
+      "loss": 3.0584,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.968441814595661e-05,
+      "loss": 2.9469,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96646942800789e-05,
+      "loss": 3.0459,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96449704142012e-05,
+      "loss": 2.9426,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962524654832349e-05,
+      "loss": 2.9672,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960552268244577e-05,
+      "loss": 2.9189,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958579881656806e-05,
+      "loss": 2.914,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956607495069034e-05,
+      "loss": 2.8608,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954635108481263e-05,
+      "loss": 2.8851,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952662721893492e-05,
+      "loss": 2.9098,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95069033530572e-05,
+      "loss": 2.7703,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.948717948717949e-05,
+      "loss": 2.7624,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946745562130177e-05,
+      "loss": 2.7823,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944773175542406e-05,
+      "loss": 2.7804,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.942800788954635e-05,
+      "loss": 2.706,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.940828402366865e-05,
+      "loss": 2.7654,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.938856015779094e-05,
+      "loss": 2.7789,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.936883629191322e-05,
+      "loss": 2.7501,
+      "step": 32
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934911242603551e-05,
+      "loss": 2.7309,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93293885601578e-05,
+      "loss": 2.7298,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930966469428008e-05,
+      "loss": 2.7167,
+      "step": 35
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928994082840237e-05,
+      "loss": 2.7242,
+      "step": 36
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927021696252466e-05,
+      "loss": 2.6855,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925049309664694e-05,
+      "loss": 2.7618,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.923076923076923e-05,
+      "loss": 2.6294,
+      "step": 39
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.921104536489151e-05,
+      "loss": 2.5916,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91913214990138e-05,
+      "loss": 2.5775,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91715976331361e-05,
+      "loss": 2.633,
+      "step": 42
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.915187376725839e-05,
+      "loss": 2.6634,
+      "step": 43
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.913214990138068e-05,
+      "loss": 2.5683,
+      "step": 44
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.911242603550296e-05,
+      "loss": 2.6114,
+      "step": 45
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.909270216962525e-05,
+      "loss": 2.6691,
+      "step": 46
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.907297830374754e-05,
+      "loss": 2.6478,
+      "step": 47
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.905325443786982e-05,
+      "loss": 2.6553,
+      "step": 48
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.903353057199212e-05,
+      "loss": 2.531,
+      "step": 49
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.90138067061144e-05,
+      "loss": 2.565,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.899408284023669e-05,
+      "loss": 2.514,
+      "step": 51
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.897435897435898e-05,
+      "loss": 2.5883,
+      "step": 52
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.895463510848127e-05,
+      "loss": 2.499,
+      "step": 53
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.893491124260356e-05,
+      "loss": 2.5677,
+      "step": 54
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891518737672584e-05,
+      "loss": 2.5398,
+      "step": 55
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.889546351084813e-05,
+      "loss": 2.5657,
+      "step": 56
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.887573964497042e-05,
+      "loss": 2.5397,
+      "step": 57
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88560157790927e-05,
+      "loss": 2.5902,
+      "step": 58
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.8836291913215e-05,
+      "loss": 2.4764,
+      "step": 59
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.881656804733729e-05,
+      "loss": 2.6158,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.879684418145957e-05,
+      "loss": 2.6273,
+      "step": 61
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.877712031558186e-05,
+      "loss": 2.4931,
+      "step": 62
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.875739644970414e-05,
+      "loss": 2.4349,
+      "step": 63
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.873767258382643e-05,
+      "loss": 2.3798,
+      "step": 64
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 2.6425,
+      "step": 65
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.869822485207101e-05,
+      "loss": 2.4093,
+      "step": 66
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.86785009861933e-05,
+      "loss": 2.4182,
+      "step": 67
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.865877712031558e-05,
+      "loss": 2.4,
+      "step": 68
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.863905325443788e-05,
+      "loss": 2.5141,
+      "step": 69
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.861932938856017e-05,
+      "loss": 2.5375,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.859960552268245e-05,
+      "loss": 2.5405,
+      "step": 71
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.857988165680474e-05,
+      "loss": 2.498,
+      "step": 72
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.856015779092702e-05,
+      "loss": 2.3824,
+      "step": 73
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.854043392504931e-05,
+      "loss": 2.3929,
+      "step": 74
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.85207100591716e-05,
+      "loss": 2.5451,
+      "step": 75
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.85009861932939e-05,
+      "loss": 2.4987,
+      "step": 76
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.848126232741619e-05,
+      "loss": 2.3975,
+      "step": 77
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.846153846153848e-05,
+      "loss": 2.4368,
+      "step": 78
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.844181459566076e-05,
+      "loss": 2.4468,
+      "step": 79
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.842209072978305e-05,
+      "loss": 2.3,
+      "step": 80
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.840236686390533e-05,
+      "loss": 2.4105,
+      "step": 81
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.838264299802762e-05,
+      "loss": 2.471,
+      "step": 82
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.836291913214991e-05,
+      "loss": 2.354,
+      "step": 83
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.834319526627219e-05,
+      "loss": 2.4156,
+      "step": 84
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.832347140039448e-05,
+      "loss": 2.3962,
+      "step": 85
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.830374753451676e-05,
+      "loss": 2.3593,
+      "step": 86
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.828402366863905e-05,
+      "loss": 2.433,
+      "step": 87
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.826429980276134e-05,
+      "loss": 2.4312,
+      "step": 88
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.824457593688364e-05,
+      "loss": 2.3218,
+      "step": 89
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.822485207100593e-05,
+      "loss": 2.4151,
+      "step": 90
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.820512820512821e-05,
+      "loss": 2.4811,
+      "step": 91
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.81854043392505e-05,
+      "loss": 2.3644,
+      "step": 92
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.816568047337279e-05,
+      "loss": 2.3723,
+      "step": 93
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.814595660749507e-05,
+      "loss": 2.4164,
+      "step": 94
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.812623274161736e-05,
+      "loss": 2.3547,
+      "step": 95
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.810650887573964e-05,
+      "loss": 2.3637,
+      "step": 96
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.808678500986193e-05,
+      "loss": 2.3211,
+      "step": 97
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.806706114398423e-05,
+      "loss": 2.4678,
+      "step": 98
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.80473372781065e-05,
+      "loss": 2.4384,
+      "step": 99
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.80276134122288e-05,
+      "loss": 2.3458,
+      "step": 100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.800788954635109e-05,
+      "loss": 2.3613,
+      "step": 101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.798816568047338e-05,
+      "loss": 2.3403,
+      "step": 102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.796844181459567e-05,
+      "loss": 2.3316,
+      "step": 103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.794871794871795e-05,
+      "loss": 2.2789,
+      "step": 104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.792899408284024e-05,
+      "loss": 2.3241,
+      "step": 105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.790927021696254e-05,
+      "loss": 2.4394,
+      "step": 106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.788954635108481e-05,
+      "loss": 2.3581,
+      "step": 107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.78698224852071e-05,
+      "loss": 2.3995,
+      "step": 108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.785009861932938e-05,
+      "loss": 2.3988,
+      "step": 109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.783037475345168e-05,
+      "loss": 2.3599,
+      "step": 110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.781065088757397e-05,
+      "loss": 2.3956,
+      "step": 111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.779092702169626e-05,
+      "loss": 2.3661,
+      "step": 112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.777120315581855e-05,
+      "loss": 2.3244,
+      "step": 113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.775147928994083e-05,
+      "loss": 2.3445,
+      "step": 114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.773175542406312e-05,
+      "loss": 2.3377,
+      "step": 115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.771203155818542e-05,
+      "loss": 2.3339,
+      "step": 116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.76923076923077e-05,
+      "loss": 2.4014,
+      "step": 117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.767258382642999e-05,
+      "loss": 2.2225,
+      "step": 118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.765285996055226e-05,
+      "loss": 2.3873,
+      "step": 119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.763313609467456e-05,
+      "loss": 2.2908,
+      "step": 120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.761341222879685e-05,
+      "loss": 2.3221,
+      "step": 121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.759368836291913e-05,
+      "loss": 2.3277,
+      "step": 122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.757396449704142e-05,
+      "loss": 2.2136,
+      "step": 123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.755424063116371e-05,
+      "loss": 2.3274,
+      "step": 124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.7534516765286e-05,
+      "loss": 2.2523,
+      "step": 125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.75147928994083e-05,
+      "loss": 2.3595,
+      "step": 126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.749506903353057e-05,
+      "loss": 2.4048,
+      "step": 127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.747534516765287e-05,
+      "loss": 2.343,
+      "step": 128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.745562130177516e-05,
+      "loss": 2.2225,
+      "step": 129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 2.2596,
+      "step": 130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.741617357001973e-05,
+      "loss": 2.3328,
+      "step": 131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.739644970414201e-05,
+      "loss": 2.2945,
+      "step": 132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.73767258382643e-05,
+      "loss": 2.3803,
+      "step": 133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.735700197238659e-05,
+      "loss": 2.3361,
+      "step": 134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.733727810650888e-05,
+      "loss": 2.3168,
+      "step": 135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.731755424063118e-05,
+      "loss": 2.2209,
+      "step": 136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.729783037475346e-05,
+      "loss": 2.2811,
+      "step": 137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.727810650887575e-05,
+      "loss": 2.2663,
+      "step": 138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.725838264299804e-05,
+      "loss": 2.3283,
+      "step": 139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.723865877712032e-05,
+      "loss": 2.2832,
+      "step": 140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.721893491124261e-05,
+      "loss": 2.1317,
+      "step": 141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.719921104536489e-05,
+      "loss": 2.2627,
+      "step": 142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.717948717948718e-05,
+      "loss": 2.26,
+      "step": 143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.715976331360947e-05,
+      "loss": 2.3603,
+      "step": 144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.714003944773175e-05,
+      "loss": 2.2045,
+      "step": 145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.712031558185404e-05,
+      "loss": 2.2285,
+      "step": 146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.710059171597634e-05,
+      "loss": 2.3495,
+      "step": 147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.708086785009863e-05,
+      "loss": 2.297,
+      "step": 148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.706114398422092e-05,
+      "loss": 2.1912,
+      "step": 149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.70414201183432e-05,
+      "loss": 2.2025,
+      "step": 150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.702169625246549e-05,
+      "loss": 2.2518,
+      "step": 151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.700197238658778e-05,
+      "loss": 2.2001,
+      "step": 152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.698224852071006e-05,
+      "loss": 2.2375,
+      "step": 153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.696252465483235e-05,
+      "loss": 2.3301,
+      "step": 154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.694280078895463e-05,
+      "loss": 2.2667,
+      "step": 155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.692307692307692e-05,
+      "loss": 2.2469,
+      "step": 156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.690335305719922e-05,
+      "loss": 2.2954,
+      "step": 157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.68836291913215e-05,
+      "loss": 2.2033,
+      "step": 158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.686390532544379e-05,
+      "loss": 2.2853,
+      "step": 159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.684418145956608e-05,
+      "loss": 2.1632,
+      "step": 160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.682445759368837e-05,
+      "loss": 2.3111,
+      "step": 161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.680473372781066e-05,
+      "loss": 2.2767,
+      "step": 162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.678500986193294e-05,
+      "loss": 2.3181,
+      "step": 163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.676528599605523e-05,
+      "loss": 2.2527,
+      "step": 164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.674556213017751e-05,
+      "loss": 2.2613,
+      "step": 165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.67258382642998e-05,
+      "loss": 2.2212,
+      "step": 166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.67061143984221e-05,
+      "loss": 2.2292,
+      "step": 167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.668639053254438e-05,
+      "loss": 2.2337,
+      "step": 168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 2.2329,
+      "step": 169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.664694280078896e-05,
+      "loss": 2.3045,
+      "step": 170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.662721893491125e-05,
+      "loss": 2.3365,
+      "step": 171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.660749506903354e-05,
+      "loss": 2.2489,
+      "step": 172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.658777120315582e-05,
+      "loss": 2.2947,
+      "step": 173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.656804733727811e-05,
+      "loss": 2.2058,
+      "step": 174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.65483234714004e-05,
+      "loss": 2.1562,
+      "step": 175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.652859960552268e-05,
+      "loss": 2.2771,
+      "step": 176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.650887573964498e-05,
+      "loss": 2.2572,
+      "step": 177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648915187376726e-05,
+      "loss": 2.2854,
+      "step": 178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646942800788955e-05,
+      "loss": 2.2783,
+      "step": 179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644970414201184e-05,
+      "loss": 2.1126,
+      "step": 180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.642998027613412e-05,
+      "loss": 2.1158,
+      "step": 181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.641025641025641e-05,
+      "loss": 2.2107,
+      "step": 182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.63905325443787e-05,
+      "loss": 2.1353,
+      "step": 183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.6370808678501e-05,
+      "loss": 2.0108,
+      "step": 184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.635108481262329e-05,
+      "loss": 2.1017,
+      "step": 185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.633136094674557e-05,
+      "loss": 2.1909,
+      "step": 186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.631163708086786e-05,
+      "loss": 2.0382,
+      "step": 187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.629191321499015e-05,
+      "loss": 2.2003,
+      "step": 188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.627218934911243e-05,
+      "loss": 2.2472,
+      "step": 189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.625246548323472e-05,
+      "loss": 2.1721,
+      "step": 190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.6232741617357e-05,
+      "loss": 2.1999,
+      "step": 191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.621301775147929e-05,
+      "loss": 2.2523,
+      "step": 192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.619329388560158e-05,
+      "loss": 2.1618,
+      "step": 193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.617357001972388e-05,
+      "loss": 2.2756,
+      "step": 194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 2.2559,
+      "step": 195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.613412228796845e-05,
+      "loss": 2.2252,
+      "step": 196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.611439842209074e-05,
+      "loss": 2.1343,
+      "step": 197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.609467455621303e-05,
+      "loss": 2.1231,
+      "step": 198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.607495069033531e-05,
+      "loss": 2.2587,
+      "step": 199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.60552268244576e-05,
+      "loss": 2.0799,
+      "step": 200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.603550295857988e-05,
+      "loss": 2.1718,
+      "step": 201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.601577909270217e-05,
+      "loss": 2.2614,
+      "step": 202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.599605522682446e-05,
+      "loss": 2.1406,
+      "step": 203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.597633136094674e-05,
+      "loss": 2.1099,
+      "step": 204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.595660749506903e-05,
+      "loss": 2.1277,
+      "step": 205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.593688362919133e-05,
+      "loss": 2.1857,
+      "step": 206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.591715976331362e-05,
+      "loss": 2.1206,
+      "step": 207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.589743589743591e-05,
+      "loss": 2.1587,
+      "step": 208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.587771203155819e-05,
+      "loss": 2.14,
+      "step": 209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.585798816568048e-05,
+      "loss": 2.1706,
+      "step": 210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.583826429980277e-05,
+      "loss": 2.1826,
+      "step": 211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.581854043392505e-05,
+      "loss": 2.0951,
+      "step": 212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.579881656804734e-05,
+      "loss": 2.1804,
+      "step": 213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.577909270216962e-05,
+      "loss": 2.1596,
+      "step": 214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.575936883629191e-05,
+      "loss": 2.1171,
+      "step": 215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.573964497041421e-05,
+      "loss": 2.0727,
+      "step": 216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.571992110453649e-05,
+      "loss": 2.2219,
+      "step": 217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.570019723865878e-05,
+      "loss": 2.1494,
+      "step": 218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.568047337278107e-05,
+      "loss": 2.184,
+      "step": 219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.566074950690336e-05,
+      "loss": 2.1904,
+      "step": 220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.564102564102565e-05,
+      "loss": 2.306,
+      "step": 221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.562130177514793e-05,
+      "loss": 2.1098,
+      "step": 222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.560157790927022e-05,
+      "loss": 2.1654,
+      "step": 223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.55818540433925e-05,
+      "loss": 2.0218,
+      "step": 224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.55621301775148e-05,
+      "loss": 2.1706,
+      "step": 225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.554240631163709e-05,
+      "loss": 2.1258,
+      "step": 226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.552268244575937e-05,
+      "loss": 2.1647,
+      "step": 227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.550295857988166e-05,
+      "loss": 2.1608,
+      "step": 228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.548323471400395e-05,
+      "loss": 2.0605,
+      "step": 229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.546351084812624e-05,
+      "loss": 2.1443,
+      "step": 230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.544378698224853e-05,
+      "loss": 2.1171,
+      "step": 231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.542406311637081e-05,
+      "loss": 2.1851,
+      "step": 232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.54043392504931e-05,
+      "loss": 2.2092,
+      "step": 233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.53846153846154e-05,
+      "loss": 2.1555,
+      "step": 234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.536489151873768e-05,
+      "loss": 2.1382,
+      "step": 235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.534516765285997e-05,
+      "loss": 2.1275,
+      "step": 236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.532544378698225e-05,
+      "loss": 2.1537,
+      "step": 237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.530571992110454e-05,
+      "loss": 2.0421,
+      "step": 238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.528599605522683e-05,
+      "loss": 1.9904,
+      "step": 239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.526627218934911e-05,
+      "loss": 2.1497,
+      "step": 240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.52465483234714e-05,
+      "loss": 2.1138,
+      "step": 241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.52268244575937e-05,
+      "loss": 2.0724,
+      "step": 242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.520710059171599e-05,
+      "loss": 1.9776,
+      "step": 243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.518737672583828e-05,
+      "loss": 2.1538,
+      "step": 244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.516765285996056e-05,
+      "loss": 2.0735,
+      "step": 245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.514792899408285e-05,
+      "loss": 2.0963,
+      "step": 246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.512820512820513e-05,
+      "loss": 2.1212,
+      "step": 247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.510848126232742e-05,
+      "loss": 2.2163,
+      "step": 248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.508875739644971e-05,
+      "loss": 2.114,
+      "step": 249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.506903353057199e-05,
+      "loss": 2.1741,
+      "step": 250
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 1.8406490087509155,
+      "eval_runtime": 69.2657,
+      "eval_samples_per_second": 321.515,
+      "eval_steps_per_second": 1.675,
+      "step": 250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.504930966469428e-05,
+      "loss": 2.1463,
+      "step": 251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.502958579881657e-05,
+      "loss": 2.1177,
+      "step": 252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.500986193293887e-05,
+      "loss": 2.1241,
+      "step": 253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.499013806706116e-05,
+      "loss": 2.0676,
+      "step": 254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.497041420118344e-05,
+      "loss": 2.2031,
+      "step": 255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.495069033530573e-05,
+      "loss": 2.0476,
+      "step": 256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.493096646942802e-05,
+      "loss": 2.1706,
+      "step": 257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.49112426035503e-05,
+      "loss": 2.1243,
+      "step": 258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.489151873767259e-05,
+      "loss": 2.0723,
+      "step": 259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 1.9851,
+      "step": 260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.485207100591716e-05,
+      "loss": 2.1657,
+      "step": 261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.483234714003945e-05,
+      "loss": 2.0537,
+      "step": 262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.481262327416173e-05,
+      "loss": 2.049,
+      "step": 263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.479289940828402e-05,
+      "loss": 2.1241,
+      "step": 264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.477317554240632e-05,
+      "loss": 2.0893,
+      "step": 265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.475345167652861e-05,
+      "loss": 2.1441,
+      "step": 266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.47337278106509e-05,
+      "loss": 2.1287,
+      "step": 267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.471400394477318e-05,
+      "loss": 2.1959,
+      "step": 268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.469428007889547e-05,
+      "loss": 2.1282,
+      "step": 269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.467455621301775e-05,
+      "loss": 1.9884,
+      "step": 270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.465483234714004e-05,
+      "loss": 2.2444,
+      "step": 271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.463510848126233e-05,
+      "loss": 2.1313,
+      "step": 272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.461538461538461e-05,
+      "loss": 2.0675,
+      "step": 273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.45956607495069e-05,
+      "loss": 2.0643,
+      "step": 274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457593688362918e-05,
+      "loss": 2.2047,
+      "step": 275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.455621301775148e-05,
+      "loss": 2.098,
+      "step": 276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.453648915187377e-05,
+      "loss": 2.1079,
+      "step": 277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.451676528599606e-05,
+      "loss": 2.0961,
+      "step": 278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.449704142011835e-05,
+      "loss": 2.2097,
+      "step": 279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.447731755424064e-05,
+      "loss": 2.0825,
+      "step": 280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.445759368836292e-05,
+      "loss": 2.0384,
+      "step": 281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.443786982248522e-05,
+      "loss": 1.98,
+      "step": 282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.44181459566075e-05,
+      "loss": 2.1016,
+      "step": 283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.439842209072979e-05,
+      "loss": 2.0015,
+      "step": 284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.437869822485208e-05,
+      "loss": 2.0672,
+      "step": 285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.435897435897436e-05,
+      "loss": 2.048,
+      "step": 286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.433925049309665e-05,
+      "loss": 2.0326,
+      "step": 287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.431952662721894e-05,
+      "loss": 2.1843,
+      "step": 288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429980276134123e-05,
+      "loss": 2.0338,
+      "step": 289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.428007889546353e-05,
+      "loss": 2.0516,
+      "step": 290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.42603550295858e-05,
+      "loss": 2.0264,
+      "step": 291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.42406311637081e-05,
+      "loss": 2.1196,
+      "step": 292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.422090729783037e-05,
+      "loss": 2.1012,
+      "step": 293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.420118343195267e-05,
+      "loss": 2.0462,
+      "step": 294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.418145956607496e-05,
+      "loss": 1.9924,
+      "step": 295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.416173570019724e-05,
+      "loss": 1.9484,
+      "step": 296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.414201183431953e-05,
+      "loss": 2.0841,
+      "step": 297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.412228796844182e-05,
+      "loss": 2.0088,
+      "step": 298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.41025641025641e-05,
+      "loss": 2.1245,
+      "step": 299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.408284023668639e-05,
+      "loss": 1.9897,
+      "step": 300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.406311637080868e-05,
+      "loss": 2.1443,
+      "step": 301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.404339250493098e-05,
+      "loss": 2.0554,
+      "step": 302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.402366863905327e-05,
+      "loss": 1.9841,
+      "step": 303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.400394477317555e-05,
+      "loss": 2.1372,
+      "step": 304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.398422090729784e-05,
+      "loss": 2.1294,
+      "step": 305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.396449704142012e-05,
+      "loss": 2.0112,
+      "step": 306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.394477317554241e-05,
+      "loss": 2.1047,
+      "step": 307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.39250493096647e-05,
+      "loss": 2.0877,
+      "step": 308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.390532544378698e-05,
+      "loss": 2.122,
+      "step": 309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.388560157790927e-05,
+      "loss": 2.0349,
+      "step": 310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.386587771203156e-05,
+      "loss": 2.1113,
+      "step": 311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.384615384615386e-05,
+      "loss": 2.1833,
+      "step": 312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.382642998027615e-05,
+      "loss": 2.1157,
+      "step": 313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.380670611439843e-05,
+      "loss": 1.9738,
+      "step": 314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.378698224852072e-05,
+      "loss": 1.9967,
+      "step": 315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.3767258382643e-05,
+      "loss": 2.0019,
+      "step": 316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.374753451676529e-05,
+      "loss": 1.9836,
+      "step": 317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.372781065088758e-05,
+      "loss": 2.0119,
+      "step": 318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.370808678500986e-05,
+      "loss": 2.0737,
+      "step": 319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.368836291913215e-05,
+      "loss": 2.1024,
+      "step": 320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.366863905325445e-05,
+      "loss": 2.0557,
+      "step": 321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.364891518737672e-05,
+      "loss": 2.1436,
+      "step": 322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.362919132149902e-05,
+      "loss": 2.0604,
+      "step": 323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.360946745562131e-05,
+      "loss": 2.0031,
+      "step": 324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 2.0288,
+      "step": 325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.357001972386589e-05,
+      "loss": 2.0268,
+      "step": 326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.355029585798817e-05,
+      "loss": 2.0515,
+      "step": 327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.353057199211046e-05,
+      "loss": 2.0537,
+      "step": 328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.351084812623274e-05,
+      "loss": 1.9182,
+      "step": 329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.349112426035503e-05,
+      "loss": 1.9767,
+      "step": 330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.347140039447733e-05,
+      "loss": 1.9775,
+      "step": 331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.34516765285996e-05,
+      "loss": 2.0768,
+      "step": 332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.34319526627219e-05,
+      "loss": 2.1042,
+      "step": 333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.341222879684417e-05,
+      "loss": 2.0881,
+      "step": 334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.339250493096647e-05,
+      "loss": 2.0554,
+      "step": 335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.337278106508876e-05,
+      "loss": 2.0446,
+      "step": 336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.335305719921105e-05,
+      "loss": 1.9072,
+      "step": 337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 1.9938,
+      "step": 338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.331360946745562e-05,
+      "loss": 1.9686,
+      "step": 339
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.329388560157791e-05,
+      "loss": 2.0474,
+      "step": 340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.32741617357002e-05,
+      "loss": 1.9905,
+      "step": 341
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.325443786982248e-05,
+      "loss": 1.9465,
+      "step": 342
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.323471400394478e-05,
+      "loss": 1.9013,
+      "step": 343
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.321499013806707e-05,
+      "loss": 2.0105,
+      "step": 344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.319526627218935e-05,
+      "loss": 2.0188,
+      "step": 345
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.317554240631164e-05,
+      "loss": 1.9706,
+      "step": 346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.315581854043393e-05,
+      "loss": 1.957,
+      "step": 347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.313609467455622e-05,
+      "loss": 2.0473,
+      "step": 348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.311637080867852e-05,
+      "loss": 1.9829,
+      "step": 349
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.30966469428008e-05,
+      "loss": 1.8897,
+      "step": 350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.307692307692309e-05,
+      "loss": 1.9492,
+      "step": 351
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.305719921104537e-05,
+      "loss": 2.0767,
+      "step": 352
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.303747534516766e-05,
+      "loss": 2.0562,
+      "step": 353
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.301775147928995e-05,
+      "loss": 1.9344,
+      "step": 354
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.299802761341223e-05,
+      "loss": 2.0683,
+      "step": 355
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.297830374753452e-05,
+      "loss": 1.9956,
+      "step": 356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.29585798816568e-05,
+      "loss": 1.9246,
+      "step": 357
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.293885601577909e-05,
+      "loss": 1.9615,
+      "step": 358
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.291913214990138e-05,
+      "loss": 2.0332,
+      "step": 359
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.289940828402367e-05,
+      "loss": 1.9608,
+      "step": 360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.287968441814597e-05,
+      "loss": 2.1074,
+      "step": 361
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.285996055226825e-05,
+      "loss": 2.0347,
+      "step": 362
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.284023668639054e-05,
+      "loss": 2.0537,
+      "step": 363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.282051282051283e-05,
+      "loss": 2.0514,
+      "step": 364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.280078895463511e-05,
+      "loss": 2.0581,
+      "step": 365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.27810650887574e-05,
+      "loss": 2.047,
+      "step": 366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.276134122287969e-05,
+      "loss": 2.1184,
+      "step": 367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.274161735700197e-05,
+      "loss": 1.9546,
+      "step": 368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.272189349112426e-05,
+      "loss": 2.0367,
+      "step": 369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.270216962524656e-05,
+      "loss": 1.9045,
+      "step": 370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.268244575936885e-05,
+      "loss": 2.0171,
+      "step": 371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.266272189349114e-05,
+      "loss": 2.0078,
+      "step": 372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.264299802761342e-05,
+      "loss": 2.0523,
+      "step": 373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.262327416173571e-05,
+      "loss": 2.0317,
+      "step": 374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.260355029585799e-05,
+      "loss": 1.9758,
+      "step": 375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.258382642998028e-05,
+      "loss": 2.065,
+      "step": 376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.256410256410257e-05,
+      "loss": 2.0664,
+      "step": 377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.254437869822485e-05,
+      "loss": 2.0338,
+      "step": 378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.252465483234714e-05,
+      "loss": 2.0023,
+      "step": 379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.250493096646942e-05,
+      "loss": 1.9733,
+      "step": 380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.248520710059171e-05,
+      "loss": 1.9596,
+      "step": 381
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.2465483234714e-05,
+      "loss": 1.9707,
+      "step": 382
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.24457593688363e-05,
+      "loss": 1.9494,
+      "step": 383
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.242603550295859e-05,
+      "loss": 2.0083,
+      "step": 384
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.240631163708087e-05,
+      "loss": 2.0268,
+      "step": 385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.238658777120316e-05,
+      "loss": 2.0073,
+      "step": 386
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.236686390532545e-05,
+      "loss": 1.9929,
+      "step": 387
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.234714003944773e-05,
+      "loss": 1.9789,
+      "step": 388
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.232741617357002e-05,
+      "loss": 1.9,
+      "step": 389
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 2.0509,
+      "step": 390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.22879684418146e-05,
+      "loss": 2.017,
+      "step": 391
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.226824457593689e-05,
+      "loss": 1.8802,
+      "step": 392
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.224852071005917e-05,
+      "loss": 2.0176,
+      "step": 393
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.222879684418146e-05,
+      "loss": 2.028,
+      "step": 394
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.220907297830376e-05,
+      "loss": 1.9025,
+      "step": 395
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.218934911242604e-05,
+      "loss": 1.9234,
+      "step": 396
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.216962524654833e-05,
+      "loss": 2.0351,
+      "step": 397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.214990138067061e-05,
+      "loss": 1.9522,
+      "step": 398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.21301775147929e-05,
+      "loss": 1.95,
+      "step": 399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.21104536489152e-05,
+      "loss": 2.0061,
+      "step": 400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.209072978303748e-05,
+      "loss": 2.0925,
+      "step": 401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.207100591715977e-05,
+      "loss": 1.892,
+      "step": 402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.205128205128205e-05,
+      "loss": 2.0573,
+      "step": 403
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.203155818540434e-05,
+      "loss": 1.9609,
+      "step": 404
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.201183431952663e-05,
+      "loss": 2.0249,
+      "step": 405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.199211045364892e-05,
+      "loss": 1.9952,
+      "step": 406
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.197238658777121e-05,
+      "loss": 1.9616,
+      "step": 407
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.19526627218935e-05,
+      "loss": 2.0223,
+      "step": 408
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.193293885601579e-05,
+      "loss": 2.023,
+      "step": 409
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.191321499013808e-05,
+      "loss": 2.0975,
+      "step": 410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.189349112426036e-05,
+      "loss": 1.9645,
+      "step": 411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.187376725838265e-05,
+      "loss": 1.992,
+      "step": 412
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.185404339250494e-05,
+      "loss": 2.0163,
+      "step": 413
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.183431952662722e-05,
+      "loss": 2.0367,
+      "step": 414
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.181459566074951e-05,
+      "loss": 2.0698,
+      "step": 415
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.179487179487179e-05,
+      "loss": 1.9803,
+      "step": 416
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.177514792899408e-05,
+      "loss": 1.9054,
+      "step": 417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.175542406311637e-05,
+      "loss": 2.0156,
+      "step": 418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.173570019723867e-05,
+      "loss": 1.9809,
+      "step": 419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.171597633136096e-05,
+      "loss": 1.9561,
+      "step": 420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.169625246548324e-05,
+      "loss": 1.9079,
+      "step": 421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.167652859960553e-05,
+      "loss": 2.0044,
+      "step": 422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.165680473372782e-05,
+      "loss": 1.9986,
+      "step": 423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.16370808678501e-05,
+      "loss": 1.9406,
+      "step": 424
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.161735700197239e-05,
+      "loss": 1.9057,
+      "step": 425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.159763313609467e-05,
+      "loss": 1.7755,
+      "step": 426
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.157790927021696e-05,
+      "loss": 1.8075,
+      "step": 427
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.155818540433925e-05,
+      "loss": 2.0759,
+      "step": 428
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.153846153846155e-05,
+      "loss": 1.9309,
+      "step": 429
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.151873767258384e-05,
+      "loss": 2.0779,
+      "step": 430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.149901380670613e-05,
+      "loss": 2.0214,
+      "step": 431
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.147928994082841e-05,
+      "loss": 2.0151,
+      "step": 432
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.14595660749507e-05,
+      "loss": 1.9855,
+      "step": 433
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.143984220907298e-05,
+      "loss": 1.9801,
+      "step": 434
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.142011834319527e-05,
+      "loss": 1.92,
+      "step": 435
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.140039447731756e-05,
+      "loss": 1.8712,
+      "step": 436
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.138067061143984e-05,
+      "loss": 2.0598,
+      "step": 437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.136094674556213e-05,
+      "loss": 1.9474,
+      "step": 438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.134122287968441e-05,
+      "loss": 1.9577,
+      "step": 439
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.13214990138067e-05,
+      "loss": 1.9435,
+      "step": 440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.1301775147929e-05,
+      "loss": 1.9947,
+      "step": 441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.128205128205129e-05,
+      "loss": 2.0715,
+      "step": 442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.126232741617358e-05,
+      "loss": 1.8831,
+      "step": 443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.124260355029586e-05,
+      "loss": 2.0133,
+      "step": 444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.122287968441815e-05,
+      "loss": 2.001,
+      "step": 445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.120315581854044e-05,
+      "loss": 2.0648,
+      "step": 446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.118343195266272e-05,
+      "loss": 1.9433,
+      "step": 447
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.116370808678501e-05,
+      "loss": 1.9728,
+      "step": 448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.11439842209073e-05,
+      "loss": 1.9932,
+      "step": 449
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.112426035502959e-05,
+      "loss": 1.9782,
+      "step": 450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.110453648915188e-05,
+      "loss": 1.8738,
+      "step": 451
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.108481262327416e-05,
+      "loss": 1.9597,
+      "step": 452
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.106508875739645e-05,
+      "loss": 1.9394,
+      "step": 453
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.104536489151875e-05,
+      "loss": 2.0332,
+      "step": 454
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 1.8407,
+      "step": 455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.100591715976332e-05,
+      "loss": 2.0598,
+      "step": 456
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.09861932938856e-05,
+      "loss": 1.9703,
+      "step": 457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.09664694280079e-05,
+      "loss": 2.0203,
+      "step": 458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.094674556213019e-05,
+      "loss": 2.0046,
+      "step": 459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.092702169625247e-05,
+      "loss": 1.9535,
+      "step": 460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.090729783037476e-05,
+      "loss": 1.9881,
+      "step": 461
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.088757396449704e-05,
+      "loss": 1.9218,
+      "step": 462
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.086785009861933e-05,
+      "loss": 1.9943,
+      "step": 463
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.084812623274162e-05,
+      "loss": 2.036,
+      "step": 464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.082840236686391e-05,
+      "loss": 1.9648,
+      "step": 465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.08086785009862e-05,
+      "loss": 2.0132,
+      "step": 466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.078895463510848e-05,
+      "loss": 1.8734,
+      "step": 467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.076923076923078e-05,
+      "loss": 1.9128,
+      "step": 468
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.074950690335307e-05,
+      "loss": 1.9224,
+      "step": 469
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.072978303747535e-05,
+      "loss": 1.8659,
+      "step": 470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.071005917159764e-05,
+      "loss": 1.8589,
+      "step": 471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.069033530571992e-05,
+      "loss": 2.0101,
+      "step": 472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.067061143984221e-05,
+      "loss": 1.9228,
+      "step": 473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.06508875739645e-05,
+      "loss": 1.937,
+      "step": 474
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.063116370808678e-05,
+      "loss": 1.9795,
+      "step": 475
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.061143984220907e-05,
+      "loss": 1.9465,
+      "step": 476
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.059171597633136e-05,
+      "loss": 1.7728,
+      "step": 477
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.057199211045366e-05,
+      "loss": 2.0266,
+      "step": 478
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.055226824457595e-05,
+      "loss": 1.8395,
+      "step": 479
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.053254437869823e-05,
+      "loss": 1.9741,
+      "step": 480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.051282051282052e-05,
+      "loss": 1.9765,
+      "step": 481
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.049309664694281e-05,
+      "loss": 2.001,
+      "step": 482
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.047337278106509e-05,
+      "loss": 1.8838,
+      "step": 483
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.045364891518738e-05,
+      "loss": 1.9306,
+      "step": 484
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.043392504930966e-05,
+      "loss": 2.0794,
+      "step": 485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.041420118343195e-05,
+      "loss": 1.9838,
+      "step": 486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.039447731755424e-05,
+      "loss": 1.9172,
+      "step": 487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.037475345167654e-05,
+      "loss": 2.0079,
+      "step": 488
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.035502958579883e-05,
+      "loss": 1.9569,
+      "step": 489
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.033530571992111e-05,
+      "loss": 1.8171,
+      "step": 490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.03155818540434e-05,
+      "loss": 1.8922,
+      "step": 491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.029585798816569e-05,
+      "loss": 1.97,
+      "step": 492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.027613412228797e-05,
+      "loss": 1.9989,
+      "step": 493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.025641025641026e-05,
+      "loss": 2.0203,
+      "step": 494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.023668639053254e-05,
+      "loss": 1.9162,
+      "step": 495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.021696252465483e-05,
+      "loss": 1.972,
+      "step": 496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.019723865877713e-05,
+      "loss": 1.8764,
+      "step": 497
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.01775147928994e-05,
+      "loss": 2.0029,
+      "step": 498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.01577909270217e-05,
+      "loss": 1.9803,
+      "step": 499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.013806706114399e-05,
+      "loss": 1.9237,
+      "step": 500
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.6951730251312256,
+      "eval_runtime": 69.4,
+      "eval_samples_per_second": 320.894,
+      "eval_steps_per_second": 1.671,
+      "step": 500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.011834319526628e-05,
+      "loss": 1.8567,
+      "step": 501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.009861932938857e-05,
+      "loss": 1.9247,
+      "step": 502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.007889546351085e-05,
+      "loss": 1.9948,
+      "step": 503
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.005917159763314e-05,
+      "loss": 1.9587,
+      "step": 504
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.003944773175543e-05,
+      "loss": 1.917,
+      "step": 505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.001972386587771e-05,
+      "loss": 1.8527,
+      "step": 506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9e-05,
+      "loss": 1.9483,
+      "step": 507
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.998027613412228e-05,
+      "loss": 1.9796,
+      "step": 508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.996055226824458e-05,
+      "loss": 2.0263,
+      "step": 509
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.994082840236687e-05,
+      "loss": 1.8482,
+      "step": 510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.992110453648915e-05,
+      "loss": 1.9704,
+      "step": 511
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.990138067061144e-05,
+      "loss": 1.9237,
+      "step": 512
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.988165680473373e-05,
+      "loss": 1.9591,
+      "step": 513
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.986193293885602e-05,
+      "loss": 2.0286,
+      "step": 514
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.984220907297832e-05,
+      "loss": 1.9616,
+      "step": 515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.98224852071006e-05,
+      "loss": 1.8996,
+      "step": 516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.980276134122289e-05,
+      "loss": 1.937,
+      "step": 517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.978303747534516e-05,
+      "loss": 1.8483,
+      "step": 518
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.976331360946746e-05,
+      "loss": 1.9662,
+      "step": 519
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 1.8575,
+      "step": 520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.972386587771203e-05,
+      "loss": 1.9057,
+      "step": 521
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.970414201183432e-05,
+      "loss": 1.9537,
+      "step": 522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.968441814595661e-05,
+      "loss": 1.9032,
+      "step": 523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.96646942800789e-05,
+      "loss": 1.8777,
+      "step": 524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.96449704142012e-05,
+      "loss": 1.974,
+      "step": 525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.962524654832347e-05,
+      "loss": 1.9356,
+      "step": 526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.960552268244577e-05,
+      "loss": 1.9666,
+      "step": 527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.958579881656806e-05,
+      "loss": 1.9192,
+      "step": 528
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.956607495069034e-05,
+      "loss": 1.9061,
+      "step": 529
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.954635108481263e-05,
+      "loss": 1.9993,
+      "step": 530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.952662721893491e-05,
+      "loss": 1.8754,
+      "step": 531
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.95069033530572e-05,
+      "loss": 1.8501,
+      "step": 532
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.948717948717949e-05,
+      "loss": 1.9235,
+      "step": 533
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.946745562130177e-05,
+      "loss": 1.9945,
+      "step": 534
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.944773175542406e-05,
+      "loss": 1.9604,
+      "step": 535
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.942800788954635e-05,
+      "loss": 1.8508,
+      "step": 536
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.940828402366865e-05,
+      "loss": 1.8619,
+      "step": 537
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.938856015779094e-05,
+      "loss": 1.8698,
+      "step": 538
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.936883629191322e-05,
+      "loss": 1.971,
+      "step": 539
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.934911242603551e-05,
+      "loss": 1.8223,
+      "step": 540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.93293885601578e-05,
+      "loss": 1.8814,
+      "step": 541
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.930966469428008e-05,
+      "loss": 1.8664,
+      "step": 542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.928994082840237e-05,
+      "loss": 1.929,
+      "step": 543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.927021696252465e-05,
+      "loss": 1.8556,
+      "step": 544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.925049309664694e-05,
+      "loss": 1.8419,
+      "step": 545
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.923076923076924e-05,
+      "loss": 1.9459,
+      "step": 546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.921104536489153e-05,
+      "loss": 1.8525,
+      "step": 547
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.919132149901382e-05,
+      "loss": 1.8501,
+      "step": 548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.91715976331361e-05,
+      "loss": 1.8935,
+      "step": 549
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.915187376725839e-05,
+      "loss": 1.9246,
+      "step": 550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.913214990138068e-05,
+      "loss": 2.0124,
+      "step": 551
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.911242603550296e-05,
+      "loss": 1.9211,
+      "step": 552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.909270216962525e-05,
+      "loss": 1.8827,
+      "step": 553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.907297830374753e-05,
+      "loss": 1.8695,
+      "step": 554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.905325443786982e-05,
+      "loss": 1.9441,
+      "step": 555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.903353057199212e-05,
+      "loss": 1.8444,
+      "step": 556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90138067061144e-05,
+      "loss": 1.8964,
+      "step": 557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899408284023669e-05,
+      "loss": 1.7572,
+      "step": 558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897435897435898e-05,
+      "loss": 1.9514,
+      "step": 559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.895463510848127e-05,
+      "loss": 1.9352,
+      "step": 560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.893491124260356e-05,
+      "loss": 1.8888,
+      "step": 561
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.891518737672584e-05,
+      "loss": 1.9046,
+      "step": 562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.889546351084813e-05,
+      "loss": 1.8477,
+      "step": 563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.887573964497043e-05,
+      "loss": 1.9415,
+      "step": 564
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.88560157790927e-05,
+      "loss": 1.9449,
+      "step": 565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.8836291913215e-05,
+      "loss": 1.9514,
+      "step": 566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.881656804733727e-05,
+      "loss": 1.896,
+      "step": 567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.879684418145957e-05,
+      "loss": 1.8957,
+      "step": 568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.877712031558186e-05,
+      "loss": 1.9475,
+      "step": 569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.875739644970414e-05,
+      "loss": 1.8998,
+      "step": 570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.873767258382644e-05,
+      "loss": 1.7727,
+      "step": 571
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.871794871794872e-05,
+      "loss": 1.8896,
+      "step": 572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.869822485207101e-05,
+      "loss": 1.9018,
+      "step": 573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.86785009861933e-05,
+      "loss": 1.9115,
+      "step": 574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.865877712031558e-05,
+      "loss": 1.8884,
+      "step": 575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.863905325443788e-05,
+      "loss": 1.897,
+      "step": 576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.861932938856016e-05,
+      "loss": 1.8468,
+      "step": 577
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.859960552268245e-05,
+      "loss": 1.9185,
+      "step": 578
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.857988165680474e-05,
+      "loss": 1.9223,
+      "step": 579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.856015779092702e-05,
+      "loss": 1.8192,
+      "step": 580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.854043392504931e-05,
+      "loss": 2.0479,
+      "step": 581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.85207100591716e-05,
+      "loss": 1.9365,
+      "step": 582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.85009861932939e-05,
+      "loss": 1.8941,
+      "step": 583
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.848126232741619e-05,
+      "loss": 1.9021,
+      "step": 584
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 1.9156,
+      "step": 585
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.844181459566076e-05,
+      "loss": 1.8298,
+      "step": 586
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.842209072978305e-05,
+      "loss": 1.8583,
+      "step": 587
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.840236686390533e-05,
+      "loss": 1.8808,
+      "step": 588
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.838264299802762e-05,
+      "loss": 1.8663,
+      "step": 589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.83629191321499e-05,
+      "loss": 1.8732,
+      "step": 590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.834319526627219e-05,
+      "loss": 1.9054,
+      "step": 591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.832347140039448e-05,
+      "loss": 1.8904,
+      "step": 592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.830374753451676e-05,
+      "loss": 1.8218,
+      "step": 593
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.828402366863905e-05,
+      "loss": 2.0163,
+      "step": 594
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.826429980276135e-05,
+      "loss": 1.8797,
+      "step": 595
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.824457593688364e-05,
+      "loss": 1.9516,
+      "step": 596
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.822485207100593e-05,
+      "loss": 1.9858,
+      "step": 597
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.820512820512821e-05,
+      "loss": 1.7547,
+      "step": 598
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.81854043392505e-05,
+      "loss": 1.8128,
+      "step": 599
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.816568047337278e-05,
+      "loss": 1.8551,
+      "step": 600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.814595660749507e-05,
+      "loss": 1.7966,
+      "step": 601
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.812623274161736e-05,
+      "loss": 1.9422,
+      "step": 602
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.810650887573964e-05,
+      "loss": 1.8748,
+      "step": 603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.808678500986193e-05,
+      "loss": 1.9392,
+      "step": 604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.806706114398423e-05,
+      "loss": 1.8507,
+      "step": 605
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.804733727810652e-05,
+      "loss": 1.9501,
+      "step": 606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.802761341222881e-05,
+      "loss": 1.9822,
+      "step": 607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.800788954635109e-05,
+      "loss": 1.8885,
+      "step": 608
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.798816568047338e-05,
+      "loss": 1.8534,
+      "step": 609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.796844181459567e-05,
+      "loss": 1.8256,
+      "step": 610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.794871794871795e-05,
+      "loss": 1.8355,
+      "step": 611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.792899408284024e-05,
+      "loss": 1.9757,
+      "step": 612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.790927021696252e-05,
+      "loss": 1.8404,
+      "step": 613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.788954635108481e-05,
+      "loss": 1.8528,
+      "step": 614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.78698224852071e-05,
+      "loss": 1.9774,
+      "step": 615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.785009861932939e-05,
+      "loss": 1.8588,
+      "step": 616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.783037475345168e-05,
+      "loss": 1.922,
+      "step": 617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.781065088757397e-05,
+      "loss": 1.8662,
+      "step": 618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.779092702169626e-05,
+      "loss": 1.8963,
+      "step": 619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.777120315581855e-05,
+      "loss": 1.8526,
+      "step": 620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.775147928994083e-05,
+      "loss": 1.866,
+      "step": 621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.773175542406312e-05,
+      "loss": 1.8987,
+      "step": 622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.77120315581854e-05,
+      "loss": 1.9756,
+      "step": 623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.76923076923077e-05,
+      "loss": 1.8095,
+      "step": 624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.767258382642999e-05,
+      "loss": 1.858,
+      "step": 625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.765285996055227e-05,
+      "loss": 1.9295,
+      "step": 626
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.763313609467456e-05,
+      "loss": 1.9062,
+      "step": 627
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.761341222879684e-05,
+      "loss": 1.7825,
+      "step": 628
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.759368836291913e-05,
+      "loss": 1.8462,
+      "step": 629
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.757396449704143e-05,
+      "loss": 1.891,
+      "step": 630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.755424063116371e-05,
+      "loss": 1.8922,
+      "step": 631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.7534516765286e-05,
+      "loss": 1.9233,
+      "step": 632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.75147928994083e-05,
+      "loss": 1.8959,
+      "step": 633
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.749506903353058e-05,
+      "loss": 1.9283,
+      "step": 634
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.747534516765287e-05,
+      "loss": 1.9196,
+      "step": 635
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.745562130177515e-05,
+      "loss": 1.9098,
+      "step": 636
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.743589743589744e-05,
+      "loss": 1.9496,
+      "step": 637
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.741617357001973e-05,
+      "loss": 1.8607,
+      "step": 638
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.739644970414201e-05,
+      "loss": 1.8805,
+      "step": 639
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.73767258382643e-05,
+      "loss": 1.8905,
+      "step": 640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.735700197238659e-05,
+      "loss": 1.898,
+      "step": 641
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.733727810650889e-05,
+      "loss": 1.7914,
+      "step": 642
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.731755424063118e-05,
+      "loss": 1.8333,
+      "step": 643
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.729783037475346e-05,
+      "loss": 1.9515,
+      "step": 644
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.727810650887575e-05,
+      "loss": 1.9269,
+      "step": 645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.725838264299803e-05,
+      "loss": 1.8602,
+      "step": 646
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.723865877712032e-05,
+      "loss": 1.8957,
+      "step": 647
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.721893491124261e-05,
+      "loss": 1.7962,
+      "step": 648
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.719921104536489e-05,
+      "loss": 1.7774,
+      "step": 649
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 1.8723,
+      "step": 650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.715976331360947e-05,
+      "loss": 1.8621,
