diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,30176 +1,63299 @@
 {
-  "best_metric": 1.4090298414230347,
-  "best_model_checkpoint": "/home/yuchenl/models/reverse-instruct/prefixgen_flan-small/checkpoint-5000",
-  "epoch": 14.792899408284024,
-  "global_step": 5000,
+  "best_metric": 1.1098241806030273,
+  "best_model_checkpoint": "/home/yuchenl/models/reverse-instruct/prefixgen_v2_flan-small/checkpoint-10500",
+  "epoch": 46.875,
+  "eval_steps": 300,
+  "global_step": 10500,
   "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,
+      "learning_rate": 9.999107142857143e-05,
+      "loss": 3.4465,
       "step": 1
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9.996055226824458e-05,
-      "loss": 3.674,
+      "learning_rate": 9.998214285714286e-05,
+      "loss": 3.3243,
       "step": 2
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9.994082840236687e-05,
-      "loss": 3.6946,
+      "learning_rate": 9.99732142857143e-05,
+      "loss": 3.2075,
       "step": 3
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 9.992110453648915e-05,
-      "loss": 3.5351,
+      "epoch": 0.02,
+      "learning_rate": 9.996428571428572e-05,
+      "loss": 3.1554,
       "step": 4
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 9.990138067061144e-05,
-      "loss": 3.3956,
+      "epoch": 0.02,
+      "learning_rate": 9.995535714285715e-05,
+      "loss": 3.0567,
       "step": 5
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.988165680473373e-05,
-      "loss": 3.4182,
+      "epoch": 0.03,
+      "learning_rate": 9.994642857142857e-05,
+      "loss": 2.8787,
       "step": 6
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.986193293885602e-05,
-      "loss": 3.3563,
+      "epoch": 0.03,
+      "learning_rate": 9.993750000000001e-05,
+      "loss": 2.8825,
       "step": 7
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 9.984220907297831e-05,
-      "loss": 3.3066,
+      "epoch": 0.04,
+      "learning_rate": 9.992857142857144e-05,
+      "loss": 2.8728,
       "step": 8
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.982248520710059e-05,
-      "loss": 3.2331,
+      "epoch": 0.04,
+      "learning_rate": 9.991964285714286e-05,
+      "loss": 2.8529,
       "step": 9
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.980276134122289e-05,
-      "loss": 3.2169,
+      "epoch": 0.04,
+      "learning_rate": 9.991071428571429e-05,
+      "loss": 2.6674,
       "step": 10
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.978303747534518e-05,
-      "loss": 3.1005,
+      "epoch": 0.05,
+      "learning_rate": 9.990178571428573e-05,
+      "loss": 2.7007,
       "step": 11
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.976331360946746e-05,
-      "loss": 3.0535,
+      "epoch": 0.05,
+      "learning_rate": 9.989285714285714e-05,
+      "loss": 2.6351,
       "step": 12
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.974358974358975e-05,
-      "loss": 3.1198,
+      "epoch": 0.06,
+      "learning_rate": 9.988392857142858e-05,
+      "loss": 2.6344,
       "step": 13
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.972386587771204e-05,
-      "loss": 3.0663,
+      "epoch": 0.06,
+      "learning_rate": 9.9875e-05,
+      "loss": 2.6691,
       "step": 14
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 9.970414201183432e-05,
-      "loss": 3.0584,
+      "epoch": 0.07,
+      "learning_rate": 9.986607142857143e-05,
+      "loss": 2.6258,
       "step": 15
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.968441814595661e-05,
-      "loss": 2.9469,
+      "epoch": 0.07,
+      "learning_rate": 9.985714285714287e-05,
+      "loss": 2.503,
       "step": 16
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.96646942800789e-05,
-      "loss": 3.0459,
+      "epoch": 0.08,
+      "learning_rate": 9.984821428571429e-05,
+      "loss": 2.4312,
       "step": 17
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 9.96449704142012e-05,
-      "loss": 2.9426,
+      "epoch": 0.08,
+      "learning_rate": 9.983928571428573e-05,
+      "loss": 2.4639,
       "step": 18
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.962524654832349e-05,
-      "loss": 2.9672,
+      "epoch": 0.08,
+      "learning_rate": 9.983035714285714e-05,
+      "loss": 2.4774,
       "step": 19
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.960552268244577e-05,
-      "loss": 2.9189,
+      "epoch": 0.09,
+      "learning_rate": 9.982142857142858e-05,
+      "loss": 2.5457,
       "step": 20
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.958579881656806e-05,
-      "loss": 2.914,
+      "epoch": 0.09,
+      "learning_rate": 9.981250000000001e-05,
+      "loss": 2.396,
       "step": 21
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 9.956607495069034e-05,
-      "loss": 2.8608,
+      "epoch": 0.1,
+      "learning_rate": 9.980357142857143e-05,
+      "loss": 2.5275,
       "step": 22
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 9.954635108481263e-05,
-      "loss": 2.8851,
+      "epoch": 0.1,
+      "learning_rate": 9.979464285714286e-05,
+      "loss": 2.3457,
       "step": 23
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 9.952662721893492e-05,
-      "loss": 2.9098,
+      "epoch": 0.11,
+      "learning_rate": 9.97857142857143e-05,
+      "loss": 2.4152,
       "step": 24
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 9.95069033530572e-05,
-      "loss": 2.7703,
+      "epoch": 0.11,
+      "learning_rate": 9.977678571428571e-05,
+      "loss": 2.3219,
       "step": 25
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 9.948717948717949e-05,
-      "loss": 2.7624,
+      "epoch": 0.12,
+      "learning_rate": 9.976785714285715e-05,
+      "loss": 2.3546,
       "step": 26
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 9.946745562130177e-05,
-      "loss": 2.7823,
+      "epoch": 0.12,
+      "learning_rate": 9.975892857142857e-05,
+      "loss": 2.3043,
       "step": 27
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 9.944773175542406e-05,
-      "loss": 2.7804,
+      "epoch": 0.12,
+      "learning_rate": 9.975000000000001e-05,
+      "loss": 2.3241,
       "step": 28
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.942800788954635e-05,
-      "loss": 2.706,
+      "epoch": 0.13,
+      "learning_rate": 9.974107142857142e-05,
+      "loss": 2.2633,
       "step": 29
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.940828402366865e-05,
-      "loss": 2.7654,
+      "epoch": 0.13,
+      "learning_rate": 9.973214285714286e-05,
+      "loss": 2.2742,
       "step": 30
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.938856015779094e-05,
-      "loss": 2.7789,
+      "epoch": 0.14,
+      "learning_rate": 9.972321428571429e-05,
+      "loss": 2.3126,
       "step": 31
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.936883629191322e-05,
-      "loss": 2.7501,
+      "epoch": 0.14,
+      "learning_rate": 9.971428571428571e-05,
+      "loss": 2.2722,
       "step": 32
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.934911242603551e-05,
-      "loss": 2.7309,
+      "epoch": 0.15,
+      "learning_rate": 9.970535714285715e-05,
+      "loss": 2.2867,
       "step": 33
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.93293885601578e-05,
-      "loss": 2.7298,
+      "epoch": 0.15,
+      "learning_rate": 9.969642857142858e-05,
+      "loss": 2.2959,
       "step": 34
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 9.930966469428008e-05,
-      "loss": 2.7167,
+      "epoch": 0.16,
+      "learning_rate": 9.96875e-05,
+      "loss": 2.2412,
       "step": 35
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.928994082840237e-05,
-      "loss": 2.7242,
+      "epoch": 0.16,
+      "learning_rate": 9.967857142857143e-05,
+      "loss": 2.2555,
       "step": 36
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.927021696252466e-05,
-      "loss": 2.6855,
+      "epoch": 0.17,
+      "learning_rate": 9.966964285714287e-05,
+      "loss": 2.1368,
       "step": 37
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 9.925049309664694e-05,
-      "loss": 2.7618,
+      "epoch": 0.17,
+      "learning_rate": 9.966071428571428e-05,
+      "loss": 2.2455,
       "step": 38
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 9.923076923076923e-05,
-      "loss": 2.6294,
+      "epoch": 0.17,
+      "learning_rate": 9.965178571428572e-05,
+      "loss": 2.2286,
       "step": 39
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 9.921104536489151e-05,
-      "loss": 2.5916,
+      "epoch": 0.18,
+      "learning_rate": 9.964285714285714e-05,
+      "loss": 2.2371,
       "step": 40
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 9.91913214990138e-05,
-      "loss": 2.5775,
+      "epoch": 0.18,
+      "learning_rate": 9.963392857142858e-05,
+      "loss": 2.1629,
       "step": 41
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 9.91715976331361e-05,
-      "loss": 2.633,
+      "epoch": 0.19,
+      "learning_rate": 9.9625e-05,
+      "loss": 2.1303,
       "step": 42
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 9.915187376725839e-05,
-      "loss": 2.6634,
+      "epoch": 0.19,
+      "learning_rate": 9.961607142857143e-05,
+      "loss": 2.1765,
       "step": 43
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 9.913214990138068e-05,
-      "loss": 2.5683,
+      "epoch": 0.2,
+      "learning_rate": 9.960714285714286e-05,
+      "loss": 2.0876,
       "step": 44
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 9.911242603550296e-05,
-      "loss": 2.6114,
+      "epoch": 0.2,
+      "learning_rate": 9.959821428571429e-05,
+      "loss": 2.1484,
       "step": 45
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 9.909270216962525e-05,
-      "loss": 2.6691,
+      "epoch": 0.21,
+      "learning_rate": 9.958928571428572e-05,
+      "loss": 2.073,
       "step": 46
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 9.907297830374754e-05,
-      "loss": 2.6478,
+      "epoch": 0.21,
+      "learning_rate": 9.958035714285715e-05,
+      "loss": 2.1636,
       "step": 47
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 9.905325443786982e-05,
-      "loss": 2.6553,
+      "epoch": 0.21,
+      "learning_rate": 9.957142857142858e-05,
+      "loss": 2.1171,
       "step": 48
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 9.903353057199212e-05,
-      "loss": 2.531,
+      "epoch": 0.22,
+      "learning_rate": 9.95625e-05,
+      "loss": 2.0663,
       "step": 49
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 9.90138067061144e-05,
-      "loss": 2.565,
+      "epoch": 0.22,
+      "learning_rate": 9.955357142857144e-05,
+      "loss": 2.1208,
       "step": 50
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 9.899408284023669e-05,
-      "loss": 2.514,
+      "epoch": 0.23,
+      "learning_rate": 9.954464285714287e-05,
+      "loss": 2.0977,
       "step": 51
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 9.897435897435898e-05,
-      "loss": 2.5883,
+      "epoch": 0.23,
+      "learning_rate": 9.953571428571429e-05,
+      "loss": 2.196,
       "step": 52
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.895463510848127e-05,
-      "loss": 2.499,
+      "epoch": 0.24,
+      "learning_rate": 9.952678571428572e-05,
+      "loss": 2.1439,
       "step": 53
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.893491124260356e-05,
-      "loss": 2.5677,
+      "epoch": 0.24,
+      "learning_rate": 9.951785714285716e-05,
+      "loss": 2.1526,
       "step": 54
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 9.891518737672584e-05,
-      "loss": 2.5398,
+      "epoch": 0.25,
+      "learning_rate": 9.950892857142857e-05,
+      "loss": 2.0269,
       "step": 55
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.889546351084813e-05,
-      "loss": 2.5657,
+      "epoch": 0.25,
+      "learning_rate": 9.95e-05,
+      "loss": 2.0512,
       "step": 56
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.887573964497042e-05,
-      "loss": 2.5397,
+      "epoch": 0.25,
+      "learning_rate": 9.949107142857143e-05,
+      "loss": 1.9623,
       "step": 57
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.88560157790927e-05,
-      "loss": 2.5902,
+      "epoch": 0.26,
+      "learning_rate": 9.948214285714286e-05,
+      "loss": 2.0149,
       "step": 58
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.8836291913215e-05,
-      "loss": 2.4764,
+      "epoch": 0.26,
+      "learning_rate": 9.947321428571428e-05,
+      "loss": 2.1295,
       "step": 59
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.881656804733729e-05,
-      "loss": 2.6158,
+      "epoch": 0.27,
+      "learning_rate": 9.946428571428572e-05,
+      "loss": 2.0301,
       "step": 60
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.879684418145957e-05,
-      "loss": 2.6273,
+      "epoch": 0.27,
+      "learning_rate": 9.945535714285715e-05,
+      "loss": 2.0102,
       "step": 61
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 9.877712031558186e-05,
-      "loss": 2.4931,
+      "epoch": 0.28,
+      "learning_rate": 9.944642857142857e-05,
+      "loss": 1.9514,
       "step": 62
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.875739644970414e-05,
-      "loss": 2.4349,
+      "epoch": 0.28,
+      "learning_rate": 9.943750000000001e-05,
+      "loss": 2.0029,
       "step": 63
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.873767258382643e-05,
-      "loss": 2.3798,
+      "epoch": 0.29,
+      "learning_rate": 9.942857142857144e-05,
+      "loss": 2.0539,
       "step": 64
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 9.871794871794872e-05,
-      "loss": 2.6425,
+      "epoch": 0.29,
+      "learning_rate": 9.941964285714286e-05,
+      "loss": 2.0232,
       "step": 65
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 9.869822485207101e-05,
-      "loss": 2.4093,
+      "epoch": 0.29,
+      "learning_rate": 9.941071428571429e-05,
+      "loss": 1.9952,
       "step": 66
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 9.86785009861933e-05,
-      "loss": 2.4182,
+      "epoch": 0.3,
+      "learning_rate": 9.940178571428573e-05,
+      "loss": 2.029,
       "step": 67
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 9.865877712031558e-05,
-      "loss": 2.4,
+      "epoch": 0.3,
+      "learning_rate": 9.939285714285714e-05,
+      "loss": 1.9605,
       "step": 68
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 9.863905325443788e-05,
-      "loss": 2.5141,
+      "epoch": 0.31,
+      "learning_rate": 9.938392857142858e-05,
+      "loss": 1.9807,
       "step": 69
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 9.861932938856017e-05,
-      "loss": 2.5375,
+      "epoch": 0.31,
+      "learning_rate": 9.9375e-05,
+      "loss": 1.9323,
       "step": 70
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 9.859960552268245e-05,
-      "loss": 2.5405,
+      "epoch": 0.32,
+      "learning_rate": 9.936607142857143e-05,
+      "loss": 1.9344,
       "step": 71
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 9.857988165680474e-05,
-      "loss": 2.498,
+      "epoch": 0.32,
+      "learning_rate": 9.935714285714285e-05,
+      "loss": 1.9683,
       "step": 72
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 9.856015779092702e-05,
-      "loss": 2.3824,
+      "epoch": 0.33,
+      "learning_rate": 9.934821428571429e-05,
+      "loss": 1.9295,
       "step": 73
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 9.854043392504931e-05,
-      "loss": 2.3929,
+      "epoch": 0.33,
+      "learning_rate": 9.933928571428572e-05,
+      "loss": 1.9278,
       "step": 74
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 9.85207100591716e-05,
-      "loss": 2.5451,
+      "epoch": 0.33,
+      "learning_rate": 9.933035714285714e-05,
+      "loss": 1.876,
       "step": 75
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 9.85009861932939e-05,
-      "loss": 2.4987,
+      "epoch": 0.34,
+      "learning_rate": 9.932142857142858e-05,
+      "loss": 2.0786,
       "step": 76
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 9.848126232741619e-05,
-      "loss": 2.3975,
+      "epoch": 0.34,
+      "learning_rate": 9.931250000000001e-05,
+      "loss": 1.915,
       "step": 77
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 9.846153846153848e-05,
-      "loss": 2.4368,
+      "epoch": 0.35,
+      "learning_rate": 9.930357142857143e-05,
+      "loss": 2.0101,
       "step": 78
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 9.844181459566076e-05,
-      "loss": 2.4468,
+      "epoch": 0.35,
+      "learning_rate": 9.929464285714286e-05,
+      "loss": 1.9662,
       "step": 79
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 9.842209072978305e-05,
-      "loss": 2.3,
+      "epoch": 0.36,
+      "learning_rate": 9.92857142857143e-05,
+      "loss": 1.9655,
       "step": 80
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 9.840236686390533e-05,
-      "loss": 2.4105,
+      "epoch": 0.36,
+      "learning_rate": 9.927678571428571e-05,
+      "loss": 1.9049,
       "step": 81
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 9.838264299802762e-05,
-      "loss": 2.471,
+      "epoch": 0.37,
+      "learning_rate": 9.926785714285715e-05,
+      "loss": 1.8175,
       "step": 82
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 9.836291913214991e-05,
-      "loss": 2.354,
+      "epoch": 0.37,
+      "learning_rate": 9.925892857142857e-05,
+      "loss": 1.9788,
       "step": 83
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 9.834319526627219e-05,
-      "loss": 2.4156,
+      "epoch": 0.38,
+      "learning_rate": 9.925000000000001e-05,
+      "loss": 1.9495,
       "step": 84
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 9.832347140039448e-05,
-      "loss": 2.3962,
+      "epoch": 0.38,
+      "learning_rate": 9.924107142857143e-05,
+      "loss": 1.8527,
       "step": 85
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 9.830374753451676e-05,
-      "loss": 2.3593,
+      "epoch": 0.38,
+      "learning_rate": 9.923214285714286e-05,
+      "loss": 1.8581,
       "step": 86
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 9.828402366863905e-05,
-      "loss": 2.433,
+      "epoch": 0.39,
+      "learning_rate": 9.922321428571429e-05,
+      "loss": 1.9438,
       "step": 87
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 9.826429980276134e-05,
-      "loss": 2.4312,
+      "epoch": 0.39,
+      "learning_rate": 9.921428571428572e-05,
+      "loss": 1.9026,
       "step": 88
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 9.824457593688364e-05,
-      "loss": 2.3218,
+      "epoch": 0.4,
+      "learning_rate": 9.920535714285714e-05,
+      "loss": 1.9224,
       "step": 89
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 9.822485207100593e-05,
-      "loss": 2.4151,
+      "epoch": 0.4,
+      "learning_rate": 9.919642857142858e-05,
+      "loss": 2.0467,
       "step": 90
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 9.820512820512821e-05,
-      "loss": 2.4811,
+      "epoch": 0.41,
+      "learning_rate": 9.91875e-05,
+      "loss": 2.0032,
       "step": 91
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 9.81854043392505e-05,
-      "loss": 2.3644,
+      "epoch": 0.41,
+      "learning_rate": 9.917857142857143e-05,
+      "loss": 1.85,
       "step": 92
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.816568047337279e-05,
-      "loss": 2.3723,
+      "epoch": 0.42,
+      "learning_rate": 9.916964285714287e-05,
+      "loss": 1.9714,
       "step": 93
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.814595660749507e-05,
-      "loss": 2.4164,
+      "epoch": 0.42,
+      "learning_rate": 9.91607142857143e-05,
+      "loss": 1.9631,
       "step": 94
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.812623274161736e-05,
-      "loss": 2.3547,
+      "epoch": 0.42,
+      "learning_rate": 9.915178571428572e-05,
+      "loss": 1.8787,
       "step": 95
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.810650887573964e-05,
-      "loss": 2.3637,
+      "epoch": 0.43,
+      "learning_rate": 9.914285714285715e-05,
+      "loss": 1.8234,
       "step": 96
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.808678500986193e-05,
-      "loss": 2.3211,
+      "epoch": 0.43,
+      "learning_rate": 9.913392857142859e-05,
+      "loss": 1.9052,
       "step": 97
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.806706114398423e-05,
-      "loss": 2.4678,
+      "epoch": 0.44,
+      "learning_rate": 9.9125e-05,
+      "loss": 1.9479,
       "step": 98
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.80473372781065e-05,
-      "loss": 2.4384,
+      "epoch": 0.44,
+      "learning_rate": 9.911607142857144e-05,
+      "loss": 1.8641,
       "step": 99
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.80276134122288e-05,
-      "loss": 2.3458,
+      "epoch": 0.45,
+      "learning_rate": 9.910714285714286e-05,
+      "loss": 1.8745,
       "step": 100
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.800788954635109e-05,
-      "loss": 2.3613,
+      "epoch": 0.45,
+      "learning_rate": 9.909821428571429e-05,
+      "loss": 1.8795,
       "step": 101
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.798816568047338e-05,
-      "loss": 2.3403,
+      "epoch": 0.46,
+      "learning_rate": 9.908928571428571e-05,
+      "loss": 1.9413,
       "step": 102
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 9.796844181459567e-05,
-      "loss": 2.3316,
+      "epoch": 0.46,
+      "learning_rate": 9.908035714285715e-05,
+      "loss": 1.8843,
       "step": 103
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.794871794871795e-05,
-      "loss": 2.2789,
+      "epoch": 0.46,
+      "learning_rate": 9.907142857142858e-05,
+      "loss": 1.8977,
       "step": 104
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.792899408284024e-05,
-      "loss": 2.3241,
+      "epoch": 0.47,
+      "learning_rate": 9.90625e-05,
+      "loss": 1.936,
       "step": 105
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.790927021696254e-05,
-      "loss": 2.4394,
+      "epoch": 0.47,
+      "learning_rate": 9.905357142857143e-05,
+      "loss": 1.7745,
       "step": 106
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.788954635108481e-05,
-      "loss": 2.3581,
+      "epoch": 0.48,
+      "learning_rate": 9.904464285714287e-05,
+      "loss": 1.9546,
       "step": 107
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.78698224852071e-05,
-      "loss": 2.3995,
+      "epoch": 0.48,
+      "learning_rate": 9.903571428571429e-05,
+      "loss": 1.8414,
       "step": 108
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.785009861932938e-05,
-      "loss": 2.3988,
+      "epoch": 0.49,
+      "learning_rate": 9.902678571428572e-05,
+      "loss": 1.8475,
       "step": 109
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.783037475345168e-05,
-      "loss": 2.3599,
+      "epoch": 0.49,
+      "learning_rate": 9.901785714285716e-05,
+      "loss": 1.9091,
       "step": 110
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.781065088757397e-05,
-      "loss": 2.3956,
+      "epoch": 0.5,
+      "learning_rate": 9.900892857142857e-05,
+      "loss": 1.8592,
       "step": 111
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.779092702169626e-05,
-      "loss": 2.3661,
+      "epoch": 0.5,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.883,
       "step": 112
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.777120315581855e-05,
-      "loss": 2.3244,
+      "epoch": 0.5,
+      "learning_rate": 9.899107142857143e-05,
+      "loss": 1.8844,
       "step": 113
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.775147928994083e-05,
-      "loss": 2.3445,
+      "epoch": 0.51,
+      "learning_rate": 9.898214285714286e-05,
+      "loss": 1.8548,
       "step": 114
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.773175542406312e-05,
-      "loss": 2.3377,
+      "epoch": 0.51,
+      "learning_rate": 9.897321428571428e-05,
+      "loss": 1.8289,
       "step": 115
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.771203155818542e-05,
-      "loss": 2.3339,
+      "epoch": 0.52,
+      "learning_rate": 9.896428571428572e-05,
+      "loss": 1.8371,
       "step": 116
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.76923076923077e-05,
-      "loss": 2.4014,
+      "epoch": 0.52,
+      "learning_rate": 9.895535714285715e-05,
+      "loss": 1.9559,
       "step": 117
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.767258382642999e-05,
-      "loss": 2.2225,
+      "epoch": 0.53,
+      "learning_rate": 9.894642857142857e-05,
+      "loss": 1.876,
       "step": 118
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.765285996055226e-05,
-      "loss": 2.3873,
+      "epoch": 0.53,
+      "learning_rate": 9.89375e-05,
+      "loss": 1.769,
       "step": 119
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.763313609467456e-05,
-      "loss": 2.2908,
+      "epoch": 0.54,
+      "learning_rate": 9.892857142857144e-05,
+      "loss": 1.8904,
       "step": 120
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.761341222879685e-05,
-      "loss": 2.3221,
+      "epoch": 0.54,
+      "learning_rate": 9.891964285714286e-05,
+      "loss": 1.7758,
       "step": 121
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.759368836291913e-05,
-      "loss": 2.3277,
+      "epoch": 0.54,
+      "learning_rate": 9.891071428571429e-05,
+      "loss": 1.826,
       "step": 122
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.757396449704142e-05,
-      "loss": 2.2136,
+      "epoch": 0.55,
+      "learning_rate": 9.890178571428573e-05,
+      "loss": 1.7525,
       "step": 123
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.755424063116371e-05,
-      "loss": 2.3274,
+      "epoch": 0.55,
+      "learning_rate": 9.889285714285714e-05,
+      "loss": 1.8546,
       "step": 124
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.7534516765286e-05,
-      "loss": 2.2523,
+      "epoch": 0.56,
+      "learning_rate": 9.888392857142858e-05,
+      "loss": 1.9542,
       "step": 125
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.75147928994083e-05,
-      "loss": 2.3595,
+      "epoch": 0.56,
+      "learning_rate": 9.8875e-05,
+      "loss": 1.8939,
       "step": 126
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.749506903353057e-05,
-      "loss": 2.4048,
+      "epoch": 0.57,
+      "learning_rate": 9.886607142857144e-05,
+      "loss": 1.8003,
       "step": 127
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.747534516765287e-05,
-      "loss": 2.343,
+      "epoch": 0.57,
+      "learning_rate": 9.885714285714286e-05,
+      "loss": 1.8129,
       "step": 128
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.745562130177516e-05,
-      "loss": 2.2225,
+      "epoch": 0.58,
+      "learning_rate": 9.88482142857143e-05,
+      "loss": 1.8309,
       "step": 129
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.743589743589744e-05,
-      "loss": 2.2596,
+      "epoch": 0.58,
+      "learning_rate": 9.883928571428572e-05,
+      "loss": 1.8912,
       "step": 130
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.741617357001973e-05,
-      "loss": 2.3328,
+      "epoch": 0.58,
+      "learning_rate": 9.883035714285715e-05,
+      "loss": 1.7931,
       "step": 131
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.739644970414201e-05,
-      "loss": 2.2945,
+      "epoch": 0.59,
+      "learning_rate": 9.882142857142857e-05,
+      "loss": 1.8464,
       "step": 132
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.73767258382643e-05,
-      "loss": 2.3803,
+      "epoch": 0.59,
+      "learning_rate": 9.881250000000001e-05,
+      "loss": 1.9576,
       "step": 133
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 9.735700197238659e-05,
-      "loss": 2.3361,
+      "epoch": 0.6,
+      "learning_rate": 9.880357142857144e-05,
+      "loss": 1.7722,
       "step": 134
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 9.733727810650888e-05,
-      "loss": 2.3168,
+      "epoch": 0.6,
+      "learning_rate": 9.879464285714286e-05,
+      "loss": 1.8855,
       "step": 135
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 9.731755424063118e-05,
-      "loss": 2.2209,
+      "epoch": 0.61,
+      "learning_rate": 9.878571428571429e-05,
+      "loss": 1.7794,
       "step": 136
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.729783037475346e-05,
-      "loss": 2.2811,
+      "epoch": 0.61,
+      "learning_rate": 9.877678571428573e-05,
+      "loss": 1.8352,
       "step": 137
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.727810650887575e-05,
-      "loss": 2.2663,
+      "epoch": 0.62,
+      "learning_rate": 9.876785714285715e-05,
+      "loss": 1.8449,
       "step": 138
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.725838264299804e-05,
-      "loss": 2.3283,
+      "epoch": 0.62,
+      "learning_rate": 9.875892857142858e-05,
+      "loss": 1.6909,
       "step": 139
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.723865877712032e-05,
-      "loss": 2.2832,
+      "epoch": 0.62,
+      "learning_rate": 9.875000000000002e-05,
+      "loss": 1.786,
       "step": 140
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 9.721893491124261e-05,
-      "loss": 2.1317,
+      "epoch": 0.63,
+      "learning_rate": 9.874107142857143e-05,
+      "loss": 1.7382,
       "step": 141
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 9.719921104536489e-05,
-      "loss": 2.2627,
+      "epoch": 0.63,
+      "learning_rate": 9.873214285714287e-05,
+      "loss": 1.7377,
       "step": 142
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 9.717948717948718e-05,
-      "loss": 2.26,
+      "epoch": 0.64,
+      "learning_rate": 9.872321428571429e-05,
+      "loss": 1.766,
       "step": 143
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.715976331360947e-05,
-      "loss": 2.3603,
+      "epoch": 0.64,
+      "learning_rate": 9.871428571428572e-05,