+      "step": 651
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.714003944773175e-05,
+      "loss": 1.7616,
+      "step": 652
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.712031558185404e-05,
+      "loss": 1.8763,
+      "step": 653
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.710059171597634e-05,
+      "loss": 1.9573,
+      "step": 654
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.708086785009863e-05,
+      "loss": 1.9369,
+      "step": 655
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.706114398422092e-05,
+      "loss": 1.9322,
+      "step": 656
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.70414201183432e-05,
+      "loss": 1.8727,
+      "step": 657
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.702169625246549e-05,
+      "loss": 1.8231,
+      "step": 658
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.700197238658777e-05,
+      "loss": 1.8983,
+      "step": 659
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.698224852071006e-05,
+      "loss": 1.897,
+      "step": 660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.696252465483235e-05,
+      "loss": 1.8838,
+      "step": 661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.694280078895463e-05,
+      "loss": 1.9052,
+      "step": 662
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.692307692307692e-05,
+      "loss": 1.7743,
+      "step": 663
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.690335305719922e-05,
+      "loss": 1.9483,
+      "step": 664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.688362919132151e-05,
+      "loss": 1.8029,
+      "step": 665
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.68639053254438e-05,
+      "loss": 1.8913,
+      "step": 666
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.684418145956608e-05,
+      "loss": 1.9851,
+      "step": 667
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.682445759368837e-05,
+      "loss": 1.8395,
+      "step": 668
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.680473372781065e-05,
+      "loss": 1.8264,
+      "step": 669
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.678500986193294e-05,
+      "loss": 1.9773,
+      "step": 670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.676528599605523e-05,
+      "loss": 1.8092,
+      "step": 671
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.674556213017751e-05,
+      "loss": 1.8509,
+      "step": 672
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.67258382642998e-05,
+      "loss": 1.8169,
+      "step": 673
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.67061143984221e-05,
+      "loss": 1.8342,
+      "step": 674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.668639053254438e-05,
+      "loss": 1.8355,
+      "step": 675
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 1.7397,
+      "step": 676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.664694280078896e-05,
+      "loss": 1.8435,
+      "step": 677
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.662721893491125e-05,
+      "loss": 1.8252,
+      "step": 678
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.660749506903354e-05,
+      "loss": 1.8396,
+      "step": 679
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.658777120315582e-05,
+      "loss": 1.8642,
+      "step": 680
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.656804733727812e-05,
+      "loss": 1.8033,
+      "step": 681
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.65483234714004e-05,
+      "loss": 1.7613,
+      "step": 682
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.652859960552269e-05,
+      "loss": 1.8704,
+      "step": 683
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.650887573964498e-05,
+      "loss": 1.8097,
+      "step": 684
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.648915187376726e-05,
+      "loss": 1.8729,
+      "step": 685
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.646942800788955e-05,
+      "loss": 1.8226,
+      "step": 686
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.644970414201183e-05,
+      "loss": 1.7739,
+      "step": 687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.642998027613412e-05,
+      "loss": 1.9138,
+      "step": 688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.641025641025642e-05,
+      "loss": 1.8903,
+      "step": 689
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.63905325443787e-05,
+      "loss": 1.785,
+      "step": 690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.6370808678501e-05,
+      "loss": 1.8039,
+      "step": 691
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.635108481262327e-05,
+      "loss": 1.9043,
+      "step": 692
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.633136094674557e-05,
+      "loss": 1.7887,
+      "step": 693
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.631163708086786e-05,
+      "loss": 1.8792,
+      "step": 694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.629191321499014e-05,
+      "loss": 1.8832,
+      "step": 695
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.627218934911243e-05,
+      "loss": 1.8354,
+      "step": 696
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.625246548323472e-05,
+      "loss": 1.9712,
+      "step": 697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.6232741617357e-05,
+      "loss": 1.8695,
+      "step": 698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.621301775147929e-05,
+      "loss": 1.7741,
+      "step": 699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.619329388560158e-05,
+      "loss": 1.8993,
+      "step": 700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.617357001972388e-05,
+      "loss": 1.8352,
+      "step": 701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.615384615384617e-05,
+      "loss": 1.942,
+      "step": 702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.613412228796845e-05,
+      "loss": 1.9322,
+      "step": 703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.611439842209074e-05,
+      "loss": 1.8666,
+      "step": 704
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.609467455621302e-05,
+      "loss": 1.8041,
+      "step": 705
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.607495069033531e-05,
+      "loss": 1.8838,
+      "step": 706
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.60552268244576e-05,
+      "loss": 1.8415,
+      "step": 707
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.603550295857988e-05,
+      "loss": 1.8423,
+      "step": 708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.601577909270217e-05,
+      "loss": 1.8116,
+      "step": 709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.599605522682445e-05,
+      "loss": 1.8176,
+      "step": 710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.597633136094674e-05,
+      "loss": 1.8452,
+      "step": 711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.595660749506904e-05,
+      "loss": 1.7979,
+      "step": 712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.593688362919133e-05,
+      "loss": 1.8683,
+      "step": 713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.591715976331362e-05,
+      "loss": 1.9563,
+      "step": 714
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 1.7894,
+      "step": 715
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.587771203155819e-05,
+      "loss": 1.8279,
+      "step": 716
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.585798816568048e-05,
+      "loss": 1.9507,
+      "step": 717
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.583826429980276e-05,
+      "loss": 1.8223,
+      "step": 718
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.581854043392505e-05,
+      "loss": 1.8711,
+      "step": 719
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.579881656804734e-05,
+      "loss": 1.8511,
+      "step": 720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.577909270216962e-05,
+      "loss": 1.8264,
+      "step": 721
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.575936883629192e-05,
+      "loss": 1.784,
+      "step": 722
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.573964497041421e-05,
+      "loss": 1.964,
+      "step": 723
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.57199211045365e-05,
+      "loss": 1.8444,
+      "step": 724
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.570019723865879e-05,
+      "loss": 1.8286,
+      "step": 725
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.568047337278107e-05,
+      "loss": 1.7989,
+      "step": 726
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.566074950690336e-05,
+      "loss": 2.0226,
+      "step": 727
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.564102564102564e-05,
+      "loss": 1.7123,
+      "step": 728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.562130177514793e-05,
+      "loss": 1.8194,
+      "step": 729
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.560157790927023e-05,
+      "loss": 1.8714,
+      "step": 730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.55818540433925e-05,
+      "loss": 1.7932,
+      "step": 731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.55621301775148e-05,
+      "loss": 1.7563,
+      "step": 732
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.554240631163707e-05,
+      "loss": 1.8045,
+      "step": 733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.552268244575937e-05,
+      "loss": 1.7721,
+      "step": 734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.550295857988166e-05,
+      "loss": 1.7691,
+      "step": 735
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.548323471400395e-05,
+      "loss": 1.8112,
+      "step": 736
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.546351084812624e-05,
+      "loss": 1.8721,
+      "step": 737
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.544378698224852e-05,
+      "loss": 1.8397,
+      "step": 738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.542406311637081e-05,
+      "loss": 1.7743,
+      "step": 739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.54043392504931e-05,
+      "loss": 1.8321,
+      "step": 740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.538461538461538e-05,
+      "loss": 1.7407,
+      "step": 741
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.536489151873768e-05,
+      "loss": 1.8339,
+      "step": 742
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.534516765285997e-05,
+      "loss": 1.9314,
+      "step": 743
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.532544378698225e-05,
+      "loss": 1.8765,
+      "step": 744
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.530571992110454e-05,
+      "loss": 1.8289,
+      "step": 745
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.528599605522682e-05,
+      "loss": 1.8368,
+      "step": 746
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.526627218934911e-05,
+      "loss": 1.8324,
+      "step": 747
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.524654832347142e-05,
+      "loss": 1.8121,
+      "step": 748
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.52268244575937e-05,
+      "loss": 1.8064,
+      "step": 749
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.520710059171599e-05,
+      "loss": 1.8275,
+      "step": 750
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 1.6339038610458374,
+      "eval_runtime": 69.4278,
+      "eval_samples_per_second": 320.765,
+      "eval_steps_per_second": 1.671,
+      "step": 750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.518737672583826e-05,
+      "loss": 1.8673,
+      "step": 751
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.516765285996056e-05,
+      "loss": 1.8409,
+      "step": 752
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.514792899408285e-05,
+      "loss": 1.7949,
+      "step": 753
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.512820512820513e-05,
+      "loss": 1.8676,
+      "step": 754
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.510848126232742e-05,
+      "loss": 1.7834,
+      "step": 755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.50887573964497e-05,
+      "loss": 1.768,
+      "step": 756
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.506903353057199e-05,
+      "loss": 1.8354,
+      "step": 757
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.504930966469428e-05,
+      "loss": 1.8592,
+      "step": 758
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.502958579881657e-05,
+      "loss": 1.8032,
+      "step": 759
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.500986193293887e-05,
+      "loss": 1.8346,
+      "step": 760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.499013806706116e-05,
+      "loss": 1.8764,
+      "step": 761
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.497041420118344e-05,
+      "loss": 1.8203,
+      "step": 762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.495069033530573e-05,
+      "loss": 1.8368,
+      "step": 763
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.493096646942801e-05,
+      "loss": 1.839,
+      "step": 764
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.49112426035503e-05,
+      "loss": 1.9035,
+      "step": 765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.489151873767259e-05,
+      "loss": 1.8462,
+      "step": 766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.487179487179487e-05,
+      "loss": 1.8428,
+      "step": 767
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.485207100591716e-05,
+      "loss": 1.8107,
+      "step": 768
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.483234714003944e-05,
+      "loss": 1.696,
+      "step": 769
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.481262327416173e-05,
+      "loss": 1.7684,
+      "step": 770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.479289940828403e-05,
+      "loss": 1.953,
+      "step": 771
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.477317554240632e-05,
+      "loss": 1.7695,
+      "step": 772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.475345167652861e-05,
+      "loss": 1.8572,
+      "step": 773
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.473372781065089e-05,
+      "loss": 1.8382,
+      "step": 774
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.471400394477318e-05,
+      "loss": 1.8085,
+      "step": 775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.469428007889547e-05,
+      "loss": 1.7531,
+      "step": 776
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.467455621301775e-05,
+      "loss": 1.7632,
+      "step": 777
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.465483234714004e-05,
+      "loss": 1.8419,
+      "step": 778
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.463510848126232e-05,
+      "loss": 1.814,
+      "step": 779
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 1.7156,
+      "step": 780
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.45956607495069e-05,
+      "loss": 1.7796,
+      "step": 781
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.45759368836292e-05,
+      "loss": 1.7029,
+      "step": 782
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.455621301775149e-05,
+      "loss": 1.8749,
+      "step": 783
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.453648915187378e-05,
+      "loss": 1.8533,
+      "step": 784
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.451676528599606e-05,
+      "loss": 1.811,
+      "step": 785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.449704142011835e-05,
+      "loss": 1.8801,
+      "step": 786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.447731755424063e-05,
+      "loss": 1.8771,
+      "step": 787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.445759368836292e-05,
+      "loss": 1.8036,
+      "step": 788
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.443786982248522e-05,
+      "loss": 1.7258,
+      "step": 789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.44181459566075e-05,
+      "loss": 1.869,
+      "step": 790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.439842209072979e-05,
+      "loss": 1.8855,
+      "step": 791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.437869822485207e-05,
+      "loss": 1.7894,
+      "step": 792
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.435897435897436e-05,
+      "loss": 1.7423,
+      "step": 793
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.433925049309665e-05,
+      "loss": 1.7918,
+      "step": 794
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.431952662721894e-05,
+      "loss": 1.8198,
+      "step": 795
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.429980276134123e-05,
+      "loss": 1.8674,
+      "step": 796
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.428007889546351e-05,
+      "loss": 1.847,
+      "step": 797
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.42603550295858e-05,
+      "loss": 1.8015,
+      "step": 798
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.42406311637081e-05,
+      "loss": 1.7316,
+      "step": 799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.422090729783038e-05,
+      "loss": 1.8473,
+      "step": 800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.420118343195267e-05,
+      "loss": 1.8447,
+      "step": 801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.418145956607495e-05,
+      "loss": 1.7289,
+      "step": 802
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.416173570019724e-05,
+      "loss": 1.8003,
+      "step": 803
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.414201183431953e-05,
+      "loss": 1.8253,
+      "step": 804
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.412228796844181e-05,
+      "loss": 1.7552,
+      "step": 805
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.410256410256411e-05,
+      "loss": 1.758,
+      "step": 806
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.40828402366864e-05,
+      "loss": 1.8004,
+      "step": 807
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.406311637080868e-05,
+      "loss": 1.7992,
+      "step": 808
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.404339250493098e-05,
+      "loss": 1.7539,
+      "step": 809
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.402366863905326e-05,
+      "loss": 1.8493,
+      "step": 810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.400394477317555e-05,
+      "loss": 1.7948,
+      "step": 811
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.398422090729784e-05,
+      "loss": 1.7966,
+      "step": 812
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.396449704142012e-05,
+      "loss": 1.9582,
+      "step": 813
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.394477317554241e-05,
+      "loss": 1.7824,
+      "step": 814
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.392504930966469e-05,
+      "loss": 1.8984,
+      "step": 815
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.390532544378698e-05,
+      "loss": 1.7662,
+      "step": 816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.388560157790927e-05,
+      "loss": 1.8165,
+      "step": 817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.386587771203157e-05,
+      "loss": 1.8998,
+      "step": 818
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.384615384615386e-05,
+      "loss": 1.8401,
+      "step": 819
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.382642998027614e-05,
+      "loss": 1.6952,
+      "step": 820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.380670611439843e-05,
+      "loss": 1.7531,
+      "step": 821
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.378698224852072e-05,
+      "loss": 1.7884,
+      "step": 822
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.3767258382643e-05,
+      "loss": 1.7808,
+      "step": 823
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.374753451676529e-05,
+      "loss": 1.7647,
+      "step": 824
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.372781065088757e-05,
+      "loss": 1.824,
+      "step": 825
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.370808678500986e-05,
+      "loss": 1.7022,
+      "step": 826
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.368836291913215e-05,
+      "loss": 1.8566,
+      "step": 827
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.366863905325443e-05,
+      "loss": 1.8913,
+      "step": 828
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.364891518737672e-05,
+      "loss": 1.8368,
+      "step": 829
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.362919132149902e-05,
+      "loss": 1.882,
+      "step": 830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.360946745562131e-05,
+      "loss": 1.7246,
+      "step": 831
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.35897435897436e-05,
+      "loss": 1.7837,
+      "step": 832
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.357001972386588e-05,
+      "loss": 1.8231,
+      "step": 833
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.355029585798817e-05,
+      "loss": 1.7347,
+      "step": 834
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.353057199211046e-05,
+      "loss": 1.8323,
+      "step": 835
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.351084812623274e-05,
+      "loss": 1.9104,
+      "step": 836
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.349112426035503e-05,
+      "loss": 1.7589,
+      "step": 837
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.347140039447731e-05,
+      "loss": 1.8092,
+      "step": 838
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.34516765285996e-05,
+      "loss": 1.8409,
+      "step": 839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.34319526627219e-05,
+      "loss": 1.7901,
+      "step": 840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.341222879684419e-05,
+      "loss": 1.8029,
+      "step": 841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.339250493096648e-05,
+      "loss": 1.808,
+      "step": 842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.337278106508876e-05,
+      "loss": 1.8475,
+      "step": 843
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.335305719921105e-05,
+      "loss": 1.8472,
+      "step": 844
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.8616,
+      "step": 845
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.331360946745562e-05,
+      "loss": 1.8659,
+      "step": 846
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.329388560157791e-05,
+      "loss": 1.8338,
+      "step": 847
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.32741617357002e-05,
+      "loss": 1.7658,
+      "step": 848
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.325443786982249e-05,
+      "loss": 1.813,
+      "step": 849
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.323471400394478e-05,
+      "loss": 1.9043,
+      "step": 850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.321499013806706e-05,
+      "loss": 1.8518,
+      "step": 851
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.319526627218935e-05,
+      "loss": 1.7663,
+      "step": 852
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.317554240631164e-05,
+      "loss": 1.864,
+      "step": 853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.315581854043393e-05,
+      "loss": 1.6727,
+      "step": 854
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.313609467455622e-05,
+      "loss": 1.8551,
+      "step": 855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.31163708086785e-05,
+      "loss": 1.8731,
+      "step": 856
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.30966469428008e-05,
+      "loss": 1.8687,
+      "step": 857
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.307692307692309e-05,
+      "loss": 1.926,
+      "step": 858
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.305719921104537e-05,
+      "loss": 1.8884,
+      "step": 859
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.303747534516766e-05,
+      "loss": 1.8735,
+      "step": 860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.301775147928994e-05,
+      "loss": 1.7344,
+      "step": 861
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.299802761341223e-05,
+      "loss": 1.7983,
+      "step": 862
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.297830374753452e-05,
+      "loss": 1.8672,
+      "step": 863
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.29585798816568e-05,
+      "loss": 1.8451,
+      "step": 864
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.29388560157791e-05,
+      "loss": 1.7965,
+      "step": 865
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.291913214990138e-05,
+      "loss": 1.8435,
+      "step": 866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.289940828402368e-05,
+      "loss": 1.8487,
+      "step": 867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.287968441814597e-05,
+      "loss": 1.7227,
+      "step": 868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.285996055226825e-05,
+      "loss": 1.8392,
+      "step": 869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.284023668639054e-05,
+      "loss": 1.8813,
+      "step": 870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.282051282051283e-05,
+      "loss": 1.7726,
+      "step": 871
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.280078895463511e-05,
+      "loss": 1.7962,
+      "step": 872
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.27810650887574e-05,
+      "loss": 1.8174,
+      "step": 873
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.276134122287968e-05,
+      "loss": 1.8915,
+      "step": 874
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.274161735700197e-05,
+      "loss": 1.8268,
+      "step": 875
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.272189349112426e-05,
+      "loss": 1.7125,
+      "step": 876
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.270216962524656e-05,
+      "loss": 1.8697,
+      "step": 877
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.268244575936885e-05,
+      "loss": 1.8499,
+      "step": 878
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.266272189349113e-05,
+      "loss": 1.8127,
+      "step": 879
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.264299802761342e-05,
+      "loss": 1.7703,
+      "step": 880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.262327416173571e-05,
+      "loss": 1.844,
+      "step": 881
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.260355029585799e-05,
+      "loss": 1.8973,
+      "step": 882
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.258382642998028e-05,
+      "loss": 1.8305,
+      "step": 883
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.256410256410256e-05,
+      "loss": 1.7808,
+      "step": 884
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.254437869822485e-05,
+      "loss": 1.7664,
+      "step": 885
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.252465483234714e-05,
+      "loss": 1.7969,
+      "step": 886
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250493096646942e-05,
+      "loss": 1.8044,
+      "step": 887
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.248520710059172e-05,
+      "loss": 1.8633,
+      "step": 888
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.246548323471401e-05,
+      "loss": 1.855,
+      "step": 889
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.24457593688363e-05,
+      "loss": 1.7586,
+      "step": 890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.242603550295859e-05,
+      "loss": 1.8667,
+      "step": 891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.240631163708087e-05,
+      "loss": 1.8961,
+      "step": 892
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.238658777120316e-05,
+      "loss": 1.8361,
+      "step": 893
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.236686390532545e-05,
+      "loss": 1.7352,
+      "step": 894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.234714003944773e-05,
+      "loss": 1.83,
+      "step": 895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.232741617357003e-05,
+      "loss": 1.7314,
+      "step": 896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.23076923076923e-05,
+      "loss": 1.8091,
+      "step": 897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.22879684418146e-05,
+      "loss": 1.7807,
+      "step": 898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.226824457593689e-05,
+      "loss": 1.8446,
+      "step": 899
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.224852071005918e-05,
+      "loss": 1.7649,
+      "step": 900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.222879684418147e-05,
+      "loss": 1.7751,
+      "step": 901
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.220907297830375e-05,
+      "loss": 1.7602,
+      "step": 902
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.218934911242604e-05,
+      "loss": 1.8633,
+      "step": 903
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.216962524654833e-05,
+      "loss": 1.8336,
+      "step": 904
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.214990138067061e-05,
+      "loss": 1.8159,
+      "step": 905
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.21301775147929e-05,
+      "loss": 1.7459,
+      "step": 906
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.211045364891518e-05,
+      "loss": 1.8947,
+      "step": 907
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.209072978303748e-05,
+      "loss": 1.789,
+      "step": 908
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.207100591715977e-05,
+      "loss": 1.7894,
+      "step": 909
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 1.783,
+      "step": 910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.203155818540434e-05,
+      "loss": 1.7893,
+      "step": 911
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.201183431952663e-05,
+      "loss": 1.7181,
+      "step": 912
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.199211045364892e-05,
+      "loss": 1.7898,
+      "step": 913
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.197238658777122e-05,
+      "loss": 1.7524,
+      "step": 914
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.19526627218935e-05,
+      "loss": 1.8185,
+      "step": 915
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.193293885601579e-05,
+      "loss": 1.846,
+      "step": 916
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.191321499013808e-05,
+      "loss": 1.8301,
+      "step": 917
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.189349112426036e-05,
+      "loss": 1.8143,
+      "step": 918
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.187376725838265e-05,
+      "loss": 1.7469,
+      "step": 919
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.185404339250493e-05,
+      "loss": 1.7839,
+      "step": 920
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.183431952662722e-05,
+      "loss": 1.7911,
+      "step": 921
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.181459566074951e-05,
+      "loss": 1.8097,
+      "step": 922
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.179487179487179e-05,
+      "loss": 1.8084,
+      "step": 923
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.17751479289941e-05,
+      "loss": 1.7076,
+      "step": 924
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.175542406311637e-05,
+      "loss": 1.8234,
+      "step": 925
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.173570019723867e-05,
+      "loss": 1.8137,
+      "step": 926
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.171597633136096e-05,
+      "loss": 1.7534,
+      "step": 927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.169625246548324e-05,
+      "loss": 1.6798,
+      "step": 928
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.167652859960553e-05,
+      "loss": 1.8602,
+      "step": 929
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.165680473372781e-05,
+      "loss": 1.8014,
+      "step": 930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.16370808678501e-05,
+      "loss": 1.75,
+      "step": 931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.161735700197239e-05,
+      "loss": 1.8285,
+      "step": 932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.159763313609467e-05,
+      "loss": 1.8733,
+      "step": 933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.157790927021696e-05,
+      "loss": 1.8037,
+      "step": 934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.155818540433925e-05,
+      "loss": 1.8907,
+      "step": 935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.153846153846155e-05,
+      "loss": 1.7746,
+      "step": 936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.151873767258384e-05,
+      "loss": 1.8329,
+      "step": 937
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.149901380670612e-05,
+      "loss": 1.904,
+      "step": 938
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.147928994082841e-05,
+      "loss": 1.8483,
+      "step": 939
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.14595660749507e-05,
+      "loss": 1.7707,
+      "step": 940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.143984220907298e-05,
+      "loss": 1.7699,
+      "step": 941
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.142011834319527e-05,
+      "loss": 1.8091,
+      "step": 942
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.140039447731755e-05,
+      "loss": 1.8439,
+      "step": 943
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.138067061143984e-05,
+      "loss": 1.9225,
+      "step": 944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.136094674556214e-05,
+      "loss": 1.74,
+      "step": 945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.134122287968441e-05,
+      "loss": 1.7734,
+      "step": 946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.13214990138067e-05,
+      "loss": 1.6983,
+      "step": 947
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.1301775147929e-05,
+      "loss": 1.8118,
+      "step": 948
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.128205128205129e-05,
+      "loss": 1.755,
+      "step": 949
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.126232741617358e-05,
+      "loss": 1.7313,
+      "step": 950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.124260355029586e-05,
+      "loss": 1.6845,
+      "step": 951
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.122287968441815e-05,
+      "loss": 1.7866,
+      "step": 952
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.120315581854043e-05,
+      "loss": 1.8977,
+      "step": 953
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.118343195266272e-05,
+      "loss": 1.7944,
+      "step": 954
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.116370808678502e-05,
+      "loss": 1.7997,
+      "step": 955
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.11439842209073e-05,
+      "loss": 1.7754,
+      "step": 956
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.112426035502959e-05,
+      "loss": 1.7726,
+      "step": 957
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.110453648915188e-05,
+      "loss": 1.806,
+      "step": 958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.108481262327417e-05,
+      "loss": 1.8488,
+      "step": 959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.106508875739646e-05,
+      "loss": 1.7413,
+      "step": 960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.104536489151874e-05,
+      "loss": 1.7158,
+      "step": 961
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.102564102564103e-05,
+      "loss": 1.7229,
+      "step": 962
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.100591715976333e-05,
+      "loss": 1.8413,
+      "step": 963
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.09861932938856e-05,
+      "loss": 1.7297,
+      "step": 964
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.09664694280079e-05,
+      "loss": 1.772,
+      "step": 965
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.094674556213017e-05,
+      "loss": 1.8933,
+      "step": 966
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.092702169625247e-05,
+      "loss": 1.8436,
+      "step": 967
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.090729783037476e-05,
+      "loss": 1.6602,
+      "step": 968
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.088757396449704e-05,
+      "loss": 1.8474,
+      "step": 969
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.086785009861933e-05,
+      "loss": 1.6688,
+      "step": 970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.084812623274162e-05,
+      "loss": 1.8355,
+      "step": 971
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.082840236686391e-05,
+      "loss": 1.7951,
+      "step": 972
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.08086785009862e-05,
+      "loss": 1.7564,
+      "step": 973
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.078895463510848e-05,
+      "loss": 1.7483,
+      "step": 974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 1.8262,
+      "step": 975
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.074950690335306e-05,
+      "loss": 1.7467,
+      "step": 976
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.072978303747535e-05,
+      "loss": 1.7623,
+      "step": 977
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.071005917159764e-05,
+      "loss": 1.7806,
+      "step": 978
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.069033530571992e-05,
+      "loss": 1.7233,
+      "step": 979
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.067061143984221e-05,
+      "loss": 1.7297,
+      "step": 980
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.06508875739645e-05,
+      "loss": 1.9552,
+      "step": 981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.063116370808678e-05,
+      "loss": 1.7473,
+      "step": 982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.061143984220909e-05,
+      "loss": 1.7356,
+      "step": 983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.059171597633137e-05,
+      "loss": 1.7811,
+      "step": 984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.057199211045366e-05,
+      "loss": 1.6885,
+      "step": 985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.055226824457595e-05,
+      "loss": 1.8778,
+      "step": 986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.053254437869823e-05,
+      "loss": 1.6362,
+      "step": 987
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.051282051282052e-05,
+      "loss": 1.7879,
+      "step": 988
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.04930966469428e-05,
+      "loss": 1.7555,
+      "step": 989
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.047337278106509e-05,
+      "loss": 1.8987,
+      "step": 990
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.045364891518738e-05,
+      "loss": 1.852,
+      "step": 991
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.043392504930966e-05,
+      "loss": 1.8562,
+      "step": 992
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.041420118343195e-05,
+      "loss": 1.901,
+      "step": 993
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.039447731755425e-05,
+      "loss": 1.7828,
+      "step": 994
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.037475345167654e-05,
+      "loss": 1.7779,
+      "step": 995
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.035502958579883e-05,
+      "loss": 1.7152,
+      "step": 996
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.033530571992111e-05,
+      "loss": 1.6811,
+      "step": 997
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.03155818540434e-05,
+      "loss": 1.6979,
+      "step": 998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.029585798816568e-05,
+      "loss": 1.7603,
+      "step": 999
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.027613412228797e-05,
+      "loss": 1.693,
+      "step": 1000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 1.5821160078048706,
+      "eval_runtime": 68.9388,
+      "eval_samples_per_second": 323.04,
+      "eval_steps_per_second": 1.683,
+      "step": 1000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.025641025641026e-05,
+      "loss": 1.875,
+      "step": 1001
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.023668639053254e-05,
+      "loss": 1.8114,
+      "step": 1002
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.021696252465483e-05,
+      "loss": 1.7824,
+      "step": 1003
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.019723865877713e-05,
+      "loss": 1.7263,
+      "step": 1004
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.01775147928994e-05,
+      "loss": 1.7795,
+      "step": 1005
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.01577909270217e-05,
+      "loss": 1.8166,
+      "step": 1006
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.013806706114399e-05,
+      "loss": 1.7061,
+      "step": 1007
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.011834319526628e-05,
+      "loss": 1.9,
+      "step": 1008
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.009861932938857e-05,
+      "loss": 1.9002,
+      "step": 1009
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.007889546351085e-05,
+      "loss": 1.7215,
+      "step": 1010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.005917159763314e-05,
+      "loss": 1.8256,
+      "step": 1011
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.003944773175542e-05,
+      "loss": 1.784,
+      "step": 1012
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.001972386587771e-05,
+      "loss": 1.7676,
+      "step": 1013
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8e-05,
+      "loss": 1.6152,
+      "step": 1014
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.998027613412229e-05,
+      "loss": 1.6694,
+      "step": 1015
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.996055226824458e-05,
+      "loss": 1.6094,
+      "step": 1016
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.994082840236687e-05,
+      "loss": 1.7174,
+      "step": 1017
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.992110453648916e-05,
+      "loss": 1.7866,
+      "step": 1018
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.990138067061145e-05,
+      "loss": 1.7978,
+      "step": 1019
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.988165680473373e-05,
+      "loss": 1.7342,
+      "step": 1020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.986193293885602e-05,
+      "loss": 1.7476,
+      "step": 1021
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.98422090729783e-05,
+      "loss": 1.6855,
+      "step": 1022
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.98224852071006e-05,
+      "loss": 1.7545,
+      "step": 1023
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.980276134122289e-05,
+      "loss": 1.7573,
+      "step": 1024
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.978303747534517e-05,
+      "loss": 1.7819,
+      "step": 1025
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.976331360946746e-05,
+      "loss": 1.756,
+      "step": 1026
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.974358974358975e-05,
+      "loss": 1.8341,
+      "step": 1027
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.972386587771203e-05,
+      "loss": 1.8413,
+      "step": 1028
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.970414201183432e-05,
+      "loss": 1.8856,
+      "step": 1029
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.968441814595661e-05,
+      "loss": 1.7549,
+      "step": 1030
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.96646942800789e-05,
+      "loss": 1.7956,
+      "step": 1031
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.96449704142012e-05,
+      "loss": 1.7822,
+      "step": 1032
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.962524654832348e-05,
+      "loss": 1.8146,
+      "step": 1033
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.960552268244577e-05,
+      "loss": 1.7863,
+      "step": 1034
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.958579881656805e-05,
+      "loss": 1.7761,
+      "step": 1035
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.956607495069034e-05,
+      "loss": 1.7215,
+      "step": 1036
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.954635108481263e-05,
+      "loss": 1.7974,
+      "step": 1037
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.952662721893491e-05,
+      "loss": 1.7175,
+      "step": 1038
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.95069033530572e-05,
+      "loss": 1.7995,
+      "step": 1039
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 1.8055,
+      "step": 1040
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.946745562130179e-05,
+      "loss": 1.77,
+      "step": 1041
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.944773175542408e-05,
+      "loss": 1.7737,
+      "step": 1042
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.942800788954636e-05,
+      "loss": 1.7763,
+      "step": 1043
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.940828402366865e-05,
+      "loss": 1.7532,
+      "step": 1044
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.938856015779093e-05,
+      "loss": 1.7509,
+      "step": 1045
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.936883629191322e-05,
+      "loss": 1.791,
+      "step": 1046
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.934911242603551e-05,
+      "loss": 1.7791,
+      "step": 1047
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.932938856015779e-05,
+      "loss": 1.7826,
+      "step": 1048
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.930966469428008e-05,
+      "loss": 1.7109,
+      "step": 1049
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.928994082840237e-05,
+      "loss": 1.7137,
+      "step": 1050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.927021696252465e-05,
+      "loss": 1.7577,
+      "step": 1051
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.925049309664694e-05,
+      "loss": 1.659,
+      "step": 1052
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.923076923076924e-05,
+      "loss": 1.8969,
+      "step": 1053
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.921104536489153e-05,
+      "loss": 1.7322,
+      "step": 1054
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.919132149901382e-05,
+      "loss": 1.832,
+      "step": 1055
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.91715976331361e-05,
+      "loss": 1.8411,
+      "step": 1056
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.915187376725839e-05,
+      "loss": 1.7345,
+      "step": 1057
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.913214990138067e-05,
+      "loss": 1.7086,
+      "step": 1058
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.911242603550296e-05,
+      "loss": 1.6368,
+      "step": 1059
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.909270216962525e-05,
+      "loss": 1.7429,
+      "step": 1060
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.907297830374753e-05,
+      "loss": 1.6527,
+      "step": 1061
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.905325443786982e-05,
+      "loss": 1.6841,
+      "step": 1062
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.90335305719921e-05,
+      "loss": 1.751,
+      "step": 1063
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.90138067061144e-05,
+      "loss": 1.798,
+      "step": 1064
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.899408284023669e-05,
+      "loss": 1.7646,
+      "step": 1065
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.897435897435898e-05,
+      "loss": 1.8508,
+      "step": 1066
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.895463510848127e-05,
+      "loss": 1.8614,
+      "step": 1067
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.893491124260355e-05,
+      "loss": 1.7231,
+      "step": 1068
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.891518737672584e-05,
+      "loss": 1.7853,
+      "step": 1069
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.889546351084813e-05,
+      "loss": 1.6603,
+      "step": 1070
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.887573964497041e-05,
+      "loss": 1.7255,
+      "step": 1071
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.88560157790927e-05,
+      "loss": 1.6954,
+      "step": 1072
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.8836291913215e-05,
+      "loss": 1.7138,
+      "step": 1073
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.881656804733728e-05,
+      "loss": 1.8525,
+      "step": 1074
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.879684418145957e-05,
+      "loss": 1.657,
+      "step": 1075
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.877712031558186e-05,
+      "loss": 1.8444,
+      "step": 1076
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.875739644970415e-05,
+      "loss": 1.7405,
+      "step": 1077
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.873767258382644e-05,
+      "loss": 1.6666,
+      "step": 1078
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.871794871794872e-05,
+      "loss": 1.654,
+      "step": 1079
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.869822485207101e-05,
+      "loss": 1.7601,
+      "step": 1080
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.86785009861933e-05,
+      "loss": 1.8334,
+      "step": 1081
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.865877712031559e-05,
+      "loss": 1.8329,
+      "step": 1082
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.863905325443788e-05,
+      "loss": 1.6851,
+      "step": 1083
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.861932938856016e-05,
+      "loss": 1.7404,
+      "step": 1084
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.859960552268245e-05,
+      "loss": 1.7274,
+      "step": 1085
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.857988165680473e-05,
+      "loss": 1.7554,
+      "step": 1086
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.856015779092702e-05,
+      "loss": 1.7663,
+      "step": 1087
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.854043392504931e-05,
+      "loss": 1.7319,
+      "step": 1088
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.85207100591716e-05,
+      "loss": 1.6156,
+      "step": 1089
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.85009861932939e-05,
+      "loss": 1.7827,
+      "step": 1090
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.848126232741617e-05,
+      "loss": 1.7047,
+      "step": 1091
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.846153846153847e-05,
+      "loss": 1.8116,
+      "step": 1092
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.844181459566076e-05,
+      "loss": 1.6917,
+      "step": 1093
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.842209072978304e-05,
+      "loss": 1.6991,
+      "step": 1094
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.840236686390533e-05,
+      "loss": 1.7751,
+      "step": 1095
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.838264299802762e-05,
+      "loss": 1.9192,
+      "step": 1096
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.83629191321499e-05,
+      "loss": 1.7745,
+      "step": 1097
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.834319526627219e-05,
+      "loss": 1.7642,
+      "step": 1098
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.832347140039447e-05,
+      "loss": 1.7567,
+      "step": 1099
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.830374753451678e-05,
+      "loss": 1.7559,
+      "step": 1100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.828402366863907e-05,
+      "loss": 1.7699,
+      "step": 1101
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.826429980276135e-05,
+      "loss": 1.6309,
+      "step": 1102
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.824457593688364e-05,
+      "loss": 1.8006,
+      "step": 1103
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.822485207100592e-05,
+      "loss": 1.7759,
+      "step": 1104
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 1.7288,
+      "step": 1105
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.81854043392505e-05,
+      "loss": 1.7954,
+      "step": 1106
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.816568047337278e-05,
+      "loss": 1.7897,
+      "step": 1107
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.814595660749507e-05,
+      "loss": 1.7113,
+      "step": 1108
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.812623274161735e-05,
+      "loss": 1.7672,
+      "step": 1109
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.810650887573964e-05,
+      "loss": 1.8214,
+      "step": 1110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.808678500986193e-05,
+      "loss": 1.7691,
+      "step": 1111
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.806706114398423e-05,
+      "loss": 1.6605,
+      "step": 1112
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.804733727810652e-05,
+      "loss": 1.7907,
+      "step": 1113
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.802761341222881e-05,
+      "loss": 1.7381,
+      "step": 1114
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.800788954635109e-05,
+      "loss": 1.6696,
+      "step": 1115
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.798816568047338e-05,
+      "loss": 1.8012,
+      "step": 1116
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.796844181459566e-05,
+      "loss": 1.8031,
+      "step": 1117
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.794871794871795e-05,
+      "loss": 1.8076,
+      "step": 1118
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.792899408284024e-05,
+      "loss": 1.7319,
+      "step": 1119
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.790927021696252e-05,
+      "loss": 1.7055,
+      "step": 1120
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.788954635108482e-05,
+      "loss": 1.7679,
+      "step": 1121
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.78698224852071e-05,
+      "loss": 1.7192,
+      "step": 1122
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.785009861932939e-05,
+      "loss": 1.7648,
+      "step": 1123
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.783037475345168e-05,
+      "loss": 1.7176,
+      "step": 1124
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.781065088757397e-05,
+      "loss": 1.7045,
+      "step": 1125
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.779092702169626e-05,
+      "loss": 1.751,
+      "step": 1126
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.777120315581854e-05,
+      "loss": 1.7799,
+      "step": 1127
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.775147928994083e-05,
+      "loss": 1.8094,
+      "step": 1128
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.773175542406313e-05,
+      "loss": 1.8088,
+      "step": 1129
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.77120315581854e-05,
+      "loss": 1.7716,
+      "step": 1130
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.76923076923077e-05,
+      "loss": 1.7479,
+      "step": 1131
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.767258382642997e-05,
+      "loss": 1.8139,
+      "step": 1132
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.765285996055227e-05,
+      "loss": 1.7373,
+      "step": 1133
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.763313609467456e-05,
+      "loss": 1.8012,
+      "step": 1134
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.761341222879685e-05,
+      "loss": 1.7971,
+      "step": 1135
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.759368836291914e-05,
+      "loss": 1.7722,
+      "step": 1136
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.757396449704144e-05,
+      "loss": 1.7686,
+      "step": 1137
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.755424063116371e-05,
+      "loss": 1.7738,
+      "step": 1138
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.7534516765286e-05,
+      "loss": 1.6721,
+      "step": 1139
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.751479289940828e-05,
+      "loss": 1.696,
+      "step": 1140
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.749506903353058e-05,
+      "loss": 1.6853,
+      "step": 1141
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.747534516765287e-05,
+      "loss": 1.6562,
+      "step": 1142
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.745562130177515e-05,
+      "loss": 1.782,
+      "step": 1143
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.743589743589744e-05,
+      "loss": 1.8166,
+      "step": 1144
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.741617357001972e-05,
+      "loss": 1.7867,
+      "step": 1145
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.739644970414201e-05,
+      "loss": 1.7471,
+      "step": 1146
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.73767258382643e-05,
+      "loss": 1.6674,
+      "step": 1147
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.73570019723866e-05,
+      "loss": 1.7256,
+      "step": 1148
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.733727810650889e-05,
+      "loss": 1.7153,
+      "step": 1149
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.731755424063116e-05,
+      "loss": 1.7559,
+      "step": 1150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.729783037475346e-05,
+      "loss": 1.7951,
+      "step": 1151
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.727810650887575e-05,
+      "loss": 1.743,
+      "step": 1152
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.725838264299803e-05,
+      "loss": 1.6439,
+      "step": 1153
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.723865877712032e-05,
+      "loss": 1.7145,
+      "step": 1154
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.72189349112426e-05,
+      "loss": 1.7168,
+      "step": 1155
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.719921104536489e-05,
+      "loss": 1.6069,
+      "step": 1156
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.717948717948718e-05,
+      "loss": 1.7337,
+      "step": 1157
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.715976331360946e-05,
+      "loss": 1.6981,
+      "step": 1158
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.714003944773177e-05,
+      "loss": 1.777,
+      "step": 1159
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.712031558185406e-05,
+      "loss": 1.8702,
+      "step": 1160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.710059171597634e-05,
+      "loss": 1.8264,
+      "step": 1161
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.708086785009863e-05,
+      "loss": 1.7243,
+      "step": 1162
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.706114398422091e-05,
+      "loss": 1.7814,
+      "step": 1163
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.70414201183432e-05,
+      "loss": 1.7619,
+      "step": 1164
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.702169625246549e-05,
+      "loss": 1.7174,
+      "step": 1165
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.700197238658777e-05,
+      "loss": 1.7622,
+      "step": 1166
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.698224852071006e-05,
+      "loss": 1.8018,
+      "step": 1167
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.696252465483234e-05,
+      "loss": 1.7336,
+      "step": 1168
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.694280078895463e-05,
+      "loss": 1.7565,
+      "step": 1169
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 1.8548,
+      "step": 1170
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.690335305719922e-05,
+      "loss": 1.6861,
+      "step": 1171
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.688362919132151e-05,
+      "loss": 1.6818,
+      "step": 1172
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.686390532544379e-05,
+      "loss": 1.6572,
+      "step": 1173
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.684418145956608e-05,
+      "loss": 1.8161,
+      "step": 1174
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.682445759368837e-05,
+      "loss": 1.6673,
+      "step": 1175
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.680473372781065e-05,
+      "loss": 1.7914,
+      "step": 1176
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.678500986193294e-05,
+      "loss": 1.7063,
+      "step": 1177
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.676528599605522e-05,
+      "loss": 1.7158,
+      "step": 1178
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.674556213017751e-05,
+      "loss": 1.743,
+      "step": 1179
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.67258382642998e-05,
+      "loss": 1.8005,
+      "step": 1180
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.670611439842208e-05,
+      "loss": 1.7206,
+      "step": 1181
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.668639053254438e-05,
+      "loss": 1.7477,
+      "step": 1182
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.666666666666667e-05,
+      "loss": 1.8617,
+      "step": 1183
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.664694280078896e-05,
+      "loss": 1.7711,
+      "step": 1184
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.662721893491125e-05,
+      "loss": 1.7947,
+      "step": 1185
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.660749506903353e-05,
+      "loss": 1.6845,
+      "step": 1186
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.658777120315582e-05,
+      "loss": 1.7788,
+      "step": 1187
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.656804733727812e-05,
+      "loss": 1.7097,
+      "step": 1188
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.65483234714004e-05,
+      "loss": 1.6946,
+      "step": 1189
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.652859960552269e-05,
+      "loss": 1.7359,
+      "step": 1190
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.650887573964497e-05,
+      "loss": 1.841,
+      "step": 1191
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.648915187376726e-05,
+      "loss": 1.6834,
+      "step": 1192
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.646942800788955e-05,
+      "loss": 1.8451,
+      "step": 1193
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.644970414201184e-05,
+      "loss": 1.6765,
+      "step": 1194
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.642998027613413e-05,
+      "loss": 1.6993,
+      "step": 1195
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.641025641025641e-05,
+      "loss": 1.7287,
+      "step": 1196
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.63905325443787e-05,
+      "loss": 1.6839,
+      "step": 1197
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.6370808678501e-05,
+      "loss": 1.6964,
+      "step": 1198
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.635108481262328e-05,
+      "loss": 1.6909,
+      "step": 1199
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.633136094674557e-05,
+      "loss": 1.7666,
+      "step": 1200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.631163708086785e-05,
+      "loss": 1.807,
+      "step": 1201
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.629191321499014e-05,
+      "loss": 1.6543,
+      "step": 1202
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.627218934911243e-05,
+      "loss": 1.7786,
+      "step": 1203
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.625246548323471e-05,
+      "loss": 1.6372,
+      "step": 1204
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.6232741617357e-05,
+      "loss": 1.733,
+      "step": 1205
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.621301775147929e-05,
+      "loss": 1.7511,
+      "step": 1206
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.619329388560158e-05,
+      "loss": 1.7956,
+      "step": 1207
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.617357001972388e-05,
+      "loss": 1.7664,
+      "step": 1208
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.615384615384616e-05,
+      "loss": 1.7942,
+      "step": 1209
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.613412228796845e-05,
+      "loss": 1.6379,
+      "step": 1210
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.611439842209074e-05,
+      "loss": 1.7422,
+      "step": 1211
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.609467455621302e-05,
+      "loss": 1.6734,
+      "step": 1212
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.607495069033531e-05,
+      "loss": 1.762,
+      "step": 1213
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.605522682445759e-05,
+      "loss": 1.7996,
+      "step": 1214
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.603550295857988e-05,
+      "loss": 1.6864,
+      "step": 1215
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.601577909270217e-05,
+      "loss": 1.8164,
+      "step": 1216
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.599605522682447e-05,
+      "loss": 1.7541,
+      "step": 1217
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.597633136094676e-05,
+      "loss": 1.7381,
+      "step": 1218
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.595660749506904e-05,
+      "loss": 1.6901,
+      "step": 1219
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.593688362919133e-05,
+      "loss": 1.7559,
+      "step": 1220
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.591715976331362e-05,
+      "loss": 1.7055,
+      "step": 1221
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.58974358974359e-05,
+      "loss": 1.7109,
+      "step": 1222
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.587771203155819e-05,
+      "loss": 1.7121,
+      "step": 1223
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.585798816568048e-05,
+      "loss": 1.6606,
+      "step": 1224
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.583826429980276e-05,
+      "loss": 1.6536,
+      "step": 1225
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.581854043392505e-05,
+      "loss": 1.7262,
+      "step": 1226
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.579881656804733e-05,
+      "loss": 1.7705,
+      "step": 1227
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.577909270216962e-05,
+      "loss": 1.7405,
+      "step": 1228
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.575936883629192e-05,
+      "loss": 1.6101,
+      "step": 1229
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.573964497041421e-05,
+      "loss": 1.6613,
+      "step": 1230
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.57199211045365e-05,
+      "loss": 1.8146,
+      "step": 1231
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.570019723865878e-05,
+      "loss": 1.8195,
+      "step": 1232
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.568047337278107e-05,
+      "loss": 1.6891,
+      "step": 1233
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.566074950690336e-05,
+      "loss": 1.7344,
+      "step": 1234
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 1.768,
+      "step": 1235
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.562130177514793e-05,
+      "loss": 1.7537,
+      "step": 1236
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.560157790927021e-05,
+      "loss": 1.7904,
+      "step": 1237
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.55818540433925e-05,
+      "loss": 1.6626,
+      "step": 1238
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 7.55621301775148e-05,
+      "loss": 1.6876,
+      "step": 1239
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 7.554240631163708e-05,
+      "loss": 1.7696,
+      "step": 1240
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 7.552268244575937e-05,
+      "loss": 1.7005,
+      "step": 1241
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 7.550295857988166e-05,
+      "loss": 1.6214,
+      "step": 1242
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 7.548323471400395e-05,
+      "loss": 1.8126,
+      "step": 1243
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 7.546351084812624e-05,
+      "loss": 1.7336,
+      "step": 1244
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 7.544378698224852e-05,
+      "loss": 1.7196,
+      "step": 1245
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.542406311637081e-05,
+      "loss": 1.6762,
+      "step": 1246
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.54043392504931e-05,
+      "loss": 1.7382,
+      "step": 1247
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.538461538461539e-05,
+      "loss": 1.8384,
+      "step": 1248
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.536489151873768e-05,
+      "loss": 1.6117,
+      "step": 1249
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.534516765285996e-05,
+      "loss": 1.5995,
+      "step": 1250
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 1.555130958557129,
+      "eval_runtime": 68.9379,
+      "eval_samples_per_second": 323.044,
+      "eval_steps_per_second": 1.683,
+      "step": 1250
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.532544378698225e-05,
+      "loss": 1.7283,
+      "step": 1251
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.530571992110454e-05,
+      "loss": 1.7611,
+      "step": 1252
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.528599605522683e-05,
+      "loss": 1.7252,
+      "step": 1253
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.526627218934912e-05,
+      "loss": 1.7686,
+      "step": 1254
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.52465483234714e-05,
+      "loss": 1.8094,
+      "step": 1255
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.52268244575937e-05,
+      "loss": 1.7976,
+      "step": 1256
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.520710059171599e-05,
+      "loss": 1.7618,
+      "step": 1257
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.518737672583827e-05,
+      "loss": 1.8102,
+      "step": 1258
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.516765285996056e-05,
+      "loss": 1.7747,
+      "step": 1259
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.514792899408284e-05,
+      "loss": 1.6812,
+      "step": 1260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.512820512820513e-05,
+      "loss": 1.712,
+      "step": 1261
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.510848126232742e-05,
+      "loss": 1.7454,
+      "step": 1262
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.50887573964497e-05,
+      "loss": 1.6957,
+      "step": 1263
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.506903353057199e-05,
+      "loss": 1.6991,
+      "step": 1264
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.504930966469428e-05,
+      "loss": 1.8196,
+      "step": 1265
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.502958579881658e-05,
+      "loss": 1.6657,
+      "step": 1266
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.500986193293887e-05,
+      "loss": 1.7202,
+      "step": 1267
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.499013806706115e-05,
+      "loss": 1.7254,
+      "step": 1268
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.497041420118344e-05,
+      "loss": 1.6749,
+      "step": 1269
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.495069033530573e-05,
+      "loss": 1.7471,
+      "step": 1270
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.493096646942801e-05,
+      "loss": 1.7241,
+      "step": 1271
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.49112426035503e-05,
+      "loss": 1.6611,
+      "step": 1272
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.489151873767258e-05,
+      "loss": 1.7788,
+      "step": 1273
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.487179487179487e-05,
+      "loss": 1.8294,
+      "step": 1274
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.485207100591716e-05,