+      "loss": 1.8498,
       "step": 144
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.714003944773175e-05,
-      "loss": 2.2045,
+      "epoch": 0.65,
+      "learning_rate": 9.870535714285714e-05,
+      "loss": 1.7399,
       "step": 145
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.712031558185404e-05,
-      "loss": 2.2285,
+      "epoch": 0.65,
+      "learning_rate": 9.869642857142858e-05,
+      "loss": 1.7725,
       "step": 146
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.710059171597634e-05,
-      "loss": 2.3495,
+      "epoch": 0.66,
+      "learning_rate": 9.868749999999999e-05,
+      "loss": 1.7133,
       "step": 147
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.708086785009863e-05,
-      "loss": 2.297,
+      "epoch": 0.66,
+      "learning_rate": 9.867857142857143e-05,
+      "loss": 1.7924,
       "step": 148
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.706114398422092e-05,
-      "loss": 2.1912,
+      "epoch": 0.67,
+      "learning_rate": 9.866964285714286e-05,
+      "loss": 1.7256,
       "step": 149
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.70414201183432e-05,
-      "loss": 2.2025,
+      "epoch": 0.67,
+      "learning_rate": 9.86607142857143e-05,
+      "loss": 1.74,
       "step": 150
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.702169625246549e-05,
-      "loss": 2.2518,
+      "epoch": 0.67,
+      "learning_rate": 9.865178571428572e-05,
+      "loss": 1.7053,
       "step": 151
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.700197238658778e-05,
-      "loss": 2.2001,
+      "epoch": 0.68,
+      "learning_rate": 9.864285714285715e-05,
+      "loss": 1.7665,
       "step": 152
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.698224852071006e-05,
-      "loss": 2.2375,
+      "epoch": 0.68,
+      "learning_rate": 9.863392857142859e-05,
+      "loss": 1.8191,
       "step": 153
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 9.696252465483235e-05,
-      "loss": 2.3301,
+      "epoch": 0.69,
+      "learning_rate": 9.8625e-05,
+      "loss": 1.8412,
       "step": 154
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 9.694280078895463e-05,
-      "loss": 2.2667,
+      "epoch": 0.69,
+      "learning_rate": 9.861607142857144e-05,
+      "loss": 1.7585,
       "step": 155
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 9.692307692307692e-05,
-      "loss": 2.2469,
+      "epoch": 0.7,
+      "learning_rate": 9.860714285714286e-05,
+      "loss": 1.8933,
       "step": 156
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 9.690335305719922e-05,
-      "loss": 2.2954,
+      "epoch": 0.7,
+      "learning_rate": 9.859821428571429e-05,
+      "loss": 1.745,
       "step": 157
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 9.68836291913215e-05,
-      "loss": 2.2033,
+      "epoch": 0.71,
+      "learning_rate": 9.858928571428571e-05,
+      "loss": 1.8199,
       "step": 158
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 9.686390532544379e-05,
-      "loss": 2.2853,
+      "epoch": 0.71,
+      "learning_rate": 9.858035714285715e-05,
+      "loss": 1.8103,
       "step": 159
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 9.684418145956608e-05,
-      "loss": 2.1632,
+      "epoch": 0.71,
+      "learning_rate": 9.857142857142858e-05,
+      "loss": 1.7397,
       "step": 160
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 9.682445759368837e-05,
-      "loss": 2.3111,
+      "epoch": 0.72,
+      "learning_rate": 9.85625e-05,
+      "loss": 1.7858,
       "step": 161
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 9.680473372781066e-05,
-      "loss": 2.2767,
+      "epoch": 0.72,
+      "learning_rate": 9.855357142857143e-05,
+      "loss": 1.8313,
       "step": 162
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 9.678500986193294e-05,
-      "loss": 2.3181,
+      "epoch": 0.73,
+      "learning_rate": 9.854464285714287e-05,
+      "loss": 1.7601,
       "step": 163
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.676528599605523e-05,
-      "loss": 2.2527,
+      "epoch": 0.73,
+      "learning_rate": 9.853571428571428e-05,
+      "loss": 1.8498,
       "step": 164
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.674556213017751e-05,
-      "loss": 2.2613,
+      "epoch": 0.74,
+      "learning_rate": 9.852678571428572e-05,
+      "loss": 1.7414,
       "step": 165
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.67258382642998e-05,
-      "loss": 2.2212,
+      "epoch": 0.74,
+      "learning_rate": 9.851785714285714e-05,
+      "loss": 1.8285,
       "step": 166
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.67061143984221e-05,
-      "loss": 2.2292,
+      "epoch": 0.75,
+      "learning_rate": 9.850892857142857e-05,
+      "loss": 1.7506,
       "step": 167
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 9.668639053254438e-05,
-      "loss": 2.2337,
+      "epoch": 0.75,
+      "learning_rate": 9.850000000000001e-05,
+      "loss": 1.7793,
       "step": 168
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 9.666666666666667e-05,
-      "loss": 2.2329,
+      "epoch": 0.75,
+      "learning_rate": 9.849107142857143e-05,
+      "loss": 1.8425,
       "step": 169
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 9.664694280078896e-05,
-      "loss": 2.3045,
+      "epoch": 0.76,
+      "learning_rate": 9.848214285714287e-05,
+      "loss": 1.7683,
       "step": 170
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.662721893491125e-05,
-      "loss": 2.3365,
+      "epoch": 0.76,
+      "learning_rate": 9.847321428571429e-05,
+      "loss": 1.7422,
       "step": 171
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.660749506903354e-05,
-      "loss": 2.2489,
+      "epoch": 0.77,
+      "learning_rate": 9.846428571428572e-05,
+      "loss": 1.7159,
       "step": 172
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.658777120315582e-05,
-      "loss": 2.2947,
+      "epoch": 0.77,
+      "learning_rate": 9.845535714285715e-05,
+      "loss": 1.7232,
       "step": 173
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 9.656804733727811e-05,
-      "loss": 2.2058,
+      "epoch": 0.78,
+      "learning_rate": 9.844642857142858e-05,
+      "loss": 1.6529,
       "step": 174
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.65483234714004e-05,
-      "loss": 2.1562,
+      "epoch": 0.78,
+      "learning_rate": 9.84375e-05,
+      "loss": 1.7201,
       "step": 175
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.652859960552268e-05,
-      "loss": 2.2771,
+      "epoch": 0.79,
+      "learning_rate": 9.842857142857144e-05,
+      "loss": 1.7292,
       "step": 176
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.650887573964498e-05,
-      "loss": 2.2572,
+      "epoch": 0.79,
+      "learning_rate": 9.841964285714285e-05,
+      "loss": 1.6659,
       "step": 177
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.648915187376726e-05,
-      "loss": 2.2854,
+      "epoch": 0.79,
+      "learning_rate": 9.841071428571429e-05,
+      "loss": 1.6975,
       "step": 178
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.646942800788955e-05,
-      "loss": 2.2783,
+      "epoch": 0.8,
+      "learning_rate": 9.840178571428572e-05,
+      "loss": 1.6929,
       "step": 179
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.644970414201184e-05,
-      "loss": 2.1126,
+      "epoch": 0.8,
+      "learning_rate": 9.839285714285714e-05,
+      "loss": 1.7031,
       "step": 180
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.642998027613412e-05,
-      "loss": 2.1158,
+      "epoch": 0.81,
+      "learning_rate": 9.838392857142858e-05,
+      "loss": 1.8641,
       "step": 181
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.641025641025641e-05,
-      "loss": 2.2107,
+      "epoch": 0.81,
+      "learning_rate": 9.8375e-05,
+      "loss": 1.697,
       "step": 182
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.63905325443787e-05,
-      "loss": 2.1353,
+      "epoch": 0.82,
+      "learning_rate": 9.836607142857143e-05,
+      "loss": 1.7685,
       "step": 183
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.6370808678501e-05,
-      "loss": 2.0108,
+      "epoch": 0.82,
+      "learning_rate": 9.835714285714286e-05,
+      "loss": 1.7946,
       "step": 184
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.635108481262329e-05,
-      "loss": 2.1017,
+      "epoch": 0.83,
+      "learning_rate": 9.83482142857143e-05,
+      "loss": 1.7414,
       "step": 185
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.633136094674557e-05,
-      "loss": 2.1909,
+      "epoch": 0.83,
+      "learning_rate": 9.833928571428572e-05,
+      "loss": 1.8219,
       "step": 186
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.631163708086786e-05,
-      "loss": 2.0382,
+      "epoch": 0.83,
+      "learning_rate": 9.833035714285715e-05,
+      "loss": 1.7517,
       "step": 187
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.629191321499015e-05,
-      "loss": 2.2003,
+      "epoch": 0.84,
+      "learning_rate": 9.832142857142857e-05,
+      "loss": 1.7143,
       "step": 188
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.627218934911243e-05,
-      "loss": 2.2472,
+      "epoch": 0.84,
+      "learning_rate": 9.831250000000001e-05,
+      "loss": 1.7945,
       "step": 189
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.625246548323472e-05,
-      "loss": 2.1721,
+      "epoch": 0.85,
+      "learning_rate": 9.830357142857142e-05,
+      "loss": 1.7604,
       "step": 190
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.6232741617357e-05,
-      "loss": 2.1999,
+      "epoch": 0.85,
+      "learning_rate": 9.829464285714286e-05,
+      "loss": 1.7589,
       "step": 191
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.621301775147929e-05,
-      "loss": 2.2523,
+      "epoch": 0.86,
+      "learning_rate": 9.828571428571429e-05,
+      "loss": 1.7658,
       "step": 192
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.619329388560158e-05,
-      "loss": 2.1618,
+      "epoch": 0.86,
+      "learning_rate": 9.827678571428573e-05,
+      "loss": 1.7609,
       "step": 193
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.617357001972388e-05,
-      "loss": 2.2756,
+      "epoch": 0.87,
+      "learning_rate": 9.826785714285714e-05,
+      "loss": 1.6651,
       "step": 194
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 9.615384615384617e-05,
-      "loss": 2.2559,
+      "epoch": 0.87,
+      "learning_rate": 9.825892857142858e-05,
+      "loss": 1.6759,
       "step": 195
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 9.613412228796845e-05,
-      "loss": 2.2252,
+      "epoch": 0.88,
+      "learning_rate": 9.825e-05,
+      "loss": 1.7188,
       "step": 196
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 9.611439842209074e-05,
-      "loss": 2.1343,
+      "epoch": 0.88,
+      "learning_rate": 9.824107142857143e-05,
+      "loss": 1.7832,
       "step": 197
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 9.609467455621303e-05,
-      "loss": 2.1231,
+      "epoch": 0.88,
+      "learning_rate": 9.823214285714287e-05,
+      "loss": 1.8073,
       "step": 198
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 9.607495069033531e-05,
-      "loss": 2.2587,
+      "epoch": 0.89,
+      "learning_rate": 9.822321428571429e-05,
+      "loss": 1.7354,
       "step": 199
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 9.60552268244576e-05,
-      "loss": 2.0799,
+      "epoch": 0.89,
+      "learning_rate": 9.821428571428572e-05,
+      "loss": 1.7167,
       "step": 200
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 9.603550295857988e-05,
-      "loss": 2.1718,
+      "epoch": 0.9,
+      "learning_rate": 9.820535714285714e-05,
+      "loss": 1.7375,
       "step": 201
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 9.601577909270217e-05,
-      "loss": 2.2614,
+      "epoch": 0.9,
+      "learning_rate": 9.819642857142858e-05,
+      "loss": 1.6587,
       "step": 202
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 9.599605522682446e-05,
-      "loss": 2.1406,
+      "epoch": 0.91,
+      "learning_rate": 9.818750000000001e-05,
+      "loss": 1.7075,
       "step": 203
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 9.597633136094674e-05,
-      "loss": 2.1099,
+      "epoch": 0.91,
+      "learning_rate": 9.817857142857143e-05,
+      "loss": 1.7524,
       "step": 204
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 9.595660749506903e-05,
-      "loss": 2.1277,
+      "epoch": 0.92,
+      "learning_rate": 9.816964285714286e-05,
+      "loss": 1.6369,
       "step": 205
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 9.593688362919133e-05,
-      "loss": 2.1857,
+      "epoch": 0.92,
+      "learning_rate": 9.81607142857143e-05,
+      "loss": 1.6869,
       "step": 206
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 9.591715976331362e-05,
-      "loss": 2.1206,
+      "epoch": 0.92,
+      "learning_rate": 9.815178571428571e-05,
+      "loss": 1.7826,
       "step": 207
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.589743589743591e-05,
-      "loss": 2.1587,
+      "epoch": 0.93,
+      "learning_rate": 9.814285714285715e-05,
+      "loss": 1.738,
       "step": 208
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.587771203155819e-05,
-      "loss": 2.14,
+      "epoch": 0.93,
+      "learning_rate": 9.813392857142857e-05,
+      "loss": 1.8027,
       "step": 209
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.585798816568048e-05,
-      "loss": 2.1706,
+      "epoch": 0.94,
+      "learning_rate": 9.8125e-05,
+      "loss": 1.6249,
       "step": 210
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.583826429980277e-05,
-      "loss": 2.1826,
+      "epoch": 0.94,
+      "learning_rate": 9.811607142857144e-05,
+      "loss": 1.6769,
       "step": 211
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 9.581854043392505e-05,
-      "loss": 2.0951,
+      "epoch": 0.95,
+      "learning_rate": 9.810714285714286e-05,
+      "loss": 1.6087,
       "step": 212
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 9.579881656804734e-05,
-      "loss": 2.1804,
+      "epoch": 0.95,
+      "learning_rate": 9.809821428571429e-05,
+      "loss": 1.7278,
       "step": 213
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 9.577909270216962e-05,
-      "loss": 2.1596,
+      "epoch": 0.96,
+      "learning_rate": 9.808928571428572e-05,
+      "loss": 1.7259,
       "step": 214
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.575936883629191e-05,
-      "loss": 2.1171,
+      "epoch": 0.96,
+      "learning_rate": 9.808035714285715e-05,
+      "loss": 1.8856,
       "step": 215
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.573964497041421e-05,
-      "loss": 2.0727,
+      "epoch": 0.96,
+      "learning_rate": 9.807142857142858e-05,
+      "loss": 1.7266,
       "step": 216
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.571992110453649e-05,
-      "loss": 2.2219,
+      "epoch": 0.97,
+      "learning_rate": 9.80625e-05,
+      "loss": 1.7725,
       "step": 217
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.570019723865878e-05,
-      "loss": 2.1494,
+      "epoch": 0.97,
+      "learning_rate": 9.805357142857143e-05,
+      "loss": 1.6654,
       "step": 218
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 9.568047337278107e-05,
-      "loss": 2.184,
+      "epoch": 0.98,
+      "learning_rate": 9.804464285714287e-05,
+      "loss": 1.7091,
       "step": 219
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 9.566074950690336e-05,
-      "loss": 2.1904,
+      "epoch": 0.98,
+      "learning_rate": 9.803571428571428e-05,
+      "loss": 1.7223,
       "step": 220
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 9.564102564102565e-05,
-      "loss": 2.306,
+      "epoch": 0.99,
+      "learning_rate": 9.802678571428572e-05,
+      "loss": 1.6884,
       "step": 221
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 9.562130177514793e-05,
-      "loss": 2.1098,
+      "epoch": 0.99,
+      "learning_rate": 9.801785714285715e-05,
+      "loss": 1.8075,
       "step": 222
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 9.560157790927022e-05,
-      "loss": 2.1654,
+      "epoch": 1.0,
+      "learning_rate": 9.800892857142857e-05,
+      "loss": 1.6754,
       "step": 223
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 9.55818540433925e-05,
-      "loss": 2.0218,
+      "epoch": 1.0,
+      "learning_rate": 9.8e-05,
+      "loss": 1.7734,
       "step": 224
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 9.55621301775148e-05,
-      "loss": 2.1706,
+      "epoch": 1.0,
+      "learning_rate": 9.799107142857144e-05,
+      "loss": 1.7212,
       "step": 225
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 9.554240631163709e-05,
-      "loss": 2.1258,
+      "epoch": 1.01,
+      "learning_rate": 9.798214285714286e-05,
+      "loss": 1.6717,
       "step": 226
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 9.552268244575937e-05,
-      "loss": 2.1647,
+      "epoch": 1.01,
+      "learning_rate": 9.797321428571429e-05,
+      "loss": 1.6191,
       "step": 227
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 9.550295857988166e-05,
-      "loss": 2.1608,
+      "epoch": 1.02,
+      "learning_rate": 9.796428571428573e-05,
+      "loss": 1.6445,
       "step": 228
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 9.548323471400395e-05,
-      "loss": 2.0605,
+      "epoch": 1.02,
+      "learning_rate": 9.795535714285715e-05,
+      "loss": 1.7129,
       "step": 229
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 9.546351084812624e-05,
-      "loss": 2.1443,
+      "epoch": 1.03,
+      "learning_rate": 9.794642857142858e-05,
+      "loss": 1.725,
       "step": 230
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 9.544378698224853e-05,
-      "loss": 2.1171,
+      "epoch": 1.03,
+      "learning_rate": 9.79375e-05,
+      "loss": 1.754,
       "step": 231
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 9.542406311637081e-05,
-      "loss": 2.1851,
+      "epoch": 1.04,
+      "learning_rate": 9.792857142857144e-05,
+      "loss": 1.6899,
       "step": 232
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 9.54043392504931e-05,
-      "loss": 2.2092,
+      "epoch": 1.04,
+      "learning_rate": 9.791964285714285e-05,
+      "loss": 1.5518,
       "step": 233
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 9.53846153846154e-05,
-      "loss": 2.1555,
+      "epoch": 1.04,
+      "learning_rate": 9.791071428571429e-05,
+      "loss": 1.6672,
       "step": 234
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.536489151873768e-05,
-      "loss": 2.1382,
+      "epoch": 1.05,
+      "learning_rate": 9.790178571428572e-05,
+      "loss": 1.734,
       "step": 235
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.534516765285997e-05,
-      "loss": 2.1275,
+      "epoch": 1.05,
+      "learning_rate": 9.789285714285716e-05,
+      "loss": 1.7947,
       "step": 236
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.532544378698225e-05,
-      "loss": 2.1537,
+      "epoch": 1.06,
+      "learning_rate": 9.788392857142857e-05,
+      "loss": 1.7411,
       "step": 237
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.530571992110454e-05,
-      "loss": 2.0421,
+      "epoch": 1.06,
+      "learning_rate": 9.787500000000001e-05,
+      "loss": 1.7147,
       "step": 238
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 9.528599605522683e-05,
-      "loss": 1.9904,
+      "epoch": 1.07,
+      "learning_rate": 9.786607142857143e-05,
+      "loss": 1.7243,
       "step": 239
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 9.526627218934911e-05,
-      "loss": 2.1497,
+      "epoch": 1.07,
+      "learning_rate": 9.785714285714286e-05,
+      "loss": 1.7008,
       "step": 240
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 9.52465483234714e-05,
-      "loss": 2.1138,
+      "epoch": 1.08,
+      "learning_rate": 9.784821428571428e-05,
+      "loss": 1.5886,
       "step": 241
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 9.52268244575937e-05,
-      "loss": 2.0724,
+      "epoch": 1.08,
+      "learning_rate": 9.783928571428572e-05,
+      "loss": 1.6462,
       "step": 242
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 9.520710059171599e-05,
-      "loss": 1.9776,
+      "epoch": 1.08,
+      "learning_rate": 9.783035714285715e-05,
+      "loss": 1.7188,
       "step": 243
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 9.518737672583828e-05,
-      "loss": 2.1538,
+      "epoch": 1.09,
+      "learning_rate": 9.782142857142857e-05,
+      "loss": 1.5504,
       "step": 244
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 9.516765285996056e-05,
-      "loss": 2.0735,
+      "epoch": 1.09,
+      "learning_rate": 9.781250000000001e-05,
+      "loss": 1.6929,
       "step": 245
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 9.514792899408285e-05,
-      "loss": 2.0963,
+      "epoch": 1.1,
+      "learning_rate": 9.780357142857142e-05,
+      "loss": 1.6694,
       "step": 246
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 9.512820512820513e-05,
-      "loss": 2.1212,
+      "epoch": 1.1,
+      "learning_rate": 9.779464285714286e-05,
+      "loss": 1.6871,
       "step": 247
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 9.510848126232742e-05,
-      "loss": 2.2163,
+      "epoch": 1.11,
+      "learning_rate": 9.778571428571429e-05,
+      "loss": 1.6956,
       "step": 248
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 9.508875739644971e-05,
-      "loss": 2.114,
+      "epoch": 1.11,
+      "learning_rate": 9.777678571428573e-05,
+      "loss": 1.5772,
       "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,
+      "epoch": 1.12,
+      "learning_rate": 9.776785714285714e-05,
+      "loss": 1.7435,
       "step": 250
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 9.504930966469428e-05,
-      "loss": 2.1463,
+      "epoch": 1.12,
+      "learning_rate": 9.775892857142858e-05,
+      "loss": 1.6399,
       "step": 251
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 9.502958579881657e-05,
-      "loss": 2.1177,
+      "epoch": 1.12,
+      "learning_rate": 9.775e-05,
+      "loss": 1.6177,
       "step": 252
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 9.500986193293887e-05,
-      "loss": 2.1241,
+      "epoch": 1.13,
+      "learning_rate": 9.774107142857143e-05,
+      "loss": 1.5724,
       "step": 253
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 9.499013806706116e-05,
-      "loss": 2.0676,
+      "epoch": 1.13,
+      "learning_rate": 9.773214285714286e-05,
+      "loss": 1.6712,
       "step": 254
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 9.497041420118344e-05,
-      "loss": 2.2031,
+      "epoch": 1.14,
+      "learning_rate": 9.77232142857143e-05,
+      "loss": 1.719,
       "step": 255
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 9.495069033530573e-05,
-      "loss": 2.0476,
+      "epoch": 1.14,
+      "learning_rate": 9.771428571428572e-05,
+      "loss": 1.6236,
       "step": 256
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 9.493096646942802e-05,
-      "loss": 2.1706,
+      "epoch": 1.15,
+      "learning_rate": 9.770535714285715e-05,
+      "loss": 1.6054,
       "step": 257
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 9.49112426035503e-05,
-      "loss": 2.1243,
+      "epoch": 1.15,
+      "learning_rate": 9.769642857142858e-05,
+      "loss": 1.6851,
       "step": 258
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 9.489151873767259e-05,
-      "loss": 2.0723,
+      "epoch": 1.16,
+      "learning_rate": 9.768750000000001e-05,
+      "loss": 1.6303,
       "step": 259
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 9.487179487179487e-05,
-      "loss": 1.9851,
+      "epoch": 1.16,
+      "learning_rate": 9.767857142857144e-05,
+      "loss": 1.7172,
       "step": 260
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 9.485207100591716e-05,
-      "loss": 2.1657,
+      "epoch": 1.17,
+      "learning_rate": 9.766964285714286e-05,
+      "loss": 1.7118,
       "step": 261
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 9.483234714003945e-05,
-      "loss": 2.0537,
+      "epoch": 1.17,
+      "learning_rate": 9.76607142857143e-05,
+      "loss": 1.6168,
       "step": 262
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 9.481262327416173e-05,
-      "loss": 2.049,
+      "epoch": 1.17,
+      "learning_rate": 9.765178571428571e-05,
+      "loss": 1.6748,
       "step": 263
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 9.479289940828402e-05,
-      "loss": 2.1241,
+      "epoch": 1.18,
+      "learning_rate": 9.764285714285715e-05,
+      "loss": 1.6154,
       "step": 264
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 9.477317554240632e-05,
-      "loss": 2.0893,
+      "epoch": 1.18,
+      "learning_rate": 9.763392857142858e-05,
+      "loss": 1.6281,
       "step": 265
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 9.475345167652861e-05,
-      "loss": 2.1441,
+      "epoch": 1.19,
+      "learning_rate": 9.7625e-05,
+      "loss": 1.6931,
       "step": 266
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 9.47337278106509e-05,
-      "loss": 2.1287,
+      "epoch": 1.19,
+      "learning_rate": 9.761607142857143e-05,
+      "loss": 1.7477,
       "step": 267
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 9.471400394477318e-05,
-      "loss": 2.1959,
+      "epoch": 1.2,
+      "learning_rate": 9.760714285714287e-05,
+      "loss": 1.6712,
       "step": 268
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.469428007889547e-05,
-      "loss": 2.1282,
+      "epoch": 1.2,
+      "learning_rate": 9.759821428571429e-05,
+      "loss": 1.6087,
       "step": 269
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.467455621301775e-05,
-      "loss": 1.9884,
+      "epoch": 1.21,
+      "learning_rate": 9.758928571428572e-05,
+      "loss": 1.6669,
       "step": 270
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.465483234714004e-05,
-      "loss": 2.2444,
+      "epoch": 1.21,
+      "learning_rate": 9.758035714285714e-05,
+      "loss": 1.7401,
       "step": 271
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.463510848126233e-05,
-      "loss": 2.1313,
+      "epoch": 1.21,
+      "learning_rate": 9.757142857142858e-05,
+      "loss": 1.694,
       "step": 272
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 9.461538461538461e-05,
-      "loss": 2.0675,
+      "epoch": 1.22,
+      "learning_rate": 9.75625e-05,
+      "loss": 1.6483,
       "step": 273
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 9.45956607495069e-05,
-      "loss": 2.0643,
+      "epoch": 1.22,
+      "learning_rate": 9.755357142857143e-05,
+      "loss": 1.5035,
       "step": 274
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 9.457593688362918e-05,
-      "loss": 2.2047,
+      "epoch": 1.23,
+      "learning_rate": 9.754464285714287e-05,
+      "loss": 1.6335,
       "step": 275
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 9.455621301775148e-05,
-      "loss": 2.098,
+      "epoch": 1.23,
+      "learning_rate": 9.753571428571428e-05,
+      "loss": 1.5788,
       "step": 276
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 9.453648915187377e-05,
-      "loss": 2.1079,
+      "epoch": 1.24,
+      "learning_rate": 9.752678571428572e-05,
+      "loss": 1.6158,
       "step": 277
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 9.451676528599606e-05,
-      "loss": 2.0961,
+      "epoch": 1.24,
+      "learning_rate": 9.751785714285715e-05,
+      "loss": 1.6611,
       "step": 278
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.449704142011835e-05,
-      "loss": 2.2097,
+      "epoch": 1.25,
+      "learning_rate": 9.750892857142859e-05,
+      "loss": 1.679,
       "step": 279
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.447731755424064e-05,
-      "loss": 2.0825,
+      "epoch": 1.25,
+      "learning_rate": 9.75e-05,
+      "loss": 1.5826,
       "step": 280
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.445759368836292e-05,
-      "loss": 2.0384,
+      "epoch": 1.25,
+      "learning_rate": 9.749107142857144e-05,
+      "loss": 1.6117,
       "step": 281
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.443786982248522e-05,
-      "loss": 1.98,
+      "epoch": 1.26,
+      "learning_rate": 9.748214285714286e-05,
+      "loss": 1.667,
       "step": 282
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 9.44181459566075e-05,
-      "loss": 2.1016,
+      "epoch": 1.26,
+      "learning_rate": 9.747321428571429e-05,
+      "loss": 1.7044,
       "step": 283
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 9.439842209072979e-05,
-      "loss": 2.0015,
+      "epoch": 1.27,
+      "learning_rate": 9.746428571428571e-05,
+      "loss": 1.7012,
       "step": 284
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 9.437869822485208e-05,
-      "loss": 2.0672,
+      "epoch": 1.27,
+      "learning_rate": 9.745535714285715e-05,
+      "loss": 1.6946,
       "step": 285
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 9.435897435897436e-05,
-      "loss": 2.048,
+      "epoch": 1.28,
+      "learning_rate": 9.744642857142858e-05,
+      "loss": 1.6581,
       "step": 286
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 9.433925049309665e-05,
-      "loss": 2.0326,
+      "epoch": 1.28,
+      "learning_rate": 9.74375e-05,
+      "loss": 1.6502,
       "step": 287
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 9.431952662721894e-05,
-      "loss": 2.1843,
+      "epoch": 1.29,
+      "learning_rate": 9.742857142857143e-05,
+      "loss": 1.7106,
       "step": 288
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.429980276134123e-05,
-      "loss": 2.0338,
+      "epoch": 1.29,
+      "learning_rate": 9.741964285714285e-05,
+      "loss": 1.6387,
       "step": 289