+      "loss": 1.6911,
+      "step": 1275
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.483234714003946e-05,
+      "loss": 1.8059,
+      "step": 1276
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.481262327416175e-05,
+      "loss": 1.739,
+      "step": 1277
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.479289940828403e-05,
+      "loss": 1.6999,
+      "step": 1278
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.477317554240632e-05,
+      "loss": 1.7439,
+      "step": 1279
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.475345167652861e-05,
+      "loss": 1.7018,
+      "step": 1280
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.473372781065089e-05,
+      "loss": 1.7103,
+      "step": 1281
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.471400394477318e-05,
+      "loss": 1.7542,
+      "step": 1282
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.469428007889546e-05,
+      "loss": 1.7068,
+      "step": 1283
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.467455621301775e-05,
+      "loss": 1.7451,
+      "step": 1284
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.465483234714004e-05,
+      "loss": 1.7675,
+      "step": 1285
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.463510848126232e-05,
+      "loss": 1.597,
+      "step": 1286
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.461538461538462e-05,
+      "loss": 1.7752,
+      "step": 1287
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.459566074950691e-05,
+      "loss": 1.8056,
+      "step": 1288
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.45759368836292e-05,
+      "loss": 1.7452,
+      "step": 1289
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.455621301775149e-05,
+      "loss": 1.7858,
+      "step": 1290
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.453648915187377e-05,
+      "loss": 1.772,
+      "step": 1291
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.451676528599606e-05,
+      "loss": 1.7693,
+      "step": 1292
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.449704142011835e-05,
+      "loss": 1.7307,
+      "step": 1293
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.447731755424063e-05,
+      "loss": 1.6464,
+      "step": 1294
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.445759368836292e-05,
+      "loss": 1.6068,
+      "step": 1295
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.44378698224852e-05,
+      "loss": 1.7304,
+      "step": 1296
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.44181459566075e-05,
+      "loss": 1.6769,
+      "step": 1297
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.439842209072979e-05,
+      "loss": 1.675,
+      "step": 1298
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.437869822485207e-05,
+      "loss": 1.6344,
+      "step": 1299
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 1.6211,
+      "step": 1300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.433925049309665e-05,
+      "loss": 1.6819,
+      "step": 1301
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.431952662721894e-05,
+      "loss": 1.7593,
+      "step": 1302
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.429980276134123e-05,
+      "loss": 1.7579,
+      "step": 1303
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.428007889546351e-05,
+      "loss": 1.7311,
+      "step": 1304
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.42603550295858e-05,
+      "loss": 1.6763,
+      "step": 1305
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.424063116370808e-05,
+      "loss": 1.7589,
+      "step": 1306
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.422090729783038e-05,
+      "loss": 1.6833,
+      "step": 1307
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.420118343195267e-05,
+      "loss": 1.7089,
+      "step": 1308
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.418145956607495e-05,
+      "loss": 1.7003,
+      "step": 1309
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 7.416173570019724e-05,
+      "loss": 1.7385,
+      "step": 1310
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 7.414201183431953e-05,
+      "loss": 1.6371,
+      "step": 1311
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 7.412228796844182e-05,
+      "loss": 1.8231,
+      "step": 1312
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 7.410256410256412e-05,
+      "loss": 1.7619,
+      "step": 1313
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.40828402366864e-05,
+      "loss": 1.8075,
+      "step": 1314
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.406311637080869e-05,
+      "loss": 1.6752,
+      "step": 1315
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.404339250493098e-05,
+      "loss": 1.8428,
+      "step": 1316
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.402366863905326e-05,
+      "loss": 1.7997,
+      "step": 1317
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.400394477317555e-05,
+      "loss": 1.644,
+      "step": 1318
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.398422090729783e-05,
+      "loss": 1.6914,
+      "step": 1319
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.396449704142012e-05,
+      "loss": 1.7578,
+      "step": 1320
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.394477317554241e-05,
+      "loss": 1.7929,
+      "step": 1321
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.392504930966469e-05,
+      "loss": 1.699,
+      "step": 1322
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.390532544378698e-05,
+      "loss": 1.6683,
+      "step": 1323
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.388560157790927e-05,
+      "loss": 1.7566,
+      "step": 1324
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.386587771203157e-05,
+      "loss": 1.7643,
+      "step": 1325
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.384615384615386e-05,
+      "loss": 1.6878,
+      "step": 1326
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.382642998027614e-05,
+      "loss": 1.7265,
+      "step": 1327
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.380670611439843e-05,
+      "loss": 1.6404,
+      "step": 1328
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.378698224852071e-05,
+      "loss": 1.6106,
+      "step": 1329
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.3767258382643e-05,
+      "loss": 1.6446,
+      "step": 1330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.374753451676529e-05,
+      "loss": 1.6853,
+      "step": 1331
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.372781065088757e-05,
+      "loss": 1.6794,
+      "step": 1332
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.370808678500986e-05,
+      "loss": 1.7357,
+      "step": 1333
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.368836291913215e-05,
+      "loss": 1.7443,
+      "step": 1334
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.366863905325445e-05,
+      "loss": 1.702,
+      "step": 1335
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.364891518737674e-05,
+      "loss": 1.6853,
+      "step": 1336
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.362919132149902e-05,
+      "loss": 1.6812,
+      "step": 1337
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.360946745562131e-05,
+      "loss": 1.6566,
+      "step": 1338
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.35897435897436e-05,
+      "loss": 1.7234,
+      "step": 1339
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.357001972386588e-05,
+      "loss": 1.6996,
+      "step": 1340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.355029585798817e-05,
+      "loss": 1.7073,
+      "step": 1341
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.353057199211045e-05,
+      "loss": 1.692,
+      "step": 1342
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.351084812623274e-05,
+      "loss": 1.6757,
+      "step": 1343
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.349112426035504e-05,
+      "loss": 1.67,
+      "step": 1344
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.347140039447731e-05,
+      "loss": 1.7981,
+      "step": 1345
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.34516765285996e-05,
+      "loss": 1.7532,
+      "step": 1346
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.34319526627219e-05,
+      "loss": 1.8358,
+      "step": 1347
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.341222879684419e-05,
+      "loss": 1.6859,
+      "step": 1348
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.339250493096648e-05,
+      "loss": 1.7356,
+      "step": 1349
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.337278106508876e-05,
+      "loss": 1.7954,
+      "step": 1350
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.335305719921105e-05,
+      "loss": 1.793,
+      "step": 1351
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 1.6172,
+      "step": 1352
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.331360946745562e-05,
+      "loss": 1.6901,
+      "step": 1353
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.329388560157792e-05,
+      "loss": 1.698,
+      "step": 1354
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.32741617357002e-05,
+      "loss": 1.7814,
+      "step": 1355
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.325443786982249e-05,
+      "loss": 1.6698,
+      "step": 1356
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.323471400394478e-05,
+      "loss": 1.6042,
+      "step": 1357
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.321499013806706e-05,
+      "loss": 1.6576,
+      "step": 1358
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.319526627218935e-05,
+      "loss": 1.8264,
+      "step": 1359
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.317554240631164e-05,
+      "loss": 1.7129,
+      "step": 1360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.315581854043393e-05,
+      "loss": 1.7531,
+      "step": 1361
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.313609467455623e-05,
+      "loss": 1.6028,
+      "step": 1362
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.31163708086785e-05,
+      "loss": 1.7758,
+      "step": 1363
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.30966469428008e-05,
+      "loss": 1.6812,
+      "step": 1364
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 1.7393,
+      "step": 1365
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.305719921104537e-05,
+      "loss": 1.6751,
+      "step": 1366
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.303747534516766e-05,
+      "loss": 1.7712,
+      "step": 1367
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.301775147928994e-05,
+      "loss": 1.7143,
+      "step": 1368
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.299802761341223e-05,
+      "loss": 1.7282,
+      "step": 1369
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.297830374753452e-05,
+      "loss": 1.7009,
+      "step": 1370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.295857988165681e-05,
+      "loss": 1.7833,
+      "step": 1371
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.29388560157791e-05,
+      "loss": 1.7464,
+      "step": 1372
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.291913214990138e-05,
+      "loss": 1.8077,
+      "step": 1373
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.289940828402368e-05,
+      "loss": 1.6965,
+      "step": 1374
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.287968441814596e-05,
+      "loss": 1.6784,
+      "step": 1375
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.285996055226825e-05,
+      "loss": 1.6507,
+      "step": 1376
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.284023668639054e-05,
+      "loss": 1.6128,
+      "step": 1377
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.282051282051282e-05,
+      "loss": 1.7519,
+      "step": 1378
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.280078895463511e-05,
+      "loss": 1.5589,
+      "step": 1379
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.27810650887574e-05,
+      "loss": 1.7534,
+      "step": 1380
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.276134122287968e-05,
+      "loss": 1.658,
+      "step": 1381
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.274161735700197e-05,
+      "loss": 1.7008,
+      "step": 1382
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.272189349112426e-05,
+      "loss": 1.647,
+      "step": 1383
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.270216962524656e-05,
+      "loss": 1.6741,
+      "step": 1384
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.268244575936885e-05,
+      "loss": 1.6591,
+      "step": 1385
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.266272189349113e-05,
+      "loss": 1.6904,
+      "step": 1386
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.264299802761342e-05,
+      "loss": 1.7178,
+      "step": 1387
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.26232741617357e-05,
+      "loss": 1.6854,
+      "step": 1388
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.260355029585799e-05,
+      "loss": 1.7031,
+      "step": 1389
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.258382642998028e-05,
+      "loss": 1.6972,
+      "step": 1390
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.256410256410256e-05,
+      "loss": 1.5943,
+      "step": 1391
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.254437869822485e-05,
+      "loss": 1.6556,
+      "step": 1392
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.252465483234713e-05,
+      "loss": 1.6773,
+      "step": 1393
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.250493096646944e-05,
+      "loss": 1.7108,
+      "step": 1394
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.248520710059173e-05,
+      "loss": 1.6129,
+      "step": 1395
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.246548323471401e-05,
+      "loss": 1.725,
+      "step": 1396
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.24457593688363e-05,
+      "loss": 1.67,
+      "step": 1397
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.242603550295858e-05,
+      "loss": 1.665,
+      "step": 1398
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.240631163708087e-05,
+      "loss": 1.7091,
+      "step": 1399
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.238658777120316e-05,
+      "loss": 1.703,
+      "step": 1400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.236686390532544e-05,
+      "loss": 1.7505,
+      "step": 1401
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.234714003944773e-05,
+      "loss": 1.749,
+      "step": 1402
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.232741617357003e-05,
+      "loss": 1.6112,
+      "step": 1403
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.23076923076923e-05,
+      "loss": 1.6849,
+      "step": 1404
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.22879684418146e-05,
+      "loss": 1.6101,
+      "step": 1405
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.226824457593689e-05,
+      "loss": 1.7052,
+      "step": 1406
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.224852071005918e-05,
+      "loss": 1.693,
+      "step": 1407
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.222879684418147e-05,
+      "loss": 1.7472,
+      "step": 1408
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.220907297830375e-05,
+      "loss": 1.5884,
+      "step": 1409
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.218934911242604e-05,
+      "loss": 1.6703,
+      "step": 1410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.216962524654832e-05,
+      "loss": 1.7915,
+      "step": 1411
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.214990138067061e-05,
+      "loss": 1.7117,
+      "step": 1412
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.21301775147929e-05,
+      "loss": 1.7337,
+      "step": 1413
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.211045364891518e-05,
+      "loss": 1.6531,
+      "step": 1414
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.209072978303748e-05,
+      "loss": 1.6375,
+      "step": 1415
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.207100591715976e-05,
+      "loss": 1.8062,
+      "step": 1416
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.205128205128205e-05,
+      "loss": 1.71,
+      "step": 1417
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.203155818540434e-05,
+      "loss": 1.7602,
+      "step": 1418
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.201183431952663e-05,
+      "loss": 1.6197,
+      "step": 1419
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.199211045364892e-05,
+      "loss": 1.7183,
+      "step": 1420
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.19723865877712e-05,
+      "loss": 1.7505,
+      "step": 1421
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.19526627218935e-05,
+      "loss": 1.6458,
+      "step": 1422
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.193293885601579e-05,
+      "loss": 1.7194,
+      "step": 1423
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.191321499013807e-05,
+      "loss": 1.7151,
+      "step": 1424
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.189349112426036e-05,
+      "loss": 1.607,
+      "step": 1425
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.187376725838265e-05,
+      "loss": 1.6577,
+      "step": 1426
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.185404339250493e-05,
+      "loss": 1.7345,
+      "step": 1427
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.183431952662722e-05,
+      "loss": 1.643,
+      "step": 1428
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.181459566074951e-05,
+      "loss": 1.6539,
+      "step": 1429
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 1.6924,
+      "step": 1430
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.17751479289941e-05,
+      "loss": 1.774,
+      "step": 1431
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.175542406311638e-05,
+      "loss": 1.6964,
+      "step": 1432
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.173570019723867e-05,
+      "loss": 1.706,
+      "step": 1433
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.171597633136095e-05,
+      "loss": 1.7393,
+      "step": 1434
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.169625246548324e-05,
+      "loss": 1.7168,
+      "step": 1435
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.167652859960553e-05,
+      "loss": 1.6651,
+      "step": 1436
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.165680473372781e-05,
+      "loss": 1.6461,
+      "step": 1437
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.16370808678501e-05,
+      "loss": 1.612,
+      "step": 1438
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.161735700197238e-05,
+      "loss": 1.7587,
+      "step": 1439
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.159763313609467e-05,
+      "loss": 1.6489,
+      "step": 1440
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.157790927021696e-05,
+      "loss": 1.817,
+      "step": 1441
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.155818540433926e-05,
+      "loss": 1.6523,
+      "step": 1442
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.153846153846155e-05,
+      "loss": 1.6976,
+      "step": 1443
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.151873767258384e-05,
+      "loss": 1.609,
+      "step": 1444
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.149901380670612e-05,
+      "loss": 1.7544,
+      "step": 1445
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.147928994082841e-05,
+      "loss": 1.7576,
+      "step": 1446
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.145956607495069e-05,
+      "loss": 1.6113,
+      "step": 1447
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.143984220907298e-05,
+      "loss": 1.7401,
+      "step": 1448
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.142011834319527e-05,
+      "loss": 1.6532,
+      "step": 1449
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.140039447731755e-05,
+      "loss": 1.7223,
+      "step": 1450
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.138067061143984e-05,
+      "loss": 1.7033,
+      "step": 1451
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.136094674556214e-05,
+      "loss": 1.785,
+      "step": 1452
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.134122287968443e-05,
+      "loss": 1.6098,
+      "step": 1453
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.132149901380672e-05,
+      "loss": 1.6419,
+      "step": 1454
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.1301775147929e-05,
+      "loss": 1.6428,
+      "step": 1455
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.128205128205129e-05,
+      "loss": 1.7238,
+      "step": 1456
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.126232741617357e-05,
+      "loss": 1.7275,
+      "step": 1457
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.124260355029586e-05,
+      "loss": 1.656,
+      "step": 1458
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.122287968441815e-05,
+      "loss": 1.7101,
+      "step": 1459
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.120315581854043e-05,
+      "loss": 1.6485,
+      "step": 1460
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.118343195266272e-05,
+      "loss": 1.6507,
+      "step": 1461
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.1163708086785e-05,
+      "loss": 1.6879,
+      "step": 1462
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.11439842209073e-05,
+      "loss": 1.6495,
+      "step": 1463
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.112426035502959e-05,
+      "loss": 1.7404,
+      "step": 1464
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.110453648915188e-05,
+      "loss": 1.643,
+      "step": 1465
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.108481262327417e-05,
+      "loss": 1.6463,
+      "step": 1466
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.106508875739646e-05,
+      "loss": 1.7981,
+      "step": 1467
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.104536489151874e-05,
+      "loss": 1.6862,
+      "step": 1468
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.102564102564103e-05,
+      "loss": 1.6814,
+      "step": 1469
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.100591715976331e-05,
+      "loss": 1.8018,
+      "step": 1470
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.09861932938856e-05,
+      "loss": 1.5749,
+      "step": 1471
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.09664694280079e-05,
+      "loss": 1.6312,
+      "step": 1472
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.094674556213018e-05,
+      "loss": 1.764,
+      "step": 1473
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.092702169625247e-05,
+      "loss": 1.7037,
+      "step": 1474
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.090729783037475e-05,
+      "loss": 1.5864,
+      "step": 1475
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.088757396449704e-05,
+      "loss": 1.7195,
+      "step": 1476
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.086785009861933e-05,
+      "loss": 1.6965,
+      "step": 1477
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.084812623274162e-05,
+      "loss": 1.6735,
+      "step": 1478
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.082840236686391e-05,
+      "loss": 1.7213,
+      "step": 1479
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.08086785009862e-05,
+      "loss": 1.6963,
+      "step": 1480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.078895463510849e-05,
+      "loss": 1.7314,
+      "step": 1481
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.076923076923078e-05,
+      "loss": 1.5856,
+      "step": 1482
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.074950690335306e-05,
+      "loss": 1.7434,
+      "step": 1483
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.072978303747535e-05,
+      "loss": 1.5951,
+      "step": 1484
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.071005917159763e-05,
+      "loss": 1.6218,
+      "step": 1485
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.069033530571992e-05,
+      "loss": 1.6218,
+      "step": 1486
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.067061143984221e-05,
+      "loss": 1.6577,
+      "step": 1487
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.06508875739645e-05,
+      "loss": 1.6189,
+      "step": 1488
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.06311637080868e-05,
+      "loss": 1.7413,
+      "step": 1489
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.061143984220909e-05,
+      "loss": 1.6668,
+      "step": 1490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.059171597633137e-05,
+      "loss": 1.6773,
+      "step": 1491
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.057199211045366e-05,
+      "loss": 1.675,
+      "step": 1492
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.055226824457594e-05,
+      "loss": 1.7214,
+      "step": 1493
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.053254437869823e-05,
+      "loss": 1.6433,
+      "step": 1494
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 1.6909,
+      "step": 1495
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.04930966469428e-05,
+      "loss": 1.6688,
+      "step": 1496
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.047337278106509e-05,
+      "loss": 1.6395,
+      "step": 1497
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.045364891518737e-05,
+      "loss": 1.6815,
+      "step": 1498
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.043392504930966e-05,
+      "loss": 1.6835,
+      "step": 1499
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.041420118343195e-05,
+      "loss": 1.6929,
+      "step": 1500
+    },
+    {
+      "epoch": 4.44,
+      "eval_loss": 1.5261839628219604,
+      "eval_runtime": 69.3076,
+      "eval_samples_per_second": 321.321,
+      "eval_steps_per_second": 1.674,
+      "step": 1500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.039447731755425e-05,
+      "loss": 1.695,
+      "step": 1501
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.037475345167654e-05,
+      "loss": 1.7585,
+      "step": 1502
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.035502958579882e-05,
+      "loss": 1.5668,
+      "step": 1503
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.033530571992111e-05,
+      "loss": 1.7283,
+      "step": 1504
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.03155818540434e-05,
+      "loss": 1.7085,
+      "step": 1505
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.029585798816568e-05,
+      "loss": 1.8013,
+      "step": 1506
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.027613412228797e-05,
+      "loss": 1.6614,
+      "step": 1507
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.025641025641025e-05,
+      "loss": 1.7083,
+      "step": 1508
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.023668639053254e-05,
+      "loss": 1.6591,
+      "step": 1509
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.021696252465483e-05,
+      "loss": 1.6939,
+      "step": 1510
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.019723865877713e-05,
+      "loss": 1.6936,
+      "step": 1511
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.017751479289942e-05,
+      "loss": 1.6156,
+      "step": 1512
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.015779092702171e-05,
+      "loss": 1.6188,
+      "step": 1513
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.013806706114399e-05,
+      "loss": 1.7037,
+      "step": 1514
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.011834319526628e-05,
+      "loss": 1.7514,
+      "step": 1515
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.009861932938856e-05,
+      "loss": 1.6374,
+      "step": 1516
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.007889546351085e-05,
+      "loss": 1.7444,
+      "step": 1517
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.005917159763314e-05,
+      "loss": 1.6806,
+      "step": 1518
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.003944773175542e-05,
+      "loss": 1.7,
+      "step": 1519
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.001972386587772e-05,
+      "loss": 1.6271,
+      "step": 1520
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7e-05,
+      "loss": 1.6974,
+      "step": 1521
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.998027613412229e-05,
+      "loss": 1.655,
+      "step": 1522
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.996055226824458e-05,
+      "loss": 1.602,
+      "step": 1523
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.994082840236687e-05,
+      "loss": 1.7667,
+      "step": 1524
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.992110453648916e-05,
+      "loss": 1.7123,
+      "step": 1525
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.990138067061144e-05,
+      "loss": 1.8451,
+      "step": 1526
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.988165680473373e-05,
+      "loss": 1.6255,
+      "step": 1527
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.986193293885603e-05,
+      "loss": 1.7041,
+      "step": 1528
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.98422090729783e-05,
+      "loss": 1.7547,
+      "step": 1529
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.98224852071006e-05,
+      "loss": 1.604,
+      "step": 1530
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.980276134122287e-05,
+      "loss": 1.6659,
+      "step": 1531
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.978303747534517e-05,
+      "loss": 1.7572,
+      "step": 1532
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.976331360946746e-05,
+      "loss": 1.7103,
+      "step": 1533
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.974358974358974e-05,
+      "loss": 1.6534,
+      "step": 1534
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.972386587771203e-05,
+      "loss": 1.6458,
+      "step": 1535
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.970414201183432e-05,
+      "loss": 1.6916,
+      "step": 1536
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.968441814595661e-05,
+      "loss": 1.6783,
+      "step": 1537
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.96646942800789e-05,
+      "loss": 1.7412,
+      "step": 1538
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.964497041420118e-05,
+      "loss": 1.7834,
+      "step": 1539
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.962524654832348e-05,
+      "loss": 1.7327,
+      "step": 1540
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.960552268244577e-05,
+      "loss": 1.6625,
+      "step": 1541
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.958579881656805e-05,
+      "loss": 1.6921,
+      "step": 1542
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.956607495069034e-05,
+      "loss": 1.6634,
+      "step": 1543
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.954635108481262e-05,
+      "loss": 1.6871,
+      "step": 1544
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.952662721893491e-05,
+      "loss": 1.706,
+      "step": 1545
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.95069033530572e-05,
+      "loss": 1.7756,
+      "step": 1546
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.94871794871795e-05,
+      "loss": 1.7524,
+      "step": 1547
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.946745562130179e-05,
+      "loss": 1.6669,
+      "step": 1548
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.944773175542406e-05,
+      "loss": 1.7623,
+      "step": 1549
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.942800788954636e-05,
+      "loss": 1.6281,
+      "step": 1550
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.940828402366865e-05,
+      "loss": 1.7002,
+      "step": 1551
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.938856015779093e-05,
+      "loss": 1.6835,
+      "step": 1552
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.936883629191322e-05,
+      "loss": 1.7039,
+      "step": 1553
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.93491124260355e-05,
+      "loss": 1.6842,
+      "step": 1554
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.932938856015779e-05,
+      "loss": 1.617,
+      "step": 1555
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.930966469428008e-05,
+      "loss": 1.5942,
+      "step": 1556
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.928994082840236e-05,
+      "loss": 1.6847,
+      "step": 1557
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.927021696252465e-05,
+      "loss": 1.6843,
+      "step": 1558
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.925049309664695e-05,
+      "loss": 1.6311,
+      "step": 1559
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 1.7402,
+      "step": 1560
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.921104536489153e-05,
+      "loss": 1.8036,
+      "step": 1561
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.919132149901381e-05,
+      "loss": 1.6092,
+      "step": 1562
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.91715976331361e-05,
+      "loss": 1.634,
+      "step": 1563
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.915187376725839e-05,
+      "loss": 1.7229,
+      "step": 1564
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.913214990138067e-05,
+      "loss": 1.6745,
+      "step": 1565
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.911242603550296e-05,
+      "loss": 1.6808,
+      "step": 1566
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.909270216962524e-05,
+      "loss": 1.6811,
+      "step": 1567
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.907297830374753e-05,
+      "loss": 1.6214,
+      "step": 1568
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.905325443786983e-05,
+      "loss": 1.6334,
+      "step": 1569
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.903353057199212e-05,
+      "loss": 1.7347,
+      "step": 1570
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.901380670611441e-05,
+      "loss": 1.586,
+      "step": 1571
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.899408284023669e-05,
+      "loss": 1.6061,
+      "step": 1572
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.897435897435898e-05,
+      "loss": 1.6794,
+      "step": 1573
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.895463510848127e-05,
+      "loss": 1.6595,
+      "step": 1574
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.893491124260355e-05,
+      "loss": 1.8368,
+      "step": 1575
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.891518737672584e-05,
+      "loss": 1.5694,
+      "step": 1576
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.889546351084814e-05,
+      "loss": 1.7532,
+      "step": 1577
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.887573964497041e-05,
+      "loss": 1.7016,
+      "step": 1578
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.88560157790927e-05,
+      "loss": 1.6331,
+      "step": 1579
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.883629191321498e-05,
+      "loss": 1.7002,
+      "step": 1580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.881656804733728e-05,
+      "loss": 1.7429,
+      "step": 1581
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.879684418145957e-05,
+      "loss": 1.5845,
+      "step": 1582
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.877712031558186e-05,
+      "loss": 1.6967,
+      "step": 1583
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.875739644970415e-05,
+      "loss": 1.671,
+      "step": 1584
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.873767258382643e-05,
+      "loss": 1.6277,
+      "step": 1585
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.871794871794872e-05,
+      "loss": 1.7199,
+      "step": 1586
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.869822485207102e-05,
+      "loss": 1.7113,
+      "step": 1587
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.86785009861933e-05,
+      "loss": 1.5845,
+      "step": 1588
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.865877712031559e-05,
+      "loss": 1.6749,
+      "step": 1589
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.863905325443787e-05,
+      "loss": 1.5795,
+      "step": 1590
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.861932938856016e-05,
+      "loss": 1.5712,
+      "step": 1591
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.859960552268245e-05,
+      "loss": 1.7565,
+      "step": 1592
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.857988165680473e-05,
+      "loss": 1.6717,
+      "step": 1593
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.856015779092702e-05,
+      "loss": 1.6726,
+      "step": 1594
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.854043392504931e-05,
+      "loss": 1.6378,
+      "step": 1595
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.85207100591716e-05,
+      "loss": 1.6823,
+      "step": 1596
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.85009861932939e-05,
+      "loss": 1.7148,
+      "step": 1597
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.848126232741617e-05,
+      "loss": 1.745,
+      "step": 1598
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.846153846153847e-05,
+      "loss": 1.629,
+      "step": 1599
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.844181459566076e-05,
+      "loss": 1.6951,
+      "step": 1600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.842209072978304e-05,
+      "loss": 1.6388,
+      "step": 1601
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.840236686390533e-05,
+      "loss": 1.6383,
+      "step": 1602
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.838264299802761e-05,
+      "loss": 1.5902,
+      "step": 1603
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.83629191321499e-05,
+      "loss": 1.6663,
+      "step": 1604
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.834319526627219e-05,
+      "loss": 1.7243,
+      "step": 1605
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.832347140039448e-05,
+      "loss": 1.5873,
+      "step": 1606
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.830374753451678e-05,
+      "loss": 1.6889,
+      "step": 1607
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.828402366863906e-05,
+      "loss": 1.5907,
+      "step": 1608
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.826429980276135e-05,
+      "loss": 1.6725,
+      "step": 1609
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.824457593688364e-05,
+      "loss": 1.712,
+      "step": 1610
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.822485207100592e-05,
+      "loss": 1.7375,
+      "step": 1611
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.820512820512821e-05,
+      "loss": 1.6592,
+      "step": 1612
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.818540433925049e-05,
+      "loss": 1.7399,
+      "step": 1613
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.816568047337278e-05,
+      "loss": 1.7114,
+      "step": 1614
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.814595660749507e-05,
+      "loss": 1.7714,
+      "step": 1615
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.812623274161735e-05,
+      "loss": 1.6295,
+      "step": 1616
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.810650887573964e-05,
+      "loss": 1.6547,
+      "step": 1617
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.808678500986194e-05,
+      "loss": 1.647,
+      "step": 1618
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.806706114398423e-05,
+      "loss": 1.7809,
+      "step": 1619
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.804733727810652e-05,
+      "loss": 1.5747,
+      "step": 1620
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.80276134122288e-05,
+      "loss": 1.6598,
+      "step": 1621
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.800788954635109e-05,
+      "loss": 1.6836,
+      "step": 1622
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.798816568047338e-05,
+      "loss": 1.6107,
+      "step": 1623
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.796844181459566e-05,
+      "loss": 1.7016,
+      "step": 1624
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 1.6553,
+      "step": 1625
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.792899408284023e-05,
+      "loss": 1.7569,
+      "step": 1626
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.790927021696252e-05,
+      "loss": 1.6879,
+      "step": 1627
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.788954635108482e-05,
+      "loss": 1.6139,
+      "step": 1628
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.786982248520711e-05,
+      "loss": 1.6525,
+      "step": 1629
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.78500986193294e-05,
+      "loss": 1.7435,
+      "step": 1630
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.783037475345168e-05,
+      "loss": 1.6012,
+      "step": 1631
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.781065088757397e-05,
+      "loss": 1.6758,
+      "step": 1632
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.779092702169626e-05,
+      "loss": 1.6789,
+      "step": 1633
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.777120315581854e-05,
+      "loss": 1.6851,
+      "step": 1634
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.775147928994083e-05,
+      "loss": 1.6771,
+      "step": 1635
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.773175542406311e-05,
+      "loss": 1.7053,
+      "step": 1636
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.77120315581854e-05,
+      "loss": 1.7031,
+      "step": 1637
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.76923076923077e-05,
+      "loss": 1.6993,
+      "step": 1638
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.767258382642998e-05,
+      "loss": 1.7541,
+      "step": 1639
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.765285996055227e-05,
+      "loss": 1.7491,
+      "step": 1640
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.763313609467456e-05,
+      "loss": 1.7476,
+      "step": 1641
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.761341222879685e-05,
+      "loss": 1.7134,
+      "step": 1642
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.759368836291914e-05,
+      "loss": 1.671,
+      "step": 1643
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.757396449704142e-05,
+      "loss": 1.6314,
+      "step": 1644
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.755424063116371e-05,
+      "loss": 1.7459,
+      "step": 1645
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.7534516765286e-05,
+      "loss": 1.6199,
+      "step": 1646
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.751479289940829e-05,
+      "loss": 1.7738,
+      "step": 1647
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.749506903353058e-05,
+      "loss": 1.5952,
+      "step": 1648
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.747534516765286e-05,
+      "loss": 1.5989,
+      "step": 1649
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.745562130177515e-05,
+      "loss": 1.649,
+      "step": 1650
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.743589743589744e-05,
+      "loss": 1.6976,
+      "step": 1651
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.741617357001972e-05,
+      "loss": 1.7155,
+      "step": 1652
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.739644970414201e-05,
+      "loss": 1.6374,
+      "step": 1653
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.73767258382643e-05,
+      "loss": 1.6676,
+      "step": 1654
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.73570019723866e-05,
+      "loss": 1.7835,
+      "step": 1655
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.733727810650889e-05,
+      "loss": 1.6443,
+      "step": 1656
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.731755424063117e-05,
+      "loss": 1.5772,
+      "step": 1657
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.729783037475346e-05,
+      "loss": 1.6758,
+      "step": 1658
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.727810650887574e-05,
+      "loss": 1.6709,
+      "step": 1659
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.725838264299803e-05,
+      "loss": 1.651,
+      "step": 1660
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.723865877712032e-05,
+      "loss": 1.755,
+      "step": 1661
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.72189349112426e-05,
+      "loss": 1.6905,
+      "step": 1662
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.719921104536489e-05,
+      "loss": 1.6215,
+      "step": 1663
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.717948717948718e-05,
+      "loss": 1.6125,
+      "step": 1664
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.715976331360948e-05,
+      "loss": 1.6466,
+      "step": 1665
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.714003944773177e-05,
+      "loss": 1.7112,
+      "step": 1666
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.712031558185405e-05,
+      "loss": 1.7251,
+      "step": 1667
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.710059171597634e-05,
+      "loss": 1.7104,
+      "step": 1668
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.708086785009863e-05,
+      "loss": 1.691,
+      "step": 1669
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.706114398422091e-05,
+      "loss": 1.7012,
+      "step": 1670
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.70414201183432e-05,
+      "loss": 1.6405,
+      "step": 1671
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.702169625246548e-05,
+      "loss": 1.7031,
+      "step": 1672
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.700197238658777e-05,
+      "loss": 1.7001,
+      "step": 1673
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.698224852071006e-05,
+      "loss": 1.6615,
+      "step": 1674
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.696252465483234e-05,
+      "loss": 1.6866,
+      "step": 1675
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.694280078895463e-05,
+      "loss": 1.6486,
+      "step": 1676
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.692307692307693e-05,
+      "loss": 1.7297,
+      "step": 1677
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.690335305719922e-05,
+      "loss": 1.6228,
+      "step": 1678
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.688362919132151e-05,
+      "loss": 1.6862,
+      "step": 1679
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.686390532544379e-05,
+      "loss": 1.6468,
+      "step": 1680
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.684418145956608e-05,
+      "loss": 1.652,
+      "step": 1681
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.682445759368836e-05,
+      "loss": 1.7169,
+      "step": 1682
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.680473372781065e-05,
+      "loss": 1.7838,
+      "step": 1683
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.678500986193294e-05,
+      "loss": 1.6955,
+      "step": 1684
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.676528599605522e-05,
+      "loss": 1.712,
+      "step": 1685
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.674556213017751e-05,
+      "loss": 1.6439,
+      "step": 1686
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.672583826429981e-05,
+      "loss": 1.6041,
+      "step": 1687
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.67061143984221e-05,
+      "loss": 1.75,
+      "step": 1688
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.668639053254439e-05,
+      "loss": 1.6455,
+      "step": 1689
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.5692,
+      "step": 1690
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.664694280078896e-05,
+      "loss": 1.6062,
+      "step": 1691
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.662721893491125e-05,
+      "loss": 1.6053,
+      "step": 1692
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.660749506903353e-05,
+      "loss": 1.6259,
+      "step": 1693
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.658777120315582e-05,
+      "loss": 1.6173,
+      "step": 1694
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.65680473372781e-05,
+      "loss": 1.6866,
+      "step": 1695
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.65483234714004e-05,
+      "loss": 1.7775,
+      "step": 1696
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.652859960552269e-05,
+      "loss": 1.6216,
+      "step": 1697
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.650887573964497e-05,
+      "loss": 1.7687,
+      "step": 1698
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.648915187376726e-05,
+      "loss": 1.6166,
+      "step": 1699
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.646942800788955e-05,
+      "loss": 1.6359,
+      "step": 1700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.644970414201184e-05,
+      "loss": 1.6395,
+      "step": 1701
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.642998027613413e-05,
+      "loss": 1.5865,
+      "step": 1702
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.641025641025641e-05,
+      "loss": 1.6524,
+      "step": 1703
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.63905325443787e-05,
+      "loss": 1.743,
+      "step": 1704
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.637080867850098e-05,
+      "loss": 1.5959,
+      "step": 1705
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.635108481262328e-05,
+      "loss": 1.6975,
+      "step": 1706
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.633136094674557e-05,
+      "loss": 1.5926,
+      "step": 1707
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.631163708086785e-05,
+      "loss": 1.6271,
+      "step": 1708
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.629191321499014e-05,
+      "loss": 1.6931,
+      "step": 1709
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.627218934911243e-05,
+      "loss": 1.703,
+      "step": 1710
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.625246548323471e-05,
+      "loss": 1.6567,
+      "step": 1711
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.6232741617357e-05,
+      "loss": 1.6859,
+      "step": 1712
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.62130177514793e-05,
+      "loss": 1.7783,
+      "step": 1713
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.619329388560159e-05,
+      "loss": 1.5862,
+      "step": 1714
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.617357001972388e-05,
+      "loss": 1.6993,
+      "step": 1715
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.615384615384616e-05,
+      "loss": 1.6873,
+      "step": 1716
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.613412228796845e-05,
+      "loss": 1.661,
+      "step": 1717
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.611439842209073e-05,
+      "loss": 1.5867,
+      "step": 1718
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.609467455621302e-05,
+      "loss": 1.6718,
+      "step": 1719
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.607495069033531e-05,
+      "loss": 1.7067,
+      "step": 1720
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.605522682445759e-05,
+      "loss": 1.7595,
+      "step": 1721
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.603550295857988e-05,
+      "loss": 1.7694,
+      "step": 1722
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.601577909270217e-05,
+      "loss": 1.585,
+      "step": 1723
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.599605522682447e-05,
+      "loss": 1.5637,
+      "step": 1724
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.597633136094676e-05,
+      "loss": 1.6366,
+      "step": 1725
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.595660749506904e-05,
+      "loss": 1.6975,
+      "step": 1726
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.593688362919133e-05,
+      "loss": 1.6301,
+      "step": 1727
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.591715976331361e-05,
+      "loss": 1.6111,
+      "step": 1728
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.58974358974359e-05,
+      "loss": 1.6016,
+      "step": 1729
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.587771203155819e-05,
+      "loss": 1.5586,
+      "step": 1730
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.585798816568047e-05,
+      "loss": 1.7331,
+      "step": 1731
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.583826429980276e-05,
+      "loss": 1.6426,
+      "step": 1732
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.581854043392505e-05,
+      "loss": 1.6112,
+      "step": 1733
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.579881656804733e-05,
+      "loss": 1.6679,
+      "step": 1734
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.577909270216963e-05,
+      "loss": 1.6295,
+      "step": 1735
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.575936883629192e-05,
+      "loss": 1.7395,
+      "step": 1736
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.573964497041421e-05,
+      "loss": 1.7692,
+      "step": 1737
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.57199211045365e-05,
+      "loss": 1.6747,
+      "step": 1738
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.570019723865878e-05,
+      "loss": 1.6445,
+      "step": 1739
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.568047337278107e-05,
+      "loss": 1.7905,
+      "step": 1740
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.566074950690335e-05,
+      "loss": 1.7085,
+      "step": 1741
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.564102564102564e-05,
+      "loss": 1.7785,
+      "step": 1742
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.562130177514794e-05,
+      "loss": 1.6156,
+      "step": 1743
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.560157790927021e-05,
+      "loss": 1.7902,
+      "step": 1744
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.55818540433925e-05,
+      "loss": 1.5176,
+      "step": 1745
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.55621301775148e-05,
+      "loss": 1.5902,
+      "step": 1746
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.554240631163709e-05,
+      "loss": 1.584,
+      "step": 1747
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.552268244575938e-05,
+      "loss": 1.664,
+      "step": 1748
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.550295857988166e-05,
+      "loss": 1.6822,
+      "step": 1749
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.548323471400395e-05,
+      "loss": 1.5269,
+      "step": 1750
+    },
+    {
+      "epoch": 5.18,
+      "eval_loss": 1.5034419298171997,
+      "eval_runtime": 69.3543,
+      "eval_samples_per_second": 321.105,
+      "eval_steps_per_second": 1.673,
+      "step": 1750
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.546351084812623e-05,
+      "loss": 1.6999,
+      "step": 1751
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.544378698224852e-05,
+      "loss": 1.6479,
+      "step": 1752
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.542406311637082e-05,
+      "loss": 1.6849,
+      "step": 1753
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.54043392504931e-05,
+      "loss": 1.6577,
+      "step": 1754
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 1.697,
+      "step": 1755
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.536489151873768e-05,
+      "loss": 1.6922,
+      "step": 1756
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.534516765285996e-05,
+      "loss": 1.6819,
+      "step": 1757
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.532544378698225e-05,
+      "loss": 1.6994,
+      "step": 1758
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.530571992110454e-05,
+      "loss": 1.5715,
+      "step": 1759
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.528599605522683e-05,
+      "loss": 1.7086,
+      "step": 1760
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.526627218934913e-05,
+      "loss": 1.6502,
+      "step": 1761
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.52465483234714e-05,
+      "loss": 1.6834,
+      "step": 1762
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.52268244575937e-05,
+      "loss": 1.6329,
+      "step": 1763
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.520710059171597e-05,
+      "loss": 1.6813,
+      "step": 1764
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.518737672583827e-05,
+      "loss": 1.5854,
+      "step": 1765
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.516765285996056e-05,
+      "loss": 1.6623,
+      "step": 1766
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.514792899408284e-05,
+      "loss": 1.7244,
+      "step": 1767
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.512820512820513e-05,
+      "loss": 1.7064,
+      "step": 1768
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.510848126232741e-05,
+      "loss": 1.734,
+      "step": 1769
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.50887573964497e-05,
+      "loss": 1.6468,
+      "step": 1770
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.506903353057199e-05,
+      "loss": 1.6086,
+      "step": 1771
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.504930966469428e-05,
+      "loss": 1.6016,
+      "step": 1772
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.502958579881658e-05,
+      "loss": 1.6209,
+      "step": 1773
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.500986193293886e-05,
+      "loss": 1.6177,
+      "step": 1774
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.499013806706115e-05,
+      "loss": 1.5701,
+      "step": 1775
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.497041420118344e-05,
+      "loss": 1.6554,
+      "step": 1776
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.495069033530572e-05,
+      "loss": 1.6647,
+      "step": 1777
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.493096646942801e-05,
+      "loss": 1.5974,
+      "step": 1778
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.49112426035503e-05,
+      "loss": 1.6408,
+      "step": 1779
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.489151873767258e-05,
+      "loss": 1.6932,
+      "step": 1780
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.487179487179487e-05,
+      "loss": 1.6165,
+      "step": 1781
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.485207100591716e-05,
+      "loss": 1.6844,
+      "step": 1782
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.483234714003946e-05,
+      "loss": 1.664,
+      "step": 1783
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.481262327416175e-05,
+      "loss": 1.7908,
+      "step": 1784
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.479289940828403e-05,
+      "loss": 1.7524,
+      "step": 1785
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.477317554240632e-05,
+      "loss": 1.649,
+      "step": 1786
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.47534516765286e-05,
+      "loss": 1.5624,
+      "step": 1787
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.473372781065089e-05,
+      "loss": 1.6394,
+      "step": 1788
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.471400394477318e-05,
+      "loss": 1.6399,
+      "step": 1789
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.469428007889546e-05,
+      "loss": 1.5973,
+      "step": 1790
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.467455621301775e-05,
+      "loss": 1.6413,
+      "step": 1791
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.465483234714003e-05,
+      "loss": 1.6051,
+      "step": 1792
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.463510848126232e-05,
+      "loss": 1.7984,
+      "step": 1793
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.461538461538462e-05,
+      "loss": 1.703,
+      "step": 1794
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.459566074950691e-05,
+      "loss": 1.6493,
+      "step": 1795
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.45759368836292e-05,
+      "loss": 1.6355,
+      "step": 1796
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.455621301775149e-05,
+      "loss": 1.7316,
+      "step": 1797
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.453648915187377e-05,
+      "loss": 1.5653,
+      "step": 1798
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.451676528599606e-05,
+      "loss": 1.5851,
+      "step": 1799
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.449704142011834e-05,
+      "loss": 1.7155,
+      "step": 1800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.447731755424063e-05,
+      "loss": 1.5956,
+      "step": 1801
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.445759368836293e-05,
+      "loss": 1.7257,
+      "step": 1802
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.44378698224852e-05,
+      "loss": 1.7179,
+      "step": 1803
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.44181459566075e-05,
+      "loss": 1.7321,
+      "step": 1804
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.439842209072979e-05,
+      "loss": 1.5654,
+      "step": 1805
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.437869822485208e-05,
+      "loss": 1.6664,
+      "step": 1806
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.435897435897437e-05,
+      "loss": 1.5924,
+      "step": 1807
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.433925049309665e-05,
+      "loss": 1.5771,
+      "step": 1808
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.431952662721894e-05,
+      "loss": 1.6284,
+      "step": 1809
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.429980276134122e-05,
+      "loss": 1.633,
+      "step": 1810
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.428007889546351e-05,
+      "loss": 1.6503,
+      "step": 1811
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.42603550295858e-05,
+      "loss": 1.5963,
+      "step": 1812
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.424063116370808e-05,
+      "loss": 1.6556,
+      "step": 1813
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.422090729783038e-05,
+      "loss": 1.6577,
+      "step": 1814
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.420118343195266e-05,
+      "loss": 1.6834,
+      "step": 1815
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.418145956607495e-05,
+      "loss": 1.6147,
+      "step": 1816
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.416173570019724e-05,
+      "loss": 1.7063,
+      "step": 1817
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.414201183431953e-05,
+      "loss": 1.6561,
+      "step": 1818
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.412228796844182e-05,
+      "loss": 1.6494,
+      "step": 1819
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 1.6226,
+      "step": 1820
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.40828402366864e-05,
+      "loss": 1.7249,
+      "step": 1821
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.406311637080869e-05,
+      "loss": 1.6362,
+      "step": 1822
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.404339250493097e-05,
+      "loss": 1.6458,
+      "step": 1823
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.402366863905326e-05,
+      "loss": 1.5809,
+      "step": 1824
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.400394477317555e-05,
+      "loss": 1.6508,
+      "step": 1825
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.398422090729783e-05,
+      "loss": 1.6044,
+      "step": 1826
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.396449704142012e-05,
+      "loss": 1.5575,
+      "step": 1827
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.39447731755424e-05,
+      "loss": 1.7022,
+      "step": 1828
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.392504930966469e-05,
+      "loss": 1.6891,
+      "step": 1829
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.390532544378698e-05,
+      "loss": 1.6565,
+      "step": 1830
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.388560157790928e-05,
+      "loss": 1.6505,
+      "step": 1831
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.386587771203157e-05,
+      "loss": 1.602,
+      "step": 1832
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.384615384615385e-05,
+      "loss": 1.6895,
+      "step": 1833
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.382642998027614e-05,
+      "loss": 1.6774,
+      "step": 1834
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.380670611439843e-05,
+      "loss": 1.7063,
+      "step": 1835
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.378698224852071e-05,
+      "loss": 1.6098,
+      "step": 1836
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.3767258382643e-05,
+      "loss": 1.5493,
+      "step": 1837
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.374753451676528e-05,
+      "loss": 1.7638,
+      "step": 1838
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.372781065088757e-05,
+      "loss": 1.5486,
+      "step": 1839
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.370808678500986e-05,
+      "loss": 1.5467,
+      "step": 1840
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.368836291913216e-05,
+      "loss": 1.7039,
+      "step": 1841
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.366863905325445e-05,
+      "loss": 1.6133,
+      "step": 1842
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.364891518737674e-05,
+      "loss": 1.6649,
+      "step": 1843
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.362919132149902e-05,
+      "loss": 1.619,
+      "step": 1844
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.360946745562131e-05,
+      "loss": 1.6084,
+      "step": 1845
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.358974358974359e-05,
+      "loss": 1.778,
+      "step": 1846
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.357001972386588e-05,
+      "loss": 1.5867,
+      "step": 1847
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.355029585798817e-05,
+      "loss": 1.6142,
+      "step": 1848
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.353057199211045e-05,
+      "loss": 1.5622,
+      "step": 1849
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.351084812623274e-05,
+      "loss": 1.6251,
+      "step": 1850
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.349112426035502e-05,
+      "loss": 1.6941,
+      "step": 1851
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.347140039447731e-05,
+      "loss": 1.7277,
+      "step": 1852
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.34516765285996e-05,
+      "loss": 1.6248,
+      "step": 1853
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.34319526627219e-05,
+      "loss": 1.6421,
+      "step": 1854
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.341222879684419e-05,
+      "loss": 1.6356,
+      "step": 1855
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.339250493096647e-05,
+      "loss": 1.6892,
+      "step": 1856
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.337278106508876e-05,
+      "loss": 1.6854,
+      "step": 1857
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.335305719921105e-05,
+      "loss": 1.6351,
+      "step": 1858
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 1.6677,
+      "step": 1859
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.331360946745562e-05,
+      "loss": 1.748,
+      "step": 1860
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.32938856015779e-05,
+      "loss": 1.6874,
+      "step": 1861
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.32741617357002e-05,
+      "loss": 1.5966,
+      "step": 1862
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.325443786982249e-05,
+      "loss": 1.6354,
+      "step": 1863
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.323471400394478e-05,
+      "loss": 1.6946,
+      "step": 1864
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.321499013806707e-05,
+      "loss": 1.7417,
+      "step": 1865
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.319526627218936e-05,
+      "loss": 1.6535,
+      "step": 1866
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.317554240631164e-05,
+      "loss": 1.7221,
+      "step": 1867
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.315581854043393e-05,
+      "loss": 1.7401,
+      "step": 1868
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.313609467455621e-05,
+      "loss": 1.5997,
+      "step": 1869
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.31163708086785e-05,
+      "loss": 1.6496,
+      "step": 1870
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.30966469428008e-05,
+      "loss": 1.6385,
+      "step": 1871
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.307692307692308e-05,
+      "loss": 1.5846,
+      "step": 1872
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.305719921104537e-05,
+      "loss": 1.5461,
+      "step": 1873
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.303747534516765e-05,
+      "loss": 1.7008,
+      "step": 1874
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.301775147928994e-05,
+      "loss": 1.6542,
+      "step": 1875
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.299802761341223e-05,
+      "loss": 1.6809,
+      "step": 1876
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.297830374753452e-05,
+      "loss": 1.6583,
+      "step": 1877
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.295857988165681e-05,
+      "loss": 1.6528,
+      "step": 1878
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.29388560157791e-05,
+      "loss": 1.632,
+      "step": 1879
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.291913214990139e-05,
+      "loss": 1.6759,
+      "step": 1880
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.289940828402368e-05,
+      "loss": 1.6115,
+      "step": 1881
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.287968441814596e-05,
+      "loss": 1.6452,
+      "step": 1882
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.285996055226825e-05,
+      "loss": 1.6374,
+      "step": 1883
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.284023668639053e-05,
+      "loss": 1.6466,
+      "step": 1884
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 1.6743,
+      "step": 1885
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.280078895463511e-05,
+      "loss": 1.7694,
+      "step": 1886
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.278106508875739e-05,
+      "loss": 1.6999,
+      "step": 1887
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.276134122287968e-05,
+      "loss": 1.5366,
+      "step": 1888
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.274161735700199e-05,
+      "loss": 1.6058,
+      "step": 1889
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.272189349112427e-05,
+      "loss": 1.5548,
+      "step": 1890
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.270216962524656e-05,
+      "loss": 1.574,
+      "step": 1891
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.268244575936884e-05,
+      "loss": 1.6095,
+      "step": 1892
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.266272189349113e-05,
+      "loss": 1.6208,
+      "step": 1893
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.264299802761342e-05,
+      "loss": 1.6171,
+      "step": 1894
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.26232741617357e-05,
+      "loss": 1.5873,
+      "step": 1895
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.260355029585799e-05,
+      "loss": 1.7138,
+      "step": 1896
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.258382642998027e-05,
+      "loss": 1.6191,
+      "step": 1897
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.256410256410256e-05,
+      "loss": 1.6254,
+      "step": 1898
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.254437869822485e-05,
+      "loss": 1.6237,
+      "step": 1899
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.252465483234715e-05,
+      "loss": 1.6244,
+      "step": 1900
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.250493096646944e-05,
+      "loss": 1.5734,
+      "step": 1901
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.248520710059172e-05,
+      "loss": 1.7106,
+      "step": 1902
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.246548323471401e-05,
+      "loss": 1.6269,
+      "step": 1903
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.24457593688363e-05,
+      "loss": 1.584,
+      "step": 1904
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.242603550295858e-05,
+      "loss": 1.6794,
+      "step": 1905
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.240631163708087e-05,
+      "loss": 1.7075,
+      "step": 1906
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.238658777120316e-05,
+      "loss": 1.7441,
+      "step": 1907
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.236686390532544e-05,
+      "loss": 1.56,
+      "step": 1908
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.234714003944773e-05,
+      "loss": 1.6514,
+      "step": 1909
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.232741617357001e-05,
+      "loss": 1.6301,
+      "step": 1910
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.23076923076923e-05,
+      "loss": 1.6797,
+      "step": 1911
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.22879684418146e-05,
+      "loss": 1.7106,
+      "step": 1912
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.226824457593689e-05,
+      "loss": 1.7865,
+      "step": 1913
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.224852071005918e-05,
+      "loss": 1.6141,
+      "step": 1914
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.222879684418146e-05,
+      "loss": 1.6293,
+      "step": 1915
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.220907297830375e-05,
+      "loss": 1.5419,
+      "step": 1916
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.218934911242604e-05,
+      "loss": 1.5938,
+      "step": 1917
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.216962524654832e-05,
+      "loss": 1.6748,
+      "step": 1918
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.214990138067062e-05,
+      "loss": 1.5019,
+      "step": 1919
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.21301775147929e-05,
+      "loss": 1.5735,
+      "step": 1920
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.211045364891519e-05,
+      "loss": 1.6608,
+      "step": 1921
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.209072978303748e-05,
+      "loss": 1.5384,
+      "step": 1922
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.207100591715977e-05,
+      "loss": 1.7461,
+      "step": 1923
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.205128205128206e-05,
+      "loss": 1.6565,
+      "step": 1924
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.203155818540434e-05,
+      "loss": 1.7347,
+      "step": 1925
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.201183431952663e-05,
+      "loss": 1.5339,
+      "step": 1926
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.199211045364892e-05,
+      "loss": 1.6116,
+      "step": 1927
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.19723865877712e-05,
+      "loss": 1.6182,
+      "step": 1928
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.19526627218935e-05,
+      "loss": 1.6928,
+      "step": 1929
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.193293885601579e-05,
+      "loss": 1.6215,
+      "step": 1930
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.191321499013807e-05,
+      "loss": 1.5391,
+      "step": 1931
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.189349112426036e-05,
+      "loss": 1.6265,
+      "step": 1932
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.187376725838264e-05,
+      "loss": 1.5649,
+      "step": 1933
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.185404339250493e-05,
+      "loss": 1.6132,
+      "step": 1934
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.183431952662722e-05,
+      "loss": 1.6076,
+      "step": 1935
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.181459566074951e-05,
+      "loss": 1.6126,
+      "step": 1936
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.17948717948718e-05,
+      "loss": 1.5696,
+      "step": 1937
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.177514792899408e-05,
+      "loss": 1.6056,
+      "step": 1938
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.175542406311638e-05,
+      "loss": 1.497,
+      "step": 1939
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.173570019723867e-05,
+      "loss": 1.6595,
+      "step": 1940
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.171597633136095e-05,
+      "loss": 1.6147,
+      "step": 1941
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.169625246548324e-05,
+      "loss": 1.6221,
+      "step": 1942
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.167652859960552e-05,
+      "loss": 1.4354,
+      "step": 1943
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.165680473372781e-05,
+      "loss": 1.672,
+      "step": 1944
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.16370808678501e-05,
+      "loss": 1.6458,
+      "step": 1945
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.161735700197238e-05,
+      "loss": 1.6172,
+      "step": 1946
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.159763313609467e-05,
+      "loss": 1.6525,
+      "step": 1947
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.157790927021696e-05,
+      "loss": 1.6611,
+      "step": 1948
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.155818540433926e-05,
+      "loss": 1.6012,
+      "step": 1949
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 1.688,
+      "step": 1950
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.151873767258383e-05,
+      "loss": 1.6337,
+      "step": 1951
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.149901380670612e-05,
+      "loss": 1.6499,
+      "step": 1952
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.147928994082841e-05,
+      "loss": 1.6346,
+      "step": 1953
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.145956607495069e-05,
+      "loss": 1.5665,
+      "step": 1954
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.143984220907298e-05,
+      "loss": 1.6425,
+      "step": 1955
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.142011834319526e-05,
+      "loss": 1.5487,
+      "step": 1956
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.140039447731755e-05,
+      "loss": 1.6378,
+      "step": 1957
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.138067061143984e-05,
+      "loss": 1.6321,
+      "step": 1958
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.136094674556214e-05,
+      "loss": 1.6962,
+      "step": 1959
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.134122287968443e-05,
+      "loss": 1.6388,
+      "step": 1960
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.132149901380671e-05,
+      "loss": 1.6181,
+      "step": 1961
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.1301775147929e-05,
+      "loss": 1.5708,
+      "step": 1962
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.128205128205129e-05,
+      "loss": 1.6482,
+      "step": 1963
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.126232741617357e-05,
+      "loss": 1.6112,
+      "step": 1964
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.124260355029586e-05,
+      "loss": 1.5604,
+      "step": 1965
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.122287968441814e-05,
+      "loss": 1.6859,
+      "step": 1966
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.120315581854043e-05,
+      "loss": 1.6221,
+      "step": 1967
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.118343195266273e-05,
+      "loss": 1.609,
+      "step": 1968
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.1163708086785e-05,
+      "loss": 1.6889,
+      "step": 1969
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.11439842209073e-05,
+      "loss": 1.6574,
+      "step": 1970
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.112426035502959e-05,
+      "loss": 1.698,
+      "step": 1971
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.110453648915188e-05,
+      "loss": 1.5841,
+      "step": 1972
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.108481262327417e-05,
+      "loss": 1.6886,
+      "step": 1973
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.106508875739645e-05,
+      "loss": 1.7181,
+      "step": 1974
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.104536489151874e-05,
+      "loss": 1.6755,
+      "step": 1975
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.1025641025641035e-05,
+      "loss": 1.5742,
+      "step": 1976
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.1005917159763314e-05,
+      "loss": 1.5847,
+      "step": 1977
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0986193293885606e-05,
+      "loss": 1.5676,
+      "step": 1978
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.096646942800789e-05,
+      "loss": 1.6134,
+      "step": 1979
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.094674556213018e-05,
+      "loss": 1.5594,
+      "step": 1980
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0927021696252475e-05,
+      "loss": 1.5746,
+      "step": 1981
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0907297830374754e-05,
+      "loss": 1.7408,
+      "step": 1982
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0887573964497046e-05,
+      "loss": 1.5906,
+      "step": 1983
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0867850098619325e-05,
+      "loss": 1.6684,
+      "step": 1984
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.084812623274162e-05,
+      "loss": 1.5945,
+      "step": 1985
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0828402366863916e-05,
+      "loss": 1.6329,
+      "step": 1986
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0808678500986194e-05,
+      "loss": 1.6363,
+      "step": 1987
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0788954635108486e-05,
+      "loss": 1.6287,
+      "step": 1988
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0769230769230765e-05,
+      "loss": 1.5999,
+      "step": 1989
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.074950690335306e-05,
+      "loss": 1.7113,
+      "step": 1990
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.072978303747535e-05,
+      "loss": 1.6081,
+      "step": 1991
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0710059171597635e-05,
+      "loss": 1.7062,
+      "step": 1992
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.069033530571993e-05,
+      "loss": 1.6882,
+      "step": 1993
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0670611439842205e-05,
+      "loss": 1.6784,
+      "step": 1994
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.06508875739645e-05,
+      "loss": 1.5149,
+      "step": 1995
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.063116370808679e-05,
+      "loss": 1.6024,
+      "step": 1996
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0611439842209075e-05,
+      "loss": 1.5979,
+      "step": 1997
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.059171597633137e-05,
+      "loss": 1.57,
+      "step": 1998
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.057199211045366e-05,
+      "loss": 1.5614,
+      "step": 1999
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.055226824457594e-05,
+      "loss": 1.701,
+      "step": 2000
+    },
+    {
+      "epoch": 5.92,
+      "eval_loss": 1.4841886758804321,
+      "eval_runtime": 71.2697,
+      "eval_samples_per_second": 312.475,
+      "eval_steps_per_second": 1.628,
+      "step": 2000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.053254437869823e-05,
+      "loss": 1.7501,
+      "step": 2001
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0512820512820515e-05,
+      "loss": 1.7029,
+      "step": 2002
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.049309664694281e-05,
+      "loss": 1.5721,
+      "step": 2003
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.04733727810651e-05,
+      "loss": 1.6621,
+      "step": 2004
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.045364891518738e-05,
+      "loss": 1.6392,
+      "step": 2005
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.043392504930967e-05,
+      "loss": 1.715,
+      "step": 2006
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.041420118343195e-05,
+      "loss": 1.6153,
+      "step": 2007
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.039447731755424e-05,
+      "loss": 1.7215,
+      "step": 2008
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.037475345167653e-05,
+      "loss": 1.7396,
+      "step": 2009
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.035502958579882e-05,
+      "loss": 1.6267,
+      "step": 2010
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.033530571992111e-05,
+      "loss": 1.5477,
+      "step": 2011
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.031558185404339e-05,
+      "loss": 1.6819,
+      "step": 2012
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.029585798816568e-05,
+      "loss": 1.7775,
+      "step": 2013
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.027613412228797e-05,
+      "loss": 1.6547,
+      "step": 2014
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 1.6192,
+      "step": 2015
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.023668639053255e-05,
+      "loss": 1.5231,
+      "step": 2016
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.021696252465484e-05,
+      "loss": 1.6265,
+      "step": 2017
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.019723865877712e-05,
+      "loss": 1.6888,
+      "step": 2018
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.017751479289941e-05,
+      "loss": 1.5533,
+      "step": 2019
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.01577909270217e-05,
+      "loss": 1.5677,
+      "step": 2020
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.013806706114399e-05,
+      "loss": 1.6082,
+      "step": 2021
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.011834319526628e-05,
+      "loss": 1.639,
+      "step": 2022
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.009861932938856e-05,
+      "loss": 1.6443,
+      "step": 2023
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0078895463510853e-05,
+      "loss": 1.6805,
+      "step": 2024
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.005917159763313e-05,
+      "loss": 1.5888,
+      "step": 2025
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0039447731755424e-05,
+      "loss": 1.6103,
+      "step": 2026
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0019723865877716e-05,
+      "loss": 1.5529,
+      "step": 2027
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6e-05,
+      "loss": 1.7657,
+      "step": 2028
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.9980276134122294e-05,
+      "loss": 1.5298,
+      "step": 2029
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.996055226824457e-05,
+      "loss": 1.6132,
+      "step": 2030
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.9940828402366864e-05,
+      "loss": 1.6396,
+      "step": 2031
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.9921104536489156e-05,
+      "loss": 1.6405,
+      "step": 2032
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.990138067061144e-05,
+      "loss": 1.5749,
+      "step": 2033
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9881656804733734e-05,
+      "loss": 1.6563,
+      "step": 2034
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.986193293885601e-05,
+      "loss": 1.6689,
+      "step": 2035
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9842209072978305e-05,
+      "loss": 1.683,
+      "step": 2036
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.98224852071006e-05,
+      "loss": 1.5831,
+      "step": 2037
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.980276134122288e-05,
+      "loss": 1.5859,
+      "step": 2038
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9783037475345174e-05,
+      "loss": 1.5538,
+      "step": 2039
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9763313609467466e-05,
+      "loss": 1.6337,
+      "step": 2040
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9743589743589745e-05,
+      "loss": 1.5621,
+      "step": 2041
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.972386587771204e-05,
+      "loss": 1.6455,
+      "step": 2042
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9704142011834315e-05,
+      "loss": 1.5531,
+      "step": 2043
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9684418145956614e-05,
+      "loss": 1.6827,
+      "step": 2044
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9664694280078906e-05,
+      "loss": 1.6581,
+      "step": 2045
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9644970414201185e-05,
+      "loss": 1.6266,
+      "step": 2046
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.962524654832348e-05,
+      "loss": 1.5763,
+      "step": 2047
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9605522682445756e-05,
+      "loss": 1.6249,
+      "step": 2048
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.958579881656805e-05,
+      "loss": 1.6313,
+      "step": 2049
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.956607495069034e-05,
+      "loss": 1.6493,
+      "step": 2050
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9546351084812625e-05,
+      "loss": 1.6173,
+      "step": 2051
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.952662721893492e-05,
+      "loss": 1.58,
+      "step": 2052
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9506903353057196e-05,
+      "loss": 1.7155,
+      "step": 2053
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.948717948717949e-05,
+      "loss": 1.6468,
+      "step": 2054
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.946745562130178e-05,
+      "loss": 1.5762,
+      "step": 2055
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9447731755424066e-05,
+      "loss": 1.6463,
+      "step": 2056
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.942800788954636e-05,
+      "loss": 1.5248,
+      "step": 2057
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9408284023668636e-05,
+      "loss": 1.662,
+      "step": 2058
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.938856015779093e-05,
+      "loss": 1.6276,
+      "step": 2059
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.936883629191322e-05,
+      "loss": 1.5576,
+      "step": 2060
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9349112426035506e-05,
+      "loss": 1.6583,
+      "step": 2061
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.93293885601578e-05,
+      "loss": 1.6066,
+      "step": 2062
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.930966469428009e-05,
+      "loss": 1.6566,
+      "step": 2063
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.928994082840237e-05,
+      "loss": 1.5704,
+      "step": 2064
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.927021696252466e-05,
+      "loss": 1.5443,
+      "step": 2065
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.925049309664694e-05,
+      "loss": 1.6922,
+      "step": 2066
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.923076923076923e-05,
+      "loss": 1.6437,
+      "step": 2067
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9211045364891523e-05,
+      "loss": 1.6683,
+      "step": 2068
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.919132149901381e-05,
+      "loss": 1.606,
+      "step": 2069
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.91715976331361e-05,
+      "loss": 1.5999,
+      "step": 2070
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.915187376725838e-05,
+      "loss": 1.5963,
+      "step": 2071
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.913214990138067e-05,
+      "loss": 1.6396,
+      "step": 2072
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9112426035502964e-05,
+      "loss": 1.7008,
+      "step": 2073
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.909270216962525e-05,
+      "loss": 1.5412,
+      "step": 2074
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.907297830374754e-05,
+      "loss": 1.5185,
+      "step": 2075
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.905325443786982e-05,
+      "loss": 1.5798,
+      "step": 2076
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.903353057199211e-05,
+      "loss": 1.575,
+      "step": 2077
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9013806706114404e-05,
+      "loss": 1.5787,
+      "step": 2078
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.899408284023669e-05,
+      "loss": 1.5432,
+      "step": 2079
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 1.6644,
+      "step": 2080
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.895463510848126e-05,
+      "loss": 1.7091,
+      "step": 2081
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.893491124260355e-05,
+      "loss": 1.6619,
+      "step": 2082
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.8915187376725844e-05,
+      "loss": 1.5301,
+      "step": 2083
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.889546351084812e-05,
+      "loss": 1.543,
+      "step": 2084
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.8875739644970415e-05,
+      "loss": 1.6549,
+      "step": 2085
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.885601577909271e-05,
+      "loss": 1.698,
+      "step": 2086
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.883629191321499e-05,
+      "loss": 1.6191,
+      "step": 2087
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.8816568047337284e-05,
+      "loss": 1.6626,
+      "step": 2088
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.879684418145956e-05,
+      "loss": 1.6065,
+      "step": 2089
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.8777120315581855e-05,
+      "loss": 1.6243,
+      "step": 2090
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.875739644970415e-05,
+      "loss": 1.5242,
+      "step": 2091
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.873767258382643e-05,
+      "loss": 1.6029,
+      "step": 2092
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.8717948717948725e-05,
+      "loss": 1.6208,
+      "step": 2093
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.8698224852071e-05,
+      "loss": 1.6836,
+      "step": 2094
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.8678500986193295e-05,
+      "loss": 1.5641,
+      "step": 2095
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.865877712031559e-05,
+      "loss": 1.5688,
+      "step": 2096
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.863905325443787e-05,
+      "loss": 1.6506,
+      "step": 2097
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.8619329388560165e-05,
+      "loss": 1.5874,
+      "step": 2098
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.8599605522682443e-05,
+      "loss": 1.7077,
+      "step": 2099
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.8579881656804736e-05,
+      "loss": 1.655,
+      "step": 2100
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.856015779092703e-05,
+      "loss": 1.6922,
+      "step": 2101
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.8540433925049306e-05,
+      "loss": 1.6188,
+      "step": 2102
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.8520710059171605e-05,
+      "loss": 1.5386,
+      "step": 2103
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.8500986193293884e-05,
+      "loss": 1.6067,
+      "step": 2104
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.8481262327416176e-05,
+      "loss": 1.5283,
+      "step": 2105
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.846153846153847e-05,
+      "loss": 1.588,
+      "step": 2106
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.8441814595660746e-05,
+      "loss": 1.5828,
+      "step": 2107
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.842209072978304e-05,
+      "loss": 1.609,
+      "step": 2108
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.840236686390533e-05,
+      "loss": 1.5982,
+      "step": 2109
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.8382642998027616e-05,
+      "loss": 1.6343,
+      "step": 2110
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.836291913214991e-05,
+      "loss": 1.498,
+      "step": 2111
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.834319526627219e-05,
+      "loss": 1.5619,
+      "step": 2112
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.832347140039448e-05,
+      "loss": 1.6793,
+      "step": 2113
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.830374753451677e-05,
+      "loss": 1.6226,
+      "step": 2114
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.8284023668639056e-05,
+      "loss": 1.6216,
+      "step": 2115
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.826429980276135e-05,
+      "loss": 1.5588,
+      "step": 2116
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.824457593688363e-05,
+      "loss": 1.6249,
+      "step": 2117
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.822485207100592e-05,
+      "loss": 1.6538,
+      "step": 2118
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.820512820512821e-05,
+      "loss": 1.592,
+      "step": 2119
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.8185404339250496e-05,
+      "loss": 1.6058,
+      "step": 2120
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.816568047337279e-05,
+      "loss": 1.6321,
+      "step": 2121
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.814595660749507e-05,
+      "loss": 1.5745,
+      "step": 2122
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.812623274161736e-05,
+      "loss": 1.6497,
+      "step": 2123
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.810650887573965e-05,
+      "loss": 1.629,
+      "step": 2124
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.808678500986193e-05,
+      "loss": 1.6455,
+      "step": 2125
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.806706114398422e-05,
+      "loss": 1.6286,
+      "step": 2126
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.804733727810651e-05,
+      "loss": 1.563,
+      "step": 2127
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.80276134122288e-05,
+      "loss": 1.5667,
+      "step": 2128
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.800788954635109e-05,
+      "loss": 1.5795,
+      "step": 2129
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.798816568047337e-05,
+      "loss": 1.6178,
+      "step": 2130
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.796844181459566e-05,
+      "loss": 1.6251,
+      "step": 2131
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.7948717948717954e-05,
+      "loss": 1.6072,
+      "step": 2132
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.792899408284024e-05,
+      "loss": 1.6705,
+      "step": 2133
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.790927021696253e-05,
+      "loss": 1.6947,
+      "step": 2134
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.788954635108481e-05,
+      "loss": 1.6439,
+      "step": 2135
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.78698224852071e-05,
+      "loss": 1.6282,
+      "step": 2136
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.7850098619329395e-05,
+      "loss": 1.5784,
+      "step": 2137
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.783037475345168e-05,
+      "loss": 1.6451,
+      "step": 2138
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.781065088757397e-05,
+      "loss": 1.6811,
+      "step": 2139
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.779092702169625e-05,
+      "loss": 1.7167,
+      "step": 2140
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.777120315581854e-05,
+      "loss": 1.653,
+      "step": 2141
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.7751479289940835e-05,
+      "loss": 1.6521,
+      "step": 2142
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.7731755424063113e-05,
+      "loss": 1.5802,
+      "step": 2143
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.7712031558185406e-05,
+      "loss": 1.6603,
+      "step": 2144
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 1.6422,
+      "step": 2145
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.767258382642998e-05,
+      "loss": 1.6256,
+      "step": 2146
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.7652859960552275e-05,
+      "loss": 1.561,
+      "step": 2147
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.7633136094674554e-05,
+      "loss": 1.5832,
+      "step": 2148
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.7613412228796846e-05,
+      "loss": 1.6375,
+      "step": 2149
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.759368836291914e-05,
+      "loss": 1.6037,
+      "step": 2150
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.757396449704142e-05,
+      "loss": 1.6044,
+      "step": 2151
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.7554240631163715e-05,
+      "loss": 1.6619,
+      "step": 2152
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.7534516765285994e-05,
+      "loss": 1.5219,
+      "step": 2153
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.7514792899408286e-05,
+      "loss": 1.5851,
+      "step": 2154
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.749506903353058e-05,
+      "loss": 1.6376,
+      "step": 2155
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.7475345167652863e-05,
+      "loss": 1.5914,
+      "step": 2156
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.7455621301775156e-05,
+      "loss": 1.717,
+      "step": 2157
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.7435897435897434e-05,
+      "loss": 1.6086,
+      "step": 2158
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.7416173570019726e-05,
+      "loss": 1.5754,
+      "step": 2159
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.739644970414202e-05,
+      "loss": 1.7116,
+      "step": 2160
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.73767258382643e-05,
+      "loss": 1.6536,
+      "step": 2161
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.7357001972386596e-05,
+      "loss": 1.6464,
+      "step": 2162
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.7337278106508874e-05,
+      "loss": 1.5645,
+      "step": 2163
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.7317554240631167e-05,
+      "loss": 1.5942,
+      "step": 2164
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.729783037475346e-05,
+      "loss": 1.6291,
+      "step": 2165
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.727810650887574e-05,
+      "loss": 1.6629,
+      "step": 2166
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.725838264299803e-05,
+      "loss": 1.5582,
+      "step": 2167
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.7238658777120315e-05,
+      "loss": 1.6325,
+      "step": 2168
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.721893491124261e-05,
+      "loss": 1.6262,
+      "step": 2169
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.71992110453649e-05,
+      "loss": 1.5959,
+      "step": 2170
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.717948717948718e-05,
+      "loss": 1.6321,
+      "step": 2171
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.715976331360947e-05,
+      "loss": 1.6225,
+      "step": 2172
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.714003944773176e-05,
+      "loss": 1.6231,
+      "step": 2173
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.712031558185405e-05,
+      "loss": 1.5911,
+      "step": 2174
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.710059171597634e-05,
+      "loss": 1.5305,
+      "step": 2175
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.708086785009862e-05,
+      "loss": 1.566,
+      "step": 2176
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.706114398422091e-05,
+      "loss": 1.5741,
+      "step": 2177
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.70414201183432e-05,
+      "loss": 1.6367,
+      "step": 2178
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.702169625246549e-05,
+      "loss": 1.6175,
+      "step": 2179
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.700197238658778e-05,
+      "loss": 1.5167,
+      "step": 2180
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.698224852071006e-05,
+      "loss": 1.5728,
+      "step": 2181
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.696252465483235e-05,
+      "loss": 1.6262,
+      "step": 2182
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.694280078895464e-05,
+      "loss": 1.6704,
+      "step": 2183
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.692307692307692e-05,
+      "loss": 1.6757,
+      "step": 2184
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.690335305719921e-05,
+      "loss": 1.6215,
+      "step": 2185
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.68836291913215e-05,
+      "loss": 1.5856,
+      "step": 2186
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.686390532544379e-05,
+      "loss": 1.6471,
+      "step": 2187
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.684418145956608e-05,
+      "loss": 1.613,
+      "step": 2188
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.682445759368836e-05,
+      "loss": 1.6151,
+      "step": 2189
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.680473372781065e-05,
+      "loss": 1.6174,
+      "step": 2190
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.678500986193294e-05,
+      "loss": 1.7336,
+      "step": 2191
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.676528599605523e-05,
+      "loss": 1.6003,
+      "step": 2192
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.674556213017752e-05,
+      "loss": 1.4921,
+      "step": 2193
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.67258382642998e-05,
+      "loss": 1.6245,
+      "step": 2194
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.670611439842209e-05,
+      "loss": 1.6483,
+      "step": 2195
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.6686390532544385e-05,
+      "loss": 1.6103,
+      "step": 2196
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.666666666666667e-05,
+      "loss": 1.6053,
+      "step": 2197
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.664694280078896e-05,
+      "loss": 1.7414,
+      "step": 2198
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.662721893491124e-05,
+      "loss": 1.6037,
+      "step": 2199
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.6607495069033534e-05,
+      "loss": 1.5836,
+      "step": 2200
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.6587771203155826e-05,
+      "loss": 1.5487,
+      "step": 2201
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.6568047337278104e-05,
+      "loss": 1.5598,
+      "step": 2202
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.6548323471400396e-05,
+      "loss": 1.6103,
+      "step": 2203
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.652859960552268e-05,
+      "loss": 1.6047,
+      "step": 2204
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.6508875739644974e-05,
+      "loss": 1.5465,
+      "step": 2205
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.6489151873767266e-05,
+      "loss": 1.5158,
+      "step": 2206
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.6469428007889544e-05,
+      "loss": 1.483,
+      "step": 2207
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.6449704142011837e-05,
+      "loss": 1.6578,
+      "step": 2208
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.642998027613412e-05,
+      "loss": 1.6073,
+      "step": 2209
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 1.642,
+      "step": 2210
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.6390532544378706e-05,
+      "loss": 1.6735,
+      "step": 2211
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.6370808678500985e-05,
+      "loss": 1.6256,
+      "step": 2212
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.635108481262328e-05,
+      "loss": 1.5961,
+      "step": 2213
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.633136094674556e-05,
+      "loss": 1.6871,
+      "step": 2214
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.6311637080867854e-05,
+      "loss": 1.6982,
+      "step": 2215
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.6291913214990146e-05,
+      "loss": 1.602,
+      "step": 2216
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.6272189349112425e-05,
+      "loss": 1.5987,
+      "step": 2217
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.625246548323472e-05,
+      "loss": 1.6333,
+      "step": 2218
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.623274161735701e-05,
+      "loss": 1.6466,
+      "step": 2219
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.621301775147929e-05,
+      "loss": 1.5937,
+      "step": 2220
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.6193293885601587e-05,
+      "loss": 1.7089,
+      "step": 2221
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.6173570019723865e-05,
+      "loss": 1.6593,
+      "step": 2222
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.615384615384616e-05,
+      "loss": 1.5531,
+      "step": 2223
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.613412228796845e-05,
+      "loss": 1.5572,
+      "step": 2224
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.611439842209073e-05,
+      "loss": 1.5783,
+      "step": 2225
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.609467455621302e-05,
+      "loss": 1.611,
+      "step": 2226
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.6074950690335305e-05,
+      "loss": 1.5844,
+      "step": 2227
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.60552268244576e-05,
+      "loss": 1.6808,
+      "step": 2228
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.603550295857989e-05,
+      "loss": 1.6279,
+      "step": 2229
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.601577909270217e-05,
+      "loss": 1.5226,
+      "step": 2230
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.599605522682446e-05,
+      "loss": 1.5973,
+      "step": 2231
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.5976331360946746e-05,
+      "loss": 1.6559,
+      "step": 2232
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.595660749506904e-05,
+      "loss": 1.5799,
+      "step": 2233
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.593688362919133e-05,
+      "loss": 1.5554,
+      "step": 2234
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.591715976331361e-05,
+      "loss": 1.5727,
+      "step": 2235
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.58974358974359e-05,
+      "loss": 1.5107,
+      "step": 2236
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.5877712031558186e-05,
+      "loss": 1.6866,
+      "step": 2237
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.585798816568048e-05,
+      "loss": 1.6106,
+      "step": 2238
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.583826429980277e-05,
+      "loss": 1.6143,
+      "step": 2239
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.581854043392505e-05,
+      "loss": 1.6286,
+      "step": 2240
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.579881656804734e-05,
+      "loss": 1.6135,
+      "step": 2241
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.577909270216963e-05,
+      "loss": 1.5536,
+      "step": 2242
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.575936883629191e-05,
+      "loss": 1.7219,
+      "step": 2243
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.5739644970414204e-05,
+      "loss": 1.5376,
+      "step": 2244
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.571992110453649e-05,
+      "loss": 1.6629,
+      "step": 2245
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.570019723865878e-05,
+      "loss": 1.6586,
+      "step": 2246
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.568047337278107e-05,
+      "loss": 1.5762,
+      "step": 2247
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.566074950690335e-05,
+      "loss": 1.638,
+      "step": 2248
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.5641025641025644e-05,
+      "loss": 1.6351,
+      "step": 2249
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.562130177514793e-05,
+      "loss": 1.5672,
+      "step": 2250
+    },
+    {
+      "epoch": 6.66,
+      "eval_loss": 1.4714722633361816,
+      "eval_runtime": 70.0996,
+      "eval_samples_per_second": 317.691,
+      "eval_steps_per_second": 1.655,
+      "step": 2250
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.560157790927022e-05,
+      "loss": 1.6109,
+      "step": 2251
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.558185404339251e-05,
+      "loss": 1.6852,
+      "step": 2252
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.556213017751479e-05,
+      "loss": 1.6274,
+      "step": 2253
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.5542406311637084e-05,
+      "loss": 1.5829,
+      "step": 2254
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.552268244575937e-05,
+      "loss": 1.5858,
+      "step": 2255
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.550295857988166e-05,
+      "loss": 1.6228,
+      "step": 2256
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.5483234714003954e-05,
+      "loss": 1.7294,
+      "step": 2257
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.546351084812623e-05,
+      "loss": 1.6681,
+      "step": 2258
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.5443786982248524e-05,
+      "loss": 1.6202,
+      "step": 2259
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.5424063116370816e-05,
+      "loss": 1.6302,
+      "step": 2260
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.5404339250493095e-05,
+      "loss": 1.5708,
+      "step": 2261
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.538461538461539e-05,
+      "loss": 1.5977,
+      "step": 2262
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.536489151873767e-05,
+      "loss": 1.5321,
+      "step": 2263
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.5345167652859965e-05,
+      "loss": 1.6043,
+      "step": 2264
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.532544378698226e-05,
+      "loss": 1.5737,
+      "step": 2265
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.5305719921104535e-05,
+      "loss": 1.5572,
+      "step": 2266
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.528599605522683e-05,
+      "loss": 1.5834,
+      "step": 2267
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.526627218934911e-05,
+      "loss": 1.5669,
+      "step": 2268
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.5246548323471405e-05,
+      "loss": 1.6753,
+      "step": 2269
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.52268244575937e-05,
+      "loss": 1.5949,
+      "step": 2270
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.5207100591715975e-05,
+      "loss": 1.592,
+      "step": 2271
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.518737672583827e-05,
+      "loss": 1.6393,
+      "step": 2272
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.516765285996055e-05,
+      "loss": 1.6183,
+      "step": 2273
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.5147928994082845e-05,
+      "loss": 1.5516,
+      "step": 2274
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 1.6637,
+      "step": 2275
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.5108481262327416e-05,
+      "loss": 1.5629,
+      "step": 2276
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.508875739644971e-05,
+      "loss": 1.5463,
+      "step": 2277
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.5069033530571986e-05,
+      "loss": 1.5953,
+      "step": 2278
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.5049309664694285e-05,
+      "loss": 1.6574,
+      "step": 2279
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.502958579881658e-05,
+      "loss": 1.6653,
+      "step": 2280
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.5009861932938856e-05,
+      "loss": 1.6695,
+      "step": 2281
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.499013806706115e-05,
+      "loss": 1.5958,
+      "step": 2282
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.497041420118344e-05,
+      "loss": 1.5628,
+      "step": 2283
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.495069033530572e-05,
+      "loss": 1.5852,
+      "step": 2284
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.493096646942801e-05,
+      "loss": 1.6424,
+      "step": 2285
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.4911242603550296e-05,
+      "loss": 1.6056,
+      "step": 2286
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.489151873767259e-05,
+      "loss": 1.7359,
+      "step": 2287
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.487179487179488e-05,
+      "loss": 1.586,
+      "step": 2288
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.485207100591716e-05,
+      "loss": 1.5643,
+      "step": 2289
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.483234714003945e-05,
+      "loss": 1.6736,
+      "step": 2290
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.4812623274161736e-05,
+      "loss": 1.5219,
+      "step": 2291
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.479289940828403e-05,
+      "loss": 1.8011,
+      "step": 2292
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.477317554240632e-05,
+      "loss": 1.7237,
+      "step": 2293
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.47534516765286e-05,
+      "loss": 1.6738,
+      "step": 2294
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.473372781065089e-05,
+      "loss": 1.5529,
+      "step": 2295
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.471400394477318e-05,
+      "loss": 1.5506,
+      "step": 2296
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.469428007889547e-05,
+      "loss": 1.5579,
+      "step": 2297
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.467455621301776e-05,
+      "loss": 1.5347,
+      "step": 2298
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.465483234714004e-05,
+      "loss": 1.5628,
+      "step": 2299
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.463510848126233e-05,
+      "loss": 1.5722,
+      "step": 2300
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.461538461538461e-05,
+      "loss": 1.6054,
+      "step": 2301
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.45956607495069e-05,
+      "loss": 1.6234,
+      "step": 2302
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.4575936883629194e-05,
+      "loss": 1.6373,
+      "step": 2303
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.455621301775148e-05,
+      "loss": 1.5613,
+      "step": 2304
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.453648915187377e-05,
+      "loss": 1.6901,
+      "step": 2305
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.4516765285996064e-05,
+      "loss": 1.5735,
+      "step": 2306
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.449704142011834e-05,
+      "loss": 1.5976,
+      "step": 2307
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.4477317554240635e-05,
+      "loss": 1.6329,
+      "step": 2308
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.445759368836292e-05,
+      "loss": 1.5565,
+      "step": 2309
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.443786982248521e-05,
+      "loss": 1.6049,
+      "step": 2310
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.4418145956607504e-05,
+      "loss": 1.6203,
+      "step": 2311
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.439842209072978e-05,
+      "loss": 1.6755,
+      "step": 2312
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.4378698224852075e-05,
+      "loss": 1.6191,
+      "step": 2313
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.435897435897436e-05,
+      "loss": 1.5839,
+      "step": 2314
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.433925049309665e-05,
+      "loss": 1.6274,
+      "step": 2315
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.4319526627218944e-05,
+      "loss": 1.5288,
+      "step": 2316
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.429980276134122e-05,
+      "loss": 1.6806,
+      "step": 2317
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.4280078895463515e-05,
+      "loss": 1.5244,
+      "step": 2318
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.4260355029585794e-05,
+      "loss": 1.618,
+      "step": 2319
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.4240631163708086e-05,
+      "loss": 1.6673,
+      "step": 2320
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.422090729783038e-05,
+      "loss": 1.663,
+      "step": 2321
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.420118343195266e-05,
+      "loss": 1.6004,
+      "step": 2322
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.4181459566074955e-05,
+      "loss": 1.5649,
+      "step": 2323
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.4161735700197234e-05,
+      "loss": 1.5782,
+      "step": 2324
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.4142011834319526e-05,
+      "loss": 1.6014,
+      "step": 2325
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.412228796844182e-05,
+      "loss": 1.5304,
+      "step": 2326
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.41025641025641e-05,
+      "loss": 1.5794,
+      "step": 2327
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.4082840236686395e-05,
+      "loss": 1.5849,
+      "step": 2328
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.406311637080869e-05,
+      "loss": 1.5872,
+      "step": 2329
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.4043392504930966e-05,
+      "loss": 1.5256,
+      "step": 2330
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.402366863905326e-05,
+      "loss": 1.5696,
+      "step": 2331
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.4003944773175544e-05,
+      "loss": 1.5824,
+      "step": 2332
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.3984220907297836e-05,
+      "loss": 1.568,
+      "step": 2333
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.396449704142013e-05,
+      "loss": 1.6071,
+      "step": 2334
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.3944773175542406e-05,
+      "loss": 1.5164,
+      "step": 2335
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.39250493096647e-05,
+      "loss": 1.7575,
+      "step": 2336
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.390532544378698e-05,
+      "loss": 1.6616,
+      "step": 2337
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.3885601577909276e-05,
+      "loss": 1.6099,
+      "step": 2338
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.386587771203157e-05,
+      "loss": 1.6757,
+      "step": 2339
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 1.5531,
+      "step": 2340
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.382642998027614e-05,
+      "loss": 1.609,
+      "step": 2341
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.380670611439842e-05,
+      "loss": 1.5781,
+      "step": 2342
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.378698224852071e-05,
+      "loss": 1.6105,
+      "step": 2343
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.3767258382643e-05,
+      "loss": 1.5951,
+      "step": 2344
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.374753451676529e-05,
+      "loss": 1.5704,
+      "step": 2345
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.372781065088758e-05,
+      "loss": 1.5163,
+      "step": 2346
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.370808678500986e-05,
+      "loss": 1.5897,
+      "step": 2347
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.368836291913215e-05,
+      "loss": 1.6108,
+      "step": 2348
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.366863905325444e-05,
+      "loss": 1.5421,
+      "step": 2349
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.364891518737673e-05,
+      "loss": 1.6512,
+      "step": 2350
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.362919132149902e-05,
+      "loss": 1.6235,
+      "step": 2351
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.360946745562131e-05,
+      "loss": 1.629,
+      "step": 2352
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.358974358974359e-05,
+      "loss": 1.5532,
+      "step": 2353
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.357001972386588e-05,
+      "loss": 1.7182,
+      "step": 2354
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.355029585798817e-05,
+      "loss": 1.5842,
+      "step": 2355
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.353057199211046e-05,
+      "loss": 1.6211,
+      "step": 2356
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.351084812623275e-05,
+      "loss": 1.6419,
+      "step": 2357
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.349112426035503e-05,
+      "loss": 1.4771,
+      "step": 2358
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.347140039447732e-05,
+      "loss": 1.6394,
+      "step": 2359
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.34516765285996e-05,
+      "loss": 1.5242,
+      "step": 2360
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.343195266272189e-05,
+      "loss": 1.5229,
+      "step": 2361
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.3412228796844185e-05,
+      "loss": 1.6056,
+      "step": 2362
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.339250493096647e-05,
+      "loss": 1.6066,
+      "step": 2363
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.337278106508876e-05,
+      "loss": 1.6204,
+      "step": 2364
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.335305719921104e-05,
+      "loss": 1.6586,
+      "step": 2365
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 1.6438,
+      "step": 2366
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.3313609467455625e-05,
+      "loss": 1.557,
+      "step": 2367
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.329388560157791e-05,
+      "loss": 1.6326,
+      "step": 2368
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.32741617357002e-05,
+      "loss": 1.548,
+      "step": 2369
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.3254437869822495e-05,
+      "loss": 1.6138,
+      "step": 2370
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.3234714003944773e-05,
+      "loss": 1.4844,
+      "step": 2371
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.3214990138067066e-05,
+      "loss": 1.5519,
+      "step": 2372
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.319526627218935e-05,
+      "loss": 1.5731,
+      "step": 2373
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.317554240631164e-05,
+      "loss": 1.7151,
+      "step": 2374
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.3155818540433935e-05,
+      "loss": 1.6242,
+      "step": 2375
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.3136094674556214e-05,
+      "loss": 1.554,
+      "step": 2376
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.3116370808678506e-05,
+      "loss": 1.6007,
+      "step": 2377
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.3096646942800784e-05,
+      "loss": 1.5668,
+      "step": 2378
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.3076923076923076e-05,
+      "loss": 1.6583,
+      "step": 2379
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.3057199211045375e-05,
+      "loss": 1.4355,
+      "step": 2380
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.3037475345167654e-05,
+      "loss": 1.4911,
+      "step": 2381
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.3017751479289946e-05,
+      "loss": 1.5661,
+      "step": 2382
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.2998027613412225e-05,
+      "loss": 1.6181,
+      "step": 2383
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.297830374753452e-05,
+      "loss": 1.6415,
+      "step": 2384
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.295857988165681e-05,
+      "loss": 1.6076,
+      "step": 2385
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.2938856015779094e-05,
+      "loss": 1.5595,
+      "step": 2386
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.2919132149901386e-05,
+      "loss": 1.6019,
+      "step": 2387
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.2899408284023665e-05,
+      "loss": 1.5907,
+      "step": 2388
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.287968441814596e-05,
+      "loss": 1.555,
+      "step": 2389
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.285996055226825e-05,
+      "loss": 1.5377,
+      "step": 2390
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.2840236686390534e-05,
+      "loss": 1.5717,
+      "step": 2391
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.2820512820512826e-05,
+      "loss": 1.6219,
+      "step": 2392
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.280078895463512e-05,
+      "loss": 1.5771,
+      "step": 2393
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.27810650887574e-05,
+      "loss": 1.6174,
+      "step": 2394
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.276134122287969e-05,
+      "loss": 1.6329,
+      "step": 2395
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.274161735700197e-05,
+      "loss": 1.55,
+      "step": 2396
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.272189349112427e-05,
+      "loss": 1.6396,
+      "step": 2397
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.270216962524656e-05,
+      "loss": 1.5326,
+      "step": 2398
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.268244575936884e-05,
+      "loss": 1.5296,
+      "step": 2399
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.266272189349113e-05,
+      "loss": 1.5958,
+      "step": 2400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.264299802761341e-05,
+      "loss": 1.5954,
+      "step": 2401
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.26232741617357e-05,
+      "loss": 1.5835,
+      "step": 2402
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.260355029585799e-05,
+      "loss": 1.5804,
+      "step": 2403
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.258382642998028e-05,
+      "loss": 1.5872,
+      "step": 2404
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 1.6489,
+      "step": 2405
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.254437869822485e-05,
+      "loss": 1.5896,
+      "step": 2406
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.252465483234714e-05,
+      "loss": 1.587,
+      "step": 2407
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.250493096646943e-05,
+      "loss": 1.6378,
+      "step": 2408
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.248520710059172e-05,
+      "loss": 1.6398,
+      "step": 2409
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.246548323471401e-05,
+      "loss": 1.5728,
+      "step": 2410
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.244575936883629e-05,
+      "loss": 1.5619,
+      "step": 2411
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.242603550295858e-05,
+      "loss": 1.5622,
+      "step": 2412
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.240631163708087e-05,
+      "loss": 1.5956,
+      "step": 2413
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.238658777120316e-05,
+      "loss": 1.7068,
+      "step": 2414
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.236686390532545e-05,
+      "loss": 1.5869,
+      "step": 2415
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.234714003944774e-05,
+      "loss": 1.5638,
+      "step": 2416
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.232741617357002e-05,
+      "loss": 1.5306,
+      "step": 2417
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.230769230769231e-05,
+      "loss": 1.5515,
+      "step": 2418
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.228796844181459e-05,
+      "loss": 1.5729,
+      "step": 2419
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.2268244575936884e-05,
+      "loss": 1.621,
+      "step": 2420
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.2248520710059176e-05,
+      "loss": 1.533,
+      "step": 2421
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.222879684418146e-05,
+      "loss": 1.5682,
+      "step": 2422
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.220907297830375e-05,
+      "loss": 1.6423,
+      "step": 2423
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.218934911242603e-05,
+      "loss": 1.5769,
+      "step": 2424
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.2169625246548324e-05,
+      "loss": 1.6267,
+      "step": 2425
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.2149901380670616e-05,
+      "loss": 1.5839,
+      "step": 2426
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.21301775147929e-05,
+      "loss": 1.5835,
+      "step": 2427
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.2110453648915193e-05,
+      "loss": 1.4921,
+      "step": 2428
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.209072978303747e-05,
+      "loss": 1.6028,
+      "step": 2429
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.2071005917159764e-05,
+      "loss": 1.6083,
+      "step": 2430
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.2051282051282056e-05,
+      "loss": 1.6222,
+      "step": 2431
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.203155818540434e-05,
+      "loss": 1.4607,
+      "step": 2432
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.2011834319526634e-05,
+      "loss": 1.5086,
+      "step": 2433
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.199211045364891e-05,
+      "loss": 1.5501,
+      "step": 2434
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.1972386587771204e-05,
+      "loss": 1.6566,
+      "step": 2435
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.1952662721893497e-05,
+      "loss": 1.5371,
+      "step": 2436
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.1932938856015775e-05,
+      "loss": 1.5238,
+      "step": 2437
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.191321499013807e-05,
+      "loss": 1.4676,
+      "step": 2438
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.1893491124260366e-05,
+      "loss": 1.599,
+      "step": 2439
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.1873767258382645e-05,
+      "loss": 1.5955,
+      "step": 2440
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.185404339250494e-05,
+      "loss": 1.5569,
+      "step": 2441
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.1834319526627215e-05,
+      "loss": 1.588,
+      "step": 2442
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.181459566074951e-05,
+      "loss": 1.6116,
+      "step": 2443
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.17948717948718e-05,
+      "loss": 1.5511,
+      "step": 2444
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.1775147928994085e-05,
+      "loss": 1.6564,
+      "step": 2445
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.175542406311638e-05,
+      "loss": 1.5787,
+      "step": 2446
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.1735700197238656e-05,
+      "loss": 1.7064,
+      "step": 2447
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.171597633136095e-05,
+      "loss": 1.6236,
+      "step": 2448
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.169625246548324e-05,
+      "loss": 1.6374,
+      "step": 2449
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.1676528599605525e-05,
+      "loss": 1.4994,
+      "step": 2450
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.165680473372782e-05,
+      "loss": 1.5591,
+      "step": 2451
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.1637080867850096e-05,
+      "loss": 1.511,
+      "step": 2452
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.161735700197239e-05,
+      "loss": 1.5993,
+      "step": 2453
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.159763313609468e-05,
+      "loss": 1.5864,
+      "step": 2454
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.157790927021696e-05,
+      "loss": 1.5499,
+      "step": 2455
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.155818540433926e-05,
+      "loss": 1.5621,
+      "step": 2456
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.1538461538461536e-05,
+      "loss": 1.5951,
+      "step": 2457
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.151873767258383e-05,
+      "loss": 1.5453,
+      "step": 2458
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.149901380670612e-05,
+      "loss": 1.5588,
+      "step": 2459
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.14792899408284e-05,
+      "loss": 1.6138,
+      "step": 2460
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.145956607495069e-05,
+      "loss": 1.5649,
+      "step": 2461
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.143984220907298e-05,
+      "loss": 1.5677,
+      "step": 2462
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.142011834319527e-05,
+      "loss": 1.52,
+      "step": 2463
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.140039447731756e-05,
+      "loss": 1.6177,
+      "step": 2464
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.138067061143984e-05,
+      "loss": 1.6592,
+      "step": 2465
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.136094674556213e-05,
+      "loss": 1.5731,
+      "step": 2466
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.134122287968442e-05,
+      "loss": 1.6572,
+      "step": 2467
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.132149901380671e-05,
+      "loss": 1.6608,
+      "step": 2468
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1301775147929e-05,
+      "loss": 1.535,
+      "step": 2469
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 1.4749,
+      "step": 2470
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.126232741617357e-05,
+      "loss": 1.6207,
+      "step": 2471
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.1242603550295864e-05,
+      "loss": 1.6127,
+      "step": 2472
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.122287968441815e-05,
+      "loss": 1.5796,
+      "step": 2473
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.120315581854044e-05,
+      "loss": 1.5719,
+      "step": 2474
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.118343195266272e-05,
+      "loss": 1.5169,
+      "step": 2475
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.116370808678501e-05,
+      "loss": 1.4565,
+      "step": 2476
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.1143984220907304e-05,
+      "loss": 1.738,
+      "step": 2477
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.112426035502958e-05,
+      "loss": 1.6292,
+      "step": 2478
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.1104536489151874e-05,
+      "loss": 1.6109,
+      "step": 2479
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.1084812623274167e-05,
+      "loss": 1.6101,
+      "step": 2480
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.106508875739645e-05,
+      "loss": 1.6435,
+      "step": 2481
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.1045364891518744e-05,
+      "loss": 1.5786,
+      "step": 2482
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.102564102564102e-05,
+      "loss": 1.5508,
+      "step": 2483
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.1005917159763315e-05,
+      "loss": 1.5764,
+      "step": 2484
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.098619329388561e-05,
+      "loss": 1.6565,
+      "step": 2485
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.096646942800789e-05,
+      "loss": 1.6528,
+      "step": 2486
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.0946745562130184e-05,
+      "loss": 1.5949,
+      "step": 2487
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.092702169625246e-05,
+      "loss": 1.5998,
+      "step": 2488
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.0907297830374755e-05,
+      "loss": 1.6078,
+      "step": 2489
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.088757396449705e-05,
+      "loss": 1.6205,
+      "step": 2490
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.086785009861933e-05,
+      "loss": 1.5645,
+      "step": 2491
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.0848126232741624e-05,
+      "loss": 1.6635,
+      "step": 2492
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.08284023668639e-05,
+      "loss": 1.5888,
+      "step": 2493
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.0808678500986195e-05,
+      "loss": 1.5121,
+      "step": 2494
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.078895463510849e-05,
+      "loss": 1.6034,
+      "step": 2495
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.0769230769230766e-05,
+      "loss": 1.652,
+      "step": 2496
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.074950690335306e-05,
+      "loss": 1.6463,
+      "step": 2497
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.072978303747534e-05,
+      "loss": 1.5371,
+      "step": 2498
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.0710059171597635e-05,
+      "loss": 1.5415,
+      "step": 2499
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.069033530571993e-05,
+      "loss": 1.6054,
+      "step": 2500
+    },
+    {
+      "epoch": 7.4,
+      "eval_loss": 1.4548845291137695,
+      "eval_runtime": 70.8759,
+      "eval_samples_per_second": 314.211,
+      "eval_steps_per_second": 1.637,
+      "step": 2500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.0670611439842206e-05,
+      "loss": 1.56,
+      "step": 2501
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.06508875739645e-05,
+      "loss": 1.6422,
+      "step": 2502
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.063116370808679e-05,
+      "loss": 1.6429,
+      "step": 2503
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.0611439842209076e-05,
+      "loss": 1.6198,
+      "step": 2504
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.059171597633137e-05,
+      "loss": 1.6745,
+      "step": 2505
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.0571992110453646e-05,
+      "loss": 1.6412,
+      "step": 2506
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.055226824457594e-05,
+      "loss": 1.6325,
+      "step": 2507
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.053254437869823e-05,
+      "loss": 1.5825,
+      "step": 2508
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.0512820512820516e-05,
+      "loss": 1.6123,
+      "step": 2509
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.049309664694281e-05,
+      "loss": 1.5409,
+      "step": 2510
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.0473372781065087e-05,
+      "loss": 1.7114,
+      "step": 2511
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.045364891518738e-05,
+      "loss": 1.5168,
+      "step": 2512
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.043392504930967e-05,
+      "loss": 1.6037,
+      "step": 2513
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.0414201183431956e-05,
+      "loss": 1.5234,
+      "step": 2514
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.039447731755425e-05,