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.428007889546353e-05,
-      "loss": 2.0516,
+      "epoch": 1.29,
+      "learning_rate": 9.74107142857143e-05,
+      "loss": 1.5918,
       "step": 290
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.42603550295858e-05,
-      "loss": 2.0264,
+      "epoch": 1.3,
+      "learning_rate": 9.740178571428572e-05,
+      "loss": 1.636,
       "step": 291
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.42406311637081e-05,
-      "loss": 2.1196,
+      "epoch": 1.3,
+      "learning_rate": 9.739285714285716e-05,
+      "loss": 1.5902,
       "step": 292
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.422090729783037e-05,
-      "loss": 2.1012,
+      "epoch": 1.31,
+      "learning_rate": 9.738392857142857e-05,
+      "loss": 1.6661,
       "step": 293
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.420118343195267e-05,
-      "loss": 2.0462,
+      "epoch": 1.31,
+      "learning_rate": 9.737500000000001e-05,
+      "loss": 1.692,
       "step": 294
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.418145956607496e-05,
-      "loss": 1.9924,
+      "epoch": 1.32,
+      "learning_rate": 9.736607142857143e-05,
+      "loss": 1.5044,
       "step": 295
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 9.416173570019724e-05,
-      "loss": 1.9484,
+      "epoch": 1.32,
+      "learning_rate": 9.735714285714286e-05,
+      "loss": 1.6047,
       "step": 296
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 9.414201183431953e-05,
-      "loss": 2.0841,
+      "epoch": 1.33,
+      "learning_rate": 9.734821428571429e-05,
+      "loss": 1.5788,
       "step": 297
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 9.412228796844182e-05,
-      "loss": 2.0088,
+      "epoch": 1.33,
+      "learning_rate": 9.733928571428572e-05,
+      "loss": 1.6071,
       "step": 298
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 9.41025641025641e-05,
-      "loss": 2.1245,
+      "epoch": 1.33,
+      "learning_rate": 9.733035714285714e-05,
+      "loss": 1.6741,
       "step": 299
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.408284023668639e-05,
-      "loss": 1.9897,
+      "epoch": 1.34,
+      "learning_rate": 9.732142857142858e-05,
+      "loss": 1.6093,
       "step": 300
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.406311637080868e-05,
-      "loss": 2.1443,
+      "epoch": 1.34,
+      "eval_loss": 1.4478356838226318,
+      "eval_runtime": 45.4052,
+      "eval_samples_per_second": 288.381,
+      "eval_steps_per_second": 1.52,
+      "step": 300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.73125e-05,
+      "loss": 1.7437,
       "step": 301
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.404339250493098e-05,
-      "loss": 2.0554,
+      "epoch": 1.35,
+      "learning_rate": 9.730357142857144e-05,
+      "loss": 1.5683,
       "step": 302
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 9.402366863905327e-05,
-      "loss": 1.9841,
+      "epoch": 1.35,
+      "learning_rate": 9.729464285714286e-05,
+      "loss": 1.6367,
       "step": 303
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 9.400394477317555e-05,
-      "loss": 2.1372,
+      "epoch": 1.36,
+      "learning_rate": 9.728571428571429e-05,
+      "loss": 1.637,
       "step": 304
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 9.398422090729784e-05,
-      "loss": 2.1294,
+      "epoch": 1.36,
+      "learning_rate": 9.727678571428573e-05,
+      "loss": 1.5501,
       "step": 305
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 9.396449704142012e-05,
-      "loss": 2.0112,
+      "epoch": 1.37,
+      "learning_rate": 9.726785714285714e-05,
+      "loss": 1.7519,
       "step": 306
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 9.394477317554241e-05,
-      "loss": 2.1047,
+      "epoch": 1.37,
+      "learning_rate": 9.725892857142858e-05,
+      "loss": 1.6322,
       "step": 307
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 9.39250493096647e-05,
-      "loss": 2.0877,
+      "epoch": 1.38,
+      "learning_rate": 9.725e-05,
+      "loss": 1.5562,
       "step": 308
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 9.390532544378698e-05,
-      "loss": 2.122,
+      "epoch": 1.38,
+      "learning_rate": 9.724107142857143e-05,
+      "loss": 1.5182,
       "step": 309
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 9.388560157790927e-05,
-      "loss": 2.0349,
+      "epoch": 1.38,
+      "learning_rate": 9.723214285714286e-05,
+      "loss": 1.5253,
       "step": 310
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 9.386587771203156e-05,
-      "loss": 2.1113,
+      "epoch": 1.39,
+      "learning_rate": 9.72232142857143e-05,
+      "loss": 1.7228,
       "step": 311
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 9.384615384615386e-05,
-      "loss": 2.1833,
+      "epoch": 1.39,
+      "learning_rate": 9.721428571428572e-05,
+      "loss": 1.4934,
       "step": 312
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 9.382642998027615e-05,
-      "loss": 2.1157,
+      "epoch": 1.4,
+      "learning_rate": 9.720535714285715e-05,
+      "loss": 1.6213,
       "step": 313
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 9.380670611439843e-05,
-      "loss": 1.9738,
+      "epoch": 1.4,
+      "learning_rate": 9.719642857142857e-05,
+      "loss": 1.5659,
       "step": 314
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 9.378698224852072e-05,
-      "loss": 1.9967,
+      "epoch": 1.41,
+      "learning_rate": 9.718750000000001e-05,
+      "loss": 1.6492,
       "step": 315
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 9.3767258382643e-05,
-      "loss": 2.0019,
+      "epoch": 1.41,
+      "learning_rate": 9.717857142857144e-05,
+      "loss": 1.5392,
       "step": 316
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 9.374753451676529e-05,
-      "loss": 1.9836,
-      "step": 317
+      "epoch": 1.42,
+      "learning_rate": 9.716964285714286e-05,
+      "loss": 1.651,
+      "step": 317
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 9.372781065088758e-05,
-      "loss": 2.0119,
+      "epoch": 1.42,
+      "learning_rate": 9.716071428571429e-05,
+      "loss": 1.6601,
       "step": 318
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 9.370808678500986e-05,
-      "loss": 2.0737,
+      "epoch": 1.42,
+      "learning_rate": 9.715178571428571e-05,
+      "loss": 1.6317,
       "step": 319
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 9.368836291913215e-05,
-      "loss": 2.1024,
+      "epoch": 1.43,
+      "learning_rate": 9.714285714285715e-05,
+      "loss": 1.6979,
       "step": 320
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 9.366863905325445e-05,
-      "loss": 2.0557,
+      "epoch": 1.43,
+      "learning_rate": 9.713392857142858e-05,
+      "loss": 1.6076,
       "step": 321
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 9.364891518737672e-05,
-      "loss": 2.1436,
+      "epoch": 1.44,
+      "learning_rate": 9.7125e-05,
+      "loss": 1.552,
       "step": 322
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.362919132149902e-05,
-      "loss": 2.0604,
+      "epoch": 1.44,
+      "learning_rate": 9.711607142857143e-05,
+      "loss": 1.5809,
       "step": 323
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.360946745562131e-05,
-      "loss": 2.0031,
+      "epoch": 1.45,
+      "learning_rate": 9.710714285714287e-05,
+      "loss": 1.5035,
       "step": 324
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.35897435897436e-05,
-      "loss": 2.0288,
+      "epoch": 1.45,
+      "learning_rate": 9.709821428571429e-05,
+      "loss": 1.6266,
       "step": 325
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.357001972386589e-05,
-      "loss": 2.0268,
+      "epoch": 1.46,
+      "learning_rate": 9.708928571428572e-05,
+      "loss": 1.616,
       "step": 326
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 9.355029585798817e-05,
-      "loss": 2.0515,
+      "epoch": 1.46,
+      "learning_rate": 9.708035714285714e-05,
+      "loss": 1.559,
       "step": 327
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 9.353057199211046e-05,
-      "loss": 2.0537,
+      "epoch": 1.46,
+      "learning_rate": 9.707142857142858e-05,
+      "loss": 1.6313,
       "step": 328
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 9.351084812623274e-05,
-      "loss": 1.9182,
+      "epoch": 1.47,
+      "learning_rate": 9.70625e-05,
+      "loss": 1.5168,
       "step": 329
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 9.349112426035503e-05,
-      "loss": 1.9767,
+      "epoch": 1.47,
+      "learning_rate": 9.705357142857143e-05,
+      "loss": 1.568,
       "step": 330
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 9.347140039447733e-05,
-      "loss": 1.9775,
+      "epoch": 1.48,
+      "learning_rate": 9.704464285714286e-05,
+      "loss": 1.6047,
       "step": 331
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 9.34516765285996e-05,
-      "loss": 2.0768,
+      "epoch": 1.48,
+      "learning_rate": 9.703571428571428e-05,
+      "loss": 1.6482,
       "step": 332
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 9.34319526627219e-05,
-      "loss": 2.1042,
+      "epoch": 1.49,
+      "learning_rate": 9.702678571428572e-05,
+      "loss": 1.6424,
       "step": 333
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 9.341222879684417e-05,
-      "loss": 2.0881,
+      "epoch": 1.49,
+      "learning_rate": 9.701785714285715e-05,
+      "loss": 1.6529,
       "step": 334
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 9.339250493096647e-05,
-      "loss": 2.0554,
+      "epoch": 1.5,
+      "learning_rate": 9.700892857142859e-05,
+      "loss": 1.6158,
       "step": 335
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 9.337278106508876e-05,
-      "loss": 2.0446,
+      "epoch": 1.5,
+      "learning_rate": 9.7e-05,
+      "loss": 1.6248,
       "step": 336
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 9.335305719921105e-05,
-      "loss": 1.9072,
+      "epoch": 1.5,
+      "learning_rate": 9.699107142857144e-05,
+      "loss": 1.5563,
       "step": 337
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 9.333333333333334e-05,
-      "loss": 1.9938,
+      "epoch": 1.51,
+      "learning_rate": 9.698214285714286e-05,
+      "loss": 1.5704,
       "step": 338
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 9.331360946745562e-05,
-      "loss": 1.9686,
+      "epoch": 1.51,
+      "learning_rate": 9.697321428571429e-05,
+      "loss": 1.5341,
       "step": 339
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 9.329388560157791e-05,
-      "loss": 2.0474,
+      "epoch": 1.52,
+      "learning_rate": 9.696428571428571e-05,
+      "loss": 1.528,
       "step": 340
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 9.32741617357002e-05,
-      "loss": 1.9905,
+      "epoch": 1.52,
+      "learning_rate": 9.695535714285715e-05,
+      "loss": 1.5368,
       "step": 341
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 9.325443786982248e-05,
-      "loss": 1.9465,
+      "epoch": 1.53,
+      "learning_rate": 9.694642857142857e-05,
+      "loss": 1.591,
       "step": 342
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 9.323471400394478e-05,
-      "loss": 1.9013,
+      "epoch": 1.53,
+      "learning_rate": 9.69375e-05,
+      "loss": 1.6576,
       "step": 343
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 9.321499013806707e-05,
-      "loss": 2.0105,
+      "epoch": 1.54,
+      "learning_rate": 9.692857142857143e-05,
+      "loss": 1.6042,
       "step": 344
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 9.319526627218935e-05,
-      "loss": 2.0188,
+      "epoch": 1.54,
+      "learning_rate": 9.691964285714287e-05,
+      "loss": 1.5878,
       "step": 345
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 9.317554240631164e-05,
-      "loss": 1.9706,
+      "epoch": 1.54,
+      "learning_rate": 9.691071428571428e-05,
+      "loss": 1.6226,
       "step": 346
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.315581854043393e-05,
-      "loss": 1.957,
+      "epoch": 1.55,
+      "learning_rate": 9.690178571428572e-05,
+      "loss": 1.639,
       "step": 347
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.313609467455622e-05,
-      "loss": 2.0473,
+      "epoch": 1.55,
+      "learning_rate": 9.689285714285715e-05,
+      "loss": 1.5206,
       "step": 348
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.311637080867852e-05,
-      "loss": 1.9829,
+      "epoch": 1.56,
+      "learning_rate": 9.688392857142857e-05,
+      "loss": 1.5888,
       "step": 349
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.30966469428008e-05,
-      "loss": 1.8897,
+      "epoch": 1.56,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 1.5142,
       "step": 350
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.307692307692309e-05,
-      "loss": 1.9492,
+      "epoch": 1.57,
+      "learning_rate": 9.686607142857144e-05,
+      "loss": 1.6506,
       "step": 351
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.305719921104537e-05,
-      "loss": 2.0767,
+      "epoch": 1.57,
+      "learning_rate": 9.685714285714286e-05,
+      "loss": 1.582,
       "step": 352
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.303747534516766e-05,
-      "loss": 2.0562,
+      "epoch": 1.58,
+      "learning_rate": 9.684821428571429e-05,
+      "loss": 1.5927,
       "step": 353
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.301775147928995e-05,
-      "loss": 1.9344,
+      "epoch": 1.58,
+      "learning_rate": 9.683928571428573e-05,
+      "loss": 1.4983,
       "step": 354
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.299802761341223e-05,
-      "loss": 2.0683,
+      "epoch": 1.58,
+      "learning_rate": 9.683035714285714e-05,
+      "loss": 1.5934,
       "step": 355
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.297830374753452e-05,
-      "loss": 1.9956,
+      "epoch": 1.59,
+      "learning_rate": 9.682142857142858e-05,
+      "loss": 1.6155,
       "step": 356
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.29585798816568e-05,
-      "loss": 1.9246,
+      "epoch": 1.59,
+      "learning_rate": 9.68125e-05,
+      "loss": 1.5585,
       "step": 357
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.293885601577909e-05,
-      "loss": 1.9615,
+      "epoch": 1.6,
+      "learning_rate": 9.680357142857144e-05,
+      "loss": 1.5765,
       "step": 358
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.291913214990138e-05,
-      "loss": 2.0332,
+      "epoch": 1.6,
+      "learning_rate": 9.679464285714285e-05,
+      "loss": 1.5882,
       "step": 359
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.289940828402367e-05,
-      "loss": 1.9608,
+      "epoch": 1.61,
+      "learning_rate": 9.678571428571429e-05,
+      "loss": 1.565,
       "step": 360
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.287968441814597e-05,
-      "loss": 2.1074,
+      "epoch": 1.61,
+      "learning_rate": 9.677678571428572e-05,
+      "loss": 1.5612,
       "step": 361
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.285996055226825e-05,
-      "loss": 2.0347,
+      "epoch": 1.62,
+      "learning_rate": 9.676785714285714e-05,
+      "loss": 1.6011,
       "step": 362
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.284023668639054e-05,
-      "loss": 2.0537,
+      "epoch": 1.62,
+      "learning_rate": 9.675892857142858e-05,
+      "loss": 1.6042,
       "step": 363
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.282051282051283e-05,
-      "loss": 2.0514,
+      "epoch": 1.62,
+      "learning_rate": 9.675000000000001e-05,
+      "loss": 1.7575,
       "step": 364
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.280078895463511e-05,
-      "loss": 2.0581,
+      "epoch": 1.63,
+      "learning_rate": 9.674107142857143e-05,
+      "loss": 1.526,
       "step": 365
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.27810650887574e-05,
-      "loss": 2.047,
+      "epoch": 1.63,
+      "learning_rate": 9.673214285714286e-05,
+      "loss": 1.5942,
       "step": 366
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.276134122287969e-05,
-      "loss": 2.1184,
+      "epoch": 1.64,
+      "learning_rate": 9.67232142857143e-05,
+      "loss": 1.5602,
       "step": 367
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.274161735700197e-05,
-      "loss": 1.9546,
+      "epoch": 1.64,
+      "learning_rate": 9.671428571428572e-05,
+      "loss": 1.6945,
       "step": 368
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.272189349112426e-05,
-      "loss": 2.0367,
+      "epoch": 1.65,
+      "learning_rate": 9.670535714285715e-05,
+      "loss": 1.6029,
       "step": 369
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.270216962524656e-05,
-      "loss": 1.9045,
+      "epoch": 1.65,
+      "learning_rate": 9.669642857142857e-05,
+      "loss": 1.6712,
       "step": 370
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 9.268244575936885e-05,
-      "loss": 2.0171,
+      "epoch": 1.66,
+      "learning_rate": 9.668750000000001e-05,
+      "loss": 1.4938,
       "step": 371
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 9.266272189349114e-05,
-      "loss": 2.0078,
+      "epoch": 1.66,
+      "learning_rate": 9.667857142857142e-05,
+      "loss": 1.5738,
       "step": 372
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 9.264299802761342e-05,
-      "loss": 2.0523,
+      "epoch": 1.67,
+      "learning_rate": 9.666964285714286e-05,
+      "loss": 1.5707,
       "step": 373
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 9.262327416173571e-05,
-      "loss": 2.0317,
+      "epoch": 1.67,
+      "learning_rate": 9.666071428571429e-05,
+      "loss": 1.5835,
       "step": 374
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 9.260355029585799e-05,
-      "loss": 1.9758,
+      "epoch": 1.67,
+      "learning_rate": 9.665178571428571e-05,
+      "loss": 1.6092,
       "step": 375
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 9.258382642998028e-05,
-      "loss": 2.065,
+      "epoch": 1.68,
+      "learning_rate": 9.664285714285714e-05,
+      "loss": 1.6052,
       "step": 376
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 9.256410256410257e-05,
-      "loss": 2.0664,
+      "epoch": 1.68,
+      "learning_rate": 9.663392857142858e-05,
+      "loss": 1.6369,
       "step": 377
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 9.254437869822485e-05,
-      "loss": 2.0338,
+      "epoch": 1.69,
+      "learning_rate": 9.6625e-05,
+      "loss": 1.524,
       "step": 378
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 9.252465483234714e-05,
-      "loss": 2.0023,
+      "epoch": 1.69,
+      "learning_rate": 9.661607142857143e-05,
+      "loss": 1.5802,
       "step": 379
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 9.250493096646942e-05,
-      "loss": 1.9733,
+      "epoch": 1.7,
+      "learning_rate": 9.660714285714287e-05,
+      "loss": 1.7112,
       "step": 380
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 9.248520710059171e-05,
-      "loss": 1.9596,
+      "epoch": 1.7,
+      "learning_rate": 9.65982142857143e-05,
+      "loss": 1.5482,
       "step": 381
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 9.2465483234714e-05,
-      "loss": 1.9707,
+      "epoch": 1.71,
+      "learning_rate": 9.658928571428572e-05,
+      "loss": 1.5687,
       "step": 382
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 9.24457593688363e-05,
-      "loss": 1.9494,
+      "epoch": 1.71,
+      "learning_rate": 9.658035714285714e-05,
+      "loss": 1.5757,
       "step": 383
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 9.242603550295859e-05,
-      "loss": 2.0083,
+      "epoch": 1.71,
+      "learning_rate": 9.657142857142858e-05,
+      "loss": 1.6566,
       "step": 384
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 9.240631163708087e-05,
-      "loss": 2.0268,
+      "epoch": 1.72,
+      "learning_rate": 9.65625e-05,
+      "loss": 1.5395,
       "step": 385
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 9.238658777120316e-05,
-      "loss": 2.0073,
+      "epoch": 1.72,
+      "learning_rate": 9.655357142857143e-05,
+      "loss": 1.7253,
       "step": 386
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 9.236686390532545e-05,
-      "loss": 1.9929,
+      "epoch": 1.73,
+      "learning_rate": 9.654464285714286e-05,
+      "loss": 1.6208,
       "step": 387
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 9.234714003944773e-05,
-      "loss": 1.9789,
+      "epoch": 1.73,
+      "learning_rate": 9.653571428571429e-05,
+      "loss": 1.5249,
       "step": 388
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 9.232741617357002e-05,
-      "loss": 1.9,
+      "epoch": 1.74,
+      "learning_rate": 9.652678571428571e-05,
+      "loss": 1.5867,
       "step": 389
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 9.230769230769232e-05,
-      "loss": 2.0509,
+      "epoch": 1.74,
+      "learning_rate": 9.651785714285715e-05,
+      "loss": 1.5828,
       "step": 390
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 9.22879684418146e-05,
-      "loss": 2.017,
+      "epoch": 1.75,
+      "learning_rate": 9.650892857142858e-05,
+      "loss": 1.5386,
       "step": 391
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 9.226824457593689e-05,
-      "loss": 1.8802,
+      "epoch": 1.75,
+      "learning_rate": 9.65e-05,
+      "loss": 1.6475,
       "step": 392
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 9.224852071005917e-05,
-      "loss": 2.0176,
+      "epoch": 1.75,
+      "learning_rate": 9.649107142857144e-05,
+      "loss": 1.5836,
       "step": 393
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 9.222879684418146e-05,
-      "loss": 2.028,
+      "epoch": 1.76,
+      "learning_rate": 9.648214285714287e-05,
+      "loss": 1.5395,
       "step": 394
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 9.220907297830376e-05,
-      "loss": 1.9025,
+      "epoch": 1.76,
+      "learning_rate": 9.647321428571429e-05,
+      "loss": 1.5331,
       "step": 395
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 9.218934911242604e-05,
-      "loss": 1.9234,
+      "epoch": 1.77,
+      "learning_rate": 9.646428571428572e-05,
+      "loss": 1.548,
       "step": 396
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 9.216962524654833e-05,
-      "loss": 2.0351,
+      "epoch": 1.77,
+      "learning_rate": 9.645535714285716e-05,
+      "loss": 1.6341,
       "step": 397
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 9.214990138067061e-05,
-      "loss": 1.9522,
+      "epoch": 1.78,
+      "learning_rate": 9.644642857142857e-05,
+      "loss": 1.5359,
       "step": 398
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 9.21301775147929e-05,
-      "loss": 1.95,
+      "epoch": 1.78,
+      "learning_rate": 9.64375e-05,
+      "loss": 1.5042,
       "step": 399
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 9.21104536489152e-05,
-      "loss": 2.0061,
+      "epoch": 1.79,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 1.5705,
       "step": 400
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 9.209072978303748e-05,
-      "loss": 2.0925,
+      "epoch": 1.79,
+      "learning_rate": 9.641964285714287e-05,
+      "loss": 1.4079,
       "step": 401
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 9.207100591715977e-05,
-      "loss": 1.892,
+      "epoch": 1.79,
+      "learning_rate": 9.641071428571428e-05,
+      "loss": 1.5099,
       "step": 402
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 9.205128205128205e-05,
-      "loss": 2.0573,
+      "epoch": 1.8,
+      "learning_rate": 9.640178571428572e-05,
+      "loss": 1.5978,
       "step": 403
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 9.203155818540434e-05,
-      "loss": 1.9609,
+      "epoch": 1.8,
+      "learning_rate": 9.639285714285715e-05,
+      "loss": 1.6551,
       "step": 404
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 9.201183431952663e-05,
-      "loss": 2.0249,
+      "epoch": 1.81,
+      "learning_rate": 9.638392857142857e-05,
+      "loss": 1.4556,
       "step": 405
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 9.199211045364892e-05,
-      "loss": 1.9952,
+      "epoch": 1.81,
+      "learning_rate": 9.6375e-05,
+      "loss": 1.667,
       "step": 406
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 9.197238658777121e-05,
-      "loss": 1.9616,
+      "epoch": 1.82,
+      "learning_rate": 9.636607142857144e-05,
+      "loss": 1.5597,
       "step": 407
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 9.19526627218935e-05,
-      "loss": 2.0223,
+      "epoch": 1.82,
+      "learning_rate": 9.635714285714286e-05,
+      "loss": 1.6819,
       "step": 408
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 9.193293885601579e-05,
-      "loss": 2.023,
+      "epoch": 1.83,
+      "learning_rate": 9.634821428571429e-05,
+      "loss": 1.5859,
       "step": 409
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 9.191321499013808e-05,
-      "loss": 2.0975,
+      "epoch": 1.83,
+      "learning_rate": 9.633928571428573e-05,
+      "loss": 1.5974,
       "step": 410
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 9.189349112426036e-05,
-      "loss": 1.9645,
+      "epoch": 1.83,
+      "learning_rate": 9.633035714285715e-05,
+      "loss": 1.5391,
       "step": 411
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 9.187376725838265e-05,
-      "loss": 1.992,
+      "epoch": 1.84,
+      "learning_rate": 9.632142857142858e-05,
+      "loss": 1.5681,
       "step": 412
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 9.185404339250494e-05,
-      "loss": 2.0163,
+      "epoch": 1.84,
+      "learning_rate": 9.63125e-05,
+      "loss": 1.486,
       "step": 413
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 9.183431952662722e-05,
-      "loss": 2.0367,
+      "epoch": 1.85,
+      "learning_rate": 9.630357142857144e-05,
+      "loss": 1.4467,
       "step": 414
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 9.181459566074951e-05,
-      "loss": 2.0698,
+      "epoch": 1.85,
+      "learning_rate": 9.629464285714285e-05,
+      "loss": 1.648,
       "step": 415
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 9.179487179487179e-05,
-      "loss": 1.9803,
+      "epoch": 1.86,
+      "learning_rate": 9.628571428571429e-05,
+      "loss": 1.5644,
       "step": 416
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 9.177514792899408e-05,
-      "loss": 1.9054,
+      "epoch": 1.86,
+      "learning_rate": 9.627678571428572e-05,
+      "loss": 1.558,
       "step": 417
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 9.175542406311637e-05,
-      "loss": 2.0156,
+      "epoch": 1.87,
+      "learning_rate": 9.626785714285714e-05,
+      "loss": 1.5024,
       "step": 418
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 9.173570019723867e-05,
-      "loss": 1.9809,
+      "epoch": 1.87,
+      "learning_rate": 9.625892857142857e-05,
+      "loss": 1.5983,
       "step": 419
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 9.171597633136096e-05,
-      "loss": 1.9561,
+      "epoch": 1.88,
+      "learning_rate": 9.625000000000001e-05,
+      "loss": 1.5898,
       "step": 420
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 9.169625246548324e-05,
-      "loss": 1.9079,
+      "epoch": 1.88,
+      "learning_rate": 9.624107142857143e-05,
+      "loss": 1.6062,
       "step": 421
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 9.167652859960553e-05,
-      "loss": 2.0044,
+      "epoch": 1.88,
+      "learning_rate": 9.623214285714286e-05,
+      "loss": 1.5297,
       "step": 422
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 9.165680473372782e-05,
-      "loss": 1.9986,
+      "epoch": 1.89,
+      "learning_rate": 9.622321428571428e-05,
+      "loss": 1.5026,
       "step": 423
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 9.16370808678501e-05,
-      "loss": 1.9406,
+      "epoch": 1.89,
+      "learning_rate": 9.621428571428572e-05,
+      "loss": 1.5891,
       "step": 424
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 9.161735700197239e-05,
-      "loss": 1.9057,
+      "epoch": 1.9,
+      "learning_rate": 9.620535714285715e-05,
+      "loss": 1.521,
       "step": 425
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 9.159763313609467e-05,
-      "loss": 1.7755,
+      "epoch": 1.9,
+      "learning_rate": 9.619642857142857e-05,
+      "loss": 1.4118,
       "step": 426
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 9.157790927021696e-05,
-      "loss": 1.8075,
+      "epoch": 1.91,
+      "learning_rate": 9.618750000000001e-05,
+      "loss": 1.5668,
       "step": 427
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 9.155818540433925e-05,
-      "loss": 2.0759,
+      "epoch": 1.91,
+      "learning_rate": 9.617857142857143e-05,
+      "loss": 1.7086,
       "step": 428
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 9.153846153846155e-05,
-      "loss": 1.9309,
+      "epoch": 1.92,
+      "learning_rate": 9.616964285714286e-05,
+      "loss": 1.6232,
       "step": 429
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 9.151873767258384e-05,
-      "loss": 2.0779,
+      "epoch": 1.92,
+      "learning_rate": 9.616071428571429e-05,
+      "loss": 1.5602,
       "step": 430
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 9.149901380670613e-05,
-      "loss": 2.0214,
+      "epoch": 1.92,
+      "learning_rate": 9.615178571428572e-05,
+      "loss": 1.5533,
       "step": 431
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 9.147928994082841e-05,
-      "loss": 2.0151,
+      "epoch": 1.93,
+      "learning_rate": 9.614285714285714e-05,
+      "loss": 1.5881,
       "step": 432
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 9.14595660749507e-05,
-      "loss": 1.9855,
+      "epoch": 1.93,
+      "learning_rate": 9.613392857142858e-05,
+      "loss": 1.6196,
       "step": 433