+      "loss": 1.6489,
+      "step": 2515
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.037475345167653e-05,
+      "loss": 1.5278,
+      "step": 2516
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.035502958579882e-05,
+      "loss": 1.5757,
+      "step": 2517
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.033530571992111e-05,
+      "loss": 1.5551,
+      "step": 2518
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.031558185404339e-05,
+      "loss": 1.5033,
+      "step": 2519
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.029585798816568e-05,
+      "loss": 1.5316,
+      "step": 2520
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.027613412228797e-05,
+      "loss": 1.5236,
+      "step": 2521
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.025641025641026e-05,
+      "loss": 1.6251,
+      "step": 2522
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.023668639053255e-05,
+      "loss": 1.5025,
+      "step": 2523
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.021696252465483e-05,
+      "loss": 1.526,
+      "step": 2524
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.019723865877712e-05,
+      "loss": 1.5358,
+      "step": 2525
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.0177514792899414e-05,
+      "loss": 1.5738,
+      "step": 2526
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.01577909270217e-05,
+      "loss": 1.5934,
+      "step": 2527
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.013806706114399e-05,
+      "loss": 1.5143,
+      "step": 2528
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.011834319526627e-05,
+      "loss": 1.5973,
+      "step": 2529
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.009861932938856e-05,
+      "loss": 1.5488,
+      "step": 2530
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.0078895463510854e-05,
+      "loss": 1.5898,
+      "step": 2531
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.005917159763314e-05,
+      "loss": 1.5416,
+      "step": 2532
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.003944773175543e-05,
+      "loss": 1.5512,
+      "step": 2533
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.001972386587771e-05,
+      "loss": 1.6195,
+      "step": 2534
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5e-05,
+      "loss": 1.6155,
+      "step": 2535
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.998027613412229e-05,
+      "loss": 1.5545,
+      "step": 2536
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.996055226824457e-05,
+      "loss": 1.5305,
+      "step": 2537
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9940828402366865e-05,
+      "loss": 1.6177,
+      "step": 2538
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.992110453648916e-05,
+      "loss": 1.6389,
+      "step": 2539
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.990138067061144e-05,
+      "loss": 1.5547,
+      "step": 2540
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.988165680473373e-05,
+      "loss": 1.5735,
+      "step": 2541
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.986193293885602e-05,
+      "loss": 1.6803,
+      "step": 2542
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.9842209072978305e-05,
+      "loss": 1.5673,
+      "step": 2543
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.98224852071006e-05,
+      "loss": 1.5917,
+      "step": 2544
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.980276134122288e-05,
+      "loss": 1.652,
+      "step": 2545
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.978303747534517e-05,
+      "loss": 1.5253,
+      "step": 2546
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.976331360946746e-05,
+      "loss": 1.5608,
+      "step": 2547
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.9743589743589746e-05,
+      "loss": 1.5733,
+      "step": 2548
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.972386587771203e-05,
+      "loss": 1.5754,
+      "step": 2549
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.970414201183432e-05,
+      "loss": 1.6281,
+      "step": 2550
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.968441814595661e-05,
+      "loss": 1.4695,
+      "step": 2551
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.96646942800789e-05,
+      "loss": 1.5733,
+      "step": 2552
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.9644970414201186e-05,
+      "loss": 1.5198,
+      "step": 2553
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.962524654832347e-05,
+      "loss": 1.6595,
+      "step": 2554
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.9605522682445757e-05,
+      "loss": 1.6281,
+      "step": 2555
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.958579881656805e-05,
+      "loss": 1.513,
+      "step": 2556
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.956607495069034e-05,
+      "loss": 1.5618,
+      "step": 2557
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.9546351084812626e-05,
+      "loss": 1.5309,
+      "step": 2558
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.952662721893491e-05,
+      "loss": 1.632,
+      "step": 2559
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.95069033530572e-05,
+      "loss": 1.663,
+      "step": 2560
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.948717948717949e-05,
+      "loss": 1.583,
+      "step": 2561
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.946745562130178e-05,
+      "loss": 1.5299,
+      "step": 2562
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.9447731755424066e-05,
+      "loss": 1.6134,
+      "step": 2563
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.942800788954635e-05,
+      "loss": 1.4874,
+      "step": 2564
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.9408284023668644e-05,
+      "loss": 1.6443,
+      "step": 2565
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.938856015779093e-05,
+      "loss": 1.5747,
+      "step": 2566
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.9368836291913215e-05,
+      "loss": 1.5026,
+      "step": 2567
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.934911242603551e-05,
+      "loss": 1.5959,
+      "step": 2568
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.932938856015779e-05,
+      "loss": 1.6472,
+      "step": 2569
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.9309664694280084e-05,
+      "loss": 1.5928,
+      "step": 2570
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.928994082840237e-05,
+      "loss": 1.6743,
+      "step": 2571
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.9270216962524655e-05,
+      "loss": 1.5463,
+      "step": 2572
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.925049309664695e-05,
+      "loss": 1.5808,
+      "step": 2573
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.923076923076924e-05,
+      "loss": 1.5411,
+      "step": 2574
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.9211045364891524e-05,
+      "loss": 1.5731,
+      "step": 2575
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.919132149901381e-05,
+      "loss": 1.5319,
+      "step": 2576
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.9171597633136095e-05,
+      "loss": 1.6105,
+      "step": 2577
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.915187376725838e-05,
+      "loss": 1.5488,
+      "step": 2578
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.913214990138067e-05,
+      "loss": 1.5737,
+      "step": 2579
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.9112426035502965e-05,
+      "loss": 1.5535,
+      "step": 2580
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.909270216962525e-05,
+      "loss": 1.6118,
+      "step": 2581
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.9072978303747535e-05,
+      "loss": 1.6119,
+      "step": 2582
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.905325443786982e-05,
+      "loss": 1.5245,
+      "step": 2583
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.903353057199211e-05,
+      "loss": 1.5387,
+      "step": 2584
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.90138067061144e-05,
+      "loss": 1.5857,
+      "step": 2585
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.899408284023669e-05,
+      "loss": 1.5631,
+      "step": 2586
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.8974358974358975e-05,
+      "loss": 1.6353,
+      "step": 2587
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.895463510848127e-05,
+      "loss": 1.5697,
+      "step": 2588
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.893491124260355e-05,
+      "loss": 1.6323,
+      "step": 2589
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.891518737672584e-05,
+      "loss": 1.6082,
+      "step": 2590
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.889546351084813e-05,
+      "loss": 1.6102,
+      "step": 2591
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.8875739644970416e-05,
+      "loss": 1.543,
+      "step": 2592
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.885601577909271e-05,
+      "loss": 1.5658,
+      "step": 2593
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.883629191321499e-05,
+      "loss": 1.611,
+      "step": 2594
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.881656804733728e-05,
+      "loss": 1.5021,
+      "step": 2595
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.8796844181459564e-05,
+      "loss": 1.5667,
+      "step": 2596
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.8777120315581856e-05,
+      "loss": 1.5666,
+      "step": 2597
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.875739644970415e-05,
+      "loss": 1.6753,
+      "step": 2598
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.873767258382643e-05,
+      "loss": 1.592,
+      "step": 2599
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 1.5678,
+      "step": 2600
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.8698224852071004e-05,
+      "loss": 1.5199,
+      "step": 2601
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.8678500986193296e-05,
+      "loss": 1.594,
+      "step": 2602
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.865877712031559e-05,
+      "loss": 1.6114,
+      "step": 2603
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.8639053254437874e-05,
+      "loss": 1.6616,
+      "step": 2604
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.861932938856016e-05,
+      "loss": 1.6126,
+      "step": 2605
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.8599605522682444e-05,
+      "loss": 1.5815,
+      "step": 2606
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.8579881656804736e-05,
+      "loss": 1.5875,
+      "step": 2607
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.856015779092702e-05,
+      "loss": 1.607,
+      "step": 2608
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.8540433925049314e-05,
+      "loss": 1.4878,
+      "step": 2609
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.85207100591716e-05,
+      "loss": 1.5974,
+      "step": 2610
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.850098619329389e-05,
+      "loss": 1.58,
+      "step": 2611
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.848126232741618e-05,
+      "loss": 1.5516,
+      "step": 2612
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.846153846153846e-05,
+      "loss": 1.6167,
+      "step": 2613
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.844181459566075e-05,
+      "loss": 1.5412,
+      "step": 2614
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.842209072978304e-05,
+      "loss": 1.5168,
+      "step": 2615
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.840236686390533e-05,
+      "loss": 1.5186,
+      "step": 2616
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.838264299802762e-05,
+      "loss": 1.6266,
+      "step": 2617
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.83629191321499e-05,
+      "loss": 1.5566,
+      "step": 2618
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.834319526627219e-05,
+      "loss": 1.5933,
+      "step": 2619
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.832347140039448e-05,
+      "loss": 1.6123,
+      "step": 2620
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.830374753451677e-05,
+      "loss": 1.6646,
+      "step": 2621
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.828402366863906e-05,
+      "loss": 1.6354,
+      "step": 2622
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.826429980276134e-05,
+      "loss": 1.6297,
+      "step": 2623
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.824457593688363e-05,
+      "loss": 1.541,
+      "step": 2624
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.822485207100592e-05,
+      "loss": 1.6938,
+      "step": 2625
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.8205128205128205e-05,
+      "loss": 1.6025,
+      "step": 2626
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.81854043392505e-05,
+      "loss": 1.5547,
+      "step": 2627
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.816568047337278e-05,
+      "loss": 1.6419,
+      "step": 2628
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.8145956607495075e-05,
+      "loss": 1.6041,
+      "step": 2629
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.812623274161736e-05,
+      "loss": 1.5637,
+      "step": 2630
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.8106508875739645e-05,
+      "loss": 1.5592,
+      "step": 2631
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.808678500986194e-05,
+      "loss": 1.668,
+      "step": 2632
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.806706114398422e-05,
+      "loss": 1.6336,
+      "step": 2633
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.8047337278106515e-05,
+      "loss": 1.5639,
+      "step": 2634
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.80276134122288e-05,
+      "loss": 1.6436,
+      "step": 2635
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.8007889546351086e-05,
+      "loss": 1.6108,
+      "step": 2636
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.798816568047337e-05,
+      "loss": 1.583,
+      "step": 2637
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.796844181459566e-05,
+      "loss": 1.5617,
+      "step": 2638
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.7948717948717955e-05,
+      "loss": 1.6691,
+      "step": 2639
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.792899408284024e-05,
+      "loss": 1.6223,
+      "step": 2640
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.7909270216962526e-05,
+      "loss": 1.6264,
+      "step": 2641
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.788954635108481e-05,
+      "loss": 1.5853,
+      "step": 2642
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.7869822485207103e-05,
+      "loss": 1.5986,
+      "step": 2643
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.785009861932939e-05,
+      "loss": 1.5375,
+      "step": 2644
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.783037475345168e-05,
+      "loss": 1.5646,
+      "step": 2645
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.7810650887573966e-05,
+      "loss": 1.5274,
+      "step": 2646
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.779092702169625e-05,
+      "loss": 1.6055,
+      "step": 2647
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.7771203155818544e-05,
+      "loss": 1.5748,
+      "step": 2648
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.775147928994083e-05,
+      "loss": 1.5891,
+      "step": 2649
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.773175542406312e-05,
+      "loss": 1.5155,
+      "step": 2650
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.7712031558185406e-05,
+      "loss": 1.6189,
+      "step": 2651
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.76923076923077e-05,
+      "loss": 1.5857,
+      "step": 2652
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.7672583826429984e-05,
+      "loss": 1.6701,
+      "step": 2653
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.765285996055227e-05,
+      "loss": 1.5406,
+      "step": 2654
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.7633136094674555e-05,
+      "loss": 1.5513,
+      "step": 2655
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.761341222879685e-05,
+      "loss": 1.6146,
+      "step": 2656
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.759368836291914e-05,
+      "loss": 1.5121,
+      "step": 2657
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.7573964497041424e-05,
+      "loss": 1.5404,
+      "step": 2658
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.755424063116371e-05,
+      "loss": 1.5345,
+      "step": 2659
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.7534516765285995e-05,
+      "loss": 1.5665,
+      "step": 2660
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.751479289940829e-05,
+      "loss": 1.6441,
+      "step": 2661
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.749506903353058e-05,
+      "loss": 1.6337,
+      "step": 2662
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.7475345167652864e-05,
+      "loss": 1.59,
+      "step": 2663
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.745562130177515e-05,
+      "loss": 1.636,
+      "step": 2664
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 1.6231,
+      "step": 2665
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.741617357001973e-05,
+      "loss": 1.5769,
+      "step": 2666
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.739644970414201e-05,
+      "loss": 1.5699,
+      "step": 2667
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.7376725838264305e-05,
+      "loss": 1.5383,
+      "step": 2668
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.735700197238659e-05,
+      "loss": 1.5106,
+      "step": 2669
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.7337278106508875e-05,
+      "loss": 1.5785,
+      "step": 2670
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.731755424063117e-05,
+      "loss": 1.578,
+      "step": 2671
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.729783037475345e-05,
+      "loss": 1.6099,
+      "step": 2672
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.727810650887574e-05,
+      "loss": 1.5263,
+      "step": 2673
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.725838264299803e-05,
+      "loss": 1.6194,
+      "step": 2674
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.723865877712032e-05,
+      "loss": 1.6472,
+      "step": 2675
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.721893491124261e-05,
+      "loss": 1.5677,
+      "step": 2676
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.719921104536489e-05,
+      "loss": 1.5994,
+      "step": 2677
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.717948717948718e-05,
+      "loss": 1.5242,
+      "step": 2678
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.715976331360947e-05,
+      "loss": 1.5719,
+      "step": 2679
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.714003944773176e-05,
+      "loss": 1.5558,
+      "step": 2680
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.712031558185405e-05,
+      "loss": 1.4538,
+      "step": 2681
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.710059171597633e-05,
+      "loss": 1.5263,
+      "step": 2682
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.708086785009862e-05,
+      "loss": 1.5961,
+      "step": 2683
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.706114398422091e-05,
+      "loss": 1.5931,
+      "step": 2684
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.7041420118343196e-05,
+      "loss": 1.562,
+      "step": 2685
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.702169625246549e-05,
+      "loss": 1.6856,
+      "step": 2686
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.7001972386587773e-05,
+      "loss": 1.589,
+      "step": 2687
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.698224852071006e-05,
+      "loss": 1.5622,
+      "step": 2688
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.696252465483235e-05,
+      "loss": 1.6076,
+      "step": 2689
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6942800788954636e-05,
+      "loss": 1.5278,
+      "step": 2690
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.692307692307693e-05,
+      "loss": 1.6589,
+      "step": 2691
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6903353057199214e-05,
+      "loss": 1.6152,
+      "step": 2692
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.68836291913215e-05,
+      "loss": 1.5556,
+      "step": 2693
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.686390532544379e-05,
+      "loss": 1.4502,
+      "step": 2694
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6844181459566076e-05,
+      "loss": 1.6657,
+      "step": 2695
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.682445759368836e-05,
+      "loss": 1.5439,
+      "step": 2696
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6804733727810654e-05,
+      "loss": 1.5983,
+      "step": 2697
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6785009861932946e-05,
+      "loss": 1.64,
+      "step": 2698
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.676528599605523e-05,
+      "loss": 1.6293,
+      "step": 2699
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.674556213017752e-05,
+      "loss": 1.6083,
+      "step": 2700
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.67258382642998e-05,
+      "loss": 1.5188,
+      "step": 2701
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.670611439842209e-05,
+      "loss": 1.645,
+      "step": 2702
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.668639053254438e-05,
+      "loss": 1.5841,
+      "step": 2703
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 1.66,
+      "step": 2704
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.664694280078896e-05,
+      "loss": 1.6348,
+      "step": 2705
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.662721893491124e-05,
+      "loss": 1.6642,
+      "step": 2706
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.6607495069033534e-05,
+      "loss": 1.6226,
+      "step": 2707
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.658777120315582e-05,
+      "loss": 1.5641,
+      "step": 2708
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.656804733727811e-05,
+      "loss": 1.5417,
+      "step": 2709
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.65483234714004e-05,
+      "loss": 1.5438,
+      "step": 2710
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.652859960552268e-05,
+      "loss": 1.5069,
+      "step": 2711
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.6508875739644975e-05,
+      "loss": 1.5806,
+      "step": 2712
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.648915187376726e-05,
+      "loss": 1.5709,
+      "step": 2713
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.6469428007889545e-05,
+      "loss": 1.489,
+      "step": 2714
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.644970414201184e-05,
+      "loss": 1.487,
+      "step": 2715
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.642998027613412e-05,
+      "loss": 1.5618,
+      "step": 2716
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.6410256410256415e-05,
+      "loss": 1.6095,
+      "step": 2717
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.63905325443787e-05,
+      "loss": 1.4637,
+      "step": 2718
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.6370808678500986e-05,
+      "loss": 1.6213,
+      "step": 2719
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.635108481262328e-05,
+      "loss": 1.612,
+      "step": 2720
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.633136094674557e-05,
+      "loss": 1.6858,
+      "step": 2721
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.6311637080867855e-05,
+      "loss": 1.5221,
+      "step": 2722
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.629191321499014e-05,
+      "loss": 1.602,
+      "step": 2723
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.6272189349112426e-05,
+      "loss": 1.5824,
+      "step": 2724
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.625246548323471e-05,
+      "loss": 1.5974,
+      "step": 2725
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.6232741617357e-05,
+      "loss": 1.4844,
+      "step": 2726
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.6213017751479295e-05,
+      "loss": 1.5841,
+      "step": 2727
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.619329388560158e-05,
+      "loss": 1.6699,
+      "step": 2728
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.6173570019723866e-05,
+      "loss": 1.5632,
+      "step": 2729
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.5179,
+      "step": 2730
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.6134122287968443e-05,
+      "loss": 1.5051,
+      "step": 2731
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.611439842209073e-05,
+      "loss": 1.5874,
+      "step": 2732
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.609467455621302e-05,
+      "loss": 1.5707,
+      "step": 2733
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.6074950690335306e-05,
+      "loss": 1.502,
+      "step": 2734
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.60552268244576e-05,
+      "loss": 1.5094,
+      "step": 2735
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.6035502958579884e-05,
+      "loss": 1.5704,
+      "step": 2736
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.601577909270217e-05,
+      "loss": 1.6142,
+      "step": 2737
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.599605522682446e-05,
+      "loss": 1.6074,
+      "step": 2738
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.597633136094675e-05,
+      "loss": 1.563,
+      "step": 2739
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.595660749506904e-05,
+      "loss": 1.5575,
+      "step": 2740
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.5936883629191324e-05,
+      "loss": 1.5769,
+      "step": 2741
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.591715976331361e-05,
+      "loss": 1.503,
+      "step": 2742
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.5897435897435895e-05,
+      "loss": 1.533,
+      "step": 2743
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.587771203155819e-05,
+      "loss": 1.5731,
+      "step": 2744
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.585798816568048e-05,
+      "loss": 1.5514,
+      "step": 2745
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.5838264299802764e-05,
+      "loss": 1.5874,
+      "step": 2746
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.581854043392505e-05,
+      "loss": 1.6034,
+      "step": 2747
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.5798816568047335e-05,
+      "loss": 1.5345,
+      "step": 2748
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.577909270216963e-05,
+      "loss": 1.5036,
+      "step": 2749
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.575936883629192e-05,
+      "loss": 1.5935,
+      "step": 2750
+    },
+    {
+      "epoch": 8.14,
+      "eval_loss": 1.4533615112304688,
+      "eval_runtime": 69.3456,
+      "eval_samples_per_second": 321.145,
+      "eval_steps_per_second": 1.673,
+      "step": 2750
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5739644970414204e-05,
+      "loss": 1.5437,
+      "step": 2751
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.571992110453649e-05,
+      "loss": 1.5314,
+      "step": 2752
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.570019723865878e-05,
+      "loss": 1.4844,
+      "step": 2753
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.568047337278107e-05,
+      "loss": 1.6147,
+      "step": 2754
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.566074950690335e-05,
+      "loss": 1.6048,
+      "step": 2755
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5641025641025645e-05,
+      "loss": 1.6355,
+      "step": 2756
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.562130177514793e-05,
+      "loss": 1.5588,
+      "step": 2757
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.560157790927022e-05,
+      "loss": 1.638,
+      "step": 2758
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.558185404339251e-05,
+      "loss": 1.6198,
+      "step": 2759
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.556213017751479e-05,
+      "loss": 1.5291,
+      "step": 2760
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.554240631163708e-05,
+      "loss": 1.568,
+      "step": 2761
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.552268244575938e-05,
+      "loss": 1.5123,
+      "step": 2762
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.550295857988166e-05,
+      "loss": 1.5304,
+      "step": 2763
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.548323471400395e-05,
+      "loss": 1.5373,
+      "step": 2764
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.546351084812623e-05,
+      "loss": 1.5204,
+      "step": 2765
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.544378698224852e-05,
+      "loss": 1.5549,
+      "step": 2766
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.542406311637081e-05,
+      "loss": 1.5293,
+      "step": 2767
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.54043392504931e-05,
+      "loss": 1.6036,
+      "step": 2768
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.538461538461539e-05,
+      "loss": 1.5415,
+      "step": 2769
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.536489151873767e-05,
+      "loss": 1.4765,
+      "step": 2770
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.534516765285996e-05,
+      "loss": 1.6607,
+      "step": 2771
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.532544378698225e-05,
+      "loss": 1.5772,
+      "step": 2772
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5305719921104536e-05,
+      "loss": 1.6257,
+      "step": 2773
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.528599605522683e-05,
+      "loss": 1.4935,
+      "step": 2774
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.5266272189349114e-05,
+      "loss": 1.5563,
+      "step": 2775
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.5246548323471406e-05,
+      "loss": 1.6774,
+      "step": 2776
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.522682445759369e-05,
+      "loss": 1.5793,
+      "step": 2777
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5207100591715976e-05,
+      "loss": 1.6289,
+      "step": 2778
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.518737672583827e-05,
+      "loss": 1.631,
+      "step": 2779
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5167652859960554e-05,
+      "loss": 1.5763,
+      "step": 2780
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5147928994082846e-05,
+      "loss": 1.5766,
+      "step": 2781
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.512820512820513e-05,
+      "loss": 1.6216,
+      "step": 2782
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5108481262327417e-05,
+      "loss": 1.5911,
+      "step": 2783
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.50887573964497e-05,
+      "loss": 1.6,
+      "step": 2784
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5069033530571994e-05,
+      "loss": 1.5748,
+      "step": 2785
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5049309664694286e-05,
+      "loss": 1.6229,
+      "step": 2786
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.502958579881657e-05,
+      "loss": 1.6346,
+      "step": 2787
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.500986193293886e-05,
+      "loss": 1.5731,
+      "step": 2788
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.499013806706114e-05,
+      "loss": 1.5048,
+      "step": 2789
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.4970414201183434e-05,
+      "loss": 1.5813,
+      "step": 2790
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.495069033530572e-05,
+      "loss": 1.6381,
+      "step": 2791
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.493096646942801e-05,
+      "loss": 1.5002,
+      "step": 2792
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.49112426035503e-05,
+      "loss": 1.5372,
+      "step": 2793
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.489151873767258e-05,
+      "loss": 1.5644,
+      "step": 2794
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 1.6434,
+      "step": 2795
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.485207100591716e-05,
+      "loss": 1.5682,
+      "step": 2796
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.483234714003945e-05,
+      "loss": 1.6144,
+      "step": 2797
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.481262327416174e-05,
+      "loss": 1.4969,
+      "step": 2798
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.479289940828403e-05,
+      "loss": 1.5724,
+      "step": 2799
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.4773175542406315e-05,
+      "loss": 1.558,
+      "step": 2800
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.47534516765286e-05,
+      "loss": 1.6054,
+      "step": 2801
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.4733727810650885e-05,
+      "loss": 1.5862,
+      "step": 2802
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.471400394477318e-05,
+      "loss": 1.5659,
+      "step": 2803
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.469428007889547e-05,
+      "loss": 1.5996,
+      "step": 2804
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.4674556213017755e-05,
+      "loss": 1.5771,
+      "step": 2805
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.465483234714004e-05,
+      "loss": 1.6214,
+      "step": 2806
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.4635108481262326e-05,
+      "loss": 1.6571,
+      "step": 2807
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.461538461538462e-05,
+      "loss": 1.6018,
+      "step": 2808
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.459566074950691e-05,
+      "loss": 1.548,
+      "step": 2809
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.4575936883629195e-05,
+      "loss": 1.4964,
+      "step": 2810
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.455621301775148e-05,
+      "loss": 1.489,
+      "step": 2811
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.4536489151873766e-05,
+      "loss": 1.5784,
+      "step": 2812
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.451676528599606e-05,
+      "loss": 1.4835,
+      "step": 2813
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.449704142011834e-05,
+      "loss": 1.5982,
+      "step": 2814
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.4477317554240635e-05,
+      "loss": 1.6174,
+      "step": 2815
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.445759368836292e-05,
+      "loss": 1.6509,
+      "step": 2816
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.443786982248521e-05,
+      "loss": 1.5494,
+      "step": 2817
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.44181459566075e-05,
+      "loss": 1.6078,
+      "step": 2818
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.4398422090729784e-05,
+      "loss": 1.685,
+      "step": 2819
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.437869822485207e-05,
+      "loss": 1.6272,
+      "step": 2820
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.435897435897436e-05,
+      "loss": 1.5914,
+      "step": 2821
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.433925049309665e-05,
+      "loss": 1.645,
+      "step": 2822
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.431952662721894e-05,
+      "loss": 1.582,
+      "step": 2823
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.4299802761341224e-05,
+      "loss": 1.5965,
+      "step": 2824
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.428007889546351e-05,
+      "loss": 1.4944,
+      "step": 2825
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.42603550295858e-05,
+      "loss": 1.6258,
+      "step": 2826
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.424063116370809e-05,
+      "loss": 1.5329,
+      "step": 2827
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.422090729783038e-05,
+      "loss": 1.5792,
+      "step": 2828
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.4201183431952664e-05,
+      "loss": 1.6054,
+      "step": 2829
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.418145956607495e-05,
+      "loss": 1.5927,
+      "step": 2830
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.416173570019724e-05,
+      "loss": 1.6204,
+      "step": 2831
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.414201183431953e-05,
+      "loss": 1.6657,
+      "step": 2832
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.412228796844182e-05,
+      "loss": 1.6263,
+      "step": 2833
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.4102564102564104e-05,
+      "loss": 1.6835,
+      "step": 2834
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.408284023668639e-05,
+      "loss": 1.6444,
+      "step": 2835
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.406311637080868e-05,
+      "loss": 1.4505,
+      "step": 2836
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.404339250493097e-05,
+      "loss": 1.5611,
+      "step": 2837
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.402366863905326e-05,
+      "loss": 1.5271,
+      "step": 2838
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.4003944773175544e-05,
+      "loss": 1.6081,
+      "step": 2839
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.398422090729784e-05,
+      "loss": 1.5489,
+      "step": 2840
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.396449704142012e-05,
+      "loss": 1.5124,
+      "step": 2841
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.394477317554241e-05,
+      "loss": 1.4926,
+      "step": 2842
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.392504930966469e-05,
+      "loss": 1.5619,
+      "step": 2843
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.3905325443786985e-05,
+      "loss": 1.5936,
+      "step": 2844
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.388560157790928e-05,
+      "loss": 1.5662,
+      "step": 2845
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.386587771203156e-05,
+      "loss": 1.523,
+      "step": 2846
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.384615384615385e-05,
+      "loss": 1.559,
+      "step": 2847
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.382642998027613e-05,
+      "loss": 1.5933,
+      "step": 2848
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.380670611439842e-05,
+      "loss": 1.5559,
+      "step": 2849
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.378698224852072e-05,
+      "loss": 1.5346,
+      "step": 2850
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.3767258382643e-05,
+      "loss": 1.5999,
+      "step": 2851
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.374753451676529e-05,
+      "loss": 1.5506,
+      "step": 2852
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.372781065088757e-05,
+      "loss": 1.5293,
+      "step": 2853
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.3708086785009865e-05,
+      "loss": 1.5117,
+      "step": 2854
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.368836291913215e-05,
+      "loss": 1.63,
+      "step": 2855
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.366863905325444e-05,
+      "loss": 1.6238,
+      "step": 2856
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.364891518737673e-05,
+      "loss": 1.5829,
+      "step": 2857
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.362919132149901e-05,
+      "loss": 1.5539,
+      "step": 2858
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.3609467455621305e-05,
+      "loss": 1.5503,
+      "step": 2859
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 1.5541,
+      "step": 2860
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.3570019723865876e-05,
+      "loss": 1.584,
+      "step": 2861
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.355029585798817e-05,
+      "loss": 1.4504,
+      "step": 2862
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.353057199211046e-05,
+      "loss": 1.4238,
+      "step": 2863
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.3510848126232746e-05,
+      "loss": 1.5022,
+      "step": 2864
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.349112426035503e-05,
+      "loss": 1.5196,
+      "step": 2865
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.3471400394477316e-05,
+      "loss": 1.5495,
+      "step": 2866
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.345167652859961e-05,
+      "loss": 1.5647,
+      "step": 2867
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.34319526627219e-05,
+      "loss": 1.4863,
+      "step": 2868
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.3412228796844186e-05,
+      "loss": 1.6388,
+      "step": 2869
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.339250493096647e-05,
+      "loss": 1.6072,
+      "step": 2870
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.3372781065088757e-05,
+      "loss": 1.4989,
+      "step": 2871
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.335305719921105e-05,
+      "loss": 1.4485,
+      "step": 2872
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 1.5938,
+      "step": 2873
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.3313609467455626e-05,
+      "loss": 1.4972,
+      "step": 2874
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.329388560157791e-05,
+      "loss": 1.68,
+      "step": 2875
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.32741617357002e-05,
+      "loss": 1.5847,
+      "step": 2876
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.325443786982249e-05,
+      "loss": 1.4753,
+      "step": 2877
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.3234714003944774e-05,
+      "loss": 1.6212,
+      "step": 2878
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.321499013806706e-05,
+      "loss": 1.4954,
+      "step": 2879
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.319526627218935e-05,
+      "loss": 1.4881,
+      "step": 2880
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.317554240631164e-05,
+      "loss": 1.61,
+      "step": 2881
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.315581854043393e-05,
+      "loss": 1.5301,
+      "step": 2882
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.3136094674556215e-05,
+      "loss": 1.5835,
+      "step": 2883
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.31163708086785e-05,
+      "loss": 1.4849,
+      "step": 2884
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.309664694280079e-05,
+      "loss": 1.5922,
+      "step": 2885
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.3076923076923084e-05,
+      "loss": 1.6262,
+      "step": 2886
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.305719921104537e-05,
+      "loss": 1.5763,
+      "step": 2887
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.3037475345167655e-05,
+      "loss": 1.5266,
+      "step": 2888
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.301775147928994e-05,
+      "loss": 1.6227,
+      "step": 2889
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.2998027613412225e-05,
+      "loss": 1.4404,
+      "step": 2890
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.297830374753452e-05,
+      "loss": 1.5905,
+      "step": 2891
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.295857988165681e-05,
+      "loss": 1.5076,
+      "step": 2892
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.2938856015779095e-05,
+      "loss": 1.5658,
+      "step": 2893
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.291913214990138e-05,
+      "loss": 1.6037,
+      "step": 2894
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.289940828402367e-05,
+      "loss": 1.4758,
+      "step": 2895
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.287968441814596e-05,
+      "loss": 1.6158,
+      "step": 2896
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.285996055226825e-05,
+      "loss": 1.5917,
+      "step": 2897
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.2840236686390535e-05,
+      "loss": 1.5032,
+      "step": 2898
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.282051282051282e-05,
+      "loss": 1.4997,
+      "step": 2899
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.280078895463511e-05,
+      "loss": 1.5596,
+      "step": 2900
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.27810650887574e-05,
+      "loss": 1.4513,
+      "step": 2901
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.276134122287968e-05,
+      "loss": 1.607,
+      "step": 2902
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.2741617357001975e-05,
+      "loss": 1.5855,
+      "step": 2903
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.272189349112426e-05,
+      "loss": 1.6101,
+      "step": 2904
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.270216962524655e-05,
+      "loss": 1.556,
+      "step": 2905
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.268244575936884e-05,
+      "loss": 1.481,
+      "step": 2906
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.2662721893491124e-05,
+      "loss": 1.4872,
+      "step": 2907
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.264299802761341e-05,
+      "loss": 1.5988,
+      "step": 2908
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.262327416173571e-05,
+      "loss": 1.5171,
+      "step": 2909
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.260355029585799e-05,
+      "loss": 1.5211,
+      "step": 2910
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.258382642998028e-05,
+      "loss": 1.5506,
+      "step": 2911
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.2564102564102564e-05,
+      "loss": 1.5946,
+      "step": 2912
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.254437869822485e-05,
+      "loss": 1.5533,
+      "step": 2913
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.252465483234714e-05,
+      "loss": 1.5597,
+      "step": 2914
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.2504930966469433e-05,
+      "loss": 1.5213,
+      "step": 2915
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.248520710059172e-05,
+      "loss": 1.5775,
+      "step": 2916
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.2465483234714004e-05,
+      "loss": 1.6257,
+      "step": 2917
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.2445759368836296e-05,
+      "loss": 1.4139,
+      "step": 2918
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.242603550295858e-05,
+      "loss": 1.5957,
+      "step": 2919
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.240631163708087e-05,
+      "loss": 1.5521,
+      "step": 2920
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.238658777120316e-05,
+      "loss": 1.6679,
+      "step": 2921
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.2366863905325444e-05,
+      "loss": 1.573,
+      "step": 2922
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.2347140039447736e-05,
+      "loss": 1.5343,
+      "step": 2923
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.232741617357002e-05,
+      "loss": 1.5355,
+      "step": 2924
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 1.6277,
+      "step": 2925
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.22879684418146e-05,
+      "loss": 1.6213,
+      "step": 2926
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.226824457593689e-05,
+      "loss": 1.5211,
+      "step": 2927
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.224852071005918e-05,
+      "loss": 1.6396,
+      "step": 2928
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.222879684418146e-05,
+      "loss": 1.5625,
+      "step": 2929
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.220907297830375e-05,
+      "loss": 1.4434,
+      "step": 2930
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.218934911242603e-05,
+      "loss": 1.6081,
+      "step": 2931
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.2169625246548325e-05,
+      "loss": 1.6177,
+      "step": 2932
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.214990138067062e-05,
+      "loss": 1.5705,
+      "step": 2933
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.21301775147929e-05,
+      "loss": 1.6803,
+      "step": 2934
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.211045364891519e-05,
+      "loss": 1.5303,
+      "step": 2935
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.209072978303747e-05,
+      "loss": 1.5839,
+      "step": 2936
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2071005917159765e-05,
+      "loss": 1.5507,
+      "step": 2937
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.205128205128206e-05,
+      "loss": 1.5969,
+      "step": 2938
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.203155818540434e-05,
+      "loss": 1.5103,
+      "step": 2939
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.201183431952663e-05,
+      "loss": 1.562,
+      "step": 2940
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.199211045364892e-05,
+      "loss": 1.4152,
+      "step": 2941
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.1972386587771205e-05,
+      "loss": 1.5368,
+      "step": 2942
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.195266272189349e-05,
+      "loss": 1.509,
+      "step": 2943
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.193293885601578e-05,
+      "loss": 1.5795,
+      "step": 2944
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.191321499013807e-05,
+      "loss": 1.6065,
+      "step": 2945
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.189349112426036e-05,
+      "loss": 1.527,
+      "step": 2946
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1873767258382646e-05,
+      "loss": 1.4939,
+      "step": 2947
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.185404339250493e-05,
+      "loss": 1.5622,
+      "step": 2948
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1834319526627216e-05,
+      "loss": 1.5788,
+      "step": 2949
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.181459566074951e-05,
+      "loss": 1.5973,
+      "step": 2950
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.17948717948718e-05,
+      "loss": 1.5017,
+      "step": 2951
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1775147928994086e-05,
+      "loss": 1.6045,
+      "step": 2952
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.175542406311637e-05,
+      "loss": 1.6384,
+      "step": 2953
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1735700197238656e-05,
+      "loss": 1.5073,
+      "step": 2954
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.171597633136095e-05,
+      "loss": 1.5103,
+      "step": 2955
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.169625246548324e-05,
+      "loss": 1.5249,
+      "step": 2956
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.1676528599605526e-05,
+      "loss": 1.5459,
+      "step": 2957
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.165680473372781e-05,
+      "loss": 1.5855,
+      "step": 2958
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.16370808678501e-05,
+      "loss": 1.6178,
+      "step": 2959
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.161735700197239e-05,
+      "loss": 1.5156,
+      "step": 2960
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.1597633136094674e-05,
+      "loss": 1.4776,
+      "step": 2961
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.1577909270216966e-05,
+      "loss": 1.557,
+      "step": 2962
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.155818540433925e-05,
+      "loss": 1.6104,
+      "step": 2963
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.1538461538461544e-05,
+      "loss": 1.5765,
+      "step": 2964
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.151873767258383e-05,
+      "loss": 1.5608,
+      "step": 2965
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.1499013806706114e-05,
+      "loss": 1.5848,
+      "step": 2966
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.14792899408284e-05,
+      "loss": 1.5948,
+      "step": 2967
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.145956607495069e-05,
+      "loss": 1.5428,
+      "step": 2968
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.1439842209072984e-05,
+      "loss": 1.5041,
+      "step": 2969
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.142011834319527e-05,
+      "loss": 1.5674,
+      "step": 2970
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.1400394477317555e-05,
+      "loss": 1.5568,
+      "step": 2971
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.138067061143984e-05,
+      "loss": 1.5431,
+      "step": 2972
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.136094674556213e-05,
+      "loss": 1.5963,
+      "step": 2973
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.1341222879684424e-05,
+      "loss": 1.5339,
+      "step": 2974
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.132149901380671e-05,
+      "loss": 1.5813,
+      "step": 2975
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.1301775147928995e-05,
+      "loss": 1.619,
+      "step": 2976
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.128205128205128e-05,
+      "loss": 1.4316,
+      "step": 2977
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.126232741617357e-05,
+      "loss": 1.5671,
+      "step": 2978
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.124260355029586e-05,
+      "loss": 1.4708,
+      "step": 2979
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.122287968441815e-05,
+      "loss": 1.5883,
+      "step": 2980
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.1203155818540435e-05,
+      "loss": 1.5604,
+      "step": 2981
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.118343195266273e-05,
+      "loss": 1.5283,
+      "step": 2982
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.116370808678501e-05,
+      "loss": 1.6185,
+      "step": 2983
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.11439842209073e-05,
+      "loss": 1.4773,
+      "step": 2984
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.112426035502959e-05,
+      "loss": 1.5274,
+      "step": 2985
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.1104536489151875e-05,
+      "loss": 1.6186,
+      "step": 2986
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.108481262327417e-05,
+      "loss": 1.6322,
+      "step": 2987
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.106508875739645e-05,
+      "loss": 1.5911,
+      "step": 2988
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.104536489151874e-05,
+      "loss": 1.5531,
+      "step": 2989
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 1.4761,
+      "step": 2990
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.1005917159763316e-05,
+      "loss": 1.4432,
+      "step": 2991
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.098619329388561e-05,
+      "loss": 1.5329,
+      "step": 2992
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.096646942800789e-05,
+      "loss": 1.5849,
+      "step": 2993
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.094674556213018e-05,
+      "loss": 1.626,
+      "step": 2994
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.0927021696252464e-05,
+      "loss": 1.519,
+      "step": 2995
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.0907297830374756e-05,
+      "loss": 1.5259,
+      "step": 2996
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.088757396449705e-05,
+      "loss": 1.5248,
+      "step": 2997
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.086785009861933e-05,
+      "loss": 1.6588,
+      "step": 2998
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.084812623274162e-05,
+      "loss": 1.5237,
+      "step": 2999
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.0828402366863904e-05,
+      "loss": 1.5347,
+      "step": 3000
+    },
+    {
+      "epoch": 8.88,
+      "eval_loss": 1.4382072687149048,
+      "eval_runtime": 69.3939,
+      "eval_samples_per_second": 320.922,
+      "eval_steps_per_second": 1.672,
+      "step": 3000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.0808678500986196e-05,
+      "loss": 1.5793,
+      "step": 3001
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.078895463510848e-05,
+      "loss": 1.6102,
+      "step": 3002
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.0769230769230773e-05,
+      "loss": 1.446,
+      "step": 3003
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.074950690335306e-05,
+      "loss": 1.5826,
+      "step": 3004
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.072978303747535e-05,
+      "loss": 1.5602,
+      "step": 3005
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.0710059171597636e-05,
+      "loss": 1.5756,
+      "step": 3006
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.069033530571992e-05,
+      "loss": 1.5139,
+      "step": 3007
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.067061143984221e-05,
+      "loss": 1.6895,
+      "step": 3008
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.06508875739645e-05,
+      "loss": 1.5488,
+      "step": 3009
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.063116370808679e-05,
+      "loss": 1.5559,
+      "step": 3010
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.0611439842209077e-05,
+      "loss": 1.4673,
+      "step": 3011
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.059171597633136e-05,
+      "loss": 1.5717,
+      "step": 3012
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.057199211045365e-05,
+      "loss": 1.4654,
+      "step": 3013
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.055226824457594e-05,
+      "loss": 1.5716,
+      "step": 3014
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.053254437869823e-05,
+      "loss": 1.5812,
+      "step": 3015
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.051282051282052e-05,
+      "loss": 1.6005,
+      "step": 3016
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.04930966469428e-05,
+      "loss": 1.5633,
+      "step": 3017
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.047337278106509e-05,
+      "loss": 1.6101,
+      "step": 3018
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.045364891518738e-05,
+      "loss": 1.5105,
+      "step": 3019
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.0433925049309665e-05,
+      "loss": 1.6047,
+      "step": 3020
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.041420118343196e-05,
+      "loss": 1.4727,
+      "step": 3021
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.039447731755424e-05,
+      "loss": 1.6329,
+      "step": 3022
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.037475345167653e-05,
+      "loss": 1.5104,
+      "step": 3023
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.035502958579882e-05,
+      "loss": 1.5747,
+      "step": 3024
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.0335305719921105e-05,
+      "loss": 1.555,
+      "step": 3025
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.031558185404339e-05,
+      "loss": 1.5769,
+      "step": 3026
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.029585798816568e-05,
+      "loss": 1.6514,
+      "step": 3027
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.0276134122287975e-05,
+      "loss": 1.5405,
+      "step": 3028
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.025641025641026e-05,
+      "loss": 1.5859,
+      "step": 3029
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.0236686390532545e-05,
+      "loss": 1.5197,
+      "step": 3030
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.021696252465483e-05,
+      "loss": 1.5322,
+      "step": 3031
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.019723865877712e-05,
+      "loss": 1.5638,
+      "step": 3032
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.0177514792899415e-05,
+      "loss": 1.5157,
+      "step": 3033
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.01577909270217e-05,
+      "loss": 1.5597,
+      "step": 3034
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.0138067061143986e-05,
+      "loss": 1.5804,
+      "step": 3035
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.011834319526627e-05,
+      "loss": 1.5735,
+      "step": 3036
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.009861932938856e-05,
+      "loss": 1.6214,
+      "step": 3037
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.007889546351085e-05,
+      "loss": 1.5735,
+      "step": 3038
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.005917159763314e-05,
+      "loss": 1.5336,
+      "step": 3039
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.0039447731755426e-05,
+      "loss": 1.4175,
+      "step": 3040
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.001972386587771e-05,
+      "loss": 1.6547,
+      "step": 3041
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4e-05,
+      "loss": 1.5879,
+      "step": 3042
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.998027613412229e-05,
+      "loss": 1.6097,
+      "step": 3043
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.996055226824458e-05,
+      "loss": 1.7091,
+      "step": 3044
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.9940828402366866e-05,
+      "loss": 1.5003,
+      "step": 3045
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.992110453648915e-05,
+      "loss": 1.4911,
+      "step": 3046
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.9901380670611444e-05,
+      "loss": 1.5997,
+      "step": 3047
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.988165680473373e-05,
+      "loss": 1.5976,
+      "step": 3048
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.9861932938856014e-05,
+      "loss": 1.5035,
+      "step": 3049
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.9842209072978306e-05,
+      "loss": 1.5185,
+      "step": 3050
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.98224852071006e-05,
+      "loss": 1.6087,
+      "step": 3051
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.9802761341222884e-05,
+      "loss": 1.4409,
+      "step": 3052
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.978303747534517e-05,
+      "loss": 1.4535,
+      "step": 3053
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.9763313609467454e-05,
+      "loss": 1.5369,
+      "step": 3054
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 1.5544,
+      "step": 3055
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.972386587771204e-05,
+      "loss": 1.5959,
+      "step": 3056
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.9704142011834324e-05,
+      "loss": 1.5999,
+      "step": 3057
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.968441814595661e-05,
+      "loss": 1.5381,
+      "step": 3058
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.9664694280078895e-05,
+      "loss": 1.5741,
+      "step": 3059
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.964497041420119e-05,
+      "loss": 1.5688,
+      "step": 3060
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.962524654832347e-05,
+      "loss": 1.4473,
+      "step": 3061
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.9605522682445764e-05,
+      "loss": 1.5307,
+      "step": 3062
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.958579881656805e-05,
+      "loss": 1.6018,
+      "step": 3063
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.9566074950690335e-05,
+      "loss": 1.4977,
+      "step": 3064
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.954635108481263e-05,
+      "loss": 1.4538,
+      "step": 3065
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.952662721893491e-05,
+      "loss": 1.5896,
+      "step": 3066
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.95069033530572e-05,
+      "loss": 1.5537,
+      "step": 3067
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.948717948717949e-05,
+      "loss": 1.5163,
+      "step": 3068
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.9467455621301775e-05,
+      "loss": 1.5602,
+      "step": 3069
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.944773175542407e-05,
+      "loss": 1.5689,
+      "step": 3070
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.942800788954635e-05,
+      "loss": 1.6069,
+      "step": 3071
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.940828402366864e-05,
+      "loss": 1.5643,
+      "step": 3072
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.938856015779093e-05,
+      "loss": 1.4874,
+      "step": 3073
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.936883629191322e-05,
+      "loss": 1.5542,
+      "step": 3074
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.934911242603551e-05,
+      "loss": 1.5281,
+      "step": 3075
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.932938856015779e-05,
+      "loss": 1.4737,
+      "step": 3076
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.930966469428008e-05,
+      "loss": 1.62,
+      "step": 3077
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.9289940828402364e-05,
+      "loss": 1.5474,
+      "step": 3078
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.9270216962524656e-05,
+      "loss": 1.5544,
+      "step": 3079
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.925049309664695e-05,
+      "loss": 1.5887,
+      "step": 3080
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.923076923076923e-05,
+      "loss": 1.548,
+      "step": 3081
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.921104536489152e-05,
+      "loss": 1.5453,
+      "step": 3082
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.919132149901381e-05,
+      "loss": 1.6304,
+      "step": 3083
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.9171597633136096e-05,
+      "loss": 1.5454,
+      "step": 3084
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.915187376725839e-05,
+      "loss": 1.545,
+      "step": 3085
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.913214990138067e-05,
+      "loss": 1.5868,
+      "step": 3086
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.911242603550296e-05,
+      "loss": 1.6242,
+      "step": 3087
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.909270216962525e-05,
+      "loss": 1.5425,
+      "step": 3088
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.9072978303747536e-05,
+      "loss": 1.5394,
+      "step": 3089
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.905325443786982e-05,
+      "loss": 1.5882,
+      "step": 3090
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.9033530571992114e-05,
+      "loss": 1.6046,
+      "step": 3091
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.9013806706114406e-05,
+      "loss": 1.58,
+      "step": 3092
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.899408284023669e-05,
+      "loss": 1.5476,
+      "step": 3093
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.8974358974358976e-05,
+      "loss": 1.5051,
+      "step": 3094
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.895463510848126e-05,
+      "loss": 1.5057,
+      "step": 3095
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.893491124260355e-05,
+      "loss": 1.575,
+      "step": 3096
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.891518737672584e-05,
+      "loss": 1.5495,
+      "step": 3097
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.889546351084813e-05,
+      "loss": 1.4767,
+      "step": 3098
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.8875739644970417e-05,
+      "loss": 1.5655,
+      "step": 3099
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.88560157790927e-05,
+      "loss": 1.5465,
+      "step": 3100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.883629191321499e-05,
+      "loss": 1.6056,
+      "step": 3101
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.881656804733728e-05,
+      "loss": 1.4579,
+      "step": 3102
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.879684418145957e-05,
+      "loss": 1.6231,
+      "step": 3103
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.877712031558186e-05,
+      "loss": 1.5228,
+      "step": 3104
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.875739644970414e-05,
+      "loss": 1.5912,
+      "step": 3105
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.8737672583826434e-05,
+      "loss": 1.572,
+      "step": 3106
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.871794871794872e-05,
+      "loss": 1.5337,
+      "step": 3107
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.8698224852071005e-05,
+      "loss": 1.4812,
+      "step": 3108
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.86785009861933e-05,
+      "loss": 1.5358,
+      "step": 3109
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.865877712031558e-05,
+      "loss": 1.5502,
+      "step": 3110
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.8639053254437874e-05,
+      "loss": 1.6087,
+      "step": 3111
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.861932938856016e-05,
+      "loss": 1.5066,
+      "step": 3112
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.8599605522682445e-05,
+      "loss": 1.5532,
+      "step": 3113
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.857988165680473e-05,
+      "loss": 1.6146,
+      "step": 3114
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.856015779092703e-05,
+      "loss": 1.4853,
+      "step": 3115
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.8540433925049315e-05,
+      "loss": 1.4852,
+      "step": 3116
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.85207100591716e-05,
+      "loss": 1.6478,
+      "step": 3117
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.8500986193293885e-05,
+      "loss": 1.5359,
+      "step": 3118
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.848126232741617e-05,
+      "loss": 1.5902,
+      "step": 3119
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.5082,
+      "step": 3120
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.8441814595660755e-05,
+      "loss": 1.4772,
+      "step": 3121
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.842209072978304e-05,
+      "loss": 1.5694,
+      "step": 3122
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.8402366863905326e-05,
+      "loss": 1.6885,
+      "step": 3123
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.838264299802761e-05,
+      "loss": 1.6287,
+      "step": 3124
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.83629191321499e-05,
+      "loss": 1.4981,
+      "step": 3125
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.834319526627219e-05,
+      "loss": 1.5669,
+      "step": 3126
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.832347140039448e-05,
+      "loss": 1.603,
+      "step": 3127
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.8303747534516766e-05,
+      "loss": 1.6242,
+      "step": 3128
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.828402366863906e-05,
+      "loss": 1.5587,
+      "step": 3129
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.826429980276134e-05,
+      "loss": 1.6115,
+      "step": 3130
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.824457593688363e-05,
+      "loss": 1.4985,
+      "step": 3131
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.822485207100592e-05,
+      "loss": 1.6299,
+      "step": 3132
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.8205128205128206e-05,
+      "loss": 1.5685,
+      "step": 3133
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.81854043392505e-05,
+      "loss": 1.5173,
+      "step": 3134
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.8165680473372784e-05,
+      "loss": 1.532,
+      "step": 3135
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.814595660749507e-05,
+      "loss": 1.4838,
+      "step": 3136
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.8126232741617354e-05,
+      "loss": 1.5372,
+      "step": 3137
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.8106508875739646e-05,
+      "loss": 1.6158,
+      "step": 3138
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.808678500986194e-05,
+      "loss": 1.5015,
+      "step": 3139
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.8067061143984224e-05,
+      "loss": 1.5996,
+      "step": 3140
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.804733727810651e-05,
+      "loss": 1.5879,
+      "step": 3141
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.8027613412228794e-05,
+      "loss": 1.5172,
+      "step": 3142
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.8007889546351087e-05,
+      "loss": 1.4851,
+      "step": 3143
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.798816568047338e-05,
+      "loss": 1.4757,
+      "step": 3144
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.7968441814595664e-05,
+      "loss": 1.4725,
+      "step": 3145
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.794871794871795e-05,
+      "loss": 1.4971,
+      "step": 3146
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.792899408284024e-05,
+      "loss": 1.5986,
+      "step": 3147
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.790927021696253e-05,
+      "loss": 1.5853,
+      "step": 3148
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.788954635108481e-05,
+      "loss": 1.5863,
+      "step": 3149
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.7869822485207104e-05,
+      "loss": 1.3966,
+      "step": 3150
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.785009861932939e-05,
+      "loss": 1.5716,
+      "step": 3151
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.783037475345168e-05,
+      "loss": 1.6478,
+      "step": 3152
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.781065088757397e-05,
+      "loss": 1.633,
+      "step": 3153
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.779092702169625e-05,
+      "loss": 1.5774,
+      "step": 3154
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.777120315581854e-05,
+      "loss": 1.5661,
+      "step": 3155
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.775147928994083e-05,
+      "loss": 1.4983,
+      "step": 3156
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.773175542406312e-05,
+      "loss": 1.6791,
+      "step": 3157
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.771203155818541e-05,
+      "loss": 1.5948,
+      "step": 3158
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.769230769230769e-05,
+      "loss": 1.5055,
+      "step": 3159
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.767258382642998e-05,
+      "loss": 1.4671,
+      "step": 3160
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.765285996055227e-05,
+      "loss": 1.5678,
+      "step": 3161
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.763313609467456e-05,
+      "loss": 1.4923,
+      "step": 3162
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.761341222879685e-05,
+      "loss": 1.6175,
+      "step": 3163
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.759368836291913e-05,
+      "loss": 1.4966,
+      "step": 3164
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.757396449704142e-05,
+      "loss": 1.4948,
+      "step": 3165
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.755424063116371e-05,
+      "loss": 1.6119,
+      "step": 3166
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.7534516765285996e-05,
+      "loss": 1.5598,
+      "step": 3167
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.751479289940829e-05,
+      "loss": 1.577,
+      "step": 3168
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.749506903353057e-05,
+      "loss": 1.5998,
+      "step": 3169
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.7475345167652865e-05,
+      "loss": 1.5736,
+      "step": 3170
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.745562130177515e-05,
+      "loss": 1.5183,
+      "step": 3171
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.7435897435897436e-05,
+      "loss": 1.5407,
+      "step": 3172
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.741617357001973e-05,
+      "loss": 1.5436,
+      "step": 3173
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.739644970414201e-05,
+      "loss": 1.5078,
+      "step": 3174
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.7376725838264305e-05,
+      "loss": 1.5411,
+      "step": 3175
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.735700197238659e-05,
+      "loss": 1.5082,
+      "step": 3176
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.7337278106508876e-05,
+      "loss": 1.5585,
+      "step": 3177
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.731755424063116e-05,
+      "loss": 1.5902,
+      "step": 3178
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.7297830374753454e-05,
+      "loss": 1.596,
+      "step": 3179
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.7278106508875746e-05,
+      "loss": 1.422,
+      "step": 3180
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.725838264299803e-05,
+      "loss": 1.5789,
+      "step": 3181
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.7238658777120316e-05,
+      "loss": 1.6676,
+      "step": 3182
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.72189349112426e-05,
+      "loss": 1.5298,
+      "step": 3183
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.7199211045364894e-05,
+      "loss": 1.4417,
+      "step": 3184
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 1.5404,
+      "step": 3185
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.715976331360947e-05,
+      "loss": 1.5469,
+      "step": 3186
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.714003944773176e-05,
+      "loss": 1.552,
+      "step": 3187
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.712031558185404e-05,
+      "loss": 1.5144,
+      "step": 3188
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.7100591715976334e-05,
+      "loss": 1.5324,
+      "step": 3189
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.708086785009862e-05,
+      "loss": 1.5837,
+      "step": 3190
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.706114398422091e-05,
+      "loss": 1.5437,
+      "step": 3191
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.70414201183432e-05,
+      "loss": 1.5105,
+      "step": 3192
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.702169625246549e-05,
+      "loss": 1.559,
+      "step": 3193
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.7001972386587774e-05,
+      "loss": 1.5477,
+      "step": 3194
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.698224852071006e-05,
+      "loss": 1.4668,
+      "step": 3195
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.6962524654832345e-05,
+      "loss": 1.4858,
+      "step": 3196
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.694280078895464e-05,
+      "loss": 1.5414,
+      "step": 3197
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 1.504,
+      "step": 3198
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.6903353057199215e-05,
+      "loss": 1.5778,
+      "step": 3199
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.68836291913215e-05,
+      "loss": 1.4725,
+      "step": 3200
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.6863905325443785e-05,
+      "loss": 1.5154,
+      "step": 3201
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.684418145956608e-05,
+      "loss": 1.5659,
+      "step": 3202
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.682445759368837e-05,
+      "loss": 1.6065,
+      "step": 3203
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.6804733727810655e-05,
+      "loss": 1.5441,
+      "step": 3204
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.678500986193294e-05,
+      "loss": 1.4903,
+      "step": 3205
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.6765285996055225e-05,
+      "loss": 1.6062,
+      "step": 3206
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.674556213017752e-05,
+      "loss": 1.4839,
+      "step": 3207
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.67258382642998e-05,
+      "loss": 1.5547,
+      "step": 3208
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.6706114398422095e-05,
+      "loss": 1.4894,
+      "step": 3209
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.668639053254438e-05,
+      "loss": 1.5241,
+      "step": 3210
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 1.5718,
+      "step": 3211
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.664694280078896e-05,
+      "loss": 1.4922,
+      "step": 3212
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.662721893491124e-05,
+      "loss": 1.4938,
+      "step": 3213
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.660749506903353e-05,
+      "loss": 1.6023,
+      "step": 3214
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.658777120315582e-05,
+      "loss": 1.6036,
+      "step": 3215
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.656804733727811e-05,
+      "loss": 1.5665,
+      "step": 3216
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.65483234714004e-05,
+      "loss": 1.4976,
+      "step": 3217
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.6528599605522683e-05,
+      "loss": 1.5764,
+      "step": 3218
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.650887573964497e-05,
+      "loss": 1.572,
+      "step": 3219
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.648915187376726e-05,
+      "loss": 1.5082,
+      "step": 3220
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.646942800788955e-05,
+      "loss": 1.5624,
+      "step": 3221
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.644970414201184e-05,
+      "loss": 1.4584,
+      "step": 3222
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.6429980276134124e-05,
+      "loss": 1.5318,
+      "step": 3223
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.641025641025641e-05,
+      "loss": 1.4668,
+      "step": 3224
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.63905325443787e-05,
+      "loss": 1.558,
+      "step": 3225
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.6370808678500986e-05,
+      "loss": 1.6236,
+      "step": 3226
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.635108481262328e-05,
+      "loss": 1.5503,
+      "step": 3227
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.6331360946745564e-05,
+      "loss": 1.4751,
+      "step": 3228
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.631163708086785e-05,
+      "loss": 1.4487,
+      "step": 3229
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.629191321499014e-05,
+      "loss": 1.6519,
+      "step": 3230
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.627218934911243e-05,
+      "loss": 1.5876,
+      "step": 3231
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.625246548323472e-05,
+      "loss": 1.5318,
+      "step": 3232
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.6232741617357004e-05,
+      "loss": 1.5113,
+      "step": 3233
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.621301775147929e-05,
+      "loss": 1.4911,
+      "step": 3234
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.619329388560158e-05,
+      "loss": 1.51,
+      "step": 3235
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.617357001972387e-05,
+      "loss": 1.6253,
+      "step": 3236
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.615384615384615e-05,
+      "loss": 1.4886,
+      "step": 3237
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.6134122287968444e-05,
+      "loss": 1.4994,
+      "step": 3238
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.6114398422090736e-05,
+      "loss": 1.5102,
+      "step": 3239
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.609467455621302e-05,
+      "loss": 1.4806,
+      "step": 3240
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.607495069033531e-05,
+      "loss": 1.5059,
+      "step": 3241
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.605522682445759e-05,
+      "loss": 1.4678,
+      "step": 3242
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.603550295857988e-05,
+      "loss": 1.5654,
+      "step": 3243
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.601577909270217e-05,
+      "loss": 1.4643,
+      "step": 3244
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.599605522682446e-05,
+      "loss": 1.4392,
+      "step": 3245
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.597633136094675e-05,
+      "loss": 1.5416,
+      "step": 3246
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.595660749506903e-05,
+      "loss": 1.62,
+      "step": 3247
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.5936883629191325e-05,
+      "loss": 1.6068,
+      "step": 3248
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.591715976331361e-05,
+      "loss": 1.6062,
+      "step": 3249
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.5317,
+      "step": 3250
+    },
+    {
+      "epoch": 9.62,
+      "eval_loss": 1.4340853691101074,
+      "eval_runtime": 69.6206,
+      "eval_samples_per_second": 319.876,
+      "eval_steps_per_second": 1.666,
+      "step": 3250
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.587771203155819e-05,
+      "loss": 1.5883,
+      "step": 3251
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.585798816568047e-05,
+      "loss": 1.4992,
+      "step": 3252
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.5838264299802765e-05,
+      "loss": 1.6249,
+      "step": 3253
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.581854043392505e-05,
+      "loss": 1.4722,
+      "step": 3254
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.5798816568047336e-05,
+      "loss": 1.5256,
+      "step": 3255
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.577909270216963e-05,
+      "loss": 1.4851,
+      "step": 3256
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.575936883629192e-05,
+      "loss": 1.4547,
+      "step": 3257
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.5739644970414205e-05,
+      "loss": 1.5067,
+      "step": 3258
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.571992110453649e-05,
+      "loss": 1.5155,
+      "step": 3259
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.5700197238658776e-05,
+      "loss": 1.5306,
+      "step": 3260
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.568047337278107e-05,
+      "loss": 1.5105,
+      "step": 3261
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.566074950690336e-05,
+      "loss": 1.5331,
+      "step": 3262
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.5641025641025646e-05,
+      "loss": 1.5691,
+      "step": 3263
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.562130177514793e-05,
+      "loss": 1.5833,
+      "step": 3264
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.5601577909270216e-05,
+      "loss": 1.4245,
+      "step": 3265
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.55818540433925e-05,
+      "loss": 1.5589,
+      "step": 3266
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.5562130177514794e-05,
+      "loss": 1.4587,
+      "step": 3267
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.5542406311637086e-05,
+      "loss": 1.5965,
+      "step": 3268
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.552268244575937e-05,
+      "loss": 1.562,
+      "step": 3269
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.5502958579881656e-05,
+      "loss": 1.5733,
+      "step": 3270
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.548323471400395e-05,
+      "loss": 1.4886,
+      "step": 3271
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.5463510848126234e-05,
+      "loss": 1.5501,
+      "step": 3272
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.544378698224852e-05,
+      "loss": 1.6245,
+      "step": 3273
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.542406311637081e-05,
+      "loss": 1.5399,
+      "step": 3274
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.54043392504931e-05,
+      "loss": 1.633,
+      "step": 3275
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.538461538461539e-05,
+      "loss": 1.5672,
+      "step": 3276
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.5364891518737674e-05,
+      "loss": 1.5234,
+      "step": 3277
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.534516765285996e-05,
+      "loss": 1.5191,
+      "step": 3278
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.532544378698225e-05,
+      "loss": 1.5722,
+      "step": 3279
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.5305719921104544e-05,
+      "loss": 1.5787,
+      "step": 3280
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.528599605522683e-05,
+      "loss": 1.5575,
+      "step": 3281
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.5266272189349114e-05,
+      "loss": 1.5621,
+      "step": 3282
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.52465483234714e-05,
+      "loss": 1.4546,
+      "step": 3283
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.5226824457593685e-05,
+      "loss": 1.5618,
+      "step": 3284
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.520710059171598e-05,
+      "loss": 1.5162,
+      "step": 3285
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.518737672583827e-05,
+      "loss": 1.5031,
+      "step": 3286
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.5167652859960555e-05,
+      "loss": 1.5204,
+      "step": 3287
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.514792899408284e-05,
+      "loss": 1.5367,
+      "step": 3288
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.5128205128205125e-05,
+      "loss": 1.5163,
+      "step": 3289
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.510848126232742e-05,
+      "loss": 1.5895,
+      "step": 3290
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.508875739644971e-05,
+      "loss": 1.4683,
+      "step": 3291
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.5069033530571995e-05,
+      "loss": 1.6268,
+      "step": 3292
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.504930966469428e-05,
+      "loss": 1.5348,
+      "step": 3293
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.502958579881657e-05,
+      "loss": 1.5806,
+      "step": 3294
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.500986193293886e-05,
+      "loss": 1.5552,
+      "step": 3295
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.499013806706114e-05,
+      "loss": 1.4595,
+      "step": 3296
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.4970414201183435e-05,
+      "loss": 1.542,
+      "step": 3297
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.495069033530572e-05,
+      "loss": 1.5846,
+      "step": 3298
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.493096646942801e-05,
+      "loss": 1.5057,
+      "step": 3299
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.49112426035503e-05,
+      "loss": 1.5888,
+      "step": 3300
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.489151873767258e-05,
+      "loss": 1.4485,
+      "step": 3301
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.487179487179487e-05,
+      "loss": 1.5359,
+      "step": 3302
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.485207100591716e-05,
+      "loss": 1.6716,
+      "step": 3303
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.483234714003945e-05,
+      "loss": 1.5512,
+      "step": 3304
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.481262327416174e-05,
+      "loss": 1.5448,
+      "step": 3305
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.4792899408284023e-05,
+      "loss": 1.5472,
+      "step": 3306
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.477317554240631e-05,
+      "loss": 1.5648,
+      "step": 3307
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.47534516765286e-05,
+      "loss": 1.4858,
+      "step": 3308
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.473372781065089e-05,
+      "loss": 1.5814,
+      "step": 3309
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.471400394477318e-05,
+      "loss": 1.522,
+      "step": 3310
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.4694280078895464e-05,
+      "loss": 1.5704,
+      "step": 3311
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.467455621301775e-05,
+      "loss": 1.5132,
+      "step": 3312
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.465483234714004e-05,
+      "loss": 1.5651,
+      "step": 3313
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.4635108481262326e-05,
+      "loss": 1.4726,
+      "step": 3314
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.5136,
+      "step": 3315
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.4595660749506904e-05,
+      "loss": 1.4024,
+      "step": 3316
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.4575936883629196e-05,
+      "loss": 1.5449,
+      "step": 3317
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.455621301775148e-05,
+      "loss": 1.6117,
+      "step": 3318
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.453648915187377e-05,
+      "loss": 1.5558,
+      "step": 3319
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.451676528599606e-05,
+      "loss": 1.5656,
+      "step": 3320
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.4497041420118344e-05,
+      "loss": 1.4923,
+      "step": 3321
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.4477317554240636e-05,
+      "loss": 1.573,
+      "step": 3322
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.445759368836292e-05,
+      "loss": 1.526,
+      "step": 3323
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.443786982248521e-05,
+      "loss": 1.5943,
+      "step": 3324
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.441814595660749e-05,
+      "loss": 1.6335,
+      "step": 3325
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.4398422090729784e-05,
+      "loss": 1.5846,
+      "step": 3326
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.4378698224852077e-05,
+      "loss": 1.642,
+      "step": 3327
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.435897435897436e-05,
+      "loss": 1.5129,
+      "step": 3328
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.433925049309665e-05,
+      "loss": 1.6131,
+      "step": 3329
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.431952662721893e-05,
+      "loss": 1.5212,
+      "step": 3330
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.4299802761341225e-05,
+      "loss": 1.586,
+      "step": 3331
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.428007889546351e-05,
+      "loss": 1.5487,
+      "step": 3332
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.42603550295858e-05,
+      "loss": 1.5574,
+      "step": 3333
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.424063116370809e-05,
+      "loss": 1.5476,
+      "step": 3334
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.422090729783038e-05,
+      "loss": 1.4726,
+      "step": 3335
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.4201183431952665e-05,
+      "loss": 1.556,
+      "step": 3336
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.418145956607495e-05,
+      "loss": 1.5672,
+      "step": 3337
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.416173570019724e-05,
+      "loss": 1.5332,
+      "step": 3338
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.414201183431953e-05,
+      "loss": 1.5252,
+      "step": 3339
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.412228796844182e-05,
+      "loss": 1.4874,
+      "step": 3340
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.4102564102564105e-05,
+      "loss": 1.484,
+      "step": 3341
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.408284023668639e-05,
+      "loss": 1.5331,
+      "step": 3342
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.4063116370808676e-05,
+      "loss": 1.4808,
+      "step": 3343
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.404339250493097e-05,
+      "loss": 1.4746,
+      "step": 3344
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.402366863905326e-05,
+      "loss": 1.5255,
+      "step": 3345
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.4003944773175545e-05,
+      "loss": 1.5398,
+      "step": 3346
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.398422090729783e-05,
+      "loss": 1.5862,
+      "step": 3347
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3964497041420116e-05,
+      "loss": 1.5257,
+      "step": 3348
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.394477317554241e-05,
+      "loss": 1.5618,
+      "step": 3349
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.39250493096647e-05,
+      "loss": 1.5621,
+      "step": 3350
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3905325443786986e-05,
+      "loss": 1.6478,
+      "step": 3351
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.388560157790927e-05,
+      "loss": 1.5761,
+      "step": 3352
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3865877712031556e-05,
+      "loss": 1.4994,
+      "step": 3353
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 1.5439,
+      "step": 3354
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3826429980276134e-05,
+      "loss": 1.4953,
+      "step": 3355
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3806706114398426e-05,
+      "loss": 1.5531,
+      "step": 3356
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.378698224852071e-05,
+      "loss": 1.6189,
+      "step": 3357
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3767258382643e-05,
+      "loss": 1.5825,
+      "step": 3358
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.374753451676529e-05,
+      "loss": 1.651,
+      "step": 3359
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3727810650887574e-05,
+      "loss": 1.4302,
+      "step": 3360
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.370808678500986e-05,
+      "loss": 1.5408,
+      "step": 3361
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.368836291913215e-05,
+      "loss": 1.5861,
+      "step": 3362
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3668639053254444e-05,
+      "loss": 1.5981,
+      "step": 3363
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.364891518737673e-05,
+      "loss": 1.5869,
+      "step": 3364
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3629191321499014e-05,
+      "loss": 1.4853,
+      "step": 3365
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.36094674556213e-05,
+      "loss": 1.5196,
+      "step": 3366
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.358974358974359e-05,
+      "loss": 1.5741,
+      "step": 3367
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3570019723865884e-05,
+      "loss": 1.4911,
+      "step": 3368
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.355029585798817e-05,
+      "loss": 1.5875,
+      "step": 3369
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3530571992110454e-05,
+      "loss": 1.5059,
+      "step": 3370
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.351084812623274e-05,
+      "loss": 1.4908,
+      "step": 3371
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.349112426035503e-05,
+      "loss": 1.5752,
+      "step": 3372
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.347140039447732e-05,
+      "loss": 1.4817,
+      "step": 3373
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.345167652859961e-05,
+      "loss": 1.5595,
+      "step": 3374
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3431952662721895e-05,
+      "loss": 1.5416,
+      "step": 3375
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.341222879684418e-05,
+      "loss": 1.6161,
+      "step": 3376
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.339250493096647e-05,
+      "loss": 1.5472,
+      "step": 3377
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337278106508876e-05,
+      "loss": 1.5452,
+      "step": 3378
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335305719921105e-05,
+      "loss": 1.53,
+      "step": 3379
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.2721,
+      "step": 3380
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.331360946745563e-05,
+      "loss": 1.5989,
+      "step": 3381
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.329388560157791e-05,
+      "loss": 1.6194,
+      "step": 3382
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.32741617357002e-05,
+      "loss": 1.5551,
+      "step": 3383
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.325443786982248e-05,
+      "loss": 1.5339,
+      "step": 3384
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3234714003944775e-05,
+      "loss": 1.6192,
+      "step": 3385
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.321499013806707e-05,
+      "loss": 1.5154,
+      "step": 3386
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.319526627218935e-05,
+      "loss": 1.4981,
+      "step": 3387
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.317554240631164e-05,
+      "loss": 1.5273,
+      "step": 3388
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.315581854043392e-05,
+      "loss": 1.5956,
+      "step": 3389
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3136094674556215e-05,
+      "loss": 1.4963,
+      "step": 3390
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.31163708086785e-05,
+      "loss": 1.51,
+      "step": 3391
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.309664694280079e-05,
+      "loss": 1.5607,
+      "step": 3392
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.307692307692308e-05,
+      "loss": 1.628,
+      "step": 3393
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3057199211045364e-05,
+      "loss": 1.5161,
+      "step": 3394
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3037475345167656e-05,
+      "loss": 1.5041,
+      "step": 3395
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.301775147928994e-05,
+      "loss": 1.5866,
+      "step": 3396
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.299802761341223e-05,
+      "loss": 1.4476,
+      "step": 3397
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.297830374753452e-05,
+      "loss": 1.4254,
+      "step": 3398
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.2958579881656804e-05,
+      "loss": 1.5533,
+      "step": 3399
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.2938856015779096e-05,
+      "loss": 1.4887,
+      "step": 3400
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.291913214990138e-05,
+      "loss": 1.5489,
+      "step": 3401
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.2899408284023667e-05,
+      "loss": 1.4956,
+      "step": 3402
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.287968441814596e-05,
+      "loss": 1.4985,
+      "step": 3403
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.285996055226825e-05,
+      "loss": 1.4674,
+      "step": 3404
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.2840236686390536e-05,
+      "loss": 1.5,
+      "step": 3405
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.282051282051282e-05,
+      "loss": 1.5355,
+      "step": 3406
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.280078895463511e-05,
+      "loss": 1.4869,
+      "step": 3407
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.27810650887574e-05,
+      "loss": 1.5488,
+      "step": 3408
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.276134122287969e-05,
+      "loss": 1.5946,
+      "step": 3409
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.2741617357001976e-05,
+      "loss": 1.5261,
+      "step": 3410
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.272189349112426e-05,
+      "loss": 1.545,
+      "step": 3411
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.270216962524655e-05,
+      "loss": 1.5686,
+      "step": 3412
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.268244575936884e-05,
+      "loss": 1.5634,
+      "step": 3413
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.2662721893491124e-05,
+      "loss": 1.4827,
+      "step": 3414
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.2642998027613417e-05,
+      "loss": 1.5042,
+      "step": 3415
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.26232741617357e-05,
+      "loss": 1.5372,
+      "step": 3416
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.260355029585799e-05,
+      "loss": 1.5907,
+      "step": 3417
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.258382642998028e-05,
+      "loss": 1.4247,
+      "step": 3418
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.2564102564102565e-05,
+      "loss": 1.5379,
+      "step": 3419
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.254437869822485e-05,
+      "loss": 1.4803,
+      "step": 3420
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.252465483234714e-05,
+      "loss": 1.5734,
+      "step": 3421
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.250493096646943e-05,
+      "loss": 1.4542,
+      "step": 3422
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.248520710059172e-05,
+      "loss": 1.3683,
+      "step": 3423
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.2465483234714005e-05,
+      "loss": 1.5039,
+      "step": 3424
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.244575936883629e-05,
+      "loss": 1.5679,
+      "step": 3425
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.242603550295858e-05,
+      "loss": 1.6378,
+      "step": 3426
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.2406311637080875e-05,
+      "loss": 1.4933,
+      "step": 3427
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.238658777120316e-05,
+      "loss": 1.5176,
+      "step": 3428
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.2366863905325445e-05,
+      "loss": 1.4542,
+      "step": 3429
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.234714003944773e-05,
+      "loss": 1.5378,
+      "step": 3430
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.2327416173570016e-05,
+      "loss": 1.5982,
+      "step": 3431
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 1.5111,
+      "step": 3432
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.22879684418146e-05,
+      "loss": 1.5593,
+      "step": 3433
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.2268244575936885e-05,
+      "loss": 1.5253,
+      "step": 3434
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.224852071005917e-05,
+      "loss": 1.5219,
+      "step": 3435
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.222879684418146e-05,
+      "loss": 1.5267,
+      "step": 3436
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.220907297830375e-05,
+      "loss": 1.5461,
+      "step": 3437
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.218934911242604e-05,
+      "loss": 1.5837,
+      "step": 3438
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.2169625246548326e-05,
+      "loss": 1.5521,
+      "step": 3439
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.214990138067061e-05,
+      "loss": 1.516,
+      "step": 3440
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.21301775147929e-05,
+      "loss": 1.5494,
+      "step": 3441
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.211045364891519e-05,
+      "loss": 1.5013,
+      "step": 3442
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.2090729783037474e-05,
+      "loss": 1.5469,
+      "step": 3443
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.2071005917159766e-05,
+      "loss": 1.5494,
+      "step": 3444
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 1.5031,
+      "step": 3445
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.203155818540434e-05,
+      "loss": 1.4928,
+      "step": 3446
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.201183431952663e-05,
+      "loss": 1.4752,
+      "step": 3447
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.1992110453648914e-05,
+      "loss": 1.5546,
+      "step": 3448
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.19723865877712e-05,
+      "loss": 1.491,
+      "step": 3449
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.195266272189349e-05,
+      "loss": 1.4241,
+      "step": 3450
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.1932938856015784e-05,
+      "loss": 1.5031,
+      "step": 3451
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.191321499013807e-05,
+      "loss": 1.4482,
+      "step": 3452
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.1893491124260354e-05,
+      "loss": 1.6384,
+      "step": 3453
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.187376725838264e-05,
+      "loss": 1.5158,
+      "step": 3454
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.185404339250493e-05,
+      "loss": 1.5904,
+      "step": 3455
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.1834319526627224e-05,
+      "loss": 1.5985,
+      "step": 3456
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.181459566074951e-05,
+      "loss": 1.5117,
+      "step": 3457
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.1794871794871795e-05,
+      "loss": 1.5607,
+      "step": 3458
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.177514792899409e-05,
+      "loss": 1.5747,
+      "step": 3459
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.175542406311637e-05,
+      "loss": 1.4874,
+      "step": 3460
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.173570019723866e-05,
+      "loss": 1.5568,
+      "step": 3461
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.171597633136095e-05,
+      "loss": 1.5093,
+      "step": 3462
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.1696252465483235e-05,
+      "loss": 1.5376,
+      "step": 3463
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.167652859960553e-05,
+      "loss": 1.5561,
+      "step": 3464
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.165680473372781e-05,
+      "loss": 1.5051,
+      "step": 3465
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.16370808678501e-05,
+      "loss": 1.516,
+      "step": 3466
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.161735700197239e-05,
+      "loss": 1.4518,
+      "step": 3467
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.159763313609468e-05,
+      "loss": 1.5867,
+      "step": 3468
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.157790927021697e-05,
+      "loss": 1.5351,
+      "step": 3469
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.155818540433925e-05,
+      "loss": 1.5367,
+      "step": 3470
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.153846153846154e-05,
+      "loss": 1.5745,
+      "step": 3471
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.151873767258382e-05,
+      "loss": 1.5279,
+      "step": 3472
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.1499013806706115e-05,
+      "loss": 1.571,
+      "step": 3473
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.147928994082841e-05,
+      "loss": 1.5501,
+      "step": 3474
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.145956607495069e-05,
+      "loss": 1.5977,
+      "step": 3475
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.143984220907298e-05,
+      "loss": 1.4951,
+      "step": 3476
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.142011834319526e-05,
+      "loss": 1.525,
+      "step": 3477
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.1400394477317555e-05,
+      "loss": 1.5636,
+      "step": 3478
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.138067061143984e-05,
+      "loss": 1.519,
+      "step": 3479
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.136094674556213e-05,
+      "loss": 1.4588,
+      "step": 3480
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.134122287968442e-05,
+      "loss": 1.5145,
+      "step": 3481
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.132149901380671e-05,
+      "loss": 1.5179,
+      "step": 3482
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.1301775147928996e-05,
+      "loss": 1.4455,
+      "step": 3483
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.128205128205128e-05,
+      "loss": 1.4993,
+      "step": 3484
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.126232741617357e-05,
+      "loss": 1.5427,
+      "step": 3485
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.124260355029586e-05,
+      "loss": 1.475,
+      "step": 3486
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.122287968441815e-05,
+      "loss": 1.4151,
+      "step": 3487
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.1203155818540436e-05,
+      "loss": 1.4675,
+      "step": 3488
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.118343195266272e-05,
+      "loss": 1.5941,
+      "step": 3489
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.116370808678501e-05,
+      "loss": 1.4751,
+      "step": 3490
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.11439842209073e-05,
+      "loss": 1.4469,
+      "step": 3491
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.112426035502959e-05,
+      "loss": 1.5247,
+      "step": 3492
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.1104536489151876e-05,
+      "loss": 1.6142,
+      "step": 3493
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.108481262327416e-05,
+      "loss": 1.4219,
+      "step": 3494
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.106508875739645e-05,
+      "loss": 1.4932,
+      "step": 3495
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.104536489151874e-05,
+      "loss": 1.5612,
+      "step": 3496
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.102564102564103e-05,
+      "loss": 1.5159,
+      "step": 3497
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.1005917159763316e-05,
+      "loss": 1.613,
+      "step": 3498
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.09861932938856e-05,
+      "loss": 1.5355,
+      "step": 3499
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.0966469428007894e-05,
+      "loss": 1.5686,
+      "step": 3500
+    },
+    {
+      "epoch": 10.36,
+      "eval_loss": 1.428783655166626,
+      "eval_runtime": 69.5832,
+      "eval_samples_per_second": 320.048,
+      "eval_steps_per_second": 1.667,
+      "step": 3500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.094674556213018e-05,
+      "loss": 1.6542,
+      "step": 3501
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.0927021696252465e-05,
+      "loss": 1.5627,
+      "step": 3502
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.090729783037476e-05,
+      "loss": 1.5776,
+      "step": 3503
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.088757396449704e-05,
+      "loss": 1.5314,
+      "step": 3504
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.0867850098619334e-05,
+      "loss": 1.44,
+      "step": 3505
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.084812623274162e-05,
+      "loss": 1.6514,
+      "step": 3506
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.0828402366863905e-05,
+      "loss": 1.5413,
+      "step": 3507
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.080867850098619e-05,
+      "loss": 1.6208,
+      "step": 3508
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.078895463510848e-05,
+      "loss": 1.5506,
+      "step": 3509
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.5627,
+      "step": 3510
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.074950690335306e-05,
+      "loss": 1.4799,
+      "step": 3511
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.0729783037475345e-05,
+      "loss": 1.4755,
+      "step": 3512
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.071005917159763e-05,
+      "loss": 1.5916,
+      "step": 3513
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.069033530571992e-05,
+      "loss": 1.4316,
+      "step": 3514
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.0670611439842215e-05,
+      "loss": 1.5374,
+      "step": 3515
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.06508875739645e-05,
+      "loss": 1.5939,
+      "step": 3516
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.0631163708086785e-05,
+      "loss": 1.5836,
+      "step": 3517
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.061143984220907e-05,
+      "loss": 1.6008,
+      "step": 3518
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.059171597633136e-05,
+      "loss": 1.4781,
+      "step": 3519
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.057199211045365e-05,
+      "loss": 1.4561,
+      "step": 3520
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.055226824457594e-05,
+      "loss": 1.4335,
+      "step": 3521
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.0532544378698226e-05,
+      "loss": 1.5363,
+      "step": 3522
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.0512820512820518e-05,
+      "loss": 1.5746,
+      "step": 3523
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.0493096646942803e-05,
+      "loss": 1.5291,
+      "step": 3524
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.047337278106509e-05,
+      "loss": 1.5459,
+      "step": 3525
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.0453648915187377e-05,
+      "loss": 1.5609,
+      "step": 3526
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.0433925049309662e-05,
+      "loss": 1.5585,
+      "step": 3527
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.0414201183431958e-05,
+      "loss": 1.4702,
+      "step": 3528
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.0394477317554243e-05,
+      "loss": 1.561,
+      "step": 3529
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.037475345167653e-05,
+      "loss": 1.5363,
+      "step": 3530
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.0355029585798817e-05,
+      "loss": 1.5827,
+      "step": 3531
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.0335305719921103e-05,
+      "loss": 1.468,
+      "step": 3532
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.0315581854043395e-05,
+      "loss": 1.6421,
+      "step": 3533
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.0295857988165683e-05,
+      "loss": 1.5592,
+      "step": 3534
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.027613412228797e-05,
+      "loss": 1.5378,
+      "step": 3535
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.0256410256410257e-05,
+      "loss": 1.5508,
+      "step": 3536
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.023668639053255e-05,
+      "loss": 1.4125,
+      "step": 3537
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.0216962524654835e-05,
+      "loss": 1.5793,
+      "step": 3538
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.019723865877712e-05,
+      "loss": 1.5848,
+      "step": 3539
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.017751479289941e-05,
+      "loss": 1.4227,
+      "step": 3540
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.0157790927021694e-05,
+      "loss": 1.5197,
+      "step": 3541
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.0138067061143986e-05,
+      "loss": 1.5711,
+      "step": 3542
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.0118343195266275e-05,
+      "loss": 1.3745,
+      "step": 3543
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.009861932938856e-05,
+      "loss": 1.5391,
+      "step": 3544
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.007889546351085e-05,
+      "loss": 1.5347,
+      "step": 3545
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.005917159763314e-05,
+      "loss": 1.5624,
+      "step": 3546
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.0039447731755427e-05,
+      "loss": 1.4337,
+      "step": 3547
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.0019723865877712e-05,
+      "loss": 1.5516,
+      "step": 3548
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3e-05,
+      "loss": 1.4535,
+      "step": 3549
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.9980276134122286e-05,
+      "loss": 1.5161,
+      "step": 3550
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.9960552268244578e-05,
+      "loss": 1.5192,
+      "step": 3551
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.9940828402366867e-05,
+      "loss": 1.6124,
+      "step": 3552
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.9921104536489152e-05,
+      "loss": 1.5605,
+      "step": 3553
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.990138067061144e-05,
+      "loss": 1.5195,
+      "step": 3554
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.9881656804733733e-05,
+      "loss": 1.5202,
+      "step": 3555
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.986193293885602e-05,
+      "loss": 1.5125,
+      "step": 3556
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.9842209072978307e-05,
+      "loss": 1.5164,
+      "step": 3557
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.9822485207100593e-05,
+      "loss": 1.4881,
+      "step": 3558
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.9802761341222878e-05,
+      "loss": 1.5173,
+      "step": 3559
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.978303747534517e-05,
+      "loss": 1.4304,
+      "step": 3560
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.976331360946746e-05,
+      "loss": 1.4912,
+      "step": 3561
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.9743589743589744e-05,
+      "loss": 1.5221,
+      "step": 3562
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.9723865877712033e-05,
+      "loss": 1.5276,
+      "step": 3563
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.9704142011834318e-05,
+      "loss": 1.5275,
+      "step": 3564
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.968441814595661e-05,
+      "loss": 1.5936,
+      "step": 3565
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.96646942800789e-05,
+      "loss": 1.5841,
+      "step": 3566
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.9644970414201184e-05,
+      "loss": 1.3831,
+      "step": 3567
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.962524654832347e-05,
+      "loss": 1.3958,
+      "step": 3568
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.9605522682445762e-05,
+      "loss": 1.4824,
+      "step": 3569
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.958579881656805e-05,
+      "loss": 1.5896,
+      "step": 3570
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.9566074950690336e-05,
+      "loss": 1.4243,
+      "step": 3571
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.9546351084812625e-05,
+      "loss": 1.4982,
+      "step": 3572
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.952662721893491e-05,
+      "loss": 1.5514,
+      "step": 3573
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.9506903353057202e-05,
+      "loss": 1.4735,
+      "step": 3574
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.4486,
+      "step": 3575
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.9467455621301776e-05,
+      "loss": 1.5311,
+      "step": 3576
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.944773175542406e-05,
+      "loss": 1.4919,
+      "step": 3577
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9428007889546353e-05,
+      "loss": 1.5512,
+      "step": 3578
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9408284023668642e-05,
+      "loss": 1.5282,
+      "step": 3579
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9388560157790928e-05,
+      "loss": 1.6092,
+      "step": 3580
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9368836291913216e-05,
+      "loss": 1.5604,
+      "step": 3581
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.93491124260355e-05,
+      "loss": 1.5878,
+      "step": 3582
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.9329388560157794e-05,
+      "loss": 1.5473,
+      "step": 3583
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.9309664694280082e-05,
+      "loss": 1.6231,
+      "step": 3584
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.9289940828402368e-05,
+      "loss": 1.6238,
+      "step": 3585
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.9270216962524653e-05,
+      "loss": 1.5918,
+      "step": 3586
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.9250493096646942e-05,
+      "loss": 1.5922,
+      "step": 3587
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.9230769230769234e-05,
+      "loss": 1.5254,
+      "step": 3588
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.921104536489152e-05,
+      "loss": 1.5017,
+      "step": 3589
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.9191321499013808e-05,
+      "loss": 1.6447,
+      "step": 3590
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.9171597633136093e-05,
+      "loss": 1.5452,
+      "step": 3591
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.9151873767258385e-05,
+      "loss": 1.6336,
+      "step": 3592
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.9132149901380674e-05,
+      "loss": 1.5082,
+      "step": 3593
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.911242603550296e-05,
+      "loss": 1.4503,
+      "step": 3594
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.9092702169625248e-05,
+      "loss": 1.5017,
+      "step": 3595
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.9072978303747534e-05,
+      "loss": 1.5083,
+      "step": 3596
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.9053254437869826e-05,
+      "loss": 1.5745,
+      "step": 3597
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.903353057199211e-05,
+      "loss": 1.5859,
+      "step": 3598
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.90138067061144e-05,
+      "loss": 1.4949,
+      "step": 3599
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.8994082840236685e-05,
+      "loss": 1.657,
+      "step": 3600
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.8974358974358977e-05,
+      "loss": 1.569,
+      "step": 3601
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.8954635108481266e-05,
+      "loss": 1.6132,
+      "step": 3602
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.893491124260355e-05,
+      "loss": 1.5582,
+      "step": 3603
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.891518737672584e-05,
+      "loss": 1.4496,
+      "step": 3604
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.8895463510848125e-05,
+      "loss": 1.6329,
+      "step": 3605
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.8875739644970417e-05,
+      "loss": 1.5437,
+      "step": 3606
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.8856015779092703e-05,
+      "loss": 1.4964,
+      "step": 3607
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.883629191321499e-05,
+      "loss": 1.4217,
+      "step": 3608
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.8816568047337277e-05,
+      "loss": 1.5547,
+      "step": 3609
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.879684418145957e-05,
+      "loss": 1.5228,
+      "step": 3610
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.8777120315581858e-05,
+      "loss": 1.5261,
+      "step": 3611
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.8757396449704143e-05,
+      "loss": 1.544,
+      "step": 3612
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.8737672583826432e-05,
+      "loss": 1.5515,
+      "step": 3613
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.8717948717948717e-05,
+      "loss": 1.5206,
+      "step": 3614
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.869822485207101e-05,
+      "loss": 1.5459,
+      "step": 3615
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.8678500986193298e-05,
+      "loss": 1.5795,
+      "step": 3616
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.8658777120315583e-05,
+      "loss": 1.4926,
+      "step": 3617
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.863905325443787e-05,
+      "loss": 1.4456,
+      "step": 3618
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.8619329388560157e-05,
+      "loss": 1.5498,
+      "step": 3619
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.859960552268245e-05,
+      "loss": 1.54,
+      "step": 3620
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.8579881656804735e-05,
+      "loss": 1.4699,
+      "step": 3621
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.8560157790927023e-05,
+      "loss": 1.5545,
+      "step": 3622
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.854043392504931e-05,
+      "loss": 1.5318,
+      "step": 3623
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.85207100591716e-05,
+      "loss": 1.3886,
+      "step": 3624
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.850098619329389e-05,
+      "loss": 1.4744,
+      "step": 3625
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.8481262327416175e-05,
+      "loss": 1.5006,
+      "step": 3626
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.846153846153846e-05,
+      "loss": 1.5539,
+      "step": 3627
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.844181459566075e-05,
+      "loss": 1.5892,
+      "step": 3628
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.842209072978304e-05,
+      "loss": 1.524,
+      "step": 3629
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.8402366863905327e-05,
+      "loss": 1.538,
+      "step": 3630
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.8382642998027615e-05,
+      "loss": 1.4624,
+      "step": 3631
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.83629191321499e-05,
+      "loss": 1.5244,
+      "step": 3632
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.8343195266272193e-05,
+      "loss": 1.5242,
+      "step": 3633
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.832347140039448e-05,
+      "loss": 1.5052,
+      "step": 3634
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.8303747534516767e-05,
+      "loss": 1.5849,
+      "step": 3635
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.8284023668639052e-05,
+      "loss": 1.5021,
+      "step": 3636
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.826429980276134e-05,
+      "loss": 1.5554,
+      "step": 3637
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.8244575936883633e-05,
+      "loss": 1.5515,
+      "step": 3638
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.8224852071005918e-05,
+      "loss": 1.5607,
+      "step": 3639
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 1.6001,
+      "step": 3640
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.8185404339250492e-05,
+      "loss": 1.5292,
+      "step": 3641
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.816568047337278e-05,
+      "loss": 1.574,
+      "step": 3642
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.8145956607495073e-05,
+      "loss": 1.5073,
+      "step": 3643
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.812623274161736e-05,
+      "loss": 1.5273,
+      "step": 3644
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.8106508875739644e-05,
+      "loss": 1.5525,
+      "step": 3645
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.8086785009861933e-05,
+      "loss": 1.545,
+      "step": 3646
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.8067061143984225e-05,
+      "loss": 1.4363,
+      "step": 3647
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.804733727810651e-05,
+      "loss": 1.5159,
+      "step": 3648
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.80276134122288e-05,
+      "loss": 1.5527,
+      "step": 3649
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.8007889546351084e-05,
+      "loss": 1.5477,
+      "step": 3650
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.7988165680473373e-05,
+      "loss": 1.5835,
+      "step": 3651
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.7968441814595665e-05,
+      "loss": 1.5365,
+      "step": 3652
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.794871794871795e-05,
+      "loss": 1.5521,
+      "step": 3653
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.792899408284024e-05,
+      "loss": 1.6245,
+      "step": 3654
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.7909270216962524e-05,
+      "loss": 1.5672,
+      "step": 3655
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.7889546351084816e-05,
+      "loss": 1.4541,
+      "step": 3656
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.7869822485207102e-05,
+      "loss": 1.5547,
+      "step": 3657
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.785009861932939e-05,
+      "loss": 1.5718,
+      "step": 3658
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.7830374753451676e-05,
+      "loss": 1.5141,
+      "step": 3659
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.7810650887573965e-05,
+      "loss": 1.5231,
+      "step": 3660
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.7790927021696257e-05,
+      "loss": 1.5294,
+      "step": 3661
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.7771203155818542e-05,
+      "loss": 1.5544,
+      "step": 3662
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.775147928994083e-05,
+      "loss": 1.5608,
+      "step": 3663
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.7731755424063116e-05,
+      "loss": 1.6343,
+      "step": 3664
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.7712031558185408e-05,
+      "loss": 1.5258,
+      "step": 3665
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 1.528,
+      "step": 3666
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.7672583826429982e-05,
+      "loss": 1.5529,
+      "step": 3667
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.7652859960552268e-05,
+      "loss": 1.4489,
+      "step": 3668
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.7633136094674556e-05,
+      "loss": 1.5355,
+      "step": 3669
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.761341222879685e-05,
+      "loss": 1.5133,
+      "step": 3670
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.7593688362919134e-05,
+      "loss": 1.5471,
+      "step": 3671
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.7573964497041422e-05,
+      "loss": 1.6117,
+      "step": 3672
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.7554240631163708e-05,
+      "loss": 1.5523,
+      "step": 3673
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.7534516765285993e-05,
+      "loss": 1.5119,
+      "step": 3674
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.751479289940829e-05,
+      "loss": 1.5067,
+      "step": 3675
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.7495069033530574e-05,
+      "loss": 1.4485,
+      "step": 3676
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.747534516765286e-05,
+      "loss": 1.5223,
+      "step": 3677
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.7455621301775148e-05,
+      "loss": 1.4588,
+      "step": 3678
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.743589743589744e-05,
+      "loss": 1.439,
+      "step": 3679
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.7416173570019726e-05,
+      "loss": 1.5502,
+      "step": 3680
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.7396449704142014e-05,
+      "loss": 1.4638,
+      "step": 3681
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.73767258382643e-05,
+      "loss": 1.5442,
+      "step": 3682
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.735700197238659e-05,
+      "loss": 1.566,
+      "step": 3683
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.733727810650888e-05,
+      "loss": 1.4338,
+      "step": 3684
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.7317554240631166e-05,
+      "loss": 1.4113,
+      "step": 3685
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.729783037475345e-05,
+      "loss": 1.5207,
+      "step": 3686
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.727810650887574e-05,
+      "loss": 1.5036,
+      "step": 3687
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.7258382642998032e-05,
+      "loss": 1.5414,
+      "step": 3688
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.7238658777120317e-05,
+      "loss": 1.4985,
+      "step": 3689
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.7218934911242606e-05,
+      "loss": 1.5309,
+      "step": 3690
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.719921104536489e-05,
+      "loss": 1.5537,
+      "step": 3691
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.717948717948718e-05,
+      "loss": 1.6118,
+      "step": 3692
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.7159763313609472e-05,
+      "loss": 1.597,
+      "step": 3693
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.7140039447731758e-05,
+      "loss": 1.5513,
+      "step": 3694
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.7120315581854043e-05,
+      "loss": 1.5372,
+      "step": 3695
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.710059171597633e-05,
+      "loss": 1.5694,
+      "step": 3696
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.7080867850098617e-05,
+      "loss": 1.5667,
+      "step": 3697
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.706114398422091e-05,
+      "loss": 1.5232,
+      "step": 3698
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.7041420118343198e-05,
+      "loss": 1.5013,
+      "step": 3699
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.7021696252465483e-05,
+      "loss": 1.5752,
+      "step": 3700
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.7001972386587772e-05,
+      "loss": 1.5563,
+      "step": 3701
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.6982248520710064e-05,
+      "loss": 1.5107,
+      "step": 3702
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.696252465483235e-05,
+      "loss": 1.5217,
+      "step": 3703
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.6942800788954638e-05,
+      "loss": 1.5045,
+      "step": 3704
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 1.5058,
+      "step": 3705
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.690335305719921e-05,
+      "loss": 1.5308,
+      "step": 3706
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.68836291913215e-05,
+      "loss": 1.5213,
+      "step": 3707
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.686390532544379e-05,
+      "loss": 1.5683,
+      "step": 3708
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.6844181459566075e-05,
+      "loss": 1.5111,
+      "step": 3709
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.6824457593688364e-05,
+      "loss": 1.4495,
+      "step": 3710
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.6804733727810656e-05,
+      "loss": 1.6226,
+      "step": 3711
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.678500986193294e-05,
+      "loss": 1.5459,
+      "step": 3712
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.676528599605523e-05,
+      "loss": 1.6076,
+      "step": 3713
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.6745562130177515e-05,
+      "loss": 1.5896,
+      "step": 3714
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.67258382642998e-05,
+      "loss": 1.5557,
+      "step": 3715
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.6706114398422093e-05,
+      "loss": 1.538,
+      "step": 3716
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.668639053254438e-05,
+      "loss": 1.5623,
+      "step": 3717
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.4831,
+      "step": 3718
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.6646942800788955e-05,
+      "loss": 1.524,
+      "step": 3719
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.6627218934911247e-05,
+      "loss": 1.5606,
+      "step": 3720
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.6607495069033533e-05,
+      "loss": 1.5041,
+      "step": 3721
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.658777120315582e-05,
+      "loss": 1.5058,
+      "step": 3722
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.6568047337278107e-05,
+      "loss": 1.5743,
+      "step": 3723
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.6548323471400392e-05,
+      "loss": 1.526,
+      "step": 3724
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.6528599605522688e-05,
+      "loss": 1.5911,
+      "step": 3725
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.6508875739644973e-05,
+      "loss": 1.5384,
+      "step": 3726
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.648915187376726e-05,
+      "loss": 1.4172,
+      "step": 3727
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.6469428007889547e-05,
+      "loss": 1.5522,
+      "step": 3728
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.6449704142011832e-05,
+      "loss": 1.5337,
+      "step": 3729
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.6429980276134125e-05,
+      "loss": 1.5068,
+      "step": 3730
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.6410256410256413e-05,
+      "loss": 1.3999,
+      "step": 3731
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.63905325443787e-05,
+      "loss": 1.5943,
+      "step": 3732
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.6370808678500984e-05,
+      "loss": 1.5324,
+      "step": 3733
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.635108481262328e-05,
+      "loss": 1.4795,
+      "step": 3734
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.6331360946745565e-05,
+      "loss": 1.513,
+      "step": 3735
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.631163708086785e-05,
+      "loss": 1.5976,
+      "step": 3736
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.629191321499014e-05,
+      "loss": 1.4744,
+      "step": 3737
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.6272189349112424e-05,
+      "loss": 1.4328,
+      "step": 3738
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.6252465483234716e-05,
+      "loss": 1.4651,
+      "step": 3739
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.6232741617357005e-05,
+      "loss": 1.4932,
+      "step": 3740
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.621301775147929e-05,
+      "loss": 1.5073,
+      "step": 3741
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.619329388560158e-05,
+      "loss": 1.5482,
+      "step": 3742
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.617357001972387e-05,
+      "loss": 1.5206,
+      "step": 3743
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.6153846153846157e-05,
+      "loss": 1.6045,
+      "step": 3744
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.6134122287968442e-05,
+      "loss": 1.5185,
+      "step": 3745
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.611439842209073e-05,
+      "loss": 1.504,
+      "step": 3746
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.6094674556213016e-05,
+      "loss": 1.4362,
+      "step": 3747
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.6074950690335308e-05,
+      "loss": 1.4653,
+      "step": 3748
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.6055226824457597e-05,
+      "loss": 1.5387,
+      "step": 3749
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.6035502958579882e-05,
+      "loss": 1.5569,
+      "step": 3750
+    },
+    {
+      "epoch": 11.09,
+      "eval_loss": 1.4197721481323242,
+      "eval_runtime": 72.343,
+      "eval_samples_per_second": 307.839,
+      "eval_steps_per_second": 1.603,
+      "step": 3750
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.601577909270217e-05,
+      "loss": 1.4919,
+      "step": 3751
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.5996055226824456e-05,
+      "loss": 1.5342,
+      "step": 3752
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.5976331360946748e-05,
+      "loss": 1.6541,
+      "step": 3753
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.5956607495069034e-05,