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 9.143984220907298e-05,
-      "loss": 1.9801,
+      "epoch": 1.94,
+      "learning_rate": 9.6125e-05,
+      "loss": 1.4679,
       "step": 434
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 9.142011834319527e-05,
-      "loss": 1.92,
+      "epoch": 1.94,
+      "learning_rate": 9.611607142857143e-05,
+      "loss": 1.5365,
       "step": 435
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 9.140039447731756e-05,
-      "loss": 1.8712,
+      "epoch": 1.95,
+      "learning_rate": 9.610714285714286e-05,
+      "loss": 1.4562,
       "step": 436
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 9.138067061143984e-05,
-      "loss": 2.0598,
+      "epoch": 1.95,
+      "learning_rate": 9.60982142857143e-05,
+      "loss": 1.5111,
       "step": 437
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 9.136094674556213e-05,
-      "loss": 1.9474,
+      "epoch": 1.96,
+      "learning_rate": 9.608928571428572e-05,
+      "loss": 1.5057,
       "step": 438
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 9.134122287968441e-05,
-      "loss": 1.9577,
+      "epoch": 1.96,
+      "learning_rate": 9.608035714285715e-05,
+      "loss": 1.5548,
       "step": 439
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 9.13214990138067e-05,
-      "loss": 1.9435,
+      "epoch": 1.96,
+      "learning_rate": 9.607142857142859e-05,
+      "loss": 1.5396,
       "step": 440
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 9.1301775147929e-05,
-      "loss": 1.9947,
+      "epoch": 1.97,
+      "learning_rate": 9.60625e-05,
+      "loss": 1.6063,
       "step": 441
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 9.128205128205129e-05,
-      "loss": 2.0715,
+      "epoch": 1.97,
+      "learning_rate": 9.605357142857144e-05,
+      "loss": 1.5789,
       "step": 442
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 9.126232741617358e-05,
-      "loss": 1.8831,
+      "epoch": 1.98,
+      "learning_rate": 9.604464285714286e-05,
+      "loss": 1.5433,
       "step": 443
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 9.124260355029586e-05,
-      "loss": 2.0133,
+      "epoch": 1.98,
+      "learning_rate": 9.60357142857143e-05,
+      "loss": 1.5016,
       "step": 444
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 9.122287968441815e-05,
-      "loss": 2.001,
+      "epoch": 1.99,
+      "learning_rate": 9.602678571428571e-05,
+      "loss": 1.5813,
       "step": 445
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 9.120315581854044e-05,
-      "loss": 2.0648,
+      "epoch": 1.99,
+      "learning_rate": 9.601785714285715e-05,
+      "loss": 1.6069,
       "step": 446
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 9.118343195266272e-05,
-      "loss": 1.9433,
+      "epoch": 2.0,
+      "learning_rate": 9.600892857142858e-05,
+      "loss": 1.5604,
       "step": 447
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 9.116370808678501e-05,
-      "loss": 1.9728,
+      "epoch": 2.0,
+      "learning_rate": 9.6e-05,
+      "loss": 1.5451,
       "step": 448
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 9.11439842209073e-05,
-      "loss": 1.9932,
+      "epoch": 2.0,
+      "learning_rate": 9.599107142857143e-05,
+      "loss": 1.5837,
       "step": 449
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 9.112426035502959e-05,
-      "loss": 1.9782,
+      "epoch": 2.01,
+      "learning_rate": 9.598214285714287e-05,
+      "loss": 1.4867,
       "step": 450
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 9.110453648915188e-05,
-      "loss": 1.8738,
+      "epoch": 2.01,
+      "learning_rate": 9.597321428571429e-05,
+      "loss": 1.5589,
       "step": 451
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 9.108481262327416e-05,
-      "loss": 1.9597,
+      "epoch": 2.02,
+      "learning_rate": 9.596428571428572e-05,
+      "loss": 1.6146,
       "step": 452
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 9.106508875739645e-05,
-      "loss": 1.9394,
+      "epoch": 2.02,
+      "learning_rate": 9.595535714285714e-05,
+      "loss": 1.4165,
       "step": 453
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 9.104536489151875e-05,
-      "loss": 2.0332,
+      "epoch": 2.03,
+      "learning_rate": 9.594642857142858e-05,
+      "loss": 1.4802,
       "step": 454
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 9.102564102564103e-05,
-      "loss": 1.8407,
+      "epoch": 2.03,
+      "learning_rate": 9.593750000000001e-05,
+      "loss": 1.6066,
       "step": 455
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 9.100591715976332e-05,
-      "loss": 2.0598,
+      "epoch": 2.04,
+      "learning_rate": 9.592857142857143e-05,
+      "loss": 1.4447,
       "step": 456
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 9.09861932938856e-05,
-      "loss": 1.9703,
+      "epoch": 2.04,
+      "learning_rate": 9.591964285714287e-05,
+      "loss": 1.5702,
       "step": 457
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 9.09664694280079e-05,
-      "loss": 2.0203,
+      "epoch": 2.04,
+      "learning_rate": 9.591071428571428e-05,
+      "loss": 1.5638,
       "step": 458
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 9.094674556213019e-05,
-      "loss": 2.0046,
+      "epoch": 2.05,
+      "learning_rate": 9.590178571428572e-05,
+      "loss": 1.506,
       "step": 459
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 9.092702169625247e-05,
-      "loss": 1.9535,
+      "epoch": 2.05,
+      "learning_rate": 9.589285714285715e-05,
+      "loss": 1.5285,
       "step": 460
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 9.090729783037476e-05,
-      "loss": 1.9881,
+      "epoch": 2.06,
+      "learning_rate": 9.588392857142857e-05,
+      "loss": 1.4575,
       "step": 461
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 9.088757396449704e-05,
-      "loss": 1.9218,
+      "epoch": 2.06,
+      "learning_rate": 9.5875e-05,
+      "loss": 1.6644,
       "step": 462
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 9.086785009861933e-05,
-      "loss": 1.9943,
+      "epoch": 2.07,
+      "learning_rate": 9.586607142857144e-05,
+      "loss": 1.449,
       "step": 463
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 9.084812623274162e-05,
-      "loss": 2.036,
+      "epoch": 2.07,
+      "learning_rate": 9.585714285714285e-05,
+      "loss": 1.5314,
       "step": 464
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 9.082840236686391e-05,
-      "loss": 1.9648,
+      "epoch": 2.08,
+      "learning_rate": 9.584821428571429e-05,
+      "loss": 1.4823,
       "step": 465
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 9.08086785009862e-05,
-      "loss": 2.0132,
+      "epoch": 2.08,
+      "learning_rate": 9.583928571428571e-05,
+      "loss": 1.5045,
       "step": 466
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 9.078895463510848e-05,
-      "loss": 1.8734,
+      "epoch": 2.08,
+      "learning_rate": 9.583035714285715e-05,
+      "loss": 1.4687,
       "step": 467
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 9.076923076923078e-05,
-      "loss": 1.9128,
+      "epoch": 2.09,
+      "learning_rate": 9.582142857142858e-05,
+      "loss": 1.5574,
       "step": 468
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 9.074950690335307e-05,
-      "loss": 1.9224,
+      "epoch": 2.09,
+      "learning_rate": 9.58125e-05,
+      "loss": 1.5077,
       "step": 469
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 9.072978303747535e-05,
-      "loss": 1.8659,
+      "epoch": 2.1,
+      "learning_rate": 9.580357142857144e-05,
+      "loss": 1.4896,
       "step": 470
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 9.071005917159764e-05,
-      "loss": 1.8589,
+      "epoch": 2.1,
+      "learning_rate": 9.579464285714286e-05,
+      "loss": 1.5651,
       "step": 471
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 9.069033530571992e-05,
-      "loss": 2.0101,
+      "epoch": 2.11,
+      "learning_rate": 9.57857142857143e-05,
+      "loss": 1.4311,
       "step": 472
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 9.067061143984221e-05,
-      "loss": 1.9228,
+      "epoch": 2.11,
+      "learning_rate": 9.577678571428572e-05,
+      "loss": 1.5298,
       "step": 473
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 9.06508875739645e-05,
-      "loss": 1.937,
+      "epoch": 2.12,
+      "learning_rate": 9.576785714285715e-05,
+      "loss": 1.5014,
       "step": 474
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 9.063116370808678e-05,
-      "loss": 1.9795,
+      "epoch": 2.12,
+      "learning_rate": 9.575892857142857e-05,
+      "loss": 1.4654,
       "step": 475
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 9.061143984220907e-05,
-      "loss": 1.9465,
+      "epoch": 2.12,
+      "learning_rate": 9.575000000000001e-05,
+      "loss": 1.5088,
       "step": 476
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 9.059171597633136e-05,
-      "loss": 1.7728,
+      "epoch": 2.13,
+      "learning_rate": 9.574107142857144e-05,
+      "loss": 1.5727,
       "step": 477
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 9.057199211045366e-05,
-      "loss": 2.0266,
+      "epoch": 2.13,
+      "learning_rate": 9.573214285714286e-05,
+      "loss": 1.5234,
       "step": 478
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 9.055226824457595e-05,
-      "loss": 1.8395,
+      "epoch": 2.14,
+      "learning_rate": 9.572321428571429e-05,
+      "loss": 1.451,
       "step": 479
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 9.053254437869823e-05,
-      "loss": 1.9741,
+      "epoch": 2.14,
+      "learning_rate": 9.571428571428573e-05,
+      "loss": 1.5821,
       "step": 480
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 9.051282051282052e-05,
-      "loss": 1.9765,
+      "epoch": 2.15,
+      "learning_rate": 9.570535714285714e-05,
+      "loss": 1.5135,
       "step": 481
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 9.049309664694281e-05,
-      "loss": 2.001,
+      "epoch": 2.15,
+      "learning_rate": 9.569642857142858e-05,
+      "loss": 1.4943,
       "step": 482
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 9.047337278106509e-05,
-      "loss": 1.8838,
+      "epoch": 2.16,
+      "learning_rate": 9.56875e-05,
+      "loss": 1.4669,
       "step": 483
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 9.045364891518738e-05,
-      "loss": 1.9306,
+      "epoch": 2.16,
+      "learning_rate": 9.567857142857143e-05,
+      "loss": 1.5256,
       "step": 484
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 9.043392504930966e-05,
-      "loss": 2.0794,
+      "epoch": 2.17,
+      "learning_rate": 9.566964285714287e-05,
+      "loss": 1.5985,
       "step": 485
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 9.041420118343195e-05,
-      "loss": 1.9838,
+      "epoch": 2.17,
+      "learning_rate": 9.566071428571429e-05,
+      "loss": 1.5649,
       "step": 486
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 9.039447731755424e-05,
-      "loss": 1.9172,
+      "epoch": 2.17,
+      "learning_rate": 9.565178571428573e-05,
+      "loss": 1.5177,
       "step": 487
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 9.037475345167654e-05,
-      "loss": 2.0079,
+      "epoch": 2.18,
+      "learning_rate": 9.564285714285714e-05,
+      "loss": 1.5474,
       "step": 488
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 9.035502958579883e-05,
-      "loss": 1.9569,
+      "epoch": 2.18,
+      "learning_rate": 9.563392857142858e-05,
+      "loss": 1.386,
       "step": 489
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 9.033530571992111e-05,
-      "loss": 1.8171,
+      "epoch": 2.19,
+      "learning_rate": 9.562500000000001e-05,
+      "loss": 1.4428,
       "step": 490
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 9.03155818540434e-05,
-      "loss": 1.8922,
+      "epoch": 2.19,
+      "learning_rate": 9.561607142857143e-05,
+      "loss": 1.5644,
       "step": 491
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 9.029585798816569e-05,
-      "loss": 1.97,
+      "epoch": 2.2,
+      "learning_rate": 9.560714285714286e-05,
+      "loss": 1.532,
       "step": 492
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 9.027613412228797e-05,
-      "loss": 1.9989,
+      "epoch": 2.2,
+      "learning_rate": 9.55982142857143e-05,
+      "loss": 1.4639,
       "step": 493
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 9.025641025641026e-05,
-      "loss": 2.0203,
+      "epoch": 2.21,
+      "learning_rate": 9.558928571428571e-05,
+      "loss": 1.5476,
       "step": 494
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 9.023668639053254e-05,
-      "loss": 1.9162,
+      "epoch": 2.21,
+      "learning_rate": 9.558035714285715e-05,
+      "loss": 1.4034,
       "step": 495
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 9.021696252465483e-05,
-      "loss": 1.972,
+      "epoch": 2.21,
+      "learning_rate": 9.557142857142857e-05,
+      "loss": 1.5721,
       "step": 496
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 9.019723865877713e-05,
-      "loss": 1.8764,
+      "epoch": 2.22,
+      "learning_rate": 9.55625e-05,
+      "loss": 1.5342,
       "step": 497
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 9.01775147928994e-05,
-      "loss": 2.0029,
+      "epoch": 2.22,
+      "learning_rate": 9.555357142857144e-05,
+      "loss": 1.4314,
       "step": 498
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 9.01577909270217e-05,
-      "loss": 1.9803,
+      "epoch": 2.23,
+      "learning_rate": 9.554464285714286e-05,
+      "loss": 1.508,
       "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,
+      "epoch": 2.23,
+      "learning_rate": 9.553571428571429e-05,
+      "loss": 1.4853,
       "step": 500
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 9.011834319526628e-05,
-      "loss": 1.8567,
+      "epoch": 2.24,
+      "learning_rate": 9.552678571428571e-05,
+      "loss": 1.5233,
       "step": 501
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 9.009861932938857e-05,
-      "loss": 1.9247,
+      "epoch": 2.24,
+      "learning_rate": 9.551785714285715e-05,
+      "loss": 1.5758,
       "step": 502
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 9.007889546351085e-05,
-      "loss": 1.9948,
+      "epoch": 2.25,
+      "learning_rate": 9.550892857142858e-05,
+      "loss": 1.5616,
       "step": 503
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 9.005917159763314e-05,
-      "loss": 1.9587,
+      "epoch": 2.25,
+      "learning_rate": 9.55e-05,
+      "loss": 1.4569,
       "step": 504
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 9.003944773175543e-05,
-      "loss": 1.917,
+      "epoch": 2.25,
+      "learning_rate": 9.549107142857143e-05,
+      "loss": 1.4903,
       "step": 505
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 9.001972386587771e-05,
-      "loss": 1.8527,
+      "epoch": 2.26,
+      "learning_rate": 9.548214285714287e-05,
+      "loss": 1.5379,
       "step": 506
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 9e-05,
-      "loss": 1.9483,
+      "epoch": 2.26,
+      "learning_rate": 9.547321428571428e-05,
+      "loss": 1.6826,
       "step": 507
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 8.998027613412228e-05,
-      "loss": 1.9796,
+      "epoch": 2.27,
+      "learning_rate": 9.546428571428572e-05,
+      "loss": 1.4933,
       "step": 508
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 8.996055226824458e-05,
-      "loss": 2.0263,
+      "epoch": 2.27,
+      "learning_rate": 9.545535714285714e-05,
+      "loss": 1.5434,
       "step": 509
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 8.994082840236687e-05,
-      "loss": 1.8482,
+      "epoch": 2.28,
+      "learning_rate": 9.544642857142858e-05,
+      "loss": 1.5703,
       "step": 510
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 8.992110453648915e-05,
-      "loss": 1.9704,
+      "epoch": 2.28,
+      "learning_rate": 9.54375e-05,
+      "loss": 1.4764,
       "step": 511
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 8.990138067061144e-05,
-      "loss": 1.9237,
+      "epoch": 2.29,
+      "learning_rate": 9.542857142857143e-05,
+      "loss": 1.511,
       "step": 512
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 8.988165680473373e-05,
-      "loss": 1.9591,
+      "epoch": 2.29,
+      "learning_rate": 9.541964285714286e-05,
+      "loss": 1.4423,
       "step": 513
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 8.986193293885602e-05,
-      "loss": 2.0286,
+      "epoch": 2.29,
+      "learning_rate": 9.541071428571429e-05,
+      "loss": 1.5096,
       "step": 514
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 8.984220907297832e-05,
-      "loss": 1.9616,
+      "epoch": 2.3,
+      "learning_rate": 9.540178571428572e-05,
+      "loss": 1.6123,
       "step": 515
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 8.98224852071006e-05,
-      "loss": 1.8996,
+      "epoch": 2.3,
+      "learning_rate": 9.539285714285715e-05,
+      "loss": 1.5126,
       "step": 516
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 8.980276134122289e-05,
-      "loss": 1.937,
+      "epoch": 2.31,
+      "learning_rate": 9.538392857142858e-05,
+      "loss": 1.4983,
       "step": 517
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 8.978303747534516e-05,
-      "loss": 1.8483,
+      "epoch": 2.31,
+      "learning_rate": 9.5375e-05,
+      "loss": 1.493,
       "step": 518
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 8.976331360946746e-05,
-      "loss": 1.9662,
+      "epoch": 2.32,
+      "learning_rate": 9.536607142857144e-05,
+      "loss": 1.5328,
       "step": 519
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 8.974358974358975e-05,
-      "loss": 1.8575,
+      "epoch": 2.32,
+      "learning_rate": 9.535714285714287e-05,
+      "loss": 1.5458,
       "step": 520
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 8.972386587771203e-05,
-      "loss": 1.9057,
+      "epoch": 2.33,
+      "learning_rate": 9.534821428571429e-05,
+      "loss": 1.5433,
       "step": 521
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 8.970414201183432e-05,
-      "loss": 1.9537,
+      "epoch": 2.33,
+      "learning_rate": 9.533928571428572e-05,
+      "loss": 1.4947,
       "step": 522
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 8.968441814595661e-05,
-      "loss": 1.9032,
+      "epoch": 2.33,
+      "learning_rate": 9.533035714285716e-05,
+      "loss": 1.5335,
       "step": 523
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 8.96646942800789e-05,
-      "loss": 1.8777,
+      "epoch": 2.34,
+      "learning_rate": 9.532142857142857e-05,
+      "loss": 1.421,
       "step": 524
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 8.96449704142012e-05,
-      "loss": 1.974,
+      "epoch": 2.34,
+      "learning_rate": 9.53125e-05,
+      "loss": 1.5472,
       "step": 525
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 8.962524654832347e-05,
-      "loss": 1.9356,
+      "epoch": 2.35,
+      "learning_rate": 9.530357142857143e-05,
+      "loss": 1.519,
       "step": 526
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 8.960552268244577e-05,
-      "loss": 1.9666,
+      "epoch": 2.35,
+      "learning_rate": 9.529464285714286e-05,
+      "loss": 1.5004,
       "step": 527
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 8.958579881656806e-05,
-      "loss": 1.9192,
+      "epoch": 2.36,
+      "learning_rate": 9.52857142857143e-05,
+      "loss": 1.4855,
       "step": 528
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 8.956607495069034e-05,
-      "loss": 1.9061,
+      "epoch": 2.36,
+      "learning_rate": 9.527678571428572e-05,
+      "loss": 1.5644,
       "step": 529
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 8.954635108481263e-05,
-      "loss": 1.9993,
+      "epoch": 2.37,
+      "learning_rate": 9.526785714285715e-05,
+      "loss": 1.5725,
       "step": 530
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 8.952662721893491e-05,
-      "loss": 1.8754,
+      "epoch": 2.37,
+      "learning_rate": 9.525892857142857e-05,
+      "loss": 1.5124,
       "step": 531
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 8.95069033530572e-05,
-      "loss": 1.8501,
+      "epoch": 2.38,
+      "learning_rate": 9.525000000000001e-05,
+      "loss": 1.5359,
       "step": 532
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 8.948717948717949e-05,
-      "loss": 1.9235,
+      "epoch": 2.38,
+      "learning_rate": 9.524107142857144e-05,
+      "loss": 1.4716,
       "step": 533
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 8.946745562130177e-05,
-      "loss": 1.9945,
+      "epoch": 2.38,
+      "learning_rate": 9.523214285714286e-05,
+      "loss": 1.5175,
       "step": 534
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 8.944773175542406e-05,
-      "loss": 1.9604,
+      "epoch": 2.39,
+      "learning_rate": 9.522321428571429e-05,
+      "loss": 1.5584,
       "step": 535
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 8.942800788954635e-05,
-      "loss": 1.8508,
+      "epoch": 2.39,
+      "learning_rate": 9.521428571428573e-05,
+      "loss": 1.5534,
       "step": 536
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 8.940828402366865e-05,
-      "loss": 1.8619,
+      "epoch": 2.4,
+      "learning_rate": 9.520535714285714e-05,
+      "loss": 1.4419,
       "step": 537
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 8.938856015779094e-05,
-      "loss": 1.8698,
+      "epoch": 2.4,
+      "learning_rate": 9.519642857142858e-05,
+      "loss": 1.5132,
       "step": 538
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 8.936883629191322e-05,
-      "loss": 1.971,
+      "epoch": 2.41,
+      "learning_rate": 9.51875e-05,
+      "loss": 1.4877,
       "step": 539
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 8.934911242603551e-05,
-      "loss": 1.8223,
+      "epoch": 2.41,
+      "learning_rate": 9.517857142857143e-05,
+      "loss": 1.474,
       "step": 540
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 8.93293885601578e-05,
-      "loss": 1.8814,
+      "epoch": 2.42,
+      "learning_rate": 9.516964285714285e-05,
+      "loss": 1.5971,
       "step": 541
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 8.930966469428008e-05,
-      "loss": 1.8664,
+      "epoch": 2.42,
+      "learning_rate": 9.516071428571429e-05,
+      "loss": 1.5381,
       "step": 542
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 8.928994082840237e-05,
-      "loss": 1.929,
+      "epoch": 2.42,
+      "learning_rate": 9.515178571428572e-05,
+      "loss": 1.5614,
       "step": 543
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 8.927021696252465e-05,
-      "loss": 1.8556,
+      "epoch": 2.43,
+      "learning_rate": 9.514285714285714e-05,
+      "loss": 1.3845,
       "step": 544
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 8.925049309664694e-05,
-      "loss": 1.8419,
+      "epoch": 2.43,
+      "learning_rate": 9.513392857142858e-05,
+      "loss": 1.457,
       "step": 545
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 8.923076923076924e-05,
-      "loss": 1.9459,
+      "epoch": 2.44,
+      "learning_rate": 9.512500000000001e-05,
+      "loss": 1.473,
       "step": 546
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 8.921104536489153e-05,
-      "loss": 1.8525,
+      "epoch": 2.44,
+      "learning_rate": 9.511607142857143e-05,
+      "loss": 1.4207,
       "step": 547
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 8.919132149901382e-05,
-      "loss": 1.8501,
+      "epoch": 2.45,
+      "learning_rate": 9.510714285714286e-05,
+      "loss": 1.5272,
       "step": 548
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 8.91715976331361e-05,
-      "loss": 1.8935,
+      "epoch": 2.45,
+      "learning_rate": 9.50982142857143e-05,
+      "loss": 1.5021,
       "step": 549
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 8.915187376725839e-05,
-      "loss": 1.9246,
+      "epoch": 2.46,
+      "learning_rate": 9.508928571428571e-05,
+      "loss": 1.5356,
       "step": 550
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 8.913214990138068e-05,
-      "loss": 2.0124,
+      "epoch": 2.46,
+      "learning_rate": 9.508035714285715e-05,
+      "loss": 1.4088,
       "step": 551
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 8.911242603550296e-05,
-      "loss": 1.9211,
+      "epoch": 2.46,
+      "learning_rate": 9.507142857142857e-05,
+      "loss": 1.5124,
       "step": 552
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 8.909270216962525e-05,
-      "loss": 1.8827,
+      "epoch": 2.47,
+      "learning_rate": 9.506250000000001e-05,
+      "loss": 1.4977,
       "step": 553
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 8.907297830374753e-05,
-      "loss": 1.8695,
+      "epoch": 2.47,
+      "learning_rate": 9.505357142857143e-05,
+      "loss": 1.5044,
       "step": 554
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 8.905325443786982e-05,
-      "loss": 1.9441,
+      "epoch": 2.48,
+      "learning_rate": 9.504464285714286e-05,
+      "loss": 1.4457,
       "step": 555
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 8.903353057199212e-05,
-      "loss": 1.8444,
+      "epoch": 2.48,
+      "learning_rate": 9.503571428571429e-05,
+      "loss": 1.5602,
       "step": 556
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 8.90138067061144e-05,
-      "loss": 1.8964,
+      "epoch": 2.49,
+      "learning_rate": 9.502678571428572e-05,
+      "loss": 1.5178,
       "step": 557
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 8.899408284023669e-05,
-      "loss": 1.7572,
+      "epoch": 2.49,
+      "learning_rate": 9.501785714285714e-05,
+      "loss": 1.4494,
       "step": 558
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 8.897435897435898e-05,
-      "loss": 1.9514,
+      "epoch": 2.5,
+      "learning_rate": 9.500892857142858e-05,
+      "loss": 1.5541,
       "step": 559
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 8.895463510848127e-05,
-      "loss": 1.9352,
+      "epoch": 2.5,
+      "learning_rate": 9.5e-05,
+      "loss": 1.493,
       "step": 560
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 8.893491124260356e-05,
-      "loss": 1.8888,
+      "epoch": 2.5,
+      "learning_rate": 9.499107142857143e-05,
+      "loss": 1.4513,
       "step": 561
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 8.891518737672584e-05,
-      "loss": 1.9046,
+      "epoch": 2.51,
+      "learning_rate": 9.498214285714287e-05,
+      "loss": 1.5018,
       "step": 562
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 8.889546351084813e-05,
-      "loss": 1.8477,
+      "epoch": 2.51,
+      "learning_rate": 9.497321428571428e-05,
+      "loss": 1.5491,
       "step": 563
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 8.887573964497043e-05,
-      "loss": 1.9415,
+      "epoch": 2.52,
+      "learning_rate": 9.496428571428572e-05,
+      "loss": 1.4503,
       "step": 564
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 8.88560157790927e-05,
-      "loss": 1.9449,
+      "epoch": 2.52,
+      "learning_rate": 9.495535714285715e-05,
+      "loss": 1.491,
       "step": 565
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 8.8836291913215e-05,
-      "loss": 1.9514,
+      "epoch": 2.53,
+      "learning_rate": 9.494642857142859e-05,
+      "loss": 1.5102,
       "step": 566
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 8.881656804733727e-05,
-      "loss": 1.896,
+      "epoch": 2.53,
+      "learning_rate": 9.49375e-05,
+      "loss": 1.5404,
       "step": 567
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 8.879684418145957e-05,
-      "loss": 1.8957,
+      "epoch": 2.54,
+      "learning_rate": 9.492857142857144e-05,
+      "loss": 1.4567,
       "step": 568
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 8.877712031558186e-05,
-      "loss": 1.9475,
+      "epoch": 2.54,
+      "learning_rate": 9.491964285714286e-05,
+      "loss": 1.4939,
       "step": 569
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 8.875739644970414e-05,
-      "loss": 1.8998,
+      "epoch": 2.54,
+      "learning_rate": 9.491071428571429e-05,
+      "loss": 1.5916,
       "step": 570
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 8.873767258382644e-05,
-      "loss": 1.7727,
+      "epoch": 2.55,
+      "learning_rate": 9.490178571428571e-05,
+      "loss": 1.5221,
       "step": 571
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 8.871794871794872e-05,
-      "loss": 1.8896,
+      "epoch": 2.55,
+      "learning_rate": 9.489285714285715e-05,
+      "loss": 1.5153,
       "step": 572
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 8.869822485207101e-05,
-      "loss": 1.9018,
+      "epoch": 2.56,
+      "learning_rate": 9.488392857142858e-05,
+      "loss": 1.519,
       "step": 573
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 8.86785009861933e-05,
-      "loss": 1.9115,
+      "epoch": 2.56,
+      "learning_rate": 9.4875e-05,
+      "loss": 1.5042,
       "step": 574
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 8.865877712031558e-05,
-      "loss": 1.8884,
+      "epoch": 2.57,
+      "learning_rate": 9.486607142857144e-05,
+      "loss": 1.4273,
       "step": 575
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 8.863905325443788e-05,
-      "loss": 1.897,
+      "epoch": 2.57,
+      "learning_rate": 9.485714285714287e-05,
+      "loss": 1.5437,
       "step": 576
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 8.861932938856016e-05,
-      "loss": 1.8468,
+      "epoch": 2.58,
+      "learning_rate": 9.484821428571429e-05,
+      "loss": 1.594,
       "step": 577
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 8.859960552268245e-05,
-      "loss": 1.9185,
+      "epoch": 2.58,
+      "learning_rate": 9.483928571428572e-05,
+      "loss": 1.5411,
       "step": 578
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 8.857988165680474e-05,
-      "loss": 1.9223,
+      "epoch": 2.58,
+      "learning_rate": 9.483035714285716e-05,
+      "loss": 1.5078,
       "step": 579
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 8.856015779092702e-05,
-      "loss": 1.8192,
+      "epoch": 2.59,
+      "learning_rate": 9.482142857142857e-05,
+      "loss": 1.5104,
       "step": 580
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 8.854043392504931e-05,
-      "loss": 2.0479,
+      "epoch": 2.59,
+      "learning_rate": 9.481250000000001e-05,
+      "loss": 1.5049,
       "step": 581
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 8.85207100591716e-05,
-      "loss": 1.9365,
+      "epoch": 2.6,
+      "learning_rate": 9.480357142857143e-05,
+      "loss": 1.4739,
       "step": 582
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 8.85009861932939e-05,
-      "loss": 1.8941,
+      "epoch": 2.6,
+      "learning_rate": 9.479464285714286e-05,
+      "loss": 1.5885,
       "step": 583
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 8.848126232741619e-05,
-      "loss": 1.9021,
+      "epoch": 2.61,
+      "learning_rate": 9.478571428571428e-05,
+      "loss": 1.5422,
       "step": 584
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 8.846153846153847e-05,
-      "loss": 1.9156,
+      "epoch": 2.61,
+      "learning_rate": 9.477678571428572e-05,
+      "loss": 1.5097,
       "step": 585
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 8.844181459566076e-05,
-      "loss": 1.8298,
+      "epoch": 2.62,
+      "learning_rate": 9.476785714285715e-05,
+      "loss": 1.5254,
       "step": 586
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.842209072978305e-05,
-      "loss": 1.8583,
+      "epoch": 2.62,
+      "learning_rate": 9.475892857142857e-05,
+      "loss": 1.5535,
       "step": 587
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.840236686390533e-05,
-      "loss": 1.8808,
+      "epoch": 2.62,
+      "learning_rate": 9.475e-05,
+      "loss": 1.5287,
       "step": 588
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.838264299802762e-05,
-      "loss": 1.8663,
+      "epoch": 2.63,
+      "learning_rate": 9.474107142857144e-05,
+      "loss": 1.5102,
       "step": 589
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.83629191321499e-05,
-      "loss": 1.8732,
+      "epoch": 2.63,
+      "learning_rate": 9.473214285714286e-05,
+      "loss": 1.5089,
       "step": 590
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.834319526627219e-05,
-      "loss": 1.9054,
+      "epoch": 2.64,
+      "learning_rate": 9.472321428571429e-05,
+      "loss": 1.4532,
       "step": 591
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.832347140039448e-05,
-      "loss": 1.8904,
+      "epoch": 2.64,
+      "learning_rate": 9.471428571428573e-05,
+      "loss": 1.4481,
       "step": 592
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.830374753451676e-05,
-      "loss": 1.8218,
+      "epoch": 2.65,
+      "learning_rate": 9.470535714285714e-05,
+      "loss": 1.4321,
       "step": 593
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 8.828402366863905e-05,
-      "loss": 2.0163,
+      "epoch": 2.65,
+      "learning_rate": 9.469642857142858e-05,
+      "loss": 1.5142,
       "step": 594
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 8.826429980276135e-05,
-      "loss": 1.8797,
+      "epoch": 2.66,
+      "learning_rate": 9.46875e-05,
+      "loss": 1.4837,
       "step": 595
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 8.824457593688364e-05,
-      "loss": 1.9516,
+      "epoch": 2.66,
+      "learning_rate": 9.467857142857143e-05,
+      "loss": 1.4947,
       "step": 596
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 8.822485207100593e-05,
-      "loss": 1.9858,
+      "epoch": 2.67,
+      "learning_rate": 9.466964285714286e-05,
+      "loss": 1.4761,
       "step": 597
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 8.820512820512821e-05,
-      "loss": 1.7547,
+      "epoch": 2.67,
+      "learning_rate": 9.46607142857143e-05,
+      "loss": 1.4891,
       "step": 598
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 8.81854043392505e-05,
-      "loss": 1.8128,
+      "epoch": 2.67,
+      "learning_rate": 9.465178571428572e-05,
+      "loss": 1.4812,
       "step": 599
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 8.816568047337278e-05,
-      "loss": 1.8551,
+      "epoch": 2.68,
+      "learning_rate": 9.464285714285715e-05,
+      "loss": 1.4301,
       "step": 600
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 8.814595660749507e-05,
-      "loss": 1.7966,
+      "epoch": 2.68,
+      "eval_loss": 1.3353039026260376,
+      "eval_runtime": 45.237,
+      "eval_samples_per_second": 289.453,
+      "eval_steps_per_second": 1.525,
+      "step": 600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463392857142857e-05,
+      "loss": 1.4693,
       "step": 601
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 8.812623274161736e-05,
-      "loss": 1.9422,
+      "epoch": 2.69,
+      "learning_rate": 9.462500000000001e-05,
+      "loss": 1.5493,
       "step": 602
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 8.810650887573964e-05,
-      "loss": 1.8748,
+      "epoch": 2.69,
+      "learning_rate": 9.461607142857144e-05,
+      "loss": 1.5536,
       "step": 603
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 8.808678500986193e-05,
-      "loss": 1.9392,
+      "epoch": 2.7,
+      "learning_rate": 9.460714285714286e-05,
+      "loss": 1.575,
       "step": 604
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 8.806706114398423e-05,
-      "loss": 1.8507,
+      "epoch": 2.7,
+      "learning_rate": 9.459821428571429e-05,
+      "loss": 1.4869,
       "step": 605
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 8.804733727810652e-05,
-      "loss": 1.9501,
+      "epoch": 2.71,
+      "learning_rate": 9.458928571428571e-05,
+      "loss": 1.5716,
       "step": 606
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 8.802761341222881e-05,
-      "loss": 1.9822,
+      "epoch": 2.71,
+      "learning_rate": 9.458035714285715e-05,
+      "loss": 1.4186,
       "step": 607
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 8.800788954635109e-05,
-      "loss": 1.8885,
+      "epoch": 2.71,
+      "learning_rate": 9.457142857142858e-05,
+      "loss": 1.4767,
       "step": 608
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 8.798816568047338e-05,
-      "loss": 1.8534,
+      "epoch": 2.72,
+      "learning_rate": 9.456250000000001e-05,
+      "loss": 1.5176,
       "step": 609
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 8.796844181459567e-05,
-      "loss": 1.8256,
+      "epoch": 2.72,
+      "learning_rate": 9.455357142857143e-05,
+      "loss": 1.4961,
       "step": 610
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 8.794871794871795e-05,
-      "loss": 1.8355,
+      "epoch": 2.73,
+      "learning_rate": 9.454464285714287e-05,
+      "loss": 1.4713,
       "step": 611
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 8.792899408284024e-05,
-      "loss": 1.9757,
+      "epoch": 2.73,
+      "learning_rate": 9.453571428571429e-05,
+      "loss": 1.5017,
       "step": 612
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 8.790927021696252e-05,
-      "loss": 1.8404,
+      "epoch": 2.74,
+      "learning_rate": 9.452678571428572e-05,
+      "loss": 1.5133,
       "step": 613
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 8.788954635108481e-05,
-      "loss": 1.8528,
+      "epoch": 2.74,
+      "learning_rate": 9.451785714285714e-05,
+      "loss": 1.573,
       "step": 614
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 8.78698224852071e-05,
-      "loss": 1.9774,
+      "epoch": 2.75,
+      "learning_rate": 9.450892857142858e-05,
+      "loss": 1.4087,
       "step": 615
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 8.785009861932939e-05,
-      "loss": 1.8588,
+      "epoch": 2.75,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 1.4045,
       "step": 616
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 8.783037475345168e-05,
-      "loss": 1.922,
+      "epoch": 2.75,
+      "learning_rate": 9.449107142857143e-05,
+      "loss": 1.5686,
       "step": 617
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 8.781065088757397e-05,
-      "loss": 1.8662,
+      "epoch": 2.76,
+      "learning_rate": 9.448214285714286e-05,
+      "loss": 1.464,
       "step": 618
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 8.779092702169626e-05,
-      "loss": 1.8963,
+      "epoch": 2.76,
+      "learning_rate": 9.44732142857143e-05,
+      "loss": 1.5039,
       "step": 619
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 8.777120315581855e-05,
-      "loss": 1.8526,
+      "epoch": 2.77,
+      "learning_rate": 9.446428571428572e-05,
+      "loss": 1.4554,
       "step": 620
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 8.775147928994083e-05,
-      "loss": 1.866,
+      "epoch": 2.77,
+      "learning_rate": 9.445535714285715e-05,
+      "loss": 1.4407,
       "step": 621
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 8.773175542406312e-05,
-      "loss": 1.8987,
+      "epoch": 2.78,
+      "learning_rate": 9.444642857142859e-05,
+      "loss": 1.427,
       "step": 622
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 8.77120315581854e-05,
-      "loss": 1.9756,
+      "epoch": 2.78,
+      "learning_rate": 9.44375e-05,
+      "loss": 1.4751,
       "step": 623
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 8.76923076923077e-05,
-      "loss": 1.8095,
+      "epoch": 2.79,
+      "learning_rate": 9.442857142857144e-05,
+      "loss": 1.5157,
       "step": 624
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 8.767258382642999e-05,
-      "loss": 1.858,
+      "epoch": 2.79,
+      "learning_rate": 9.441964285714286e-05,
+      "loss": 1.4338,
       "step": 625
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 8.765285996055227e-05,
-      "loss": 1.9295,
+      "epoch": 2.79,
+      "learning_rate": 9.441071428571429e-05,
+      "loss": 1.5716,
       "step": 626
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 8.763313609467456e-05,
-      "loss": 1.9062,
+      "epoch": 2.8,
+      "learning_rate": 9.440178571428571e-05,
+      "loss": 1.4638,
       "step": 627
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 8.761341222879684e-05,
-      "loss": 1.7825,
+      "epoch": 2.8,
+      "learning_rate": 9.439285714285715e-05,
+      "loss": 1.47,
       "step": 628
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 8.759368836291913e-05,
-      "loss": 1.8462,
+      "epoch": 2.81,
+      "learning_rate": 9.438392857142858e-05,
+      "loss": 1.5555,
       "step": 629
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 8.757396449704143e-05,
-      "loss": 1.891,
+      "epoch": 2.81,
+      "learning_rate": 9.4375e-05,
+      "loss": 1.5265,
       "step": 630
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 8.755424063116371e-05,
-      "loss": 1.8922,
+      "epoch": 2.82,
+      "learning_rate": 9.436607142857143e-05,
+      "loss": 1.4612,
       "step": 631
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 8.7534516765286e-05,
-      "loss": 1.9233,
+      "epoch": 2.82,
+      "learning_rate": 9.435714285714287e-05,
+      "loss": 1.5354,
       "step": 632
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 8.75147928994083e-05,
-      "loss": 1.8959,
+      "epoch": 2.83,
+      "learning_rate": 9.43482142857143e-05,
+      "loss": 1.4584,
       "step": 633
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 8.749506903353058e-05,
-      "loss": 1.9283,
+      "epoch": 2.83,
+      "learning_rate": 9.433928571428572e-05,
+      "loss": 1.4502,
       "step": 634
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 8.747534516765287e-05,
-      "loss": 1.9196,
+      "epoch": 2.83,
+      "learning_rate": 9.433035714285714e-05,
+      "loss": 1.4643,
       "step": 635
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 8.745562130177515e-05,
-      "loss": 1.9098,
+      "epoch": 2.84,
+      "learning_rate": 9.432142857142857e-05,
+      "loss": 1.4338,
       "step": 636
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 8.743589743589744e-05,
-      "loss": 1.9496,
+      "epoch": 2.84,
+      "learning_rate": 9.431250000000001e-05,
+      "loss": 1.3962,
       "step": 637
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 8.741617357001973e-05,
-      "loss": 1.8607,
+      "epoch": 2.85,
+      "learning_rate": 9.430357142857143e-05,
+      "loss": 1.4613,
       "step": 638
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 8.739644970414201e-05,
-      "loss": 1.8805,
+      "epoch": 2.85,
+      "learning_rate": 9.429464285714286e-05,
+      "loss": 1.4005,
       "step": 639
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 8.73767258382643e-05,
-      "loss": 1.8905,
+      "epoch": 2.86,
+      "learning_rate": 9.428571428571429e-05,
+      "loss": 1.4391,
       "step": 640
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 8.735700197238659e-05,
-      "loss": 1.898,
+      "epoch": 2.86,
+      "learning_rate": 9.427678571428572e-05,
+      "loss": 1.5348,
       "step": 641
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 8.733727810650889e-05,
-      "loss": 1.7914,
+      "epoch": 2.87,
+      "learning_rate": 9.426785714285715e-05,
+      "loss": 1.4378,
       "step": 642
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 8.731755424063118e-05,
-      "loss": 1.8333,
+      "epoch": 2.87,
+      "learning_rate": 9.425892857142858e-05,
+      "loss": 1.3862,
       "step": 643
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 8.729783037475346e-05,
-      "loss": 1.9515,
+      "epoch": 2.88,
+      "learning_rate": 9.425e-05,
+      "loss": 1.4234,
       "step": 644
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 8.727810650887575e-05,
-      "loss": 1.9269,
+      "epoch": 2.88,
+      "learning_rate": 9.424107142857144e-05,
+      "loss": 1.5162,
       "step": 645
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 8.725838264299803e-05,
-      "loss": 1.8602,
+      "epoch": 2.88,
+      "learning_rate": 9.423214285714285e-05,
+      "loss": 1.5046,
       "step": 646
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 8.723865877712032e-05,
-      "loss": 1.8957,
+      "epoch": 2.89,
+      "learning_rate": 9.422321428571429e-05,
+      "loss": 1.4723,
       "step": 647
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.721893491124261e-05,
-      "loss": 1.7962,
+      "epoch": 2.89,
+      "learning_rate": 9.421428571428572e-05,
+      "loss": 1.4755,
       "step": 648
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.719921104536489e-05,
-      "loss": 1.7774,
+      "epoch": 2.9,
+      "learning_rate": 9.420535714285714e-05,
+      "loss": 1.4674,
       "step": 649
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.717948717948718e-05,
-      "loss": 1.8723,
+      "epoch": 2.9,
+      "learning_rate": 9.419642857142858e-05,
+      "loss": 1.4318,
       "step": 650
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 8.715976331360947e-05,
-      "loss": 1.8621,
+      "epoch": 2.91,
+      "learning_rate": 9.41875e-05,
+      "loss": 1.4187,
       "step": 651
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 8.714003944773175e-05,
-      "loss": 1.7616,
+      "epoch": 2.91,
+      "learning_rate": 9.417857142857144e-05,
+      "loss": 1.476,
       "step": 652
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 8.712031558185404e-05,
-      "loss": 1.8763,
+      "epoch": 2.92,
+      "learning_rate": 9.416964285714286e-05,
+      "loss": 1.4114,
       "step": 653
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 8.710059171597634e-05,
-      "loss": 1.9573,
+      "epoch": 2.92,
+      "learning_rate": 9.41607142857143e-05,
+      "loss": 1.5035,
       "step": 654
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 8.708086785009863e-05,
-      "loss": 1.9369,
+      "epoch": 2.92,
+      "learning_rate": 9.415178571428572e-05,
+      "loss": 1.46,
       "step": 655
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 8.706114398422092e-05,
-      "loss": 1.9322,
+      "epoch": 2.93,
+      "learning_rate": 9.414285714285715e-05,
+      "loss": 1.4753,
       "step": 656
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 8.70414201183432e-05,
-      "loss": 1.8727,
+      "epoch": 2.93,
+      "learning_rate": 9.413392857142857e-05,
+      "loss": 1.5215,
       "step": 657
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 8.702169625246549e-05,
-      "loss": 1.8231,
+      "epoch": 2.94,
+      "learning_rate": 9.412500000000001e-05,
+      "loss": 1.5833,
       "step": 658
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 8.700197238658777e-05,
-      "loss": 1.8983,
+      "epoch": 2.94,
+      "learning_rate": 9.411607142857142e-05,
+      "loss": 1.5483,
       "step": 659
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 8.698224852071006e-05,
-      "loss": 1.897,
+      "epoch": 2.95,
+      "learning_rate": 9.410714285714286e-05,
+      "loss": 1.3421,
       "step": 660
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 8.696252465483235e-05,
-      "loss": 1.8838,
+      "epoch": 2.95,
+      "learning_rate": 9.409821428571429e-05,
+      "loss": 1.5287,
       "step": 661
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 8.694280078895463e-05,
-      "loss": 1.9052,
+      "epoch": 2.96,
+      "learning_rate": 9.408928571428573e-05,
+      "loss": 1.4806,
       "step": 662
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 8.692307692307692e-05,
-      "loss": 1.7743,
+      "epoch": 2.96,
+      "learning_rate": 9.408035714285714e-05,
+      "loss": 1.4152,
       "step": 663
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 8.690335305719922e-05,
-      "loss": 1.9483,
+      "epoch": 2.96,
+      "learning_rate": 9.407142857142858e-05,
+      "loss": 1.398,
       "step": 664
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.688362919132151e-05,
-      "loss": 1.8029,
+      "epoch": 2.97,
+      "learning_rate": 9.40625e-05,
+      "loss": 1.467,
       "step": 665
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.68639053254438e-05,
-      "loss": 1.8913,
+      "epoch": 2.97,
+      "learning_rate": 9.405357142857143e-05,
+      "loss": 1.4585,
       "step": 666
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.684418145956608e-05,
-      "loss": 1.9851,
+      "epoch": 2.98,
+      "learning_rate": 9.404464285714287e-05,
+      "loss": 1.49,
       "step": 667
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 8.682445759368837e-05,
-      "loss": 1.8395,
+      "epoch": 2.98,
+      "learning_rate": 9.403571428571429e-05,
+      "loss": 1.5219,
       "step": 668
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 8.680473372781065e-05,
-      "loss": 1.8264,
+      "epoch": 2.99,
+      "learning_rate": 9.402678571428572e-05,
+      "loss": 1.4339,
       "step": 669
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 8.678500986193294e-05,
-      "loss": 1.9773,
+      "epoch": 2.99,
+      "learning_rate": 9.401785714285714e-05,
+      "loss": 1.3661,
       "step": 670
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 8.676528599605523e-05,
-      "loss": 1.8092,
+      "epoch": 3.0,
+      "learning_rate": 9.400892857142858e-05,
+      "loss": 1.4952,
       "step": 671
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 8.674556213017751e-05,
-      "loss": 1.8509,
+      "epoch": 3.0,
+      "learning_rate": 9.4e-05,
+      "loss": 1.6228,
       "step": 672
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 8.67258382642998e-05,
-      "loss": 1.8169,
+      "epoch": 3.0,
+      "learning_rate": 9.399107142857143e-05,
+      "loss": 1.3825,
       "step": 673
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 8.67061143984221e-05,
-      "loss": 1.8342,
+      "epoch": 3.01,
+      "learning_rate": 9.398214285714286e-05,
+      "loss": 1.383,
       "step": 674
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 8.668639053254438e-05,
-      "loss": 1.8355,
+      "epoch": 3.01,
+      "learning_rate": 9.39732142857143e-05,
+      "loss": 1.4033,
       "step": 675
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 8.666666666666667e-05,
-      "loss": 1.7397,
+      "epoch": 3.02,
+      "learning_rate": 9.396428571428571e-05,
+      "loss": 1.4545,
       "step": 676
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 8.664694280078896e-05,
-      "loss": 1.8435,
+      "epoch": 3.02,
+      "learning_rate": 9.395535714285715e-05,
+      "loss": 1.3826,
       "step": 677
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 8.662721893491125e-05,
-      "loss": 1.8252,
+      "epoch": 3.03,
+      "learning_rate": 9.394642857142857e-05,
+      "loss": 1.401,
       "step": 678
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 8.660749506903354e-05,
-      "loss": 1.8396,
+      "epoch": 3.03,
+      "learning_rate": 9.39375e-05,
+      "loss": 1.3877,
       "step": 679
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 8.658777120315582e-05,
-      "loss": 1.8642,
+      "epoch": 3.04,
+      "learning_rate": 9.392857142857144e-05,
+      "loss": 1.436,
       "step": 680
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 8.656804733727812e-05,
-      "loss": 1.8033,
+      "epoch": 3.04,
+      "learning_rate": 9.391964285714286e-05,
+      "loss": 1.5234,
       "step": 681
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 8.65483234714004e-05,
-      "loss": 1.7613,
+      "epoch": 3.04,
+      "learning_rate": 9.391071428571429e-05,
+      "loss": 1.3916,
       "step": 682
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 8.652859960552269e-05,
-      "loss": 1.8704,
+      "epoch": 3.05,
+      "learning_rate": 9.390178571428572e-05,
+      "loss": 1.4657,
       "step": 683
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 8.650887573964498e-05,
-      "loss": 1.8097,
+      "epoch": 3.05,
+      "learning_rate": 9.389285714285715e-05,
+      "loss": 1.4582,
       "step": 684
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 8.648915187376726e-05,
-      "loss": 1.8729,
+      "epoch": 3.06,
+      "learning_rate": 9.388392857142858e-05,
+      "loss": 1.4595,
       "step": 685
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 8.646942800788955e-05,
-      "loss": 1.8226,
+      "epoch": 3.06,
+      "learning_rate": 9.3875e-05,
+      "loss": 1.5022,
       "step": 686
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 8.644970414201183e-05,
-      "loss": 1.7739,
+      "epoch": 3.07,
+      "learning_rate": 9.386607142857143e-05,
+      "loss": 1.4815,
       "step": 687
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.642998027613412e-05,
-      "loss": 1.9138,
+      "epoch": 3.07,
+      "learning_rate": 9.385714285714287e-05,
+      "loss": 1.4478,
       "step": 688
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.641025641025642e-05,
-      "loss": 1.8903,
+      "epoch": 3.08,
+      "learning_rate": 9.384821428571428e-05,
+      "loss": 1.4992,
       "step": 689
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.63905325443787e-05,
-      "loss": 1.785,
+      "epoch": 3.08,
+      "learning_rate": 9.383928571428572e-05,
+      "loss": 1.3871,
       "step": 690
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.6370808678501e-05,
-      "loss": 1.8039,
+      "epoch": 3.08,
+      "learning_rate": 9.383035714285715e-05,
+      "loss": 1.4233,
       "step": 691
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 8.635108481262327e-05,
-      "loss": 1.9043,
+      "epoch": 3.09,
+      "learning_rate": 9.382142857142857e-05,
+      "loss": 1.491,
       "step": 692
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 8.633136094674557e-05,
-      "loss": 1.7887,
+      "epoch": 3.09,
+      "learning_rate": 9.38125e-05,
+      "loss": 1.4282,
       "step": 693
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 8.631163708086786e-05,
-      "loss": 1.8792,
+      "epoch": 3.1,
+      "learning_rate": 9.380357142857144e-05,
+      "loss": 1.5228,
       "step": 694
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 8.629191321499014e-05,
-      "loss": 1.8832,
+      "epoch": 3.1,
+      "learning_rate": 9.379464285714286e-05,
+      "loss": 1.4551,
       "step": 695
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 8.627218934911243e-05,
-      "loss": 1.8354,
+      "epoch": 3.11,
+      "learning_rate": 9.378571428571429e-05,
+      "loss": 1.4665,
       "step": 696
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 8.625246548323472e-05,
-      "loss": 1.9712,
+      "epoch": 3.11,
+      "learning_rate": 9.377678571428573e-05,
+      "loss": 1.4547,
       "step": 697
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.6232741617357e-05,
-      "loss": 1.8695,
+      "epoch": 3.12,
+      "learning_rate": 9.376785714285715e-05,
+      "loss": 1.4398,
       "step": 698
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.621301775147929e-05,
-      "loss": 1.7741,
+      "epoch": 3.12,
+      "learning_rate": 9.375892857142858e-05,
+      "loss": 1.4247,
       "step": 699
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.619329388560158e-05,
-      "loss": 1.8993,
+      "epoch": 3.12,
+      "learning_rate": 9.375e-05,
+      "loss": 1.4869,
       "step": 700
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.617357001972388e-05,
-      "loss": 1.8352,
+      "epoch": 3.13,
+      "learning_rate": 9.374107142857144e-05,
+      "loss": 1.4764,
       "step": 701
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 8.615384615384617e-05,
-      "loss": 1.942,
+      "epoch": 3.13,
+      "learning_rate": 9.373214285714285e-05,
+      "loss": 1.3529,
       "step": 702
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 8.613412228796845e-05,
-      "loss": 1.9322,
+      "epoch": 3.14,
+      "learning_rate": 9.372321428571429e-05,
+      "loss": 1.4833,
       "step": 703
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 8.611439842209074e-05,
-      "loss": 1.8666,
+      "epoch": 3.14,
+      "learning_rate": 9.371428571428572e-05,
+      "loss": 1.3927,
       "step": 704
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 8.609467455621302e-05,
-      "loss": 1.8041,
+      "epoch": 3.15,
+      "learning_rate": 9.370535714285714e-05,
+      "loss": 1.484,
       "step": 705
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 8.607495069033531e-05,
-      "loss": 1.8838,
+      "epoch": 3.15,
+      "learning_rate": 9.369642857142857e-05,
+      "loss": 1.3368,
       "step": 706
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 8.60552268244576e-05,
-      "loss": 1.8415,
+      "epoch": 3.16,
+      "learning_rate": 9.368750000000001e-05,
+      "loss": 1.4514,
       "step": 707
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 8.603550295857988e-05,
-      "loss": 1.8423,
+      "epoch": 3.16,
+      "learning_rate": 9.367857142857143e-05,
+      "loss": 1.435,
       "step": 708
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 8.601577909270217e-05,
-      "loss": 1.8116,
+      "epoch": 3.17,
+      "learning_rate": 9.366964285714286e-05,
+      "loss": 1.4127,
       "step": 709
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 8.599605522682445e-05,
-      "loss": 1.8176,
+      "epoch": 3.17,
+      "learning_rate": 9.36607142857143e-05,
+      "loss": 1.4474,
       "step": 710
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 8.597633136094674e-05,
-      "loss": 1.8452,
+      "epoch": 3.17,
+      "learning_rate": 9.365178571428572e-05,
+      "loss": 1.389,
       "step": 711
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 8.595660749506904e-05,
-      "loss": 1.7979,
+      "epoch": 3.18,
+      "learning_rate": 9.364285714285715e-05,
+      "loss": 1.4505,
       "step": 712
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 8.593688362919133e-05,
-      "loss": 1.8683,
+      "epoch": 3.18,
+      "learning_rate": 9.363392857142857e-05,
+      "loss": 1.4442,
       "step": 713
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 8.591715976331362e-05,
-      "loss": 1.9563,
+      "epoch": 3.19,
+      "learning_rate": 9.362500000000001e-05,
+      "loss": 1.5195,
       "step": 714
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 8.58974358974359e-05,
-      "loss": 1.7894,
+      "epoch": 3.19,
+      "learning_rate": 9.361607142857142e-05,
+      "loss": 1.4078,
       "step": 715
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 8.587771203155819e-05,
-      "loss": 1.8279,
+      "epoch": 3.2,
+      "learning_rate": 9.360714285714286e-05,
+      "loss": 1.5011,
       "step": 716
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 8.585798816568048e-05,
-      "loss": 1.9507,
+      "epoch": 3.2,
+      "learning_rate": 9.359821428571429e-05,
+      "loss": 1.3988,
       "step": 717
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 8.583826429980276e-05,
-      "loss": 1.8223,
+      "epoch": 3.21,
+      "learning_rate": 9.358928571428573e-05,
+      "loss": 1.3509,
       "step": 718
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 8.581854043392505e-05,
-      "loss": 1.8711,
+      "epoch": 3.21,
+      "learning_rate": 9.358035714285714e-05,
+      "loss": 1.4718,
       "step": 719
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 8.579881656804734e-05,
-      "loss": 1.8511,
+      "epoch": 3.21,
+      "learning_rate": 9.357142857142858e-05,
+      "loss": 1.4196,
       "step": 720
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 8.577909270216962e-05,
-      "loss": 1.8264,
+      "epoch": 3.22,
+      "learning_rate": 9.35625e-05,
+      "loss": 1.4559,
       "step": 721
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 8.575936883629192e-05,
-      "loss": 1.784,
+      "epoch": 3.22,
+      "learning_rate": 9.355357142857143e-05,