+      "loss": 1.5163,
+      "step": 3754
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.5936883629191322e-05,
+      "loss": 1.4143,
+      "step": 3755
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.5917159763313608e-05,
+      "loss": 1.5188,
+      "step": 3756
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.58974358974359e-05,
+      "loss": 1.5655,
+      "step": 3757
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.587771203155819e-05,
+      "loss": 1.429,
+      "step": 3758
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.5857988165680474e-05,
+      "loss": 1.478,
+      "step": 3759
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.5838264299802763e-05,
+      "loss": 1.5319,
+      "step": 3760
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.5818540433925048e-05,
+      "loss": 1.5191,
+      "step": 3761
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.579881656804734e-05,
+      "loss": 1.567,
+      "step": 3762
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.577909270216963e-05,
+      "loss": 1.5798,
+      "step": 3763
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.5759368836291914e-05,
+      "loss": 1.5914,
+      "step": 3764
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.57396449704142e-05,
+      "loss": 1.4832,
+      "step": 3765
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.571992110453649e-05,
+      "loss": 1.5404,
+      "step": 3766
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.570019723865878e-05,
+      "loss": 1.5002,
+      "step": 3767
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.5680473372781066e-05,
+      "loss": 1.621,
+      "step": 3768
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.5660749506903354e-05,
+      "loss": 1.5687,
+      "step": 3769
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 1.4335,
+      "step": 3770
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.5621301775147932e-05,
+      "loss": 1.4665,
+      "step": 3771
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.560157790927022e-05,
+      "loss": 1.5645,
+      "step": 3772
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.5581854043392506e-05,
+      "loss": 1.6061,
+      "step": 3773
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.556213017751479e-05,
+      "loss": 1.5408,
+      "step": 3774
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.5542406311637083e-05,
+      "loss": 1.5171,
+      "step": 3775
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.5522682445759372e-05,
+      "loss": 1.4927,
+      "step": 3776
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.5502958579881657e-05,
+      "loss": 1.5321,
+      "step": 3777
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.5483234714003946e-05,
+      "loss": 1.5455,
+      "step": 3778
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.546351084812623e-05,
+      "loss": 1.5219,
+      "step": 3779
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.5443786982248524e-05,
+      "loss": 1.5673,
+      "step": 3780
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.5424063116370812e-05,
+      "loss": 1.4837,
+      "step": 3781
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.5404339250493098e-05,
+      "loss": 1.5371,
+      "step": 3782
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.5384615384615383e-05,
+      "loss": 1.4384,
+      "step": 3783
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.536489151873767e-05,
+      "loss": 1.5029,
+      "step": 3784
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.5345167652859964e-05,
+      "loss": 1.5853,
+      "step": 3785
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.532544378698225e-05,
+      "loss": 1.4937,
+      "step": 3786
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.5305719921104538e-05,
+      "loss": 1.5099,
+      "step": 3787
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.5285996055226823e-05,
+      "loss": 1.5342,
+      "step": 3788
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.5266272189349115e-05,
+      "loss": 1.5933,
+      "step": 3789
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.5246548323471404e-05,
+      "loss": 1.6151,
+      "step": 3790
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.522682445759369e-05,
+      "loss": 1.5265,
+      "step": 3791
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.5207100591715978e-05,
+      "loss": 1.459,
+      "step": 3792
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.5187376725838263e-05,
+      "loss": 1.5081,
+      "step": 3793
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.5167652859960556e-05,
+      "loss": 1.4831,
+      "step": 3794
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.514792899408284e-05,
+      "loss": 1.5611,
+      "step": 3795
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.512820512820513e-05,
+      "loss": 1.5441,
+      "step": 3796
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.5108481262327415e-05,
+      "loss": 1.4822,
+      "step": 3797
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.5088757396449707e-05,
+      "loss": 1.4833,
+      "step": 3798
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.5069033530571996e-05,
+      "loss": 1.5501,
+      "step": 3799
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.504930966469428e-05,
+      "loss": 1.4574,
+      "step": 3800
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.502958579881657e-05,
+      "loss": 1.4842,
+      "step": 3801
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.5009861932938855e-05,
+      "loss": 1.5941,
+      "step": 3802
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.4990138067061144e-05,
+      "loss": 1.5049,
+      "step": 3803
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.4970414201183433e-05,
+      "loss": 1.5371,
+      "step": 3804
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.495069033530572e-05,
+      "loss": 1.5914,
+      "step": 3805
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.493096646942801e-05,
+      "loss": 1.4083,
+      "step": 3806
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.49112426035503e-05,
+      "loss": 1.4795,
+      "step": 3807
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.4891518737672584e-05,
+      "loss": 1.4984,
+      "step": 3808
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.4871794871794873e-05,
+      "loss": 1.4559,
+      "step": 3809
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.485207100591716e-05,
+      "loss": 1.5304,
+      "step": 3810
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.483234714003945e-05,
+      "loss": 1.4901,
+      "step": 3811
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.4812623274161736e-05,
+      "loss": 1.6006,
+      "step": 3812
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.4792899408284024e-05,
+      "loss": 1.4493,
+      "step": 3813
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.4773175542406313e-05,
+      "loss": 1.442,
+      "step": 3814
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.47534516765286e-05,
+      "loss": 1.5658,
+      "step": 3815
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.473372781065089e-05,
+      "loss": 1.5164,
+      "step": 3816
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.4714003944773176e-05,
+      "loss": 1.4131,
+      "step": 3817
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.4694280078895465e-05,
+      "loss": 1.518,
+      "step": 3818
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.4674556213017753e-05,
+      "loss": 1.5181,
+      "step": 3819
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.4654832347140042e-05,
+      "loss": 1.5011,
+      "step": 3820
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.4635108481262327e-05,
+      "loss": 1.5612,
+      "step": 3821
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 1.6012,
+      "step": 3822
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.4595660749506905e-05,
+      "loss": 1.5509,
+      "step": 3823
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.457593688362919e-05,
+      "loss": 1.4795,
+      "step": 3824
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.4556213017751482e-05,
+      "loss": 1.5048,
+      "step": 3825
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.4536489151873768e-05,
+      "loss": 1.5672,
+      "step": 3826
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.4516765285996056e-05,
+      "loss": 1.4869,
+      "step": 3827
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.4497041420118345e-05,
+      "loss": 1.5516,
+      "step": 3828
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.4477317554240634e-05,
+      "loss": 1.5001,
+      "step": 3829
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.445759368836292e-05,
+      "loss": 1.486,
+      "step": 3830
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.4437869822485208e-05,
+      "loss": 1.5286,
+      "step": 3831
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.4418145956607497e-05,
+      "loss": 1.4732,
+      "step": 3832
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.4398422090729782e-05,
+      "loss": 1.5875,
+      "step": 3833
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.4378698224852074e-05,
+      "loss": 1.4337,
+      "step": 3834
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 1.6015,
+      "step": 3835
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.4339250493096648e-05,
+      "loss": 1.68,
+      "step": 3836
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.4319526627218937e-05,
+      "loss": 1.6564,
+      "step": 3837
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.4299802761341222e-05,
+      "loss": 1.5301,
+      "step": 3838
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.428007889546351e-05,
+      "loss": 1.5559,
+      "step": 3839
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.42603550295858e-05,
+      "loss": 1.467,
+      "step": 3840
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.424063116370809e-05,
+      "loss": 1.5005,
+      "step": 3841
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.4220907297830374e-05,
+      "loss": 1.6289,
+      "step": 3842
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.4201183431952666e-05,
+      "loss": 1.5174,
+      "step": 3843
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.418145956607495e-05,
+      "loss": 1.532,
+      "step": 3844
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.416173570019724e-05,
+      "loss": 1.6316,
+      "step": 3845
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.414201183431953e-05,
+      "loss": 1.4573,
+      "step": 3846
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.4122287968441814e-05,
+      "loss": 1.6001,
+      "step": 3847
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 1.5116,
+      "step": 3848
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.408284023668639e-05,
+      "loss": 1.5313,
+      "step": 3849
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.406311637080868e-05,
+      "loss": 1.5197,
+      "step": 3850
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.404339250493097e-05,
+      "loss": 1.5454,
+      "step": 3851
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.4023668639053258e-05,
+      "loss": 1.4549,
+      "step": 3852
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.4003944773175543e-05,
+      "loss": 1.4995,
+      "step": 3853
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.398422090729783e-05,
+      "loss": 1.5604,
+      "step": 3854
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.396449704142012e-05,
+      "loss": 1.4891,
+      "step": 3855
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.3944773175542406e-05,
+      "loss": 1.518,
+      "step": 3856
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.3925049309664694e-05,
+      "loss": 1.5788,
+      "step": 3857
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.3905325443786983e-05,
+      "loss": 1.5739,
+      "step": 3858
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.3885601577909272e-05,
+      "loss": 1.5237,
+      "step": 3859
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.386587771203156e-05,
+      "loss": 1.5577,
+      "step": 3860
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.384615384615385e-05,
+      "loss": 1.4586,
+      "step": 3861
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3826429980276135e-05,
+      "loss": 1.4176,
+      "step": 3862
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3806706114398423e-05,
+      "loss": 1.4003,
+      "step": 3863
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3786982248520712e-05,
+      "loss": 1.4539,
+      "step": 3864
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3767258382642997e-05,
+      "loss": 1.5073,
+      "step": 3865
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.374753451676529e-05,
+      "loss": 1.511,
+      "step": 3866
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.3727810650887575e-05,
+      "loss": 1.5356,
+      "step": 3867
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.3708086785009864e-05,
+      "loss": 1.5218,
+      "step": 3868
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.3688362919132152e-05,
+      "loss": 1.4847,
+      "step": 3869
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.3668639053254438e-05,
+      "loss": 1.5406,
+      "step": 3870
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.3648915187376726e-05,
+      "loss": 1.6504,
+      "step": 3871
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.3629191321499015e-05,
+      "loss": 1.4019,
+      "step": 3872
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.3609467455621304e-05,
+      "loss": 1.5051,
+      "step": 3873
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.358974358974359e-05,
+      "loss": 1.5612,
+      "step": 3874
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.357001972386588e-05,
+      "loss": 1.5369,
+      "step": 3875
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.3550295857988167e-05,
+      "loss": 1.4753,
+      "step": 3876
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.3530571992110455e-05,
+      "loss": 1.6106,
+      "step": 3877
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.3510848126232744e-05,
+      "loss": 1.5731,
+      "step": 3878
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.349112426035503e-05,
+      "loss": 1.5377,
+      "step": 3879
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.3471400394477318e-05,
+      "loss": 1.5032,
+      "step": 3880
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.3451676528599607e-05,
+      "loss": 1.4935,
+      "step": 3881
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.3431952662721896e-05,
+      "loss": 1.4418,
+      "step": 3882
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.341222879684418e-05,
+      "loss": 1.5347,
+      "step": 3883
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.3392504930966473e-05,
+      "loss": 1.5239,
+      "step": 3884
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.337278106508876e-05,
+      "loss": 1.4942,
+      "step": 3885
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.3353057199211044e-05,
+      "loss": 1.4287,
+      "step": 3886
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 1.555,
+      "step": 3887
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.331360946745562e-05,
+      "loss": 1.4773,
+      "step": 3888
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.329388560157791e-05,
+      "loss": 1.5395,
+      "step": 3889
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.32741617357002e-05,
+      "loss": 1.4844,
+      "step": 3890
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.3254437869822487e-05,
+      "loss": 1.4484,
+      "step": 3891
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.3234714003944773e-05,
+      "loss": 1.605,
+      "step": 3892
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.321499013806706e-05,
+      "loss": 1.5195,
+      "step": 3893
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.319526627218935e-05,
+      "loss": 1.497,
+      "step": 3894
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.317554240631164e-05,
+      "loss": 1.4831,
+      "step": 3895
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.3155818540433928e-05,
+      "loss": 1.484,
+      "step": 3896
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.3136094674556213e-05,
+      "loss": 1.5322,
+      "step": 3897
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.31163708086785e-05,
+      "loss": 1.5574,
+      "step": 3898
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.309664694280079e-05,
+      "loss": 1.4984,
+      "step": 3899
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.4911,
+      "step": 3900
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.3057199211045364e-05,
+      "loss": 1.5551,
+      "step": 3901
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.3037475345167653e-05,
+      "loss": 1.5136,
+      "step": 3902
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.3017751479289942e-05,
+      "loss": 1.6196,
+      "step": 3903
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.299802761341223e-05,
+      "loss": 1.4466,
+      "step": 3904
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.297830374753452e-05,
+      "loss": 1.4943,
+      "step": 3905
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.2958579881656805e-05,
+      "loss": 1.6048,
+      "step": 3906
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.2938856015779093e-05,
+      "loss": 1.4924,
+      "step": 3907
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.2919132149901382e-05,
+      "loss": 1.5339,
+      "step": 3908
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.2899408284023667e-05,
+      "loss": 1.4731,
+      "step": 3909
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.287968441814596e-05,
+      "loss": 1.5362,
+      "step": 3910
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.2859960552268245e-05,
+      "loss": 1.4232,
+      "step": 3911
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.2840236686390534e-05,
+      "loss": 1.5834,
+      "step": 3912
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.2820512820512822e-05,
+      "loss": 1.52,
+      "step": 3913
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.280078895463511e-05,
+      "loss": 1.538,
+      "step": 3914
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.2781065088757396e-05,
+      "loss": 1.4782,
+      "step": 3915
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.276134122287969e-05,
+      "loss": 1.5761,
+      "step": 3916
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.2741617357001974e-05,
+      "loss": 1.5675,
+      "step": 3917
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.272189349112426e-05,
+      "loss": 1.4933,
+      "step": 3918
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.270216962524655e-05,
+      "loss": 1.5684,
+      "step": 3919
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.2682445759368837e-05,
+      "loss": 1.4698,
+      "step": 3920
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.2662721893491125e-05,
+      "loss": 1.5726,
+      "step": 3921
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.2642998027613414e-05,
+      "loss": 1.5573,
+      "step": 3922
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.2623274161735703e-05,
+      "loss": 1.4633,
+      "step": 3923
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.2603550295857988e-05,
+      "loss": 1.61,
+      "step": 3924
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.2583826429980277e-05,
+      "loss": 1.5585,
+      "step": 3925
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.2564102564102566e-05,
+      "loss": 1.4016,
+      "step": 3926
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.254437869822485e-05,
+      "loss": 1.566,
+      "step": 3927
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.2524654832347143e-05,
+      "loss": 1.5971,
+      "step": 3928
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.250493096646943e-05,
+      "loss": 1.5449,
+      "step": 3929
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.2485207100591717e-05,
+      "loss": 1.5265,
+      "step": 3930
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.2465483234714006e-05,
+      "loss": 1.483,
+      "step": 3931
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.244575936883629e-05,
+      "loss": 1.4567,
+      "step": 3932
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.242603550295858e-05,
+      "loss": 1.6027,
+      "step": 3933
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.240631163708087e-05,
+      "loss": 1.4606,
+      "step": 3934
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.2386587771203157e-05,
+      "loss": 1.5066,
+      "step": 3935
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.2366863905325443e-05,
+      "loss": 1.533,
+      "step": 3936
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.2347140039447735e-05,
+      "loss": 1.5199,
+      "step": 3937
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.232741617357002e-05,
+      "loss": 1.5797,
+      "step": 3938
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.230769230769231e-05,
+      "loss": 1.5237,
+      "step": 3939
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.2287968441814598e-05,
+      "loss": 1.5205,
+      "step": 3940
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.2268244575936883e-05,
+      "loss": 1.4477,
+      "step": 3941
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.224852071005917e-05,
+      "loss": 1.5671,
+      "step": 3942
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.222879684418146e-05,
+      "loss": 1.4821,
+      "step": 3943
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.220907297830375e-05,
+      "loss": 1.4893,
+      "step": 3944
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.2189349112426034e-05,
+      "loss": 1.5215,
+      "step": 3945
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.2169625246548327e-05,
+      "loss": 1.5608,
+      "step": 3946
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.2149901380670612e-05,
+      "loss": 1.5188,
+      "step": 3947
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.21301775147929e-05,
+      "loss": 1.5168,
+      "step": 3948
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.211045364891519e-05,
+      "loss": 1.5293,
+      "step": 3949
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.2090729783037475e-05,
+      "loss": 1.4956,
+      "step": 3950
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.2071005917159763e-05,
+      "loss": 1.4969,
+      "step": 3951
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.2051282051282052e-05,
+      "loss": 1.5545,
+      "step": 3952
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.203155818540434e-05,
+      "loss": 1.5051,
+      "step": 3953
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.201183431952663e-05,
+      "loss": 1.5447,
+      "step": 3954
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.199211045364892e-05,
+      "loss": 1.5402,
+      "step": 3955
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.1972386587771204e-05,
+      "loss": 1.5198,
+      "step": 3956
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.1952662721893492e-05,
+      "loss": 1.5177,
+      "step": 3957
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.193293885601578e-05,
+      "loss": 1.4708,
+      "step": 3958
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.1913214990138066e-05,
+      "loss": 1.5379,
+      "step": 3959
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.189349112426036e-05,
+      "loss": 1.5549,
+      "step": 3960
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.1873767258382644e-05,
+      "loss": 1.5236,
+      "step": 3961
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.1854043392504933e-05,
+      "loss": 1.535,
+      "step": 3962
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.183431952662722e-05,
+      "loss": 1.5194,
+      "step": 3963
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.1814595660749507e-05,
+      "loss": 1.5852,
+      "step": 3964
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 1.4721,
+      "step": 3965
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.1775147928994084e-05,
+      "loss": 1.625,
+      "step": 3966
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.1755424063116373e-05,
+      "loss": 1.4772,
+      "step": 3967
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.1735700197238658e-05,
+      "loss": 1.5934,
+      "step": 3968
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.171597633136095e-05,
+      "loss": 1.4686,
+      "step": 3969
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.1696252465483236e-05,
+      "loss": 1.433,
+      "step": 3970
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.1676528599605524e-05,
+      "loss": 1.5019,
+      "step": 3971
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.1656804733727813e-05,
+      "loss": 1.5283,
+      "step": 3972
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.16370808678501e-05,
+      "loss": 1.5143,
+      "step": 3973
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.1617357001972387e-05,
+      "loss": 1.5033,
+      "step": 3974
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.1597633136094676e-05,
+      "loss": 1.5821,
+      "step": 3975
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.1577909270216965e-05,
+      "loss": 1.4746,
+      "step": 3976
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.155818540433925e-05,
+      "loss": 1.4856,
+      "step": 3977
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.1538461538461542e-05,
+      "loss": 1.5431,
+      "step": 3978
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.1518737672583827e-05,
+      "loss": 1.5486,
+      "step": 3979
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.1499013806706113e-05,
+      "loss": 1.4281,
+      "step": 3980
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.1479289940828405e-05,
+      "loss": 1.4135,
+      "step": 3981
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.145956607495069e-05,
+      "loss": 1.5298,
+      "step": 3982
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.143984220907298e-05,
+      "loss": 1.5711,
+      "step": 3983
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.1420118343195268e-05,
+      "loss": 1.4044,
+      "step": 3984
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.1400394477317556e-05,
+      "loss": 1.4524,
+      "step": 3985
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.138067061143984e-05,
+      "loss": 1.4978,
+      "step": 3986
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.136094674556213e-05,
+      "loss": 1.609,
+      "step": 3987
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.134122287968442e-05,
+      "loss": 1.5264,
+      "step": 3988
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.1321499013806704e-05,
+      "loss": 1.5291,
+      "step": 3989
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.1301775147928997e-05,
+      "loss": 1.4846,
+      "step": 3990
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.1282051282051282e-05,
+      "loss": 1.5262,
+      "step": 3991
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.126232741617357e-05,
+      "loss": 1.5409,
+      "step": 3992
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.124260355029586e-05,
+      "loss": 1.4966,
+      "step": 3993
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.1222879684418148e-05,
+      "loss": 1.4439,
+      "step": 3994
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.1203155818540433e-05,
+      "loss": 1.5454,
+      "step": 3995
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.1183431952662722e-05,
+      "loss": 1.5864,
+      "step": 3996
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.116370808678501e-05,
+      "loss": 1.5029,
+      "step": 3997
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.11439842209073e-05,
+      "loss": 1.5343,
+      "step": 3998
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.112426035502959e-05,
+      "loss": 1.4932,
+      "step": 3999
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.1104536489151874e-05,
+      "loss": 1.5282,
+      "step": 4000
+    },
+    {
+      "epoch": 11.83,
+      "eval_loss": 1.4160889387130737,
+      "eval_runtime": 69.0948,
+      "eval_samples_per_second": 322.311,
+      "eval_steps_per_second": 1.679,
+      "step": 4000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.1084812623274162e-05,
+      "loss": 1.5637,
+      "step": 4001
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.106508875739645e-05,
+      "loss": 1.5078,
+      "step": 4002
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.1045364891518736e-05,
+      "loss": 1.5403,
+      "step": 4003
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.102564102564103e-05,
+      "loss": 1.5753,
+      "step": 4004
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.1005917159763314e-05,
+      "loss": 1.5425,
+      "step": 4005
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.0986193293885603e-05,
+      "loss": 1.4366,
+      "step": 4006
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.096646942800789e-05,
+      "loss": 1.4725,
+      "step": 4007
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.094674556213018e-05,
+      "loss": 1.4372,
+      "step": 4008
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0927021696252465e-05,
+      "loss": 1.5661,
+      "step": 4009
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0907297830374754e-05,
+      "loss": 1.6106,
+      "step": 4010
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.0887573964497043e-05,
+      "loss": 1.5462,
+      "step": 4011
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.0867850098619328e-05,
+      "loss": 1.4595,
+      "step": 4012
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.084812623274162e-05,
+      "loss": 1.5283,
+      "step": 4013
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0828402366863906e-05,
+      "loss": 1.5471,
+      "step": 4014
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0808678500986194e-05,
+      "loss": 1.4812,
+      "step": 4015
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0788954635108483e-05,
+      "loss": 1.4654,
+      "step": 4016
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.0769230769230772e-05,
+      "loss": 1.505,
+      "step": 4017
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0749506903353057e-05,
+      "loss": 1.5331,
+      "step": 4018
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0729783037475346e-05,
+      "loss": 1.5755,
+      "step": 4019
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0710059171597635e-05,
+      "loss": 1.5443,
+      "step": 4020
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.069033530571992e-05,
+      "loss": 1.5015,
+      "step": 4021
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0670611439842212e-05,
+      "loss": 1.5939,
+      "step": 4022
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.0650887573964497e-05,
+      "loss": 1.4551,
+      "step": 4023
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0631163708086786e-05,
+      "loss": 1.5079,
+      "step": 4024
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0611439842209075e-05,
+      "loss": 1.4947,
+      "step": 4025
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0591715976331364e-05,
+      "loss": 1.4686,
+      "step": 4026
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.057199211045365e-05,
+      "loss": 1.4927,
+      "step": 4027
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0552268244575938e-05,
+      "loss": 1.5401,
+      "step": 4028
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0532544378698226e-05,
+      "loss": 1.5739,
+      "step": 4029
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 1.538,
+      "step": 4030
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0493096646942804e-05,
+      "loss": 1.4578,
+      "step": 4031
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.047337278106509e-05,
+      "loss": 1.5104,
+      "step": 4032
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0453648915187378e-05,
+      "loss": 1.4663,
+      "step": 4033
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.0433925049309667e-05,
+      "loss": 1.4785,
+      "step": 4034
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.0414201183431952e-05,
+      "loss": 1.4676,
+      "step": 4035
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.039447731755424e-05,
+      "loss": 1.6178,
+      "step": 4036
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.037475345167653e-05,
+      "loss": 1.4598,
+      "step": 4037
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0355029585798818e-05,
+      "loss": 1.4459,
+      "step": 4038
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0335305719921103e-05,
+      "loss": 1.4983,
+      "step": 4039
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.0315581854043396e-05,
+      "loss": 1.4677,
+      "step": 4040
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.029585798816568e-05,
+      "loss": 1.5457,
+      "step": 4041
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.027613412228797e-05,
+      "loss": 1.4478,
+      "step": 4042
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.025641025641026e-05,
+      "loss": 1.5569,
+      "step": 4043
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0236686390532544e-05,
+      "loss": 1.5118,
+      "step": 4044
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0216962524654832e-05,
+      "loss": 1.455,
+      "step": 4045
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.019723865877712e-05,
+      "loss": 1.5226,
+      "step": 4046
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.017751479289941e-05,
+      "loss": 1.4823,
+      "step": 4047
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0157790927021695e-05,
+      "loss": 1.4724,
+      "step": 4048
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0138067061143987e-05,
+      "loss": 1.4857,
+      "step": 4049
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0118343195266273e-05,
+      "loss": 1.6352,
+      "step": 4050
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.009861932938856e-05,
+      "loss": 1.4974,
+      "step": 4051
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.007889546351085e-05,
+      "loss": 1.4888,
+      "step": 4052
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0059171597633135e-05,
+      "loss": 1.5282,
+      "step": 4053
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0039447731755424e-05,
+      "loss": 1.5485,
+      "step": 4054
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.0019723865877713e-05,
+      "loss": 1.6068,
+      "step": 4055
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2e-05,
+      "loss": 1.5936,
+      "step": 4056
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.998027613412229e-05,
+      "loss": 1.5364,
+      "step": 4057
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9960552268244576e-05,
+      "loss": 1.438,
+      "step": 4058
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9940828402366864e-05,
+      "loss": 1.5224,
+      "step": 4059
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9921104536489153e-05,
+      "loss": 1.5321,
+      "step": 4060
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9901380670611442e-05,
+      "loss": 1.5962,
+      "step": 4061
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.9881656804733727e-05,
+      "loss": 1.3947,
+      "step": 4062
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.986193293885602e-05,
+      "loss": 1.481,
+      "step": 4063
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.9842209072978305e-05,
+      "loss": 1.6088,
+      "step": 4064
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9822485207100593e-05,
+      "loss": 1.566,
+      "step": 4065
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9802761341222882e-05,
+      "loss": 1.4306,
+      "step": 4066
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9783037475345167e-05,
+      "loss": 1.4425,
+      "step": 4067
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9763313609467456e-05,
+      "loss": 1.5176,
+      "step": 4068
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.5632,
+      "step": 4069
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9723865877712034e-05,
+      "loss": 1.4995,
+      "step": 4070
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.970414201183432e-05,
+      "loss": 1.4757,
+      "step": 4071
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.968441814595661e-05,
+      "loss": 1.5045,
+      "step": 4072
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9664694280078896e-05,
+      "loss": 1.5206,
+      "step": 4073
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9644970414201182e-05,
+      "loss": 1.4875,
+      "step": 4074
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9625246548323474e-05,
+      "loss": 1.5377,
+      "step": 4075
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.960552268244576e-05,
+      "loss": 1.4142,
+      "step": 4076
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9585798816568048e-05,
+      "loss": 1.4396,
+      "step": 4077
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9566074950690337e-05,
+      "loss": 1.4574,
+      "step": 4078
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9546351084812625e-05,
+      "loss": 1.521,
+      "step": 4079
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.952662721893491e-05,
+      "loss": 1.548,
+      "step": 4080
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.9506903353057203e-05,
+      "loss": 1.5249,
+      "step": 4081
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.5541,
+      "step": 4082
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9467455621301774e-05,
+      "loss": 1.4202,
+      "step": 4083
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9447731755424066e-05,
+      "loss": 1.4028,
+      "step": 4084
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.942800788954635e-05,
+      "loss": 1.5512,
+      "step": 4085
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.940828402366864e-05,
+      "loss": 1.4373,
+      "step": 4086
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.938856015779093e-05,
+      "loss": 1.4749,
+      "step": 4087
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9368836291913217e-05,
+      "loss": 1.4923,
+      "step": 4088
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.9349112426035502e-05,
+      "loss": 1.495,
+      "step": 4089
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.932938856015779e-05,
+      "loss": 1.5964,
+      "step": 4090
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.930966469428008e-05,
+      "loss": 1.5687,
+      "step": 4091
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9289940828402365e-05,
+      "loss": 1.4688,
+      "step": 4092
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9270216962524657e-05,
+      "loss": 1.4821,
+      "step": 4093
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9250493096646943e-05,
+      "loss": 1.5323,
+      "step": 4094
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.5593,
+      "step": 4095
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.921104536489152e-05,
+      "loss": 1.5117,
+      "step": 4096
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9191321499013805e-05,
+      "loss": 1.5514,
+      "step": 4097
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.9171597633136094e-05,
+      "loss": 1.534,
+      "step": 4098
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.9151873767258383e-05,
+      "loss": 1.4936,
+      "step": 4099
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.913214990138067e-05,
+      "loss": 1.4988,
+      "step": 4100
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.911242603550296e-05,
+      "loss": 1.5185,
+      "step": 4101
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.909270216962525e-05,
+      "loss": 1.5722,
+      "step": 4102
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9072978303747534e-05,
+      "loss": 1.597,
+      "step": 4103
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9053254437869823e-05,
+      "loss": 1.4537,
+      "step": 4104
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9033530571992112e-05,
+      "loss": 1.5169,
+      "step": 4105
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.9013806706114397e-05,
+      "loss": 1.5272,
+      "step": 4106
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.899408284023669e-05,
+      "loss": 1.5185,
+      "step": 4107
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.4632,
+      "step": 4108
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.8954635108481263e-05,
+      "loss": 1.5662,
+      "step": 4109
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.8934911242603552e-05,
+      "loss": 1.5147,
+      "step": 4110
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.891518737672584e-05,
+      "loss": 1.5018,
+      "step": 4111
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.8895463510848126e-05,
+      "loss": 1.4931,
+      "step": 4112
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.8875739644970415e-05,
+      "loss": 1.595,
+      "step": 4113
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.8856015779092704e-05,
+      "loss": 1.5018,
+      "step": 4114
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.883629191321499e-05,
+      "loss": 1.5057,
+      "step": 4115
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.881656804733728e-05,
+      "loss": 1.5517,
+      "step": 4116
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.8796844181459566e-05,
+      "loss": 1.5801,
+      "step": 4117
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.8777120315581855e-05,
+      "loss": 1.6221,
+      "step": 4118
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.8757396449704144e-05,
+      "loss": 1.496,
+      "step": 4119
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.8737672583826433e-05,
+      "loss": 1.5247,
+      "step": 4120
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.5099,
+      "step": 4121
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.8698224852071007e-05,
+      "loss": 1.5412,
+      "step": 4122
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.8678500986193295e-05,
+      "loss": 1.5734,
+      "step": 4123
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.865877712031558e-05,
+      "loss": 1.5339,
+      "step": 4124
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.8639053254437873e-05,
+      "loss": 1.5401,
+      "step": 4125
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.8619329388560158e-05,
+      "loss": 1.5508,
+      "step": 4126
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.8599605522682447e-05,
+      "loss": 1.5531,
+      "step": 4127
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.8579881656804736e-05,
+      "loss": 1.4241,
+      "step": 4128
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.856015779092702e-05,
+      "loss": 1.5448,
+      "step": 4129
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.854043392504931e-05,
+      "loss": 1.5039,
+      "step": 4130
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.85207100591716e-05,
+      "loss": 1.5151,
+      "step": 4131
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.8500986193293887e-05,
+      "loss": 1.6209,
+      "step": 4132
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.8481262327416172e-05,
+      "loss": 1.5379,
+      "step": 4133
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.5718,
+      "step": 4134
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.844181459566075e-05,
+      "loss": 1.5115,
+      "step": 4135
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.842209072978304e-05,
+      "loss": 1.6294,
+      "step": 4136
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.8402366863905327e-05,
+      "loss": 1.5502,
+      "step": 4137
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.8382642998027613e-05,
+      "loss": 1.484,
+      "step": 4138
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.83629191321499e-05,
+      "loss": 1.4592,
+      "step": 4139
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.834319526627219e-05,
+      "loss": 1.5247,
+      "step": 4140
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.832347140039448e-05,
+      "loss": 1.5581,
+      "step": 4141
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.8303747534516764e-05,
+      "loss": 1.5239,
+      "step": 4142
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.8284023668639056e-05,
+      "loss": 1.5446,
+      "step": 4143
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.8264299802761342e-05,
+      "loss": 1.552,
+      "step": 4144
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.824457593688363e-05,
+      "loss": 1.5326,
+      "step": 4145
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.822485207100592e-05,
+      "loss": 1.4967,
+      "step": 4146
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.8205128205128204e-05,
+      "loss": 1.5451,
+      "step": 4147
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.8185404339250493e-05,
+      "loss": 1.5132,
+      "step": 4148
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.8165680473372782e-05,
+      "loss": 1.5085,
+      "step": 4149
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.814595660749507e-05,
+      "loss": 1.5028,
+      "step": 4150
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.812623274161736e-05,
+      "loss": 1.5593,
+      "step": 4151
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.8106508875739645e-05,
+      "loss": 1.5202,
+      "step": 4152
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.8086785009861933e-05,
+      "loss": 1.4873,
+      "step": 4153
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.8067061143984222e-05,
+      "loss": 1.599,
+      "step": 4154
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.804733727810651e-05,
+      "loss": 1.4669,
+      "step": 4155
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.8027613412228796e-05,
+      "loss": 1.5187,
+      "step": 4156
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.8007889546351085e-05,
+      "loss": 1.4357,
+      "step": 4157
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7988165680473374e-05,
+      "loss": 1.4705,
+      "step": 4158
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7968441814595662e-05,
+      "loss": 1.4793,
+      "step": 4159
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.5264,
+      "step": 4160
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.7928994082840236e-05,
+      "loss": 1.5674,
+      "step": 4161
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.7909270216962525e-05,
+      "loss": 1.4315,
+      "step": 4162
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.7889546351084814e-05,
+      "loss": 1.5988,
+      "step": 4163
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.7869822485207103e-05,
+      "loss": 1.4935,
+      "step": 4164
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.7850098619329388e-05,
+      "loss": 1.4646,
+      "step": 4165
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.783037475345168e-05,
+      "loss": 1.4999,
+      "step": 4166
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.7810650887573965e-05,
+      "loss": 1.5417,
+      "step": 4167
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.779092702169625e-05,
+      "loss": 1.4964,
+      "step": 4168
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.7771203155818543e-05,
+      "loss": 1.5509,
+      "step": 4169
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 1.4779,
+      "step": 4170
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.7731755424063117e-05,
+      "loss": 1.4014,
+      "step": 4171
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.7712031558185406e-05,
+      "loss": 1.5449,
+      "step": 4172
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.5072,
+      "step": 4173
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.767258382642998e-05,
+      "loss": 1.498,
+      "step": 4174
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.7652859960552272e-05,
+      "loss": 1.6421,
+      "step": 4175
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.7633136094674557e-05,
+      "loss": 1.6284,
+      "step": 4176
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.7613412228796843e-05,
+      "loss": 1.5651,
+      "step": 4177
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.7593688362919135e-05,
+      "loss": 1.5575,
+      "step": 4178
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.757396449704142e-05,
+      "loss": 1.4848,
+      "step": 4179
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.755424063116371e-05,
+      "loss": 1.4944,
+      "step": 4180
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.7534516765285997e-05,
+      "loss": 1.4636,
+      "step": 4181
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.7514792899408286e-05,
+      "loss": 1.5047,
+      "step": 4182
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.749506903353057e-05,
+      "loss": 1.4512,
+      "step": 4183
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.747534516765286e-05,
+      "loss": 1.596,
+      "step": 4184
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.745562130177515e-05,
+      "loss": 1.4302,
+      "step": 4185
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.7435897435897434e-05,
+      "loss": 1.4777,
+      "step": 4186
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.7416173570019726e-05,
+      "loss": 1.4566,
+      "step": 4187
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.7396449704142012e-05,
+      "loss": 1.5828,
+      "step": 4188
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.73767258382643e-05,
+      "loss": 1.4574,
+      "step": 4189
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.735700197238659e-05,
+      "loss": 1.4208,
+      "step": 4190
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.7337278106508875e-05,
+      "loss": 1.4113,
+      "step": 4191
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.7317554240631163e-05,
+      "loss": 1.5423,
+      "step": 4192
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.7297830374753452e-05,
+      "loss": 1.4927,
+      "step": 4193
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.727810650887574e-05,
+      "loss": 1.515,
+      "step": 4194
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.725838264299803e-05,
+      "loss": 1.5461,
+      "step": 4195
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.7238658777120318e-05,
+      "loss": 1.4103,
+      "step": 4196
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.7218934911242603e-05,
+      "loss": 1.541,
+      "step": 4197
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.7199211045364892e-05,
+      "loss": 1.5678,
+      "step": 4198
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.4378,
+      "step": 4199
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.7159763313609466e-05,
+      "loss": 1.3859,
+      "step": 4200
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.7140039447731755e-05,
+      "loss": 1.5438,
+      "step": 4201
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.7120315581854044e-05,
+      "loss": 1.6406,
+      "step": 4202
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.7100591715976332e-05,
+      "loss": 1.4286,
+      "step": 4203
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.708086785009862e-05,
+      "loss": 1.6236,
+      "step": 4204
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.706114398422091e-05,
+      "loss": 1.5294,
+      "step": 4205
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.7041420118343195e-05,
+      "loss": 1.5139,
+      "step": 4206
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.7021696252465484e-05,
+      "loss": 1.5072,
+      "step": 4207
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.7001972386587773e-05,
+      "loss": 1.5861,
+      "step": 4208
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.6982248520710058e-05,
+      "loss": 1.4916,
+      "step": 4209
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.696252465483235e-05,
+      "loss": 1.5846,
+      "step": 4210
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6942800788954635e-05,
+      "loss": 1.5076,
+      "step": 4211
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.6003,
+      "step": 4212
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6903353057199213e-05,
+      "loss": 1.4954,
+      "step": 4213
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.68836291913215e-05,
+      "loss": 1.5283,
+      "step": 4214
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.6863905325443787e-05,
+      "loss": 1.5473,
+      "step": 4215
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.6844181459566076e-05,
+      "loss": 1.4538,
+      "step": 4216
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.6824457593688364e-05,
+      "loss": 1.4968,
+      "step": 4217
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.680473372781065e-05,
+      "loss": 1.5019,
+      "step": 4218
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.6785009861932942e-05,
+      "loss": 1.5245,
+      "step": 4219
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6765285996055227e-05,
+      "loss": 1.5367,
+      "step": 4220
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6745562130177516e-05,
+      "loss": 1.356,
+      "step": 4221
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6725838264299805e-05,
+      "loss": 1.5689,
+      "step": 4222
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.670611439842209e-05,
+      "loss": 1.5526,
+      "step": 4223
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.668639053254438e-05,
+      "loss": 1.5399,
+      "step": 4224
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.5434,
+      "step": 4225
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.6646942800788956e-05,
+      "loss": 1.4228,
+      "step": 4226
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.662721893491124e-05,
+      "loss": 1.5076,
+      "step": 4227
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6607495069033534e-05,
+      "loss": 1.521,
+      "step": 4228
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.658777120315582e-05,
+      "loss": 1.4152,
+      "step": 4229
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6568047337278108e-05,
+      "loss": 1.4537,
+      "step": 4230
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.6548323471400396e-05,
+      "loss": 1.4465,
+      "step": 4231
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.6528599605522682e-05,
+      "loss": 1.3849,
+      "step": 4232
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.650887573964497e-05,
+      "loss": 1.5079,
+      "step": 4233
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.648915187376726e-05,
+      "loss": 1.5592,
+      "step": 4234
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.6469428007889548e-05,
+      "loss": 1.5028,
+      "step": 4235
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.6449704142011833e-05,
+      "loss": 1.5578,
+      "step": 4236
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.6429980276134125e-05,
+      "loss": 1.5864,
+      "step": 4237
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.5559,
+      "step": 4238
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.63905325443787e-05,
+      "loss": 1.5133,
+      "step": 4239
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.6370808678500988e-05,
+      "loss": 1.5003,
+      "step": 4240
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.6351084812623274e-05,
+      "loss": 1.5315,
+      "step": 4241
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.6331360946745562e-05,
+      "loss": 1.4813,
+      "step": 4242
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.631163708086785e-05,
+      "loss": 1.5168,
+      "step": 4243
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.629191321499014e-05,
+      "loss": 1.5972,
+      "step": 4244
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6272189349112425e-05,
+      "loss": 1.5016,
+      "step": 4245
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6252465483234714e-05,
+      "loss": 1.4523,
+      "step": 4246
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6232741617357002e-05,
+      "loss": 1.5243,
+      "step": 4247
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.621301775147929e-05,
+      "loss": 1.6179,
+      "step": 4248
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.619329388560158e-05,
+      "loss": 1.6028,
+      "step": 4249
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6173570019723865e-05,
+      "loss": 1.4539,
+      "step": 4250
+    },
+    {
+      "epoch": 12.57,
+      "eval_loss": 1.4127225875854492,
+      "eval_runtime": 69.3164,
+      "eval_samples_per_second": 321.28,
+      "eval_steps_per_second": 1.673,
+      "step": 4250
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.4997,
+      "step": 4251
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.6134122287968443e-05,
+      "loss": 1.4373,
+      "step": 4252
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.611439842209073e-05,
+      "loss": 1.4793,
+      "step": 4253
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.609467455621302e-05,
+      "loss": 1.4426,
+      "step": 4254
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6074950690335306e-05,
+      "loss": 1.4371,
+      "step": 4255
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6055226824457594e-05,
+      "loss": 1.4883,
+      "step": 4256
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6035502958579883e-05,
+      "loss": 1.4728,
+      "step": 4257
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.601577909270217e-05,
+      "loss": 1.4832,
+      "step": 4258
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.5996055226824457e-05,
+      "loss": 1.4007,
+      "step": 4259
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.5976331360946746e-05,
+      "loss": 1.4617,
+      "step": 4260
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5956607495069034e-05,
+      "loss": 1.4971,
+      "step": 4261
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.593688362919132e-05,
+      "loss": 1.4827,
+      "step": 4262
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5917159763313612e-05,
+      "loss": 1.5495,
+      "step": 4263
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.5454,
+      "step": 4264
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5877712031558186e-05,
+      "loss": 1.5385,
+      "step": 4265
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5857988165680475e-05,
+      "loss": 1.4699,
+      "step": 4266
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5838264299802763e-05,
+      "loss": 1.4406,
+      "step": 4267
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.581854043392505e-05,
+      "loss": 1.5118,
+      "step": 4268
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.579881656804734e-05,
+      "loss": 1.5187,
+      "step": 4269
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.5779092702169626e-05,
+      "loss": 1.503,
+      "step": 4270
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.575936883629191e-05,
+      "loss": 1.5519,
+      "step": 4271
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.5739644970414204e-05,
+      "loss": 1.4622,
+      "step": 4272
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.571992110453649e-05,
+      "loss": 1.4861,
+      "step": 4273
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.5700197238658778e-05,
+      "loss": 1.5439,
+      "step": 4274
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.5680473372781066e-05,
+      "loss": 1.5291,
+      "step": 4275
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.5660749506903355e-05,
+      "loss": 1.5214,
+      "step": 4276
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.564102564102564e-05,
+      "loss": 1.4889,
+      "step": 4277
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.562130177514793e-05,
+      "loss": 1.4854,
+      "step": 4278
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.5601577909270218e-05,
+      "loss": 1.5007,
+      "step": 4279
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.5581854043392503e-05,
+      "loss": 1.3762,
+      "step": 4280
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.5562130177514795e-05,
+      "loss": 1.4904,
+      "step": 4281
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.554240631163708e-05,
+      "loss": 1.5093,
+      "step": 4282
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.552268244575937e-05,
+      "loss": 1.478,
+      "step": 4283
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.5502958579881658e-05,
+      "loss": 1.4785,
+      "step": 4284
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5483234714003947e-05,
+      "loss": 1.5555,
+      "step": 4285
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5463510848126232e-05,
+      "loss": 1.5286,
+      "step": 4286
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.544378698224852e-05,
+      "loss": 1.4648,
+      "step": 4287
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.542406311637081e-05,
+      "loss": 1.4908,
+      "step": 4288
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5404339250493095e-05,
+      "loss": 1.4917,
+      "step": 4289
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.525,
+      "step": 4290
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.5364891518737673e-05,
+      "loss": 1.4672,
+      "step": 4291
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.534516765285996e-05,
+      "loss": 1.4288,
+      "step": 4292
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.532544378698225e-05,
+      "loss": 1.5937,
+      "step": 4293
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.5305719921104535e-05,
+      "loss": 1.5354,
+      "step": 4294
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5285996055226824e-05,
+      "loss": 1.5589,
+      "step": 4295
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5266272189349113e-05,
+      "loss": 1.5225,
+      "step": 4296
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5246548323471401e-05,
+      "loss": 1.4931,
+      "step": 4297
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5226824457593689e-05,
+      "loss": 1.5184,
+      "step": 4298
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5207100591715979e-05,
+      "loss": 1.4811,
+      "step": 4299
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5187376725838264e-05,
+      "loss": 1.3628,
+      "step": 4300
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5167652859960551e-05,
+      "loss": 1.4928,
+      "step": 4301
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5147928994082842e-05,
+      "loss": 1.5553,
+      "step": 4302
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.5047,
+      "step": 4303
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5108481262327417e-05,
+      "loss": 1.5047,
+      "step": 4304
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.5088757396449705e-05,
+      "loss": 1.5257,
+      "step": 4305
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.5069033530571993e-05,
+      "loss": 1.4726,
+      "step": 4306
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.504930966469428e-05,
+      "loss": 1.6115,
+      "step": 4307
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.502958579881657e-05,
+      "loss": 1.4472,
+      "step": 4308
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.5009861932938856e-05,
+      "loss": 1.5375,
+      "step": 4309
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4990138067061143e-05,
+      "loss": 1.3987,
+      "step": 4310
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4970414201183433e-05,
+      "loss": 1.4239,
+      "step": 4311
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.495069033530572e-05,
+      "loss": 1.5228,
+      "step": 4312
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.493096646942801e-05,
+      "loss": 1.5467,
+      "step": 4313
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.4911242603550296e-05,
+      "loss": 1.5023,
+      "step": 4314
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.4891518737672585e-05,
+      "loss": 1.4661,
+      "step": 4315
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.4871794871794872e-05,
+      "loss": 1.5034,
+      "step": 4316
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.4852071005917159e-05,
+      "loss": 1.4954,
+      "step": 4317
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.483234714003945e-05,
+      "loss": 1.498,
+      "step": 4318
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.4812623274161735e-05,
+      "loss": 1.4703,
+      "step": 4319
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.4792899408284025e-05,
+      "loss": 1.5557,
+      "step": 4320
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.4773175542406312e-05,
+      "loss": 1.4528,
+      "step": 4321
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.4753451676528601e-05,
+      "loss": 1.4365,
+      "step": 4322
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.4733727810650888e-05,
+      "loss": 1.429,
+      "step": 4323
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.4714003944773177e-05,
+      "loss": 1.4518,
+      "step": 4324
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.4694280078895464e-05,
+      "loss": 1.4352,
+      "step": 4325
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.467455621301775e-05,
+      "loss": 1.5836,
+      "step": 4326
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.4654832347140041e-05,
+      "loss": 1.5365,
+      "step": 4327
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.4635108481262327e-05,
+      "loss": 1.5222,
+      "step": 4328
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.4615384615384617e-05,
+      "loss": 1.4345,
+      "step": 4329
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.4595660749506904e-05,
+      "loss": 1.5458,
+      "step": 4330
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.4575936883629193e-05,
+      "loss": 1.4795,
+      "step": 4331
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.455621301775148e-05,
+      "loss": 1.4989,
+      "step": 4332
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.4536489151873767e-05,
+      "loss": 1.4408,
+      "step": 4333
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.4516765285996056e-05,
+      "loss": 1.4602,
+      "step": 4334
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.4497041420118343e-05,
+      "loss": 1.5116,
+      "step": 4335
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.4477317554240633e-05,
+      "loss": 1.4874,
+      "step": 4336
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.445759368836292e-05,
+      "loss": 1.5332,
+      "step": 4337
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.4437869822485209e-05,
+      "loss": 1.5278,
+      "step": 4338
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.4418145956607496e-05,
+      "loss": 1.4024,
+      "step": 4339
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.4398422090729784e-05,
+      "loss": 1.4589,
+      "step": 4340
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.4378698224852072e-05,
+      "loss": 1.5949,
+      "step": 4341
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.4358974358974359e-05,
+      "loss": 1.3883,
+      "step": 4342
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.4339250493096649e-05,
+      "loss": 1.6363,
+      "step": 4343
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.4319526627218934e-05,
+      "loss": 1.5038,
+      "step": 4344
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.4299802761341225e-05,
+      "loss": 1.5245,
+      "step": 4345
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.4280078895463512e-05,
+      "loss": 1.5139,
+      "step": 4346
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.42603550295858e-05,
+      "loss": 1.4926,
+      "step": 4347
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.4240631163708088e-05,
+      "loss": 1.4609,
+      "step": 4348
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4220907297830375e-05,
+      "loss": 1.5655,
+      "step": 4349
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 1.5513,
+      "step": 4350
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.418145956607495e-05,
+      "loss": 1.537,
+      "step": 4351
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.416173570019724e-05,
+      "loss": 1.5518,
+      "step": 4352
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4142011834319526e-05,
+      "loss": 1.5066,
+      "step": 4353
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4122287968441816e-05,
+      "loss": 1.554,
+      "step": 4354
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.5868,
+      "step": 4355
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.408284023668639e-05,
+      "loss": 1.3856,
+      "step": 4356
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.406311637080868e-05,
+      "loss": 1.5059,
+      "step": 4357
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4043392504930966e-05,
+      "loss": 1.5634,
+      "step": 4358
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4023668639053255e-05,
+      "loss": 1.6097,
+      "step": 4359
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4003944773175542e-05,
+      "loss": 1.4543,
+      "step": 4360
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.3984220907297832e-05,
+      "loss": 1.482,
+      "step": 4361
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.396449704142012e-05,
+      "loss": 1.4936,
+      "step": 4362
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.3944773175542408e-05,
+      "loss": 1.658,
+      "step": 4363
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.3925049309664695e-05,
+      "loss": 1.4461,
+      "step": 4364
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.3905325443786982e-05,
+      "loss": 1.4767,
+      "step": 4365
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.3885601577909271e-05,
+      "loss": 1.5614,
+      "step": 4366
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.3865877712031558e-05,
+      "loss": 1.4945,
+      "step": 4367
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.6189,
+      "step": 4368
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.3826429980276134e-05,
+      "loss": 1.5819,
+      "step": 4369
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.3806706114398424e-05,
+      "loss": 1.5447,
+      "step": 4370
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.3786982248520711e-05,
+      "loss": 1.4642,
+      "step": 4371
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.3767258382642997e-05,
+      "loss": 1.483,
+      "step": 4372
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.3747534516765287e-05,
+      "loss": 1.5282,
+      "step": 4373
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.3727810650887574e-05,
+      "loss": 1.5253,
+      "step": 4374
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.3708086785009863e-05,
+      "loss": 1.4755,
+      "step": 4375
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.368836291913215e-05,
+      "loss": 1.4912,
+      "step": 4376
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.366863905325444e-05,
+      "loss": 1.5332,
+      "step": 4377
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.3648915187376726e-05,