+      "loss": 1.4958,
       "step": 722
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 8.573964497041421e-05,
-      "loss": 1.964,
+      "epoch": 3.23,
+      "learning_rate": 9.354464285714286e-05,
+      "loss": 1.4946,
       "step": 723
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 8.57199211045365e-05,
-      "loss": 1.8444,
+      "epoch": 3.23,
+      "learning_rate": 9.35357142857143e-05,
+      "loss": 1.4928,
       "step": 724
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 8.570019723865879e-05,
-      "loss": 1.8286,
+      "epoch": 3.24,
+      "learning_rate": 9.352678571428572e-05,
+      "loss": 1.4233,
       "step": 725
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 8.568047337278107e-05,
-      "loss": 1.7989,
+      "epoch": 3.24,
+      "learning_rate": 9.351785714285714e-05,
+      "loss": 1.437,
       "step": 726
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 8.566074950690336e-05,
-      "loss": 2.0226,
+      "epoch": 3.25,
+      "learning_rate": 9.350892857142858e-05,
+      "loss": 1.4298,
       "step": 727
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 8.564102564102564e-05,
-      "loss": 1.7123,
+      "epoch": 3.25,
+      "learning_rate": 9.350000000000001e-05,
+      "loss": 1.4048,
       "step": 728
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 8.562130177514793e-05,
-      "loss": 1.8194,
+      "epoch": 3.25,
+      "learning_rate": 9.349107142857143e-05,
+      "loss": 1.4781,
       "step": 729
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 8.560157790927023e-05,
-      "loss": 1.8714,
+      "epoch": 3.26,
+      "learning_rate": 9.348214285714286e-05,
+      "loss": 1.4045,
       "step": 730
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 8.55818540433925e-05,
-      "loss": 1.7932,
+      "epoch": 3.26,
+      "learning_rate": 9.34732142857143e-05,
+      "loss": 1.4718,
       "step": 731
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 8.55621301775148e-05,
-      "loss": 1.7563,
+      "epoch": 3.27,
+      "learning_rate": 9.346428571428571e-05,
+      "loss": 1.4309,
       "step": 732
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 8.554240631163707e-05,
-      "loss": 1.8045,
+      "epoch": 3.27,
+      "learning_rate": 9.345535714285715e-05,
+      "loss": 1.4129,
       "step": 733
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 8.552268244575937e-05,
-      "loss": 1.7721,
+      "epoch": 3.28,
+      "learning_rate": 9.344642857142858e-05,
+      "loss": 1.4198,
       "step": 734
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 8.550295857988166e-05,
-      "loss": 1.7691,
+      "epoch": 3.28,
+      "learning_rate": 9.34375e-05,
+      "loss": 1.3969,
       "step": 735
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 8.548323471400395e-05,
-      "loss": 1.8112,
+      "epoch": 3.29,
+      "learning_rate": 9.342857142857143e-05,
+      "loss": 1.4334,
       "step": 736
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 8.546351084812624e-05,
-      "loss": 1.8721,
+      "epoch": 3.29,
+      "learning_rate": 9.341964285714287e-05,
+      "loss": 1.4187,
       "step": 737
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 8.544378698224852e-05,
-      "loss": 1.8397,
+      "epoch": 3.29,
+      "learning_rate": 9.341071428571429e-05,
+      "loss": 1.4439,
       "step": 738
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 8.542406311637081e-05,
-      "loss": 1.7743,
+      "epoch": 3.3,
+      "learning_rate": 9.340178571428572e-05,
+      "loss": 1.5223,
       "step": 739
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 8.54043392504931e-05,
-      "loss": 1.8321,
+      "epoch": 3.3,
+      "learning_rate": 9.339285714285714e-05,
+      "loss": 1.4555,
       "step": 740
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 8.538461538461538e-05,
-      "loss": 1.7407,
+      "epoch": 3.31,
+      "learning_rate": 9.338392857142858e-05,
+      "loss": 1.4963,
       "step": 741
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 8.536489151873768e-05,
-      "loss": 1.8339,
+      "epoch": 3.31,
+      "learning_rate": 9.3375e-05,
+      "loss": 1.398,
       "step": 742
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 8.534516765285997e-05,
-      "loss": 1.9314,
+      "epoch": 3.32,
+      "learning_rate": 9.336607142857143e-05,
+      "loss": 1.3628,
       "step": 743
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 8.532544378698225e-05,
-      "loss": 1.8765,
+      "epoch": 3.32,
+      "learning_rate": 9.335714285714287e-05,
+      "loss": 1.3931,
       "step": 744
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 8.530571992110454e-05,
-      "loss": 1.8289,
+      "epoch": 3.33,
+      "learning_rate": 9.334821428571428e-05,
+      "loss": 1.4882,
       "step": 745
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 8.528599605522682e-05,
-      "loss": 1.8368,
+      "epoch": 3.33,
+      "learning_rate": 9.333928571428572e-05,
+      "loss": 1.422,
       "step": 746
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 8.526627218934911e-05,
-      "loss": 1.8324,
+      "epoch": 3.33,
+      "learning_rate": 9.333035714285715e-05,
+      "loss": 1.3909,
       "step": 747
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 8.524654832347142e-05,
-      "loss": 1.8121,
+      "epoch": 3.34,
+      "learning_rate": 9.332142857142857e-05,
+      "loss": 1.3911,
       "step": 748
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 8.52268244575937e-05,
-      "loss": 1.8064,
+      "epoch": 3.34,
+      "learning_rate": 9.33125e-05,
+      "loss": 1.4314,
       "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,
+      "epoch": 3.35,
+      "learning_rate": 9.330357142857144e-05,
+      "loss": 1.5127,
       "step": 750
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 8.518737672583826e-05,
-      "loss": 1.8673,
+      "epoch": 3.35,
+      "learning_rate": 9.329464285714286e-05,
+      "loss": 1.4411,
       "step": 751
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 8.516765285996056e-05,
-      "loss": 1.8409,
+      "epoch": 3.36,
+      "learning_rate": 9.328571428571429e-05,
+      "loss": 1.3833,
       "step": 752
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 8.514792899408285e-05,
-      "loss": 1.7949,
+      "epoch": 3.36,
+      "learning_rate": 9.327678571428571e-05,
+      "loss": 1.4195,
       "step": 753
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 8.512820512820513e-05,
-      "loss": 1.8676,
+      "epoch": 3.37,
+      "learning_rate": 9.326785714285715e-05,
+      "loss": 1.4438,
       "step": 754
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 8.510848126232742e-05,
-      "loss": 1.7834,
+      "epoch": 3.37,
+      "learning_rate": 9.325892857142858e-05,
+      "loss": 1.4746,
       "step": 755
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 8.50887573964497e-05,
-      "loss": 1.768,
+      "epoch": 3.38,
+      "learning_rate": 9.325e-05,
+      "loss": 1.4477,
       "step": 756
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 8.506903353057199e-05,
-      "loss": 1.8354,
+      "epoch": 3.38,
+      "learning_rate": 9.324107142857144e-05,
+      "loss": 1.4291,
       "step": 757
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 8.504930966469428e-05,
-      "loss": 1.8592,
+      "epoch": 3.38,
+      "learning_rate": 9.323214285714285e-05,
+      "loss": 1.4435,
       "step": 758
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 8.502958579881657e-05,
-      "loss": 1.8032,
+      "epoch": 3.39,
+      "learning_rate": 9.32232142857143e-05,
+      "loss": 1.3607,
       "step": 759
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 8.500986193293887e-05,
-      "loss": 1.8346,
+      "epoch": 3.39,
+      "learning_rate": 9.321428571428572e-05,
+      "loss": 1.4672,
       "step": 760
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 8.499013806706116e-05,
-      "loss": 1.8764,
+      "epoch": 3.4,
+      "learning_rate": 9.320535714285716e-05,
+      "loss": 1.4008,
       "step": 761
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 8.497041420118344e-05,
-      "loss": 1.8203,
+      "epoch": 3.4,
+      "learning_rate": 9.319642857142857e-05,
+      "loss": 1.4886,
       "step": 762
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 8.495069033530573e-05,
-      "loss": 1.8368,
+      "epoch": 3.41,
+      "learning_rate": 9.318750000000001e-05,
+      "loss": 1.4119,
       "step": 763
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 8.493096646942801e-05,
-      "loss": 1.839,
+      "epoch": 3.41,
+      "learning_rate": 9.317857142857143e-05,
+      "loss": 1.4415,
       "step": 764
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 8.49112426035503e-05,
-      "loss": 1.9035,
+      "epoch": 3.42,
+      "learning_rate": 9.316964285714286e-05,
+      "loss": 1.5172,
       "step": 765
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 8.489151873767259e-05,
-      "loss": 1.8462,
+      "epoch": 3.42,
+      "learning_rate": 9.316071428571428e-05,
+      "loss": 1.4229,
       "step": 766
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 8.487179487179487e-05,
-      "loss": 1.8428,
+      "epoch": 3.42,
+      "learning_rate": 9.315178571428572e-05,
+      "loss": 1.4302,
       "step": 767
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 8.485207100591716e-05,
-      "loss": 1.8107,
+      "epoch": 3.43,
+      "learning_rate": 9.314285714285715e-05,
+      "loss": 1.4407,
       "step": 768
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 8.483234714003944e-05,
-      "loss": 1.696,
+      "epoch": 3.43,
+      "learning_rate": 9.313392857142857e-05,
+      "loss": 1.3756,
       "step": 769
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 8.481262327416173e-05,
-      "loss": 1.7684,
+      "epoch": 3.44,
+      "learning_rate": 9.3125e-05,
+      "loss": 1.3862,
       "step": 770
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 8.479289940828403e-05,
-      "loss": 1.953,
+      "epoch": 3.44,
+      "learning_rate": 9.311607142857143e-05,
+      "loss": 1.5136,
       "step": 771
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 8.477317554240632e-05,
-      "loss": 1.7695,
+      "epoch": 3.45,
+      "learning_rate": 9.310714285714286e-05,
+      "loss": 1.4013,
       "step": 772
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 8.475345167652861e-05,
-      "loss": 1.8572,
+      "epoch": 3.45,
+      "learning_rate": 9.309821428571429e-05,
+      "loss": 1.3958,
       "step": 773
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 8.473372781065089e-05,
-      "loss": 1.8382,
+      "epoch": 3.46,
+      "learning_rate": 9.308928571428573e-05,
+      "loss": 1.4501,
       "step": 774
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 8.471400394477318e-05,
-      "loss": 1.8085,
+      "epoch": 3.46,
+      "learning_rate": 9.308035714285714e-05,
+      "loss": 1.4412,
       "step": 775
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 8.469428007889547e-05,
-      "loss": 1.7531,
+      "epoch": 3.46,
+      "learning_rate": 9.307142857142858e-05,
+      "loss": 1.4717,
       "step": 776
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 8.467455621301775e-05,
-      "loss": 1.7632,
+      "epoch": 3.47,
+      "learning_rate": 9.30625e-05,
+      "loss": 1.4375,
       "step": 777
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 8.465483234714004e-05,
-      "loss": 1.8419,
+      "epoch": 3.47,
+      "learning_rate": 9.305357142857143e-05,
+      "loss": 1.4371,
       "step": 778
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 8.463510848126232e-05,
-      "loss": 1.814,
+      "epoch": 3.48,
+      "learning_rate": 9.304464285714286e-05,
+      "loss": 1.389,
       "step": 779
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 8.461538461538461e-05,
-      "loss": 1.7156,
+      "epoch": 3.48,
+      "learning_rate": 9.30357142857143e-05,
+      "loss": 1.4174,
       "step": 780
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 8.45956607495069e-05,
-      "loss": 1.7796,
+      "epoch": 3.49,
+      "learning_rate": 9.302678571428571e-05,
+      "loss": 1.4134,
       "step": 781
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 8.45759368836292e-05,
-      "loss": 1.7029,
+      "epoch": 3.49,
+      "learning_rate": 9.301785714285715e-05,
+      "loss": 1.5032,
       "step": 782
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 8.455621301775149e-05,
-      "loss": 1.8749,
+      "epoch": 3.5,
+      "learning_rate": 9.300892857142857e-05,
+      "loss": 1.5088,
       "step": 783
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 8.453648915187378e-05,
-      "loss": 1.8533,
+      "epoch": 3.5,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 1.4511,
       "step": 784
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 8.451676528599606e-05,
-      "loss": 1.811,
+      "epoch": 3.5,
+      "learning_rate": 9.299107142857144e-05,
+      "loss": 1.396,
       "step": 785
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 8.449704142011835e-05,
-      "loss": 1.8801,
+      "epoch": 3.51,
+      "learning_rate": 9.298214285714286e-05,
+      "loss": 1.3508,
       "step": 786
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 8.447731755424063e-05,
-      "loss": 1.8771,
+      "epoch": 3.51,
+      "learning_rate": 9.29732142857143e-05,
+      "loss": 1.4004,
       "step": 787
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 8.445759368836292e-05,
-      "loss": 1.8036,
+      "epoch": 3.52,
+      "learning_rate": 9.296428571428571e-05,
+      "loss": 1.4199,
       "step": 788
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 8.443786982248522e-05,
-      "loss": 1.7258,
+      "epoch": 3.52,
+      "learning_rate": 9.295535714285715e-05,
+      "loss": 1.4719,
       "step": 789
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 8.44181459566075e-05,
-      "loss": 1.869,
+      "epoch": 3.53,
+      "learning_rate": 9.294642857142858e-05,
+      "loss": 1.4731,
       "step": 790
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 8.439842209072979e-05,
-      "loss": 1.8855,
+      "epoch": 3.53,
+      "learning_rate": 9.29375e-05,
+      "loss": 1.4612,
       "step": 791
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 8.437869822485207e-05,
-      "loss": 1.7894,
+      "epoch": 3.54,
+      "learning_rate": 9.292857142857143e-05,
+      "loss": 1.4505,
       "step": 792
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 8.435897435897436e-05,
-      "loss": 1.7423,
+      "epoch": 3.54,
+      "learning_rate": 9.291964285714287e-05,
+      "loss": 1.3417,
       "step": 793
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 8.433925049309665e-05,
-      "loss": 1.7918,
+      "epoch": 3.54,
+      "learning_rate": 9.291071428571429e-05,
+      "loss": 1.5073,
       "step": 794
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 8.431952662721894e-05,
-      "loss": 1.8198,
+      "epoch": 3.55,
+      "learning_rate": 9.290178571428572e-05,
+      "loss": 1.445,
       "step": 795
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 8.429980276134123e-05,
-      "loss": 1.8674,
+      "epoch": 3.55,
+      "learning_rate": 9.289285714285714e-05,
+      "loss": 1.3951,
       "step": 796
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 8.428007889546351e-05,
-      "loss": 1.847,
+      "epoch": 3.56,
+      "learning_rate": 9.288392857142858e-05,
+      "loss": 1.3862,
       "step": 797
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 8.42603550295858e-05,
-      "loss": 1.8015,
+      "epoch": 3.56,
+      "learning_rate": 9.2875e-05,
+      "loss": 1.4106,
       "step": 798
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 8.42406311637081e-05,
-      "loss": 1.7316,
+      "epoch": 3.57,
+      "learning_rate": 9.286607142857143e-05,
+      "loss": 1.399,
       "step": 799
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 8.422090729783038e-05,
-      "loss": 1.8473,
+      "epoch": 3.57,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 1.4018,
       "step": 800
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 8.420118343195267e-05,
-      "loss": 1.8447,
+      "epoch": 3.58,
+      "learning_rate": 9.284821428571428e-05,
+      "loss": 1.4007,
       "step": 801
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 8.418145956607495e-05,
-      "loss": 1.7289,
+      "epoch": 3.58,
+      "learning_rate": 9.283928571428572e-05,
+      "loss": 1.4354,
       "step": 802
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 8.416173570019724e-05,
-      "loss": 1.8003,
+      "epoch": 3.58,
+      "learning_rate": 9.283035714285715e-05,
+      "loss": 1.4738,
       "step": 803
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 8.414201183431953e-05,
-      "loss": 1.8253,
+      "epoch": 3.59,
+      "learning_rate": 9.282142857142859e-05,
+      "loss": 1.4533,
       "step": 804
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 8.412228796844181e-05,
-      "loss": 1.7552,
+      "epoch": 3.59,
+      "learning_rate": 9.28125e-05,
+      "loss": 1.4904,
       "step": 805
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 8.410256410256411e-05,
-      "loss": 1.758,
+      "epoch": 3.6,
+      "learning_rate": 9.280357142857144e-05,
+      "loss": 1.3712,
       "step": 806
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 8.40828402366864e-05,
-      "loss": 1.8004,
+      "epoch": 3.6,
+      "learning_rate": 9.279464285714286e-05,
+      "loss": 1.3532,
       "step": 807
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 8.406311637080868e-05,
-      "loss": 1.7992,
+      "epoch": 3.61,
+      "learning_rate": 9.278571428571429e-05,
+      "loss": 1.4519,
       "step": 808
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 8.404339250493098e-05,
-      "loss": 1.7539,
+      "epoch": 3.61,
+      "learning_rate": 9.277678571428571e-05,
+      "loss": 1.3473,
       "step": 809
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 8.402366863905326e-05,
-      "loss": 1.8493,
+      "epoch": 3.62,
+      "learning_rate": 9.276785714285715e-05,
+      "loss": 1.4115,
       "step": 810
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 8.400394477317555e-05,
-      "loss": 1.7948,
+      "epoch": 3.62,
+      "learning_rate": 9.275892857142857e-05,
+      "loss": 1.3918,
       "step": 811
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 8.398422090729784e-05,
-      "loss": 1.7966,
+      "epoch": 3.62,
+      "learning_rate": 9.275e-05,
+      "loss": 1.4108,
       "step": 812
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 8.396449704142012e-05,
-      "loss": 1.9582,
+      "epoch": 3.63,
+      "learning_rate": 9.274107142857143e-05,
+      "loss": 1.3929,
       "step": 813
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 8.394477317554241e-05,
-      "loss": 1.7824,
+      "epoch": 3.63,
+      "learning_rate": 9.273214285714286e-05,
+      "loss": 1.3592,
       "step": 814
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 8.392504930966469e-05,
-      "loss": 1.8984,
+      "epoch": 3.64,
+      "learning_rate": 9.27232142857143e-05,
+      "loss": 1.474,
       "step": 815
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 8.390532544378698e-05,
-      "loss": 1.7662,
+      "epoch": 3.64,
+      "learning_rate": 9.271428571428572e-05,
+      "loss": 1.3927,
       "step": 816
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 8.388560157790927e-05,
-      "loss": 1.8165,
+      "epoch": 3.65,
+      "learning_rate": 9.270535714285715e-05,
+      "loss": 1.4229,
       "step": 817
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 8.386587771203157e-05,
-      "loss": 1.8998,
+      "epoch": 3.65,
+      "learning_rate": 9.269642857142857e-05,
+      "loss": 1.3873,
       "step": 818
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 8.384615384615386e-05,
-      "loss": 1.8401,
+      "epoch": 3.66,
+      "learning_rate": 9.268750000000001e-05,
+      "loss": 1.444,
       "step": 819
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 8.382642998027614e-05,
-      "loss": 1.6952,
+      "epoch": 3.66,
+      "learning_rate": 9.267857142857144e-05,
+      "loss": 1.4468,
       "step": 820
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 8.380670611439843e-05,
-      "loss": 1.7531,
+      "epoch": 3.67,
+      "learning_rate": 9.266964285714286e-05,
+      "loss": 1.4555,
       "step": 821
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 8.378698224852072e-05,
-      "loss": 1.7884,
+      "epoch": 3.67,
+      "learning_rate": 9.266071428571429e-05,
+      "loss": 1.4762,
       "step": 822
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 8.3767258382643e-05,
-      "loss": 1.7808,
+      "epoch": 3.67,
+      "learning_rate": 9.265178571428573e-05,
+      "loss": 1.4443,
       "step": 823
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 8.374753451676529e-05,
-      "loss": 1.7647,
+      "epoch": 3.68,
+      "learning_rate": 9.264285714285714e-05,
+      "loss": 1.4049,
       "step": 824
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 8.372781065088757e-05,
-      "loss": 1.824,
+      "epoch": 3.68,
+      "learning_rate": 9.263392857142858e-05,
+      "loss": 1.4335,
       "step": 825
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 8.370808678500986e-05,
-      "loss": 1.7022,
+      "epoch": 3.69,
+      "learning_rate": 9.2625e-05,
+      "loss": 1.5376,
       "step": 826
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 8.368836291913215e-05,
-      "loss": 1.8566,
+      "epoch": 3.69,
+      "learning_rate": 9.261607142857144e-05,
+      "loss": 1.4135,
       "step": 827
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 8.366863905325443e-05,
-      "loss": 1.8913,
+      "epoch": 3.7,
+      "learning_rate": 9.260714285714285e-05,
+      "loss": 1.3573,
       "step": 828
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 8.364891518737672e-05,
-      "loss": 1.8368,
+      "epoch": 3.7,
+      "learning_rate": 9.259821428571429e-05,
+      "loss": 1.4228,
       "step": 829
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 8.362919132149902e-05,
-      "loss": 1.882,
+      "epoch": 3.71,
+      "learning_rate": 9.258928571428572e-05,
+      "loss": 1.394,
       "step": 830
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 8.360946745562131e-05,
-      "loss": 1.7246,
+      "epoch": 3.71,
+      "learning_rate": 9.258035714285714e-05,
+      "loss": 1.4224,
       "step": 831
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 8.35897435897436e-05,
-      "loss": 1.7837,
+      "epoch": 3.71,
+      "learning_rate": 9.257142857142858e-05,
+      "loss": 1.462,
       "step": 832
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 8.357001972386588e-05,
-      "loss": 1.8231,
+      "epoch": 3.72,
+      "learning_rate": 9.256250000000001e-05,
+      "loss": 1.3922,
       "step": 833
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.355029585798817e-05,
-      "loss": 1.7347,
+      "epoch": 3.72,
+      "learning_rate": 9.255357142857143e-05,
+      "loss": 1.507,
       "step": 834
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.353057199211046e-05,
-      "loss": 1.8323,
+      "epoch": 3.73,
+      "learning_rate": 9.254464285714286e-05,
+      "loss": 1.4143,
       "step": 835
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.351084812623274e-05,
-      "loss": 1.9104,
+      "epoch": 3.73,
+      "learning_rate": 9.25357142857143e-05,
+      "loss": 1.4669,
       "step": 836
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.349112426035503e-05,
-      "loss": 1.7589,
+      "epoch": 3.74,
+      "learning_rate": 9.252678571428572e-05,
+      "loss": 1.4004,
       "step": 837
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.347140039447731e-05,
-      "loss": 1.8092,
+      "epoch": 3.74,
+      "learning_rate": 9.251785714285715e-05,
+      "loss": 1.5495,
       "step": 838
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.34516765285996e-05,
-      "loss": 1.8409,
+      "epoch": 3.75,
+      "learning_rate": 9.250892857142857e-05,
+      "loss": 1.3979,
       "step": 839
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.34319526627219e-05,
-      "loss": 1.7901,
+      "epoch": 3.75,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 1.4014,
       "step": 840
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.341222879684419e-05,
-      "loss": 1.8029,
+      "epoch": 3.75,
+      "learning_rate": 9.249107142857142e-05,
+      "loss": 1.4291,
       "step": 841
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.339250493096648e-05,
-      "loss": 1.808,
+      "epoch": 3.76,
+      "learning_rate": 9.248214285714286e-05,
+      "loss": 1.4686,
       "step": 842
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.337278106508876e-05,
-      "loss": 1.8475,
+      "epoch": 3.76,
+      "learning_rate": 9.247321428571429e-05,
+      "loss": 1.4463,
       "step": 843
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.335305719921105e-05,
-      "loss": 1.8472,
+      "epoch": 3.77,
+      "learning_rate": 9.246428571428571e-05,
+      "loss": 1.449,
       "step": 844
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.333333333333334e-05,
-      "loss": 1.8616,
+      "epoch": 3.77,
+      "learning_rate": 9.245535714285715e-05,
+      "loss": 1.4609,
       "step": 845
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.331360946745562e-05,
-      "loss": 1.8659,
+      "epoch": 3.78,
+      "learning_rate": 9.244642857142858e-05,
+      "loss": 1.3803,
       "step": 846
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.329388560157791e-05,
-      "loss": 1.8338,
+      "epoch": 3.78,
+      "learning_rate": 9.24375e-05,
+      "loss": 1.4043,
       "step": 847
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.32741617357002e-05,
-      "loss": 1.7658,
+      "epoch": 3.79,
+      "learning_rate": 9.242857142857143e-05,
+      "loss": 1.4657,
       "step": 848
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.325443786982249e-05,
-      "loss": 1.813,
+      "epoch": 3.79,
+      "learning_rate": 9.241964285714287e-05,
+      "loss": 1.4049,
       "step": 849
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.323471400394478e-05,
-      "loss": 1.9043,
+      "epoch": 3.79,
+      "learning_rate": 9.24107142857143e-05,
+      "loss": 1.4552,
       "step": 850
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.321499013806706e-05,
-      "loss": 1.8518,
+      "epoch": 3.8,
+      "learning_rate": 9.240178571428572e-05,
+      "loss": 1.3818,
       "step": 851
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.319526627218935e-05,
-      "loss": 1.7663,
+      "epoch": 3.8,
+      "learning_rate": 9.239285714285714e-05,
+      "loss": 1.3051,
       "step": 852
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.317554240631164e-05,
-      "loss": 1.864,
+      "epoch": 3.81,
+      "learning_rate": 9.238392857142858e-05,
+      "loss": 1.4034,
       "step": 853
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 8.315581854043393e-05,
-      "loss": 1.6727,
+      "epoch": 3.81,
+      "learning_rate": 9.2375e-05,
+      "loss": 1.4671,
       "step": 854
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 8.313609467455622e-05,
-      "loss": 1.8551,
+      "epoch": 3.82,
+      "learning_rate": 9.236607142857143e-05,
+      "loss": 1.4329,
       "step": 855
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 8.31163708086785e-05,
-      "loss": 1.8731,
+      "epoch": 3.82,
+      "learning_rate": 9.235714285714286e-05,
+      "loss": 1.4073,
       "step": 856
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 8.30966469428008e-05,
-      "loss": 1.8687,
+      "epoch": 3.83,
+      "learning_rate": 9.234821428571429e-05,
+      "loss": 1.3775,
       "step": 857
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 8.307692307692309e-05,
-      "loss": 1.926,
+      "epoch": 3.83,
+      "learning_rate": 9.233928571428571e-05,
+      "loss": 1.414,
       "step": 858
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 8.305719921104537e-05,
-      "loss": 1.8884,
+      "epoch": 3.83,
+      "learning_rate": 9.233035714285715e-05,
+      "loss": 1.3507,
       "step": 859
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 8.303747534516766e-05,
-      "loss": 1.8735,
+      "epoch": 3.84,
+      "learning_rate": 9.232142857142858e-05,
+      "loss": 1.4134,
       "step": 860
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 8.301775147928994e-05,
-      "loss": 1.7344,
+      "epoch": 3.84,
+      "learning_rate": 9.23125e-05,
+      "loss": 1.4524,
       "step": 861
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 8.299802761341223e-05,
-      "loss": 1.7983,
+      "epoch": 3.85,
+      "learning_rate": 9.230357142857144e-05,
+      "loss": 1.4689,
       "step": 862
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 8.297830374753452e-05,
-      "loss": 1.8672,
+      "epoch": 3.85,
+      "learning_rate": 9.229464285714287e-05,
+      "loss": 1.3628,
       "step": 863
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 8.29585798816568e-05,
-      "loss": 1.8451,
+      "epoch": 3.86,
+      "learning_rate": 9.228571428571429e-05,
+      "loss": 1.402,
       "step": 864
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 8.29388560157791e-05,
-      "loss": 1.7965,
+      "epoch": 3.86,
+      "learning_rate": 9.227678571428572e-05,
+      "loss": 1.3697,
       "step": 865
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 8.291913214990138e-05,
-      "loss": 1.8435,
+      "epoch": 3.87,
+      "learning_rate": 9.226785714285716e-05,
+      "loss": 1.3718,
       "step": 866
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 8.289940828402368e-05,
-      "loss": 1.8487,
+      "epoch": 3.87,
+      "learning_rate": 9.225892857142857e-05,
+      "loss": 1.4102,
       "step": 867
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 8.287968441814597e-05,
-      "loss": 1.7227,
+      "epoch": 3.88,
+      "learning_rate": 9.225e-05,
+      "loss": 1.4024,
       "step": 868
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 8.285996055226825e-05,
-      "loss": 1.8392,
+      "epoch": 3.88,
+      "learning_rate": 9.224107142857143e-05,
+      "loss": 1.3889,
       "step": 869
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 8.284023668639054e-05,
-      "loss": 1.8813,
+      "epoch": 3.88,
+      "learning_rate": 9.223214285714287e-05,
+      "loss": 1.4559,
       "step": 870
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 8.282051282051283e-05,
-      "loss": 1.7726,
+      "epoch": 3.89,
+      "learning_rate": 9.222321428571428e-05,
+      "loss": 1.5164,
       "step": 871
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 8.280078895463511e-05,
-      "loss": 1.7962,
+      "epoch": 3.89,
+      "learning_rate": 9.221428571428572e-05,
+      "loss": 1.2942,
       "step": 872
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 8.27810650887574e-05,
-      "loss": 1.8174,
+      "epoch": 3.9,
+      "learning_rate": 9.220535714285715e-05,
+      "loss": 1.4801,
       "step": 873
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 8.276134122287968e-05,
-      "loss": 1.8915,
+      "epoch": 3.9,
+      "learning_rate": 9.219642857142857e-05,
+      "loss": 1.4663,
       "step": 874
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 8.274161735700197e-05,
-      "loss": 1.8268,
+      "epoch": 3.91,
+      "learning_rate": 9.21875e-05,
+      "loss": 1.4437,
       "step": 875
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 8.272189349112426e-05,
-      "loss": 1.7125,
+      "epoch": 3.91,
+      "learning_rate": 9.217857142857144e-05,
+      "loss": 1.4118,
       "step": 876
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 8.270216962524656e-05,
-      "loss": 1.8697,
+      "epoch": 3.92,
+      "learning_rate": 9.216964285714286e-05,
+      "loss": 1.461,
       "step": 877
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 8.268244575936885e-05,
-      "loss": 1.8499,
+      "epoch": 3.92,
+      "learning_rate": 9.216071428571429e-05,
+      "loss": 1.4386,
       "step": 878
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 8.266272189349113e-05,
-      "loss": 1.8127,
+      "epoch": 3.92,
+      "learning_rate": 9.215178571428573e-05,
+      "loss": 1.3951,
       "step": 879
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 8.264299802761342e-05,
-      "loss": 1.7703,
+      "epoch": 3.93,
+      "learning_rate": 9.214285714285714e-05,
+      "loss": 1.4186,
       "step": 880
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 8.262327416173571e-05,
-      "loss": 1.844,
+      "epoch": 3.93,
+      "learning_rate": 9.213392857142858e-05,
+      "loss": 1.343,
       "step": 881
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 8.260355029585799e-05,
-      "loss": 1.8973,
+      "epoch": 3.94,
+      "learning_rate": 9.2125e-05,
+      "loss": 1.3825,
       "step": 882
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 8.258382642998028e-05,
-      "loss": 1.8305,
+      "epoch": 3.94,
+      "learning_rate": 9.211607142857144e-05,
+      "loss": 1.4301,
       "step": 883
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 8.256410256410256e-05,
-      "loss": 1.7808,
+      "epoch": 3.95,
+      "learning_rate": 9.210714285714285e-05,
+      "loss": 1.3665,
       "step": 884
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 8.254437869822485e-05,
-      "loss": 1.7664,
+      "epoch": 3.95,
+      "learning_rate": 9.209821428571429e-05,
+      "loss": 1.4727,
       "step": 885
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 8.252465483234714e-05,
-      "loss": 1.7969,
+      "epoch": 3.96,
+      "learning_rate": 9.208928571428572e-05,
+      "loss": 1.4215,
       "step": 886
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 8.250493096646942e-05,
-      "loss": 1.8044,
+      "epoch": 3.96,
+      "learning_rate": 9.208035714285714e-05,
+      "loss": 1.453,
       "step": 887
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 8.248520710059172e-05,
-      "loss": 1.8633,
+      "epoch": 3.96,
+      "learning_rate": 9.207142857142857e-05,
+      "loss": 1.4383,
       "step": 888
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 8.246548323471401e-05,
-      "loss": 1.855,
+      "epoch": 3.97,
+      "learning_rate": 9.206250000000001e-05,
+      "loss": 1.4597,
       "step": 889
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 8.24457593688363e-05,
-      "loss": 1.7586,
+      "epoch": 3.97,
+      "learning_rate": 9.205357142857143e-05,
+      "loss": 1.4469,
       "step": 890
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 8.242603550295859e-05,
-      "loss": 1.8667,
+      "epoch": 3.98,
+      "learning_rate": 9.204464285714286e-05,
+      "loss": 1.4029,
       "step": 891
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 8.240631163708087e-05,
-      "loss": 1.8961,
+      "epoch": 3.98,
+      "learning_rate": 9.20357142857143e-05,
+      "loss": 1.3644,
       "step": 892
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 8.238658777120316e-05,
-      "loss": 1.8361,
+      "epoch": 3.99,
+      "learning_rate": 9.202678571428572e-05,
+      "loss": 1.4237,
       "step": 893
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 8.236686390532545e-05,
-      "loss": 1.7352,
+      "epoch": 3.99,
+      "learning_rate": 9.201785714285715e-05,
+      "loss": 1.4633,
       "step": 894
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 8.234714003944773e-05,
-      "loss": 1.83,
+      "epoch": 4.0,
+      "learning_rate": 9.200892857142857e-05,
+      "loss": 1.3694,
       "step": 895
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 8.232741617357003e-05,
-      "loss": 1.7314,
+      "epoch": 4.0,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.3308,
       "step": 896
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 8.23076923076923e-05,
-      "loss": 1.8091,
+      "epoch": 4.0,
+      "learning_rate": 9.199107142857143e-05,
+      "loss": 1.3775,
       "step": 897
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 8.22879684418146e-05,
-      "loss": 1.7807,
+      "epoch": 4.01,
+      "learning_rate": 9.198214285714286e-05,
+      "loss": 1.3573,
       "step": 898
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 8.226824457593689e-05,
-      "loss": 1.8446,
+      "epoch": 4.01,
+      "learning_rate": 9.197321428571429e-05,
+      "loss": 1.4226,
       "step": 899
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 8.224852071005918e-05,
-      "loss": 1.7649,
+      "epoch": 4.02,
+      "learning_rate": 9.196428571428572e-05,
+      "loss": 1.3235,
       "step": 900
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 8.222879684418147e-05,
-      "loss": 1.7751,
-      "step": 901
+      "epoch": 4.02,
+      "eval_loss": 1.2803537845611572,
+      "eval_runtime": 45.6203,
+      "eval_samples_per_second": 287.022,
+      "eval_steps_per_second": 1.512,
+      "step": 900
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 8.220907297830375e-05,
-      "loss": 1.7602,
+      "epoch": 4.02,
+      "learning_rate": 9.195535714285714e-05,
+      "loss": 1.434,
+      "step": 901
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.194642857142858e-05,
+      "loss": 1.4772,
       "step": 902
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 8.218934911242604e-05,
-      "loss": 1.8633,
+      "epoch": 4.03,
+      "learning_rate": 9.19375e-05,
+      "loss": 1.3499,
       "step": 903
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 8.216962524654833e-05,
-      "loss": 1.8336,
+      "epoch": 4.04,
+      "learning_rate": 9.192857142857143e-05,
+      "loss": 1.3896,
       "step": 904
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 8.214990138067061e-05,
-      "loss": 1.8159,
+      "epoch": 4.04,
+      "learning_rate": 9.191964285714286e-05,
+      "loss": 1.445,
       "step": 905
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 8.21301775147929e-05,
-      "loss": 1.7459,
+      "epoch": 4.04,
+      "learning_rate": 9.19107142857143e-05,
+      "loss": 1.3902,
       "step": 906
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 8.211045364891518e-05,
-      "loss": 1.8947,
+      "epoch": 4.05,
+      "learning_rate": 9.190178571428572e-05,
+      "loss": 1.4206,
       "step": 907
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 8.209072978303748e-05,
-      "loss": 1.789,
+      "epoch": 4.05,
+      "learning_rate": 9.189285714285715e-05,
+      "loss": 1.4312,
       "step": 908
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 8.207100591715977e-05,
-      "loss": 1.7894,
+      "epoch": 4.06,
+      "learning_rate": 9.188392857142859e-05,
+      "loss": 1.4362,
       "step": 909
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 8.205128205128205e-05,
-      "loss": 1.783,
+      "epoch": 4.06,
+      "learning_rate": 9.1875e-05,
+      "loss": 1.4898,
       "step": 910
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 8.203155818540434e-05,
-      "loss": 1.7893,
+      "epoch": 4.07,
+      "learning_rate": 9.186607142857144e-05,
+      "loss": 1.3594,
       "step": 911
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 8.201183431952663e-05,
-      "loss": 1.7181,
+      "epoch": 4.07,
+      "learning_rate": 9.185714285714286e-05,
+      "loss": 1.476,
       "step": 912
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 8.199211045364892e-05,
-      "loss": 1.7898,
+      "epoch": 4.08,
+      "learning_rate": 9.184821428571429e-05,
+      "loss": 1.377,
       "step": 913
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 8.197238658777122e-05,
-      "loss": 1.7524,
+      "epoch": 4.08,
+      "learning_rate": 9.183928571428571e-05,
+      "loss": 1.3857,
       "step": 914
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 8.19526627218935e-05,
-      "loss": 1.8185,
+      "epoch": 4.08,
+      "learning_rate": 9.183035714285715e-05,
+      "loss": 1.4099,
       "step": 915
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 8.193293885601579e-05,
-      "loss": 1.846,
+      "epoch": 4.09,
+      "learning_rate": 9.182142857142858e-05,
+      "loss": 1.3569,
       "step": 916
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 8.191321499013808e-05,
-      "loss": 1.8301,
+      "epoch": 4.09,
+      "learning_rate": 9.18125e-05,
+      "loss": 1.3395,
       "step": 917
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 8.189349112426036e-05,
-      "loss": 1.8143,
+      "epoch": 4.1,
+      "learning_rate": 9.180357142857143e-05,
+      "loss": 1.3949,
       "step": 918
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 8.187376725838265e-05,
-      "loss": 1.7469,
+      "epoch": 4.1,
+      "learning_rate": 9.179464285714287e-05,
+      "loss": 1.3753,
       "step": 919
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 8.185404339250493e-05,
-      "loss": 1.7839,
+      "epoch": 4.11,
+      "learning_rate": 9.178571428571429e-05,
+      "loss": 1.3775,
       "step": 920
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 8.183431952662722e-05,
-      "loss": 1.7911,
+      "epoch": 4.11,
+      "learning_rate": 9.177678571428572e-05,
+      "loss": 1.3683,
       "step": 921
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 8.181459566074951e-05,
-      "loss": 1.8097,
+      "epoch": 4.12,
+      "learning_rate": 9.176785714285714e-05,
+      "loss": 1.4846,
       "step": 922
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 8.179487179487179e-05,
-      "loss": 1.8084,
+      "epoch": 4.12,
+      "learning_rate": 9.175892857142857e-05,
+      "loss": 1.3241,
       "step": 923
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 8.17751479289941e-05,
-      "loss": 1.7076,
+      "epoch": 4.12,
+      "learning_rate": 9.175000000000001e-05,
+      "loss": 1.3959,
       "step": 924
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 8.175542406311637e-05,
-      "loss": 1.8234,
+      "epoch": 4.13,
+      "learning_rate": 9.174107142857143e-05,
+      "loss": 1.3668,
       "step": 925
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 8.173570019723867e-05,
-      "loss": 1.8137,
+      "epoch": 4.13,
+      "learning_rate": 9.173214285714287e-05,
+      "loss": 1.3215,
       "step": 926
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 8.171597633136096e-05,
-      "loss": 1.7534,
+      "epoch": 4.14,
+      "learning_rate": 9.172321428571428e-05,
+      "loss": 1.3635,
       "step": 927
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 8.169625246548324e-05,
-      "loss": 1.6798,
+      "epoch": 4.14,
+      "learning_rate": 9.171428571428572e-05,
+      "loss": 1.3956,
       "step": 928
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 8.167652859960553e-05,
-      "loss": 1.8602,
+      "epoch": 4.15,
+      "learning_rate": 9.170535714285715e-05,
+      "loss": 1.4117,
       "step": 929
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 8.165680473372781e-05,
-      "loss": 1.8014,
+      "epoch": 4.15,
+      "learning_rate": 9.169642857142857e-05,
+      "loss": 1.3735,
       "step": 930
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 8.16370808678501e-05,
-      "loss": 1.75,
+      "epoch": 4.16,
+      "learning_rate": 9.16875e-05,
+      "loss": 1.4224,
       "step": 931
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 8.161735700197239e-05,
-      "loss": 1.8285,
+      "epoch": 4.16,
+      "learning_rate": 9.167857142857144e-05,
+      "loss": 1.4684,
       "step": 932
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 8.159763313609467e-05,
-      "loss": 1.8733,
+      "epoch": 4.17,
+      "learning_rate": 9.166964285714285e-05,
+      "loss": 1.4138,
       "step": 933
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 8.157790927021696e-05,
-      "loss": 1.8037,
+      "epoch": 4.17,
+      "learning_rate": 9.166071428571429e-05,
+      "loss": 1.3149,
       "step": 934
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 8.155818540433925e-05,
-      "loss": 1.8907,
+      "epoch": 4.17,
+      "learning_rate": 9.165178571428571e-05,
+      "loss": 1.3681,
       "step": 935
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 8.153846153846155e-05,
-      "loss": 1.7746,
+      "epoch": 4.18,
+      "learning_rate": 9.164285714285715e-05,
+      "loss": 1.3734,
       "step": 936
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 8.151873767258384e-05,
-      "loss": 1.8329,
+      "epoch": 4.18,
+      "learning_rate": 9.163392857142858e-05,
+      "loss": 1.3499,
       "step": 937
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 8.149901380670612e-05,
-      "loss": 1.904,
+      "epoch": 4.19,
+      "learning_rate": 9.1625e-05,
+      "loss": 1.37,
       "step": 938
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 8.147928994082841e-05,
-      "loss": 1.8483,
+      "epoch": 4.19,
+      "learning_rate": 9.161607142857144e-05,
+      "loss": 1.2712,
       "step": 939
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 8.14595660749507e-05,
-      "loss": 1.7707,
+      "epoch": 4.2,
+      "learning_rate": 9.160714285714286e-05,
+      "loss": 1.3496,
       "step": 940
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 8.143984220907298e-05,
-      "loss": 1.7699,
+      "epoch": 4.2,
+      "learning_rate": 9.15982142857143e-05,
+      "loss": 1.4282,
       "step": 941
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 8.142011834319527e-05,
-      "loss": 1.8091,
+      "epoch": 4.21,
+      "learning_rate": 9.158928571428572e-05,
+      "loss": 1.4299,
       "step": 942
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 8.140039447731755e-05,
-      "loss": 1.8439,
+      "epoch": 4.21,
+      "learning_rate": 9.158035714285715e-05,
+      "loss": 1.397,
       "step": 943
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 8.138067061143984e-05,
-      "loss": 1.9225,
+      "epoch": 4.21,
+      "learning_rate": 9.157142857142857e-05,
+      "loss": 1.4089,
       "step": 944
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 8.136094674556214e-05,
-      "loss": 1.74,
+      "epoch": 4.22,
+      "learning_rate": 9.156250000000001e-05,
+      "loss": 1.3381,
       "step": 945
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 8.134122287968441e-05,
-      "loss": 1.7734,
+      "epoch": 4.22,
+      "learning_rate": 9.155357142857142e-05,
+      "loss": 1.3818,
       "step": 946
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 8.13214990138067e-05,
-      "loss": 1.6983,
+      "epoch": 4.23,
+      "learning_rate": 9.154464285714286e-05,
+      "loss": 1.3851,
       "step": 947
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 8.1301775147929e-05,
-      "loss": 1.8118,
+      "epoch": 4.23,
+      "learning_rate": 9.153571428571429e-05,
+      "loss": 1.4881,
       "step": 948
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 8.128205128205129e-05,
-      "loss": 1.755,
+      "epoch": 4.24,
+      "learning_rate": 9.152678571428573e-05,
+      "loss": 1.3235,
       "step": 949
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 8.126232741617358e-05,
-      "loss": 1.7313,
+      "epoch": 4.24,
+      "learning_rate": 9.151785714285715e-05,
+      "loss": 1.3575,
       "step": 950
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 8.124260355029586e-05,
-      "loss": 1.6845,
+      "epoch": 4.25,
+      "learning_rate": 9.150892857142858e-05,
+      "loss": 1.3483,
       "step": 951
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 8.122287968441815e-05,
-      "loss": 1.7866,
+      "epoch": 4.25,
+      "learning_rate": 9.15e-05,
+      "loss": 1.3223,
       "step": 952
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 8.120315581854043e-05,
-      "loss": 1.8977,
+      "epoch": 4.25,
+      "learning_rate": 9.149107142857143e-05,
+      "loss": 1.3601,
       "step": 953
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 8.118343195266272e-05,
-      "loss": 1.7944,
+      "epoch": 4.26,
+      "learning_rate": 9.148214285714287e-05,
+      "loss": 1.3683,
       "step": 954
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 8.116370808678502e-05,
-      "loss": 1.7997,
+      "epoch": 4.26,
+      "learning_rate": 9.147321428571429e-05,
+      "loss": 1.3895,
       "step": 955
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 8.11439842209073e-05,
-      "loss": 1.7754,
+      "epoch": 4.27,
+      "learning_rate": 9.146428571428572e-05,
+      "loss": 1.3327,
       "step": 956
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 8.112426035502959e-05,
-      "loss": 1.7726,
+      "epoch": 4.27,
+      "learning_rate": 9.145535714285714e-05,
+      "loss": 1.3262,
       "step": 957
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 8.110453648915188e-05,
-      "loss": 1.806,
+      "epoch": 4.28,
+      "learning_rate": 9.144642857142858e-05,
+      "loss": 1.4223,
       "step": 958
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 8.108481262327417e-05,
-      "loss": 1.8488,
+      "epoch": 4.28,
+      "learning_rate": 9.14375e-05,
+      "loss": 1.3261,
       "step": 959
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 8.106508875739646e-05,
-      "loss": 1.7413,
+      "epoch": 4.29,
+      "learning_rate": 9.142857142857143e-05,
+      "loss": 1.3947,
       "step": 960
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 8.104536489151874e-05,
-      "loss": 1.7158,
+      "epoch": 4.29,
+      "learning_rate": 9.141964285714286e-05,
+      "loss": 1.4281,
       "step": 961
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 8.102564102564103e-05,
-      "loss": 1.7229,
+      "epoch": 4.29,
+      "learning_rate": 9.14107142857143e-05,
+      "loss": 1.3025,
       "step": 962
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 8.100591715976333e-05,
-      "loss": 1.8413,
+      "epoch": 4.3,
+      "learning_rate": 9.140178571428571e-05,
+      "loss": 1.3304,
       "step": 963
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 8.09861932938856e-05,
-      "loss": 1.7297,
+      "epoch": 4.3,
+      "learning_rate": 9.139285714285715e-05,
+      "loss": 1.4289,
       "step": 964
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 8.09664694280079e-05,
-      "loss": 1.772,
+      "epoch": 4.31,
+      "learning_rate": 9.138392857142857e-05,
+      "loss": 1.349,
       "step": 965
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 8.094674556213017e-05,
-      "loss": 1.8933,
+      "epoch": 4.31,
+      "learning_rate": 9.1375e-05,
+      "loss": 1.4244,
       "step": 966
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 8.092702169625247e-05,
-      "loss": 1.8436,
+      "epoch": 4.32,
+      "learning_rate": 9.136607142857144e-05,
+      "loss": 1.3275,
       "step": 967
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 8.090729783037476e-05,
-      "loss": 1.6602,
+      "epoch": 4.32,
+      "learning_rate": 9.135714285714286e-05,
+      "loss": 1.3929,
       "step": 968
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 8.088757396449704e-05,
-      "loss": 1.8474,
+      "epoch": 4.33,
+      "learning_rate": 9.13482142857143e-05,
+      "loss": 1.393,
       "step": 969
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 8.086785009861933e-05,
-      "loss": 1.6688,
+      "epoch": 4.33,
+      "learning_rate": 9.133928571428571e-05,
+      "loss": 1.4017,
       "step": 970
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 8.084812623274162e-05,
-      "loss": 1.8355,
+      "epoch": 4.33,
+      "learning_rate": 9.133035714285715e-05,
+      "loss": 1.4737,
       "step": 971
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 8.082840236686391e-05,
-      "loss": 1.7951,
+      "epoch": 4.34,
+      "learning_rate": 9.132142857142858e-05,
+      "loss": 1.3037,
       "step": 972
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 8.08086785009862e-05,
-      "loss": 1.7564,
+      "epoch": 4.34,
+      "learning_rate": 9.13125e-05,
+      "loss": 1.4011,
       "step": 973
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 8.078895463510848e-05,
-      "loss": 1.7483,
+      "epoch": 4.35,
+      "learning_rate": 9.130357142857143e-05,
+      "loss": 1.3789,
       "step": 974
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 8.076923076923078e-05,
-      "loss": 1.8262,
+      "epoch": 4.35,
+      "learning_rate": 9.129464285714287e-05,
+      "loss": 1.3034,
       "step": 975
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 8.074950690335306e-05,
-      "loss": 1.7467,
+      "epoch": 4.36,
+      "learning_rate": 9.128571428571428e-05,
+      "loss": 1.394,
       "step": 976
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 8.072978303747535e-05,
-      "loss": 1.7623,
+      "epoch": 4.36,
+      "learning_rate": 9.127678571428572e-05,
+      "loss": 1.3902,
       "step": 977
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 8.071005917159764e-05,
-      "loss": 1.7806,
+      "epoch": 4.37,
+      "learning_rate": 9.126785714285714e-05,
+      "loss": 1.4623,
       "step": 978
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 8.069033530571992e-05,
-      "loss": 1.7233,
+      "epoch": 4.37,
+      "learning_rate": 9.125892857142858e-05,
+      "loss": 1.3545,
       "step": 979
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 8.067061143984221e-05,
-      "loss": 1.7297,
+      "epoch": 4.38,
+      "learning_rate": 9.125e-05,
+      "loss": 1.3049,
       "step": 980
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 8.06508875739645e-05,
-      "loss": 1.9552,
+      "epoch": 4.38,
+      "learning_rate": 9.124107142857143e-05,
+      "loss": 1.3897,
       "step": 981
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 8.063116370808678e-05,
-      "loss": 1.7473,
+      "epoch": 4.38,
+      "learning_rate": 9.123214285714286e-05,
+      "loss": 1.4096,
       "step": 982
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 8.061143984220909e-05,
-      "loss": 1.7356,
+      "epoch": 4.39,
+      "learning_rate": 9.122321428571429e-05,
+      "loss": 1.3983,
       "step": 983
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 8.059171597633137e-05,
-      "loss": 1.7811,
+      "epoch": 4.39,
+      "learning_rate": 9.121428571428572e-05,
+      "loss": 1.369,
       "step": 984
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 8.057199211045366e-05,
-      "loss": 1.6885,
+      "epoch": 4.4,
+      "learning_rate": 9.120535714285715e-05,
+      "loss": 1.3641,
       "step": 985
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 8.055226824457595e-05,
-      "loss": 1.8778,
+      "epoch": 4.4,
+      "learning_rate": 9.119642857142858e-05,
+      "loss": 1.4354,
       "step": 986
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 8.053254437869823e-05,
-      "loss": 1.6362,
+      "epoch": 4.41,
+      "learning_rate": 9.11875e-05,
+      "loss": 1.3533,
       "step": 987
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 8.051282051282052e-05,
-      "loss": 1.7879,
+      "epoch": 4.41,
+      "learning_rate": 9.117857142857144e-05,
+      "loss": 1.3348,
       "step": 988
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 8.04930966469428e-05,
-      "loss": 1.7555,
+      "epoch": 4.42,
+      "learning_rate": 9.116964285714285e-05,
+      "loss": 1.4055,
       "step": 989
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 8.047337278106509e-05,
-      "loss": 1.8987,
+      "epoch": 4.42,
+      "learning_rate": 9.116071428571429e-05,
+      "loss": 1.374,
       "step": 990
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 8.045364891518738e-05,
-      "loss": 1.852,
+      "epoch": 4.42,
+      "learning_rate": 9.115178571428572e-05,
+      "loss": 1.3742,
       "step": 991
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 8.043392504930966e-05,
-      "loss": 1.8562,
+      "epoch": 4.43,
+      "learning_rate": 9.114285714285716e-05,
+      "loss": 1.4204,
       "step": 992
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 8.041420118343195e-05,
-      "loss": 1.901,
+      "epoch": 4.43,
+      "learning_rate": 9.113392857142857e-05,
+      "loss": 1.4066,
       "step": 993
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 8.039447731755425e-05,
-      "loss": 1.7828,
+      "epoch": 4.44,
+      "learning_rate": 9.1125e-05,
+      "loss": 1.3075,
       "step": 994
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 8.037475345167654e-05,
-      "loss": 1.7779,
+      "epoch": 4.44,
+      "learning_rate": 9.111607142857143e-05,
+      "loss": 1.4087,
       "step": 995
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.035502958579883e-05,
-      "loss": 1.7152,
+      "epoch": 4.45,
+      "learning_rate": 9.110714285714286e-05,
+      "loss": 1.4243,
       "step": 996
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.033530571992111e-05,
-      "loss": 1.6811,
+      "epoch": 4.45,
+      "learning_rate": 9.10982142857143e-05,
+      "loss": 1.3915,
       "step": 997
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.03155818540434e-05,
-      "loss": 1.6979,
+      "epoch": 4.46,
+      "learning_rate": 9.108928571428572e-05,
+      "loss": 1.3566,
       "step": 998
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 8.029585798816568e-05,
-      "loss": 1.7603,
+      "epoch": 4.46,
+      "learning_rate": 9.108035714285715e-05,
+      "loss": 1.3003,
       "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,
+      "epoch": 4.46,
+      "learning_rate": 9.107142857142857e-05,
+      "loss": 1.5127,
       "step": 1000
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 8.025641025641026e-05,
-      "loss": 1.875,
+      "epoch": 4.47,
+      "learning_rate": 9.106250000000001e-05,
+      "loss": 1.3663,
       "step": 1001
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 8.023668639053254e-05,
-      "loss": 1.8114,
+      "epoch": 4.47,
+      "learning_rate": 9.105357142857144e-05,
+      "loss": 1.3888,
       "step": 1002
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 8.021696252465483e-05,
-      "loss": 1.7824,
+      "epoch": 4.48,
+      "learning_rate": 9.104464285714286e-05,
+      "loss": 1.448,
       "step": 1003
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 8.019723865877713e-05,
-      "loss": 1.7263,
+      "epoch": 4.48,
+      "learning_rate": 9.103571428571429e-05,
+      "loss": 1.4064,
       "step": 1004
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 8.01775147928994e-05,
-      "loss": 1.7795,
+      "epoch": 4.49,
+      "learning_rate": 9.102678571428573e-05,
+      "loss": 1.503,
       "step": 1005
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 8.01577909270217e-05,
-      "loss": 1.8166,
+      "epoch": 4.49,
+      "learning_rate": 9.101785714285714e-05,
+      "loss": 1.3267,
       "step": 1006
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 8.013806706114399e-05,
-      "loss": 1.7061,
+      "epoch": 4.5,
+      "learning_rate": 9.100892857142858e-05,
+      "loss": 1.4447,
       "step": 1007
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 8.011834319526628e-05,
-      "loss": 1.9,
+      "epoch": 4.5,
+      "learning_rate": 9.1e-05,
+      "loss": 1.3903,
       "step": 1008
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 8.009861932938857e-05,
-      "loss": 1.9002,
+      "epoch": 4.5,
+      "learning_rate": 9.099107142857143e-05,
+      "loss": 1.3968,
       "step": 1009
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 8.007889546351085e-05,
-      "loss": 1.7215,
+      "epoch": 4.51,
+      "learning_rate": 9.098214285714285e-05,
+      "loss": 1.435,
       "step": 1010
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 8.005917159763314e-05,
-      "loss": 1.8256,
+      "epoch": 4.51,
+      "learning_rate": 9.097321428571429e-05,