+      "loss": 1.5044,
+      "step": 4378
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.3629191321499016e-05,
+      "loss": 1.4001,
+      "step": 4379
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.3609467455621303e-05,
+      "loss": 1.4888,
+      "step": 4380
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.358974358974359e-05,
+      "loss": 1.6381,
+      "step": 4381
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.3570019723865879e-05,
+      "loss": 1.5089,
+      "step": 4382
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.3550295857988166e-05,
+      "loss": 1.4796,
+      "step": 4383
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.3530571992110455e-05,
+      "loss": 1.5374,
+      "step": 4384
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.3510848126232742e-05,
+      "loss": 1.4941,
+      "step": 4385
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.3491124260355032e-05,
+      "loss": 1.6142,
+      "step": 4386
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.3471400394477319e-05,
+      "loss": 1.4876,
+      "step": 4387
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.3451676528599604e-05,
+      "loss": 1.5567,
+      "step": 4388
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.3431952662721895e-05,
+      "loss": 1.5013,
+      "step": 4389
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.3412228796844182e-05,
+      "loss": 1.5129,
+      "step": 4390
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.339250493096647e-05,
+      "loss": 1.4824,
+      "step": 4391
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.3372781065088758e-05,
+      "loss": 1.4687,
+      "step": 4392
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.3353057199211046e-05,
+      "loss": 1.5359,
+      "step": 4393
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.4585,
+      "step": 4394
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.3313609467455624e-05,
+      "loss": 1.408,
+      "step": 4395
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.329388560157791e-05,
+      "loss": 1.5439,
+      "step": 4396
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.3274161735700196e-05,
+      "loss": 1.5025,
+      "step": 4397
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.3254437869822487e-05,
+      "loss": 1.4338,
+      "step": 4398
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.3234714003944774e-05,
+      "loss": 1.5286,
+      "step": 4399
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.3214990138067062e-05,
+      "loss": 1.4862,
+      "step": 4400
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.319526627218935e-05,
+      "loss": 1.5425,
+      "step": 4401
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.317554240631164e-05,
+      "loss": 1.482,
+      "step": 4402
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.3155818540433925e-05,
+      "loss": 1.5337,
+      "step": 4403
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.3136094674556212e-05,
+      "loss": 1.682,
+      "step": 4404
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.3116370808678502e-05,
+      "loss": 1.4914,
+      "step": 4405
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.309664694280079e-05,
+      "loss": 1.4931,
+      "step": 4406
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.4782,
+      "step": 4407
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.3057199211045365e-05,
+      "loss": 1.5732,
+      "step": 4408
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.3037475345167654e-05,
+      "loss": 1.5009,
+      "step": 4409
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.3017751479289941e-05,
+      "loss": 1.5962,
+      "step": 4410
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.2998027613412228e-05,
+      "loss": 1.5356,
+      "step": 4411
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.2978303747534517e-05,
+      "loss": 1.5665,
+      "step": 4412
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.2958579881656804e-05,
+      "loss": 1.4855,
+      "step": 4413
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.2938856015779094e-05,
+      "loss": 1.61,
+      "step": 4414
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.2919132149901381e-05,
+      "loss": 1.5796,
+      "step": 4415
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.289940828402367e-05,
+      "loss": 1.5221,
+      "step": 4416
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.2879684418145957e-05,
+      "loss": 1.4913,
+      "step": 4417
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.2859960552268246e-05,
+      "loss": 1.5283,
+      "step": 4418
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.2840236686390533e-05,
+      "loss": 1.5149,
+      "step": 4419
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 1.6281,
+      "step": 4420
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.280078895463511e-05,
+      "loss": 1.5614,
+      "step": 4421
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.2781065088757396e-05,
+      "loss": 1.4296,
+      "step": 4422
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.2761341222879686e-05,
+      "loss": 1.4523,
+      "step": 4423
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.2741617357001973e-05,
+      "loss": 1.4713,
+      "step": 4424
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.2721893491124262e-05,
+      "loss": 1.3836,
+      "step": 4425
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.2702169625246549e-05,
+      "loss": 1.5866,
+      "step": 4426
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.2682445759368836e-05,
+      "loss": 1.5002,
+      "step": 4427
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.2662721893491125e-05,
+      "loss": 1.5102,
+      "step": 4428
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.2642998027613412e-05,
+      "loss": 1.5146,
+      "step": 4429
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.2623274161735702e-05,
+      "loss": 1.5274,
+      "step": 4430
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.2603550295857989e-05,
+      "loss": 1.5167,
+      "step": 4431
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.2583826429980278e-05,
+      "loss": 1.588,
+      "step": 4432
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.4729,
+      "step": 4433
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.2544378698224854e-05,
+      "loss": 1.585,
+      "step": 4434
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.252465483234714e-05,
+      "loss": 1.4174,
+      "step": 4435
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.2504930966469428e-05,
+      "loss": 1.5354,
+      "step": 4436
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.2485207100591716e-05,
+      "loss": 1.4194,
+      "step": 4437
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.2465483234714005e-05,
+      "loss": 1.468,
+      "step": 4438
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.2445759368836292e-05,
+      "loss": 1.5739,
+      "step": 4439
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.242603550295858e-05,
+      "loss": 1.5346,
+      "step": 4440
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.2406311637080868e-05,
+      "loss": 1.5406,
+      "step": 4441
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.2386587771203157e-05,
+      "loss": 1.3586,
+      "step": 4442
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.2366863905325445e-05,
+      "loss": 1.5062,
+      "step": 4443
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.2347140039447732e-05,
+      "loss": 1.5051,
+      "step": 4444
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.2327416173570021e-05,
+      "loss": 1.5155,
+      "step": 4445
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.4375,
+      "step": 4446
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.2287968441814595e-05,
+      "loss": 1.4258,
+      "step": 4447
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.2268244575936884e-05,
+      "loss": 1.5,
+      "step": 4448
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.2248520710059173e-05,
+      "loss": 1.5402,
+      "step": 4449
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.222879684418146e-05,
+      "loss": 1.5751,
+      "step": 4450
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.2209072978303748e-05,
+      "loss": 1.4664,
+      "step": 4451
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.2189349112426037e-05,
+      "loss": 1.5469,
+      "step": 4452
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.2169625246548324e-05,
+      "loss": 1.5547,
+      "step": 4453
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.2149901380670611e-05,
+      "loss": 1.5317,
+      "step": 4454
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.21301775147929e-05,
+      "loss": 1.5794,
+      "step": 4455
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.2110453648915187e-05,
+      "loss": 1.4884,
+      "step": 4456
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.2090729783037476e-05,
+      "loss": 1.51,
+      "step": 4457
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.2071005917159764e-05,
+      "loss": 1.4951,
+      "step": 4458
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 1.4528,
+      "step": 4459
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.203155818540434e-05,
+      "loss": 1.4795,
+      "step": 4460
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.2011834319526629e-05,
+      "loss": 1.4589,
+      "step": 4461
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1992110453648916e-05,
+      "loss": 1.5372,
+      "step": 4462
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1972386587771203e-05,
+      "loss": 1.402,
+      "step": 4463
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1952662721893492e-05,
+      "loss": 1.5279,
+      "step": 4464
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.193293885601578e-05,
+      "loss": 1.5442,
+      "step": 4465
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1913214990138067e-05,
+      "loss": 1.5387,
+      "step": 4466
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1893491124260356e-05,
+      "loss": 1.5206,
+      "step": 4467
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1873767258382645e-05,
+      "loss": 1.5165,
+      "step": 4468
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1854043392504932e-05,
+      "loss": 1.5633,
+      "step": 4469
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1834319526627219e-05,
+      "loss": 1.4873,
+      "step": 4470
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1814595660749508e-05,
+      "loss": 1.4568,
+      "step": 4471
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1794871794871795e-05,
+      "loss": 1.5036,
+      "step": 4472
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1775147928994083e-05,
+      "loss": 1.5199,
+      "step": 4473
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1755424063116372e-05,
+      "loss": 1.4326,
+      "step": 4474
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1735700197238659e-05,
+      "loss": 1.5925,
+      "step": 4475
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1715976331360948e-05,
+      "loss": 1.4351,
+      "step": 4476
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1696252465483237e-05,
+      "loss": 1.572,
+      "step": 4477
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1676528599605522e-05,
+      "loss": 1.4266,
+      "step": 4478
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.165680473372781e-05,
+      "loss": 1.5375,
+      "step": 4479
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.16370808678501e-05,
+      "loss": 1.3816,
+      "step": 4480
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1617357001972386e-05,
+      "loss": 1.5579,
+      "step": 4481
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1597633136094675e-05,
+      "loss": 1.4604,
+      "step": 4482
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1577909270216964e-05,
+      "loss": 1.4528,
+      "step": 4483
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.155818540433925e-05,
+      "loss": 1.4225,
+      "step": 4484
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.4957,
+      "step": 4485
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1518737672583827e-05,
+      "loss": 1.3856,
+      "step": 4486
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1499013806706115e-05,
+      "loss": 1.5849,
+      "step": 4487
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1479289940828402e-05,
+      "loss": 1.5095,
+      "step": 4488
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1459566074950691e-05,
+      "loss": 1.6027,
+      "step": 4489
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.143984220907298e-05,
+      "loss": 1.5778,
+      "step": 4490
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1420118343195267e-05,
+      "loss": 1.5374,
+      "step": 4491
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1400394477317556e-05,
+      "loss": 1.5647,
+      "step": 4492
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1380670611439844e-05,
+      "loss": 1.5798,
+      "step": 4493
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.136094674556213e-05,
+      "loss": 1.5163,
+      "step": 4494
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1341222879684418e-05,
+      "loss": 1.4523,
+      "step": 4495
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1321499013806707e-05,
+      "loss": 1.5288,
+      "step": 4496
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1301775147928994e-05,
+      "loss": 1.4928,
+      "step": 4497
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.4881,
+      "step": 4498
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1262327416173572e-05,
+      "loss": 1.6119,
+      "step": 4499
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 1.4719,
+      "step": 4500
+    },
+    {
+      "epoch": 13.31,
+      "eval_loss": 1.4102731943130493,
+      "eval_runtime": 69.4852,
+      "eval_samples_per_second": 320.5,
+      "eval_steps_per_second": 1.669,
+      "step": 4500
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1222879684418146e-05,
+      "loss": 1.5319,
+      "step": 4501
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1203155818540434e-05,
+      "loss": 1.5758,
+      "step": 4502
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1183431952662721e-05,
+      "loss": 1.5193,
+      "step": 4503
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.116370808678501e-05,
+      "loss": 1.3976,
+      "step": 4504
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1143984220907299e-05,
+      "loss": 1.5623,
+      "step": 4505
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1124260355029586e-05,
+      "loss": 1.4782,
+      "step": 4506
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1104536489151875e-05,
+      "loss": 1.5643,
+      "step": 4507
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1084812623274163e-05,
+      "loss": 1.5236,
+      "step": 4508
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.106508875739645e-05,
+      "loss": 1.4397,
+      "step": 4509
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1045364891518737e-05,
+      "loss": 1.5783,
+      "step": 4510
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1025641025641026e-05,
+      "loss": 1.5263,
+      "step": 4511
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1005917159763315e-05,
+      "loss": 1.4982,
+      "step": 4512
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.0986193293885602e-05,
+      "loss": 1.5102,
+      "step": 4513
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.096646942800789e-05,
+      "loss": 1.5041,
+      "step": 4514
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.094674556213018e-05,
+      "loss": 1.5178,
+      "step": 4515
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.0927021696252466e-05,
+      "loss": 1.5584,
+      "step": 4516
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.0907297830374753e-05,
+      "loss": 1.5301,
+      "step": 4517
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.0887573964497042e-05,
+      "loss": 1.4868,
+      "step": 4518
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.0867850098619329e-05,
+      "loss": 1.4577,
+      "step": 4519
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.0848126232741618e-05,
+      "loss": 1.5632,
+      "step": 4520
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.0828402366863907e-05,
+      "loss": 1.4598,
+      "step": 4521
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.0808678500986194e-05,
+      "loss": 1.5108,
+      "step": 4522
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.0788954635108482e-05,
+      "loss": 1.5648,
+      "step": 4523
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.0769230769230771e-05,
+      "loss": 1.3821,
+      "step": 4524
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.0749506903353056e-05,
+      "loss": 1.4677,
+      "step": 4525
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.0729783037475345e-05,
+      "loss": 1.585,
+      "step": 4526
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.0710059171597634e-05,
+      "loss": 1.4728,
+      "step": 4527
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.069033530571992e-05,
+      "loss": 1.4412,
+      "step": 4528
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.067061143984221e-05,
+      "loss": 1.5083,
+      "step": 4529
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.0650887573964498e-05,
+      "loss": 1.5206,
+      "step": 4530
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.0631163708086785e-05,
+      "loss": 1.5228,
+      "step": 4531
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.0611439842209074e-05,
+      "loss": 1.5545,
+      "step": 4532
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.0591715976331361e-05,
+      "loss": 1.4624,
+      "step": 4533
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.057199211045365e-05,
+      "loss": 1.5496,
+      "step": 4534
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.0552268244575937e-05,
+      "loss": 1.5362,
+      "step": 4535
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.0532544378698226e-05,
+      "loss": 1.5171,
+      "step": 4536
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 1.5534,
+      "step": 4537
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.0493096646942801e-05,
+      "loss": 1.4954,
+      "step": 4538
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.047337278106509e-05,
+      "loss": 1.4859,
+      "step": 4539
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.0453648915187377e-05,
+      "loss": 1.4528,
+      "step": 4540
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.0433925049309664e-05,
+      "loss": 1.4255,
+      "step": 4541
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.0414201183431953e-05,
+      "loss": 1.4854,
+      "step": 4542
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.0394477317554242e-05,
+      "loss": 1.5269,
+      "step": 4543
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.0374753451676529e-05,
+      "loss": 1.5569,
+      "step": 4544
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.0355029585798817e-05,
+      "loss": 1.4417,
+      "step": 4545
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.0335305719921106e-05,
+      "loss": 1.4147,
+      "step": 4546
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.0315581854043393e-05,
+      "loss": 1.4721,
+      "step": 4547
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.0295857988165682e-05,
+      "loss": 1.4688,
+      "step": 4548
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.0276134122287969e-05,
+      "loss": 1.4414,
+      "step": 4549
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.5879,
+      "step": 4550
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.0236686390532545e-05,
+      "loss": 1.5453,
+      "step": 4551
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.0216962524654833e-05,
+      "loss": 1.5071,
+      "step": 4552
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.019723865877712e-05,
+      "loss": 1.5494,
+      "step": 4553
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.0177514792899409e-05,
+      "loss": 1.5148,
+      "step": 4554
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.0157790927021698e-05,
+      "loss": 1.4966,
+      "step": 4555
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.0138067061143985e-05,
+      "loss": 1.5086,
+      "step": 4556
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.0118343195266272e-05,
+      "loss": 1.4476,
+      "step": 4557
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.009861932938856e-05,
+      "loss": 1.5852,
+      "step": 4558
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.0078895463510848e-05,
+      "loss": 1.5884,
+      "step": 4559
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.0059171597633136e-05,
+      "loss": 1.5983,
+      "step": 4560
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.0039447731755425e-05,
+      "loss": 1.5413,
+      "step": 4561
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.0019723865877712e-05,
+      "loss": 1.5489,
+      "step": 4562
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1e-05,
+      "loss": 1.52,
+      "step": 4563
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 9.980276134122288e-06,
+      "loss": 1.481,
+      "step": 4564
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 9.960552268244577e-06,
+      "loss": 1.515,
+      "step": 4565
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 9.940828402366864e-06,
+      "loss": 1.4688,
+      "step": 4566
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 9.921104536489152e-06,
+      "loss": 1.5169,
+      "step": 4567
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 9.901380670611441e-06,
+      "loss": 1.5418,
+      "step": 4568
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 9.881656804733728e-06,
+      "loss": 1.5893,
+      "step": 4569
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 9.861932938856017e-06,
+      "loss": 1.4654,
+      "step": 4570
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 9.842209072978306e-06,
+      "loss": 1.4535,
+      "step": 4571
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 9.822485207100591e-06,
+      "loss": 1.5634,
+      "step": 4572
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 9.80276134122288e-06,
+      "loss": 1.5744,
+      "step": 4573
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 9.783037475345168e-06,
+      "loss": 1.5237,
+      "step": 4574
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 9.763313609467455e-06,
+      "loss": 1.5229,
+      "step": 4575
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.4538,
+      "step": 4576
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 9.723865877712033e-06,
+      "loss": 1.4927,
+      "step": 4577
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 9.70414201183432e-06,
+      "loss": 1.5437,
+      "step": 4578
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 9.684418145956609e-06,
+      "loss": 1.4452,
+      "step": 4579
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 9.664694280078896e-06,
+      "loss": 1.5662,
+      "step": 4580
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 9.644970414201183e-06,
+      "loss": 1.4411,
+      "step": 4581
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 9.625246548323471e-06,
+      "loss": 1.4432,
+      "step": 4582
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 9.60552268244576e-06,
+      "loss": 1.4488,
+      "step": 4583
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 9.585798816568047e-06,
+      "loss": 1.5543,
+      "step": 4584
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 9.566074950690336e-06,
+      "loss": 1.5046,
+      "step": 4585
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 9.546351084812625e-06,
+      "loss": 1.4724,
+      "step": 4586
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 9.526627218934912e-06,
+      "loss": 1.5135,
+      "step": 4587
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 9.506903353057199e-06,
+      "loss": 1.5463,
+      "step": 4588
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.4529,
+      "step": 4589
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 9.467455621301776e-06,
+      "loss": 1.6195,
+      "step": 4590
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 9.447731755424063e-06,
+      "loss": 1.6467,
+      "step": 4591
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 9.428007889546352e-06,
+      "loss": 1.4606,
+      "step": 4592
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 9.40828402366864e-06,
+      "loss": 1.4812,
+      "step": 4593
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 9.388560157790928e-06,
+      "loss": 1.408,
+      "step": 4594
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 9.368836291913216e-06,
+      "loss": 1.5475,
+      "step": 4595
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 9.349112426035503e-06,
+      "loss": 1.4236,
+      "step": 4596
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 9.32938856015779e-06,
+      "loss": 1.5355,
+      "step": 4597
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 9.309664694280079e-06,
+      "loss": 1.4159,
+      "step": 4598
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 9.289940828402368e-06,
+      "loss": 1.4632,
+      "step": 4599
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 9.270216962524655e-06,
+      "loss": 1.5385,
+      "step": 4600
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 9.250493096646944e-06,
+      "loss": 1.5154,
+      "step": 4601
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.4405,
+      "step": 4602
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 9.21104536489152e-06,
+      "loss": 1.4594,
+      "step": 4603
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 9.191321499013806e-06,
+      "loss": 1.5343,
+      "step": 4604
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 9.171597633136095e-06,
+      "loss": 1.4666,
+      "step": 4605
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 9.151873767258382e-06,
+      "loss": 1.4538,
+      "step": 4606
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 9.132149901380671e-06,
+      "loss": 1.4032,
+      "step": 4607
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 9.11242603550296e-06,
+      "loss": 1.4715,
+      "step": 4608
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 9.092702169625247e-06,
+      "loss": 1.4757,
+      "step": 4609
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 9.072978303747535e-06,
+      "loss": 1.5301,
+      "step": 4610
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 9.053254437869822e-06,
+      "loss": 1.5518,
+      "step": 4611
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 9.033530571992111e-06,
+      "loss": 1.4914,
+      "step": 4612
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 9.013806706114398e-06,
+      "loss": 1.5893,
+      "step": 4613
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 8.994082840236687e-06,
+      "loss": 1.4901,
+      "step": 4614
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.431,
+      "step": 4615
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 8.954635108481263e-06,
+      "loss": 1.5575,
+      "step": 4616
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 8.934911242603551e-06,
+      "loss": 1.6481,
+      "step": 4617
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 8.91518737672584e-06,
+      "loss": 1.5141,
+      "step": 4618
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 8.895463510848125e-06,
+      "loss": 1.5286,
+      "step": 4619
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 8.875739644970414e-06,
+      "loss": 1.4458,
+      "step": 4620
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 8.856015779092703e-06,
+      "loss": 1.5639,
+      "step": 4621
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 8.83629191321499e-06,
+      "loss": 1.4643,
+      "step": 4622
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 8.816568047337279e-06,
+      "loss": 1.4249,
+      "step": 4623
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 8.796844181459567e-06,
+      "loss": 1.5165,
+      "step": 4624
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 8.777120315581854e-06,
+      "loss": 1.441,
+      "step": 4625
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 8.757396449704143e-06,
+      "loss": 1.5218,
+      "step": 4626
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 8.73767258382643e-06,
+      "loss": 1.4988,
+      "step": 4627
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 8.717948717948717e-06,
+      "loss": 1.4345,
+      "step": 4628
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 8.698224852071006e-06,
+      "loss": 1.4325,
+      "step": 4629
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 8.678500986193295e-06,
+      "loss": 1.5832,
+      "step": 4630
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 8.658777120315582e-06,
+      "loss": 1.5201,
+      "step": 4631
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 8.63905325443787e-06,
+      "loss": 1.4544,
+      "step": 4632
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 8.619329388560159e-06,
+      "loss": 1.5565,
+      "step": 4633
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 8.599605522682446e-06,
+      "loss": 1.5176,
+      "step": 4634
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 8.579881656804733e-06,
+      "loss": 1.4058,
+      "step": 4635
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 8.560157790927022e-06,
+      "loss": 1.4833,
+      "step": 4636
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 8.54043392504931e-06,
+      "loss": 1.515,
+      "step": 4637
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 8.520710059171598e-06,
+      "loss": 1.5557,
+      "step": 4638
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 8.500986193293886e-06,
+      "loss": 1.4854,
+      "step": 4639
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 8.481262327416175e-06,
+      "loss": 1.5795,
+      "step": 4640
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.4351,
+      "step": 4641
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 8.44181459566075e-06,
+      "loss": 1.4354,
+      "step": 4642
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 8.422090729783038e-06,
+      "loss": 1.4783,
+      "step": 4643
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 8.402366863905325e-06,
+      "loss": 1.4676,
+      "step": 4644
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 8.382642998027614e-06,
+      "loss": 1.4841,
+      "step": 4645
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 8.362919132149902e-06,
+      "loss": 1.4492,
+      "step": 4646
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 8.34319526627219e-06,
+      "loss": 1.4598,
+      "step": 4647
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 8.323471400394478e-06,
+      "loss": 1.5564,
+      "step": 4648
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 8.303747534516767e-06,
+      "loss": 1.4876,
+      "step": 4649
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 1.5112,
+      "step": 4650
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 8.264299802761341e-06,
+      "loss": 1.5159,
+      "step": 4651
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 8.24457593688363e-06,
+      "loss": 1.4748,
+      "step": 4652
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 8.224852071005917e-06,
+      "loss": 1.4625,
+      "step": 4653
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.5199,
+      "step": 4654
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 8.185404339250494e-06,
+      "loss": 1.5122,
+      "step": 4655
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 8.165680473372781e-06,
+      "loss": 1.5162,
+      "step": 4656
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 8.14595660749507e-06,
+      "loss": 1.5875,
+      "step": 4657
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 8.126232741617357e-06,
+      "loss": 1.4914,
+      "step": 4658
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 8.106508875739646e-06,
+      "loss": 1.5028,
+      "step": 4659
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 8.086785009861933e-06,
+      "loss": 1.4772,
+      "step": 4660
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 8.067061143984221e-06,
+      "loss": 1.4521,
+      "step": 4661
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 8.04733727810651e-06,
+      "loss": 1.4987,
+      "step": 4662
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 8.027613412228797e-06,
+      "loss": 1.5708,
+      "step": 4663
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 8.007889546351086e-06,
+      "loss": 1.4634,
+      "step": 4664
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 7.988165680473373e-06,
+      "loss": 1.5043,
+      "step": 4665
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 7.96844181459566e-06,
+      "loss": 1.5349,
+      "step": 4666
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.4501,
+      "step": 4667
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 7.928994082840237e-06,
+      "loss": 1.4839,
+      "step": 4668
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 7.909270216962524e-06,
+      "loss": 1.4788,
+      "step": 4669
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 7.889546351084813e-06,
+      "loss": 1.4325,
+      "step": 4670
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 7.869822485207102e-06,
+      "loss": 1.5616,
+      "step": 4671
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 7.850098619329389e-06,
+      "loss": 1.5384,
+      "step": 4672
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 7.830374753451678e-06,
+      "loss": 1.5243,
+      "step": 4673
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 7.810650887573965e-06,
+      "loss": 1.5067,
+      "step": 4674
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 7.790927021696252e-06,
+      "loss": 1.4136,
+      "step": 4675
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 7.77120315581854e-06,
+      "loss": 1.5292,
+      "step": 4676
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 7.751479289940829e-06,
+      "loss": 1.4561,
+      "step": 4677
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 7.731755424063116e-06,
+      "loss": 1.5028,
+      "step": 4678
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 7.712031558185405e-06,
+      "loss": 1.5155,
+      "step": 4679
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.4598,
+      "step": 4680
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 7.67258382642998e-06,
+      "loss": 1.4999,
+      "step": 4681
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 7.652859960552268e-06,
+      "loss": 1.5121,
+      "step": 4682
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 7.633136094674556e-06,
+      "loss": 1.5301,
+      "step": 4683
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 7.613412228796844e-06,
+      "loss": 1.5079,
+      "step": 4684
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 7.593688362919132e-06,
+      "loss": 1.5107,
+      "step": 4685
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 7.573964497041421e-06,
+      "loss": 1.4796,
+      "step": 4686
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 7.554240631163709e-06,
+      "loss": 1.528,
+      "step": 4687
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 7.534516765285997e-06,
+      "loss": 1.4729,
+      "step": 4688
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 7.514792899408285e-06,
+      "loss": 1.3823,
+      "step": 4689
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 7.4950690335305715e-06,
+      "loss": 1.5994,
+      "step": 4690
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 7.47534516765286e-06,
+      "loss": 1.567,
+      "step": 4691
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 7.455621301775148e-06,
+      "loss": 1.5709,
+      "step": 4692
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 7.435897435897436e-06,
+      "loss": 1.4886,
+      "step": 4693
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 7.416173570019725e-06,
+      "loss": 1.4158,
+      "step": 4694
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 7.396449704142013e-06,
+      "loss": 1.4831,
+      "step": 4695
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 7.3767258382643005e-06,
+      "loss": 1.5123,
+      "step": 4696
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 7.357001972386588e-06,
+      "loss": 1.5019,
+      "step": 4697
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 7.337278106508875e-06,
+      "loss": 1.5614,
+      "step": 4698
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 7.317554240631163e-06,
+      "loss": 1.5035,
+      "step": 4699
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 7.297830374753452e-06,
+      "loss": 1.473,
+      "step": 4700
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 7.27810650887574e-06,
+      "loss": 1.4877,
+      "step": 4701
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 7.258382642998028e-06,
+      "loss": 1.4283,
+      "step": 4702
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 7.2386587771203165e-06,
+      "loss": 1.5233,
+      "step": 4703
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 7.218934911242604e-06,
+      "loss": 1.464,
+      "step": 4704
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 7.199211045364892e-06,
+      "loss": 1.5517,
+      "step": 4705
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 7.179487179487179e-06,
+      "loss": 1.5264,
+      "step": 4706
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 7.159763313609467e-06,
+      "loss": 1.4836,
+      "step": 4707
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 7.140039447731756e-06,
+      "loss": 1.5136,
+      "step": 4708
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 7.120315581854044e-06,
+      "loss": 1.5405,
+      "step": 4709
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 7.100591715976332e-06,
+      "loss": 1.4694,
+      "step": 4710
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 7.08086785009862e-06,
+      "loss": 1.5195,
+      "step": 4711
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 7.061143984220908e-06,
+      "loss": 1.4639,
+      "step": 4712
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 7.041420118343195e-06,
+      "loss": 1.5696,
+      "step": 4713
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 7.021696252465483e-06,
+      "loss": 1.5226,
+      "step": 4714
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 7.001972386587771e-06,
+      "loss": 1.4682,
+      "step": 4715
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.98224852071006e-06,
+      "loss": 1.5026,
+      "step": 4716
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.962524654832348e-06,
+      "loss": 1.506,
+      "step": 4717
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.9428007889546355e-06,
+      "loss": 1.5298,
+      "step": 4718
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.5621,
+      "step": 4719
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.903353057199212e-06,
+      "loss": 1.4526,
+      "step": 4720
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.883629191321498e-06,
+      "loss": 1.5005,
+      "step": 4721
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.863905325443787e-06,
+      "loss": 1.4651,
+      "step": 4722
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.844181459566075e-06,
+      "loss": 1.5505,
+      "step": 4723
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.824457593688363e-06,
+      "loss": 1.4798,
+      "step": 4724
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.8047337278106515e-06,
+      "loss": 1.5577,
+      "step": 4725
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.785009861932939e-06,
+      "loss": 1.5206,
+      "step": 4726
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.765285996055227e-06,
+      "loss": 1.5925,
+      "step": 4727
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.745562130177516e-06,
+      "loss": 1.5099,
+      "step": 4728
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.725838264299802e-06,
+      "loss": 1.5857,
+      "step": 4729
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.706114398422091e-06,
+      "loss": 1.5007,
+      "step": 4730
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.686390532544379e-06,
+      "loss": 1.4499,
+      "step": 4731
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.4894,
+      "step": 4732
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.646942800788955e-06,
+      "loss": 1.505,
+      "step": 4733
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.627218934911243e-06,
+      "loss": 1.6157,
+      "step": 4734
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.607495069033531e-06,
+      "loss": 1.5003,
+      "step": 4735
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.58777120315582e-06,
+      "loss": 1.5386,
+      "step": 4736
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.568047337278106e-06,
+      "loss": 1.5206,
+      "step": 4737
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.548323471400395e-06,
+      "loss": 1.4733,
+      "step": 4738
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.528599605522683e-06,
+      "loss": 1.5458,
+      "step": 4739
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.5088757396449705e-06,
+      "loss": 1.3759,
+      "step": 4740
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.489151873767258e-06,
+      "loss": 1.5464,
+      "step": 4741
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.469428007889547e-06,
+      "loss": 1.5275,
+      "step": 4742
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.449704142011835e-06,
+      "loss": 1.5618,
+      "step": 4743
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.429980276134123e-06,
+      "loss": 1.5161,
+      "step": 4744
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 1.5037,
+      "step": 4745
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.390532544378698e-06,
+      "loss": 1.4424,
+      "step": 4746
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.3708086785009865e-06,
+      "loss": 1.4973,
+      "step": 4747
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.351084812623274e-06,
+      "loss": 1.5167,
+      "step": 4748
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.331360946745562e-06,
+      "loss": 1.4974,
+      "step": 4749
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.311637080867851e-06,
+      "loss": 1.4992,
+      "step": 4750
+    },
+    {
+      "epoch": 14.05,
+      "eval_loss": 1.4094032049179077,
+      "eval_runtime": 69.5782,
+      "eval_samples_per_second": 320.071,
+      "eval_steps_per_second": 1.667,
+      "step": 4750
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.291913214990139e-06,
+      "loss": 1.5984,
+      "step": 4751
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.272189349112427e-06,
+      "loss": 1.405,
+      "step": 4752
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.252465483234714e-06,
+      "loss": 1.5788,
+      "step": 4753
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.2327416173570025e-06,
+      "loss": 1.4384,
+      "step": 4754
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.21301775147929e-06,
+      "loss": 1.5015,
+      "step": 4755
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.193293885601578e-06,
+      "loss": 1.5134,
+      "step": 4756
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.173570019723866e-06,
+      "loss": 1.4876,
+      "step": 4757
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.5256,
+      "step": 4758
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.134122287968442e-06,
+      "loss": 1.5403,
+      "step": 4759
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.11439842209073e-06,
+      "loss": 1.5524,
+      "step": 4760
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.0946745562130185e-06,
+      "loss": 1.5337,
+      "step": 4761
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.0749506903353055e-06,
+      "loss": 1.5065,
+      "step": 4762
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.055226824457593e-06,
+      "loss": 1.5805,
+      "step": 4763
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.035502958579882e-06,
+      "loss": 1.6528,
+      "step": 4764
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.01577909270217e-06,
+      "loss": 1.4761,
+      "step": 4765
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 5.996055226824458e-06,
+      "loss": 1.564,
+      "step": 4766
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 5.976331360946746e-06,
+      "loss": 1.5328,
+      "step": 4767
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 5.956607495069034e-06,
+      "loss": 1.5358,
+      "step": 4768
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 5.936883629191322e-06,
+      "loss": 1.6012,
+      "step": 4769
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 5.917159763313609e-06,
+      "loss": 1.5372,
+      "step": 4770
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 5.897435897435897e-06,
+      "loss": 1.4105,
+      "step": 4771
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 5.877712031558186e-06,
+      "loss": 1.4671,
+      "step": 4772
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 5.857988165680474e-06,
+      "loss": 1.467,
+      "step": 4773
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 5.838264299802761e-06,
+      "loss": 1.4765,
+      "step": 4774
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 5.81854043392505e-06,
+      "loss": 1.4821,
+      "step": 4775
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 5.7988165680473375e-06,
+      "loss": 1.5277,
+      "step": 4776
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 5.779092702169625e-06,
+      "loss": 1.4904,
+      "step": 4777
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 5.759368836291913e-06,
+      "loss": 1.4022,
+      "step": 4778
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 5.739644970414201e-06,
+      "loss": 1.4523,
+      "step": 4779
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 5.71992110453649e-06,
+      "loss": 1.4252,
+      "step": 4780
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 5.700197238658778e-06,
+      "loss": 1.5071,
+      "step": 4781
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 5.680473372781065e-06,
+      "loss": 1.4713,
+      "step": 4782
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 5.6607495069033535e-06,
+      "loss": 1.5751,
+      "step": 4783
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.467,
+      "step": 4784
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 5.621301775147929e-06,
+      "loss": 1.4518,
+      "step": 4785
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 5.601577909270217e-06,
+      "loss": 1.5738,
+      "step": 4786
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 5.581854043392505e-06,
+      "loss": 1.5338,
+      "step": 4787
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 5.562130177514793e-06,
+      "loss": 1.5466,
+      "step": 4788
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 5.542406311637082e-06,
+      "loss": 1.5026,
+      "step": 4789
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 5.522682445759369e-06,
+      "loss": 1.5082,
+      "step": 4790
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 5.502958579881657e-06,
+      "loss": 1.5016,
+      "step": 4791
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 5.483234714003945e-06,
+      "loss": 1.5297,
+      "step": 4792
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 5.463510848126233e-06,
+      "loss": 1.5414,
+      "step": 4793
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 5.443786982248521e-06,
+      "loss": 1.5132,
+      "step": 4794
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 5.424063116370809e-06,
+      "loss": 1.4368,
+      "step": 4795
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 5.404339250493097e-06,
+      "loss": 1.5607,
+      "step": 4796
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 5.3846153846153855e-06,
+      "loss": 1.5111,
+      "step": 4797
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 5.3648915187376725e-06,
+      "loss": 1.4841,
+      "step": 4798
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 5.34516765285996e-06,
+      "loss": 1.5806,
+      "step": 4799
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 1.5408,
+      "step": 4800
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 5.305719921104537e-06,
+      "loss": 1.4758,
+      "step": 4801
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 5.285996055226825e-06,
+      "loss": 1.5556,
+      "step": 4802
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 5.266272189349113e-06,
+      "loss": 1.5019,
+      "step": 4803
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 5.246548323471401e-06,
+      "loss": 1.4918,
+      "step": 4804
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 5.2268244575936885e-06,
+      "loss": 1.5137,
+      "step": 4805
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 5.207100591715976e-06,
+      "loss": 1.5335,
+      "step": 4806
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 5.187376725838264e-06,
+      "loss": 1.4084,
+      "step": 4807
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 5.167652859960553e-06,
+      "loss": 1.487,
+      "step": 4808
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 5.147928994082841e-06,
+      "loss": 1.6162,
+      "step": 4809
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.4465,
+      "step": 4810
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 5.108481262327417e-06,
+      "loss": 1.4373,
+      "step": 4811
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 5.0887573964497045e-06,
+      "loss": 1.4168,
+      "step": 4812
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 5.069033530571992e-06,
+      "loss": 1.5013,
+      "step": 4813
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 5.04930966469428e-06,
+      "loss": 1.3873,
+      "step": 4814
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 5.029585798816568e-06,
+      "loss": 1.4884,
+      "step": 4815
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 5.009861932938856e-06,
+      "loss": 1.4456,
+      "step": 4816
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.990138067061144e-06,
+      "loss": 1.4467,
+      "step": 4817
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.970414201183432e-06,
+      "loss": 1.535,
+      "step": 4818
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.9506903353057205e-06,
+      "loss": 1.4916,
+      "step": 4819
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.930966469428008e-06,
+      "loss": 1.4436,
+      "step": 4820
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.9112426035502954e-06,
+      "loss": 1.4457,
+      "step": 4821
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.891518737672584e-06,
+      "loss": 1.4539,
+      "step": 4822
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.4705,
+      "step": 4823
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.85207100591716e-06,
+      "loss": 1.4756,
+      "step": 4824
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.832347140039448e-06,
+      "loss": 1.4614,
+      "step": 4825
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.812623274161736e-06,
+      "loss": 1.4738,
+      "step": 4826
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.7928994082840236e-06,
+      "loss": 1.5051,
+      "step": 4827
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.773175542406312e-06,
+      "loss": 1.3982,
+      "step": 4828
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.753451676528599e-06,
+      "loss": 1.5996,
+      "step": 4829
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.733727810650888e-06,
+      "loss": 1.419,
+      "step": 4830
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.714003944773176e-06,
+      "loss": 1.6281,
+      "step": 4831
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.694280078895464e-06,
+      "loss": 1.5096,
+      "step": 4832
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.674556213017752e-06,
+      "loss": 1.4688,
+      "step": 4833
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.6548323471400395e-06,
+      "loss": 1.4721,
+      "step": 4834
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.6351084812623274e-06,
+      "loss": 1.5087,
+      "step": 4835
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.4755,
+      "step": 4836
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.595660749506903e-06,
+      "loss": 1.4728,
+      "step": 4837
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.575936883629191e-06,
+      "loss": 1.5102,
+      "step": 4838
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.55621301775148e-06,
+      "loss": 1.6027,
+      "step": 4839
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.536489151873768e-06,
+      "loss": 1.4724,
+      "step": 4840
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.5167652859960555e-06,
+      "loss": 1.5002,
+      "step": 4841
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.497041420118343e-06,
+      "loss": 1.4557,
+      "step": 4842
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.477317554240631e-06,
+      "loss": 1.5635,
+      "step": 4843
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.45759368836292e-06,
+      "loss": 1.46,
+      "step": 4844
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.437869822485207e-06,
+      "loss": 1.5329,
+      "step": 4845
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.418145956607495e-06,
+      "loss": 1.517,
+      "step": 4846
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.398422090729784e-06,
+      "loss": 1.5802,
+      "step": 4847
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.3786982248520715e-06,
+      "loss": 1.4897,
+      "step": 4848
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.3589743589743586e-06,
+      "loss": 1.5879,
+      "step": 4849
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.339250493096647e-06,
+      "loss": 1.4931,
+      "step": 4850
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.319526627218935e-06,
+      "loss": 1.5616,
+      "step": 4851
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.299802761341223e-06,
+      "loss": 1.509,
+      "step": 4852
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.280078895463511e-06,
+      "loss": 1.5095,
+      "step": 4853
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.260355029585799e-06,
+      "loss": 1.4551,
+      "step": 4854
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2406311637080875e-06,
+      "loss": 1.4874,
+      "step": 4855
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.220907297830375e-06,
+      "loss": 1.5681,
+      "step": 4856
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.2011834319526624e-06,
+      "loss": 1.5489,
+      "step": 4857
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.181459566074951e-06,
+      "loss": 1.464,
+      "step": 4858
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.161735700197239e-06,
+      "loss": 1.5587,
+      "step": 4859
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.142011834319527e-06,
+      "loss": 1.4997,
+      "step": 4860
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.122287968441815e-06,
+      "loss": 1.5652,
+      "step": 4861
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.5012,
+      "step": 4862
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.0828402366863906e-06,
+      "loss": 1.5211,
+      "step": 4863
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.0631163708086784e-06,
+      "loss": 1.5765,
+      "step": 4864
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.043392504930966e-06,
+      "loss": 1.5418,
+      "step": 4865
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.023668639053255e-06,
+      "loss": 1.421,
+      "step": 4866
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.003944773175543e-06,
+      "loss": 1.5385,
+      "step": 4867
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.98422090729783e-06,
+      "loss": 1.5389,
+      "step": 4868
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.964497041420119e-06,
+      "loss": 1.4719,
+      "step": 4869
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9447731755424066e-06,
+      "loss": 1.3723,
+      "step": 4870
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9250493096646944e-06,
+      "loss": 1.5203,
+      "step": 4871
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.905325443786982e-06,
+      "loss": 1.5167,
+      "step": 4872
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.88560157790927e-06,
+      "loss": 1.4802,
+      "step": 4873
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.865877712031558e-06,
+      "loss": 1.5624,
+      "step": 4874
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.4021,
+      "step": 4875
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.826429980276134e-06,
+      "loss": 1.5516,
+      "step": 4876
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.806706114398422e-06,
+      "loss": 1.5251,
+      "step": 4877
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.7869822485207104e-06,
+      "loss": 1.5286,
+      "step": 4878
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.7672583826429983e-06,
+      "loss": 1.5264,
+      "step": 4879
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.7475345167652858e-06,
+      "loss": 1.5508,
+      "step": 4880
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.727810650887574e-06,
+      "loss": 1.5248,
+      "step": 4881
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.7080867850098624e-06,
+      "loss": 1.4757,
+      "step": 4882
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.6883629191321502e-06,
+      "loss": 1.5019,
+      "step": 4883
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.6686390532544377e-06,
+      "loss": 1.5014,
+      "step": 4884
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.648915187376726e-06,
+      "loss": 1.4407,
+      "step": 4885
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.629191321499014e-06,
+      "loss": 1.4362,
+      "step": 4886
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.609467455621302e-06,
+      "loss": 1.4353,
+      "step": 4887
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.5897435897435896e-06,
+      "loss": 1.5225,
+      "step": 4888
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.570019723865878e-06,
+      "loss": 1.4921,
+      "step": 4889
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.550295857988166e-06,
+      "loss": 1.5012,
+      "step": 4890
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.530571992110454e-06,
+      "loss": 1.4543,
+      "step": 4891
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.5108481262327416e-06,
+      "loss": 1.481,
+      "step": 4892
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.49112426035503e-06,
+      "loss": 1.5091,
+      "step": 4893
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.4714003944773178e-06,
+      "loss": 1.4999,
+      "step": 4894
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.451676528599606e-06,
+      "loss": 1.3969,
+      "step": 4895
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.4319526627218935e-06,
+      "loss": 1.3679,
+      "step": 4896
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.4122287968441814e-06,
+      "loss": 1.4016,
+      "step": 4897
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.3925049309664697e-06,
+      "loss": 1.476,
+      "step": 4898
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.372781065088758e-06,
+      "loss": 1.4645,
+      "step": 4899
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.3530571992110454e-06,
+      "loss": 1.5035,
+      "step": 4900
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.452,
+      "step": 4901
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.3136094674556216e-06,
+      "loss": 1.4936,
+      "step": 4902
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.29388560157791e-06,
+      "loss": 1.4109,
+      "step": 4903
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.2741617357001974e-06,
+      "loss": 1.4575,
+      "step": 4904
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.2544378698224853e-06,
+      "loss": 1.4934,
+      "step": 4905
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.2347140039447736e-06,
+      "loss": 1.6156,
+      "step": 4906
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.2149901380670614e-06,
+      "loss": 1.4454,
+      "step": 4907
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.195266272189349e-06,
+      "loss": 1.5062,
+      "step": 4908
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.175542406311637e-06,
+      "loss": 1.4824,
+      "step": 4909
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.1558185404339255e-06,
+      "loss": 1.5364,
+      "step": 4910
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.1360946745562134e-06,
+      "loss": 1.4698,
+      "step": 4911
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.1163708086785013e-06,
+      "loss": 1.4314,
+      "step": 4912
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.096646942800789e-06,
+      "loss": 1.4891,
+      "step": 4913
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.4364,
+      "step": 4914
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.057199211045365e-06,
+      "loss": 1.3672,
+      "step": 4915
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.0374753451676528e-06,
+      "loss": 1.544,
+      "step": 4916
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.017751479289941e-06,
+      "loss": 1.6116,
+      "step": 4917
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.998027613412229e-06,
+      "loss": 1.5827,
+      "step": 4918
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.978303747534517e-06,
+      "loss": 1.5072,
+      "step": 4919
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.9585798816568047e-06,
+      "loss": 1.5127,
+      "step": 4920
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.938856015779093e-06,
+      "loss": 1.5171,
+      "step": 4921
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.9191321499013805e-06,
+      "loss": 1.54,
+      "step": 4922
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8994082840236688e-06,
+      "loss": 1.5738,
+      "step": 4923
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8796844181459566e-06,
+      "loss": 1.4842,
+      "step": 4924
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.859960552268245e-06,
+      "loss": 1.4526,
+      "step": 4925
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8402366863905324e-06,
+      "loss": 1.5009,
+      "step": 4926
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.5843,
+      "step": 4927
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8007889546351086e-06,
+      "loss": 1.5028,
+      "step": 4928
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.7810650887573965e-06,
+      "loss": 1.5506,
+      "step": 4929
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.7613412228796843e-06,
+      "loss": 1.511,
+      "step": 4930
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.7416173570019726e-06,
+      "loss": 1.4684,
+      "step": 4931
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.7218934911242605e-06,
+      "loss": 1.4829,
+      "step": 4932
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.7021696252465484e-06,
+      "loss": 1.4769,
+      "step": 4933
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.6824457593688363e-06,
+      "loss": 1.4692,
+      "step": 4934
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.6627218934911246e-06,
+      "loss": 1.4715,
+      "step": 4935
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.6429980276134125e-06,
+      "loss": 1.4625,
+      "step": 4936
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.6232741617357003e-06,
+      "loss": 1.4575,
+      "step": 4937
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.603550295857988e-06,
+      "loss": 1.4557,
+      "step": 4938
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.5838264299802765e-06,
+      "loss": 1.431,
+      "step": 4939
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.4767,
+      "step": 4940
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.5443786982248523e-06,
+      "loss": 1.564,
+      "step": 4941
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.52465483234714e-06,
+      "loss": 1.4597,
+      "step": 4942
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.504930966469428e-06,
+      "loss": 1.5002,
+      "step": 4943
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.485207100591716e-06,
+      "loss": 1.527,
+      "step": 4944
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.465483234714004e-06,
+      "loss": 1.5678,
+      "step": 4945
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.445759368836292e-06,
+      "loss": 1.5493,
+      "step": 4946
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.42603550295858e-06,
+      "loss": 1.4934,
+      "step": 4947
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.406311637080868e-06,
+      "loss": 1.5408,
+      "step": 4948
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.386587771203156e-06,
+      "loss": 1.6009,
+      "step": 4949
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 1.5665,
+      "step": 4950
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.347140039447732e-06,
+      "loss": 1.5641,
+      "step": 4951
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.3274161735700198e-06,
+      "loss": 1.4012,
+      "step": 4952
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.4796,
+      "step": 4953
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.2879684418145955e-06,
+      "loss": 1.5148,
+      "step": 4954
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.268244575936884e-06,
+      "loss": 1.6046,
+      "step": 4955
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.2485207100591717e-06,
+      "loss": 1.5349,
+      "step": 4956
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.22879684418146e-06,
+      "loss": 1.6172,
+      "step": 4957
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.2090729783037475e-06,
+      "loss": 1.5022,
+      "step": 4958
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.1893491124260358e-06,
+      "loss": 1.4739,
+      "step": 4959
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.1696252465483236e-06,
+      "loss": 1.5705,
+      "step": 4960
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.1499013806706115e-06,
+      "loss": 1.4976,
+      "step": 4961
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.1301775147928994e-06,
+      "loss": 1.4873,
+      "step": 4962
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.1104536489151877e-06,
+      "loss": 1.5008,
+      "step": 4963
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.0907297830374756e-06,
+      "loss": 1.5401,
+      "step": 4964
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.0710059171597635e-06,
+      "loss": 1.5599,
+      "step": 4965
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.4222,
+      "step": 4966
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.0315581854043392e-06,
+      "loss": 1.5207,
+      "step": 4967
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.0118343195266275e-06,
+      "loss": 1.445,
+      "step": 4968
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.992110453648915e-06,
+      "loss": 1.5467,
+      "step": 4969
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.9723865877712033e-06,
+      "loss": 1.511,
+      "step": 4970
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.952662721893491e-06,
+      "loss": 1.5844,
+      "step": 4971
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.932938856015779e-06,
+      "loss": 1.4944,
+      "step": 4972
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.913214990138067e-06,
+      "loss": 1.5418,
+      "step": 4973
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.8934911242603552e-06,
+      "loss": 1.4706,
+      "step": 4974
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.8737672583826429e-06,
+      "loss": 1.4544,
+      "step": 4975
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.8540433925049312e-06,
+      "loss": 1.4864,
+      "step": 4976
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.8343195266272188e-06,
+      "loss": 1.5465,
+      "step": 4977
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.814595660749507e-06,
+      "loss": 1.5149,
+      "step": 4978
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.7948717948717948e-06,
+      "loss": 1.4297,
+      "step": 4979
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.775147928994083e-06,
+      "loss": 1.4647,
+      "step": 4980
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.7554240631163708e-06,
+      "loss": 1.6061,
+      "step": 4981
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.7357001972386589e-06,
+      "loss": 1.4876,
+      "step": 4982
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.7159763313609468e-06,
+      "loss": 1.5115,
+      "step": 4983
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.6962524654832348e-06,
+      "loss": 1.4207,
+      "step": 4984
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.6765285996055227e-06,
+      "loss": 1.4745,
+      "step": 4985
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.6568047337278108e-06,
+      "loss": 1.5263,
+      "step": 4986
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.6370808678500987e-06,
+      "loss": 1.4185,
+      "step": 4987
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.6173570019723868e-06,
+      "loss": 1.5049,
+      "step": 4988
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.5976331360946744e-06,
+      "loss": 1.5205,
+      "step": 4989
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.5779092702169627e-06,
+      "loss": 1.5299,
+      "step": 4990
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.5581854043392506e-06,
+      "loss": 1.5808,
+      "step": 4991
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.621,
+      "step": 4992
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.5187376725838264e-06,
+      "loss": 1.4845,
+      "step": 4993
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.4990138067061145e-06,
+      "loss": 1.4918,
+      "step": 4994
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.4792899408284024e-06,
+      "loss": 1.3984,
+      "step": 4995
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.4595660749506902e-06,
+      "loss": 1.5662,
+      "step": 4996
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.4398422090729783e-06,
+      "loss": 1.4453,
+      "step": 4997
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.4201183431952662e-06,
+      "loss": 1.4596,
+      "step": 4998
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.4003944773175543e-06,
+      "loss": 1.497,
+      "step": 4999
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3806706114398422e-06,
+      "loss": 1.5055,
+      "step": 5000
+    },
+    {
+      "epoch": 14.79,
+      "eval_loss": 1.4090298414230347,
+      "eval_runtime": 70.3822,
+      "eval_samples_per_second": 316.415,
+      "eval_steps_per_second": 1.648,
+      "step": 5000
+    }
+  ],
+  "max_steps": 5070,
+  "num_train_epochs": 15,
+  "total_flos": 4.898288883108741e+17,
+  "trial_name": null,
+  "trial_params": null
+}