+      "loss": 1.3684,
       "step": 1011
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 8.003944773175542e-05,
-      "loss": 1.784,
+      "epoch": 4.52,
+      "learning_rate": 9.096428571428572e-05,
+      "loss": 1.5087,
       "step": 1012
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 8.001972386587771e-05,
-      "loss": 1.7676,
+      "epoch": 4.52,
+      "learning_rate": 9.095535714285714e-05,
+      "loss": 1.4411,
       "step": 1013
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 8e-05,
-      "loss": 1.6152,
+      "epoch": 4.53,
+      "learning_rate": 9.094642857142858e-05,
+      "loss": 1.4003,
       "step": 1014
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 7.998027613412229e-05,
-      "loss": 1.6694,
+      "epoch": 4.53,
+      "learning_rate": 9.093750000000001e-05,
+      "loss": 1.3611,
       "step": 1015
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 7.996055226824458e-05,
-      "loss": 1.6094,
+      "epoch": 4.54,
+      "learning_rate": 9.092857142857143e-05,
+      "loss": 1.3668,
       "step": 1016
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 7.994082840236687e-05,
-      "loss": 1.7174,
+      "epoch": 4.54,
+      "learning_rate": 9.091964285714286e-05,
+      "loss": 1.3226,
       "step": 1017
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 7.992110453648916e-05,
-      "loss": 1.7866,
+      "epoch": 4.54,
+      "learning_rate": 9.09107142857143e-05,
+      "loss": 1.3254,
       "step": 1018
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 7.990138067061145e-05,
-      "loss": 1.7978,
+      "epoch": 4.55,
+      "learning_rate": 9.090178571428571e-05,
+      "loss": 1.3464,
       "step": 1019
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 7.988165680473373e-05,
-      "loss": 1.7342,
+      "epoch": 4.55,
+      "learning_rate": 9.089285714285715e-05,
+      "loss": 1.4678,
       "step": 1020
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 7.986193293885602e-05,
-      "loss": 1.7476,
+      "epoch": 4.56,
+      "learning_rate": 9.088392857142857e-05,
+      "loss": 1.3281,
       "step": 1021
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 7.98422090729783e-05,
-      "loss": 1.6855,
+      "epoch": 4.56,
+      "learning_rate": 9.0875e-05,
+      "loss": 1.3767,
       "step": 1022
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 7.98224852071006e-05,
-      "loss": 1.7545,
+      "epoch": 4.57,
+      "learning_rate": 9.086607142857143e-05,
+      "loss": 1.3867,
       "step": 1023
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 7.980276134122289e-05,
-      "loss": 1.7573,
+      "epoch": 4.57,
+      "learning_rate": 9.085714285714286e-05,
+      "loss": 1.3194,
       "step": 1024
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 7.978303747534517e-05,
-      "loss": 1.7819,
+      "epoch": 4.58,
+      "learning_rate": 9.084821428571429e-05,
+      "loss": 1.3917,
       "step": 1025
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.976331360946746e-05,
-      "loss": 1.756,
+      "epoch": 4.58,
+      "learning_rate": 9.083928571428572e-05,
+      "loss": 1.377,
       "step": 1026
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.974358974358975e-05,
-      "loss": 1.8341,
+      "epoch": 4.58,
+      "learning_rate": 9.083035714285715e-05,
+      "loss": 1.426,
       "step": 1027
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.972386587771203e-05,
-      "loss": 1.8413,
+      "epoch": 4.59,
+      "learning_rate": 9.082142857142858e-05,
+      "loss": 1.4186,
       "step": 1028
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.970414201183432e-05,
-      "loss": 1.8856,
+      "epoch": 4.59,
+      "learning_rate": 9.08125e-05,
+      "loss": 1.4215,
       "step": 1029
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 7.968441814595661e-05,
-      "loss": 1.7549,
+      "epoch": 4.6,
+      "learning_rate": 9.080357142857143e-05,
+      "loss": 1.4719,
       "step": 1030
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 7.96646942800789e-05,
-      "loss": 1.7956,
+      "epoch": 4.6,
+      "learning_rate": 9.079464285714287e-05,
+      "loss": 1.3211,
       "step": 1031
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 7.96449704142012e-05,
-      "loss": 1.7822,
+      "epoch": 4.61,
+      "learning_rate": 9.078571428571428e-05,
+      "loss": 1.2643,
       "step": 1032
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 7.962524654832348e-05,
-      "loss": 1.8146,
+      "epoch": 4.61,
+      "learning_rate": 9.077678571428572e-05,
+      "loss": 1.3775,
       "step": 1033
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 7.960552268244577e-05,
-      "loss": 1.7863,
+      "epoch": 4.62,
+      "learning_rate": 9.076785714285715e-05,
+      "loss": 1.3524,
       "step": 1034
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 7.958579881656805e-05,
-      "loss": 1.7761,
+      "epoch": 4.62,
+      "learning_rate": 9.075892857142858e-05,
+      "loss": 1.3344,
       "step": 1035
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.956607495069034e-05,
-      "loss": 1.7215,
+      "epoch": 4.62,
+      "learning_rate": 9.075e-05,
+      "loss": 1.3925,
       "step": 1036
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.954635108481263e-05,
-      "loss": 1.7974,
+      "epoch": 4.63,
+      "learning_rate": 9.074107142857144e-05,
+      "loss": 1.3683,
       "step": 1037
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.952662721893491e-05,
-      "loss": 1.7175,
+      "epoch": 4.63,
+      "learning_rate": 9.073214285714286e-05,
+      "loss": 1.4368,
       "step": 1038
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.95069033530572e-05,
-      "loss": 1.7995,
+      "epoch": 4.64,
+      "learning_rate": 9.072321428571429e-05,
+      "loss": 1.335,
       "step": 1039
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 7.948717948717948e-05,
-      "loss": 1.8055,
+      "epoch": 4.64,
+      "learning_rate": 9.071428571428571e-05,
+      "loss": 1.3816,
       "step": 1040
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 7.946745562130179e-05,
-      "loss": 1.77,
+      "epoch": 4.65,
+      "learning_rate": 9.070535714285715e-05,
+      "loss": 1.3188,
       "step": 1041
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 7.944773175542408e-05,
-      "loss": 1.7737,
+      "epoch": 4.65,
+      "learning_rate": 9.069642857142858e-05,
+      "loss": 1.3701,
       "step": 1042
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 7.942800788954636e-05,
-      "loss": 1.7763,
+      "epoch": 4.66,
+      "learning_rate": 9.06875e-05,
+      "loss": 1.3887,
       "step": 1043
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 7.940828402366865e-05,
-      "loss": 1.7532,
+      "epoch": 4.66,
+      "learning_rate": 9.067857142857144e-05,
+      "loss": 1.2697,
       "step": 1044
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 7.938856015779093e-05,
-      "loss": 1.7509,
+      "epoch": 4.67,
+      "learning_rate": 9.066964285714287e-05,
+      "loss": 1.404,
       "step": 1045
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 7.936883629191322e-05,
-      "loss": 1.791,
+      "epoch": 4.67,
+      "learning_rate": 9.066071428571429e-05,
+      "loss": 1.3991,
       "step": 1046
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 7.934911242603551e-05,
-      "loss": 1.7791,
+      "epoch": 4.67,
+      "learning_rate": 9.065178571428572e-05,
+      "loss": 1.294,
       "step": 1047
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 7.932938856015779e-05,
-      "loss": 1.7826,
+      "epoch": 4.68,
+      "learning_rate": 9.064285714285716e-05,
+      "loss": 1.4079,
       "step": 1048
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 7.930966469428008e-05,
-      "loss": 1.7109,
+      "epoch": 4.68,
+      "learning_rate": 9.063392857142857e-05,
+      "loss": 1.3398,
       "step": 1049
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 7.928994082840237e-05,
-      "loss": 1.7137,
+      "epoch": 4.69,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 1.3305,
       "step": 1050
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 7.927021696252465e-05,
-      "loss": 1.7577,
+      "epoch": 4.69,
+      "learning_rate": 9.061607142857143e-05,
+      "loss": 1.4483,
       "step": 1051
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 7.925049309664694e-05,
-      "loss": 1.659,
+      "epoch": 4.7,
+      "learning_rate": 9.060714285714286e-05,
+      "loss": 1.3603,
       "step": 1052
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 7.923076923076924e-05,
-      "loss": 1.8969,
+      "epoch": 4.7,
+      "learning_rate": 9.059821428571428e-05,
+      "loss": 1.339,
       "step": 1053
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 7.921104536489153e-05,
-      "loss": 1.7322,
+      "epoch": 4.71,
+      "learning_rate": 9.058928571428572e-05,
+      "loss": 1.3751,
       "step": 1054
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 7.919132149901382e-05,
-      "loss": 1.832,
+      "epoch": 4.71,
+      "learning_rate": 9.058035714285715e-05,
+      "loss": 1.3254,
       "step": 1055
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 7.91715976331361e-05,
-      "loss": 1.8411,
+      "epoch": 4.71,
+      "learning_rate": 9.057142857142857e-05,
+      "loss": 1.3665,
       "step": 1056
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 7.915187376725839e-05,
-      "loss": 1.7345,
+      "epoch": 4.72,
+      "learning_rate": 9.05625e-05,
+      "loss": 1.3545,
       "step": 1057
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 7.913214990138067e-05,
-      "loss": 1.7086,
+      "epoch": 4.72,
+      "learning_rate": 9.055357142857144e-05,
+      "loss": 1.4644,
       "step": 1058
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 7.911242603550296e-05,
-      "loss": 1.6368,
+      "epoch": 4.73,
+      "learning_rate": 9.054464285714286e-05,
+      "loss": 1.3955,
       "step": 1059
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 7.909270216962525e-05,
-      "loss": 1.7429,
+      "epoch": 4.73,
+      "learning_rate": 9.053571428571429e-05,
+      "loss": 1.36,
       "step": 1060
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 7.907297830374753e-05,
-      "loss": 1.6527,
+      "epoch": 4.74,
+      "learning_rate": 9.052678571428573e-05,
+      "loss": 1.3636,
       "step": 1061
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 7.905325443786982e-05,
-      "loss": 1.6841,
+      "epoch": 4.74,
+      "learning_rate": 9.051785714285714e-05,
+      "loss": 1.4061,
       "step": 1062
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 7.90335305719921e-05,
-      "loss": 1.751,
+      "epoch": 4.75,
+      "learning_rate": 9.050892857142858e-05,
+      "loss": 1.4194,
       "step": 1063
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 7.90138067061144e-05,
-      "loss": 1.798,
+      "epoch": 4.75,
+      "learning_rate": 9.05e-05,
+      "loss": 1.3417,
       "step": 1064
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 7.899408284023669e-05,
-      "loss": 1.7646,
+      "epoch": 4.75,
+      "learning_rate": 9.049107142857143e-05,
+      "loss": 1.464,
       "step": 1065
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 7.897435897435898e-05,
-      "loss": 1.8508,
+      "epoch": 4.76,
+      "learning_rate": 9.048214285714286e-05,
+      "loss": 1.3381,
       "step": 1066
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 7.895463510848127e-05,
-      "loss": 1.8614,
+      "epoch": 4.76,
+      "learning_rate": 9.04732142857143e-05,
+      "loss": 1.3291,
       "step": 1067
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 7.893491124260355e-05,
-      "loss": 1.7231,
+      "epoch": 4.77,
+      "learning_rate": 9.046428571428572e-05,
+      "loss": 1.3491,
       "step": 1068
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 7.891518737672584e-05,
-      "loss": 1.7853,
+      "epoch": 4.77,
+      "learning_rate": 9.045535714285715e-05,
+      "loss": 1.3245,
       "step": 1069
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 7.889546351084813e-05,
-      "loss": 1.6603,
+      "epoch": 4.78,
+      "learning_rate": 9.044642857142857e-05,
+      "loss": 1.4048,
       "step": 1070
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 7.887573964497041e-05,
-      "loss": 1.7255,
+      "epoch": 4.78,
+      "learning_rate": 9.043750000000001e-05,
+      "loss": 1.366,
       "step": 1071
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 7.88560157790927e-05,
-      "loss": 1.6954,
+      "epoch": 4.79,
+      "learning_rate": 9.042857142857143e-05,
+      "loss": 1.4357,
       "step": 1072
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 7.8836291913215e-05,
-      "loss": 1.7138,
+      "epoch": 4.79,
+      "learning_rate": 9.041964285714286e-05,
+      "loss": 1.3923,
       "step": 1073
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 7.881656804733728e-05,
-      "loss": 1.8525,
+      "epoch": 4.79,
+      "learning_rate": 9.04107142857143e-05,
+      "loss": 1.4304,
       "step": 1074
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 7.879684418145957e-05,
-      "loss": 1.657,
+      "epoch": 4.8,
+      "learning_rate": 9.040178571428571e-05,
+      "loss": 1.2906,
       "step": 1075
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 7.877712031558186e-05,
-      "loss": 1.8444,
+      "epoch": 4.8,
+      "learning_rate": 9.039285714285715e-05,
+      "loss": 1.4415,
       "step": 1076
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 7.875739644970415e-05,
-      "loss": 1.7405,
+      "epoch": 4.81,
+      "learning_rate": 9.038392857142858e-05,
+      "loss": 1.3482,
       "step": 1077
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 7.873767258382644e-05,
-      "loss": 1.6666,
+      "epoch": 4.81,
+      "learning_rate": 9.037500000000001e-05,
+      "loss": 1.3832,
       "step": 1078
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 7.871794871794872e-05,
-      "loss": 1.654,
+      "epoch": 4.82,
+      "learning_rate": 9.036607142857143e-05,
+      "loss": 1.3536,
       "step": 1079
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 7.869822485207101e-05,
-      "loss": 1.7601,
+      "epoch": 4.82,
+      "learning_rate": 9.035714285714287e-05,
+      "loss": 1.4095,
       "step": 1080
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 7.86785009861933e-05,
-      "loss": 1.8334,
+      "epoch": 4.83,
+      "learning_rate": 9.034821428571429e-05,
+      "loss": 1.3455,
       "step": 1081
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 7.865877712031559e-05,
-      "loss": 1.8329,
+      "epoch": 4.83,
+      "learning_rate": 9.033928571428572e-05,
+      "loss": 1.3725,
       "step": 1082
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 7.863905325443788e-05,
-      "loss": 1.6851,
+      "epoch": 4.83,
+      "learning_rate": 9.033035714285714e-05,
+      "loss": 1.3722,
       "step": 1083
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 7.861932938856016e-05,
-      "loss": 1.7404,
+      "epoch": 4.84,
+      "learning_rate": 9.032142857142858e-05,
+      "loss": 1.418,
       "step": 1084
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 7.859960552268245e-05,
-      "loss": 1.7274,
+      "epoch": 4.84,
+      "learning_rate": 9.03125e-05,
+      "loss": 1.3464,
       "step": 1085
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 7.857988165680473e-05,
-      "loss": 1.7554,
+      "epoch": 4.85,
+      "learning_rate": 9.030357142857143e-05,
+      "loss": 1.36,
       "step": 1086
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 7.856015779092702e-05,
-      "loss": 1.7663,
+      "epoch": 4.85,
+      "learning_rate": 9.029464285714286e-05,
+      "loss": 1.3675,
       "step": 1087
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 7.854043392504931e-05,
-      "loss": 1.7319,
+      "epoch": 4.86,
+      "learning_rate": 9.028571428571428e-05,
+      "loss": 1.4357,
       "step": 1088
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 7.85207100591716e-05,
-      "loss": 1.6156,
+      "epoch": 4.86,
+      "learning_rate": 9.027678571428572e-05,
+      "loss": 1.3805,
       "step": 1089
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 7.85009861932939e-05,
-      "loss": 1.7827,
+      "epoch": 4.87,
+      "learning_rate": 9.026785714285715e-05,
+      "loss": 1.396,
       "step": 1090
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 7.848126232741617e-05,
-      "loss": 1.7047,
+      "epoch": 4.87,
+      "learning_rate": 9.025892857142859e-05,
+      "loss": 1.4812,
       "step": 1091
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 7.846153846153847e-05,
-      "loss": 1.8116,
+      "epoch": 4.88,
+      "learning_rate": 9.025e-05,
+      "loss": 1.412,
       "step": 1092
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 7.844181459566076e-05,
-      "loss": 1.6917,
+      "epoch": 4.88,
+      "learning_rate": 9.024107142857144e-05,
+      "loss": 1.3089,
       "step": 1093
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 7.842209072978304e-05,
-      "loss": 1.6991,
+      "epoch": 4.88,
+      "learning_rate": 9.023214285714286e-05,
+      "loss": 1.3028,
       "step": 1094
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 7.840236686390533e-05,
-      "loss": 1.7751,
+      "epoch": 4.89,
+      "learning_rate": 9.022321428571429e-05,
+      "loss": 1.3622,
       "step": 1095
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 7.838264299802762e-05,
-      "loss": 1.9192,
+      "epoch": 4.89,
+      "learning_rate": 9.021428571428571e-05,
+      "loss": 1.4296,
       "step": 1096
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 7.83629191321499e-05,
-      "loss": 1.7745,
+      "epoch": 4.9,
+      "learning_rate": 9.020535714285715e-05,
+      "loss": 1.3145,
       "step": 1097
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 7.834319526627219e-05,
-      "loss": 1.7642,
+      "epoch": 4.9,
+      "learning_rate": 9.019642857142856e-05,
+      "loss": 1.283,
       "step": 1098
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 7.832347140039447e-05,
-      "loss": 1.7567,
+      "epoch": 4.91,
+      "learning_rate": 9.01875e-05,
+      "loss": 1.3734,
       "step": 1099
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 7.830374753451678e-05,
-      "loss": 1.7559,
+      "epoch": 4.91,
+      "learning_rate": 9.017857142857143e-05,
+      "loss": 1.3255,
       "step": 1100
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 7.828402366863907e-05,
-      "loss": 1.7699,
+      "epoch": 4.92,
+      "learning_rate": 9.016964285714287e-05,
+      "loss": 1.4233,
       "step": 1101
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 7.826429980276135e-05,
-      "loss": 1.6309,
+      "epoch": 4.92,
+      "learning_rate": 9.01607142857143e-05,
+      "loss": 1.3145,
       "step": 1102
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 7.824457593688364e-05,
-      "loss": 1.8006,
+      "epoch": 4.92,
+      "learning_rate": 9.015178571428572e-05,
+      "loss": 1.3025,
       "step": 1103
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 7.822485207100592e-05,
-      "loss": 1.7759,
+      "epoch": 4.93,
+      "learning_rate": 9.014285714285716e-05,
+      "loss": 1.3936,
       "step": 1104
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 7.820512820512821e-05,
-      "loss": 1.7288,
+      "epoch": 4.93,
+      "learning_rate": 9.013392857142857e-05,
+      "loss": 1.4304,
       "step": 1105
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 7.81854043392505e-05,
-      "loss": 1.7954,
+      "epoch": 4.94,
+      "learning_rate": 9.012500000000001e-05,
+      "loss": 1.3781,
       "step": 1106
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 7.816568047337278e-05,
-      "loss": 1.7897,
+      "epoch": 4.94,
+      "learning_rate": 9.011607142857143e-05,
+      "loss": 1.3886,
       "step": 1107
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 7.814595660749507e-05,
-      "loss": 1.7113,
+      "epoch": 4.95,
+      "learning_rate": 9.010714285714286e-05,
+      "loss": 1.4073,
       "step": 1108
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 7.812623274161735e-05,
-      "loss": 1.7672,
+      "epoch": 4.95,
+      "learning_rate": 9.009821428571429e-05,
+      "loss": 1.3674,
       "step": 1109
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 7.810650887573964e-05,
-      "loss": 1.8214,
+      "epoch": 4.96,
+      "learning_rate": 9.008928571428572e-05,
+      "loss": 1.3997,
       "step": 1110
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 7.808678500986193e-05,
-      "loss": 1.7691,
+      "epoch": 4.96,
+      "learning_rate": 9.008035714285715e-05,
+      "loss": 1.3478,
       "step": 1111
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 7.806706114398423e-05,
-      "loss": 1.6605,
+      "epoch": 4.96,
+      "learning_rate": 9.007142857142857e-05,
+      "loss": 1.3641,
       "step": 1112
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 7.804733727810652e-05,
-      "loss": 1.7907,
+      "epoch": 4.97,
+      "learning_rate": 9.00625e-05,
+      "loss": 1.3233,
       "step": 1113
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 7.802761341222881e-05,
-      "loss": 1.7381,
+      "epoch": 4.97,
+      "learning_rate": 9.005357142857144e-05,
+      "loss": 1.2948,
       "step": 1114
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 7.800788954635109e-05,
-      "loss": 1.6696,
+      "epoch": 4.98,
+      "learning_rate": 9.004464285714285e-05,
+      "loss": 1.3748,
       "step": 1115
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 7.798816568047338e-05,
-      "loss": 1.8012,
+      "epoch": 4.98,
+      "learning_rate": 9.003571428571429e-05,
+      "loss": 1.3766,
       "step": 1116
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 7.796844181459566e-05,
-      "loss": 1.8031,
+      "epoch": 4.99,
+      "learning_rate": 9.002678571428572e-05,
+      "loss": 1.4065,
       "step": 1117
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 7.794871794871795e-05,
-      "loss": 1.8076,
+      "epoch": 4.99,
+      "learning_rate": 9.001785714285714e-05,
+      "loss": 1.3011,
       "step": 1118
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 7.792899408284024e-05,
-      "loss": 1.7319,
+      "epoch": 5.0,
+      "learning_rate": 9.000892857142858e-05,
+      "loss": 1.3553,
       "step": 1119
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 7.790927021696252e-05,
-      "loss": 1.7055,
+      "epoch": 5.0,
+      "learning_rate": 9e-05,
+      "loss": 1.3194,
       "step": 1120
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 7.788954635108482e-05,
-      "loss": 1.7679,
+      "epoch": 5.0,
+      "learning_rate": 8.999107142857143e-05,
+      "loss": 1.2942,
       "step": 1121
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 7.78698224852071e-05,
-      "loss": 1.7192,
+      "epoch": 5.01,
+      "learning_rate": 8.998214285714286e-05,
+      "loss": 1.3054,
       "step": 1122
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 7.785009861932939e-05,
-      "loss": 1.7648,
+      "epoch": 5.01,
+      "learning_rate": 8.99732142857143e-05,
+      "loss": 1.3481,
       "step": 1123
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 7.783037475345168e-05,
-      "loss": 1.7176,
+      "epoch": 5.02,
+      "learning_rate": 8.996428571428572e-05,
+      "loss": 1.4412,
       "step": 1124
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 7.781065088757397e-05,
-      "loss": 1.7045,
+      "epoch": 5.02,
+      "learning_rate": 8.995535714285715e-05,
+      "loss": 1.4411,
       "step": 1125
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 7.779092702169626e-05,
-      "loss": 1.751,
+      "epoch": 5.03,
+      "learning_rate": 8.994642857142857e-05,
+      "loss": 1.3326,
       "step": 1126
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 7.777120315581854e-05,
-      "loss": 1.7799,
+      "epoch": 5.03,
+      "learning_rate": 8.993750000000001e-05,
+      "loss": 1.3804,
       "step": 1127
     },
     {
-      "epoch": 3.34,
-      "learning_rate": 7.775147928994083e-05,
-      "loss": 1.8094,
+      "epoch": 5.04,
+      "learning_rate": 8.992857142857142e-05,
+      "loss": 1.4847,
       "step": 1128
     },
     {
-      "epoch": 3.34,
-      "learning_rate": 7.773175542406313e-05,
-      "loss": 1.8088,
+      "epoch": 5.04,
+      "learning_rate": 8.991964285714286e-05,
+      "loss": 1.3864,
       "step": 1129
     },
     {
-      "epoch": 3.34,
-      "learning_rate": 7.77120315581854e-05,
-      "loss": 1.7716,
+      "epoch": 5.04,
+      "learning_rate": 8.991071428571429e-05,
+      "loss": 1.3566,
       "step": 1130
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 7.76923076923077e-05,
-      "loss": 1.7479,
+      "epoch": 5.05,
+      "learning_rate": 8.990178571428571e-05,
+      "loss": 1.3478,
       "step": 1131
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 7.767258382642997e-05,
-      "loss": 1.8139,
+      "epoch": 5.05,
+      "learning_rate": 8.989285714285715e-05,
+      "loss": 1.2995,
       "step": 1132
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 7.765285996055227e-05,
-      "loss": 1.7373,
+      "epoch": 5.06,
+      "learning_rate": 8.988392857142858e-05,
+      "loss": 1.4047,
       "step": 1133
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 7.763313609467456e-05,
-      "loss": 1.8012,
+      "epoch": 5.06,
+      "learning_rate": 8.9875e-05,
+      "loss": 1.4202,
       "step": 1134
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 7.761341222879685e-05,
-      "loss": 1.7971,
+      "epoch": 5.07,
+      "learning_rate": 8.986607142857143e-05,
+      "loss": 1.3452,
       "step": 1135
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 7.759368836291914e-05,
-      "loss": 1.7722,
+      "epoch": 5.07,
+      "learning_rate": 8.985714285714287e-05,
+      "loss": 1.2398,
       "step": 1136
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 7.757396449704144e-05,
-      "loss": 1.7686,
+      "epoch": 5.08,
+      "learning_rate": 8.984821428571429e-05,
+      "loss": 1.3769,
       "step": 1137
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 7.755424063116371e-05,
-      "loss": 1.7738,
+      "epoch": 5.08,
+      "learning_rate": 8.983928571428572e-05,
+      "loss": 1.3222,
       "step": 1138
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 7.7534516765286e-05,
-      "loss": 1.6721,
+      "epoch": 5.08,
+      "learning_rate": 8.983035714285714e-05,
+      "loss": 1.3355,
       "step": 1139
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 7.751479289940828e-05,
-      "loss": 1.696,
+      "epoch": 5.09,
+      "learning_rate": 8.982142857142858e-05,
+      "loss": 1.2997,
       "step": 1140
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 7.749506903353058e-05,
-      "loss": 1.6853,
+      "epoch": 5.09,
+      "learning_rate": 8.98125e-05,
+      "loss": 1.4362,
       "step": 1141
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 7.747534516765287e-05,
-      "loss": 1.6562,
+      "epoch": 5.1,
+      "learning_rate": 8.980357142857143e-05,
+      "loss": 1.4235,
       "step": 1142
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 7.745562130177515e-05,
-      "loss": 1.782,
+      "epoch": 5.1,
+      "learning_rate": 8.979464285714286e-05,
+      "loss": 1.4044,
       "step": 1143
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 7.743589743589744e-05,
-      "loss": 1.8166,
+      "epoch": 5.11,
+      "learning_rate": 8.97857142857143e-05,
+      "loss": 1.2776,
       "step": 1144
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 7.741617357001972e-05,
-      "loss": 1.7867,
+      "epoch": 5.11,
+      "learning_rate": 8.977678571428571e-05,
+      "loss": 1.3464,
       "step": 1145
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 7.739644970414201e-05,
-      "loss": 1.7471,
+      "epoch": 5.12,
+      "learning_rate": 8.976785714285715e-05,
+      "loss": 1.3693,
       "step": 1146
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 7.73767258382643e-05,
-      "loss": 1.6674,
+      "epoch": 5.12,
+      "learning_rate": 8.975892857142857e-05,
+      "loss": 1.3538,
       "step": 1147
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 7.73570019723866e-05,
-      "loss": 1.7256,
+      "epoch": 5.12,
+      "learning_rate": 8.975e-05,
+      "loss": 1.3953,
       "step": 1148
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 7.733727810650889e-05,
-      "loss": 1.7153,
+      "epoch": 5.13,
+      "learning_rate": 8.974107142857144e-05,
+      "loss": 1.3775,
       "step": 1149
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 7.731755424063116e-05,
-      "loss": 1.7559,
+      "epoch": 5.13,
+      "learning_rate": 8.973214285714286e-05,
+      "loss": 1.294,
       "step": 1150
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 7.729783037475346e-05,
-      "loss": 1.7951,
+      "epoch": 5.14,
+      "learning_rate": 8.972321428571429e-05,
+      "loss": 1.3277,
       "step": 1151
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 7.727810650887575e-05,
-      "loss": 1.743,
+      "epoch": 5.14,
+      "learning_rate": 8.971428571428571e-05,
+      "loss": 1.3631,
       "step": 1152
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 7.725838264299803e-05,
-      "loss": 1.6439,
+      "epoch": 5.15,
+      "learning_rate": 8.970535714285715e-05,
+      "loss": 1.4032,
       "step": 1153
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 7.723865877712032e-05,
-      "loss": 1.7145,
+      "epoch": 5.15,
+      "learning_rate": 8.969642857142858e-05,
+      "loss": 1.3367,
       "step": 1154
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 7.72189349112426e-05,
-      "loss": 1.7168,
+      "epoch": 5.16,
+      "learning_rate": 8.96875e-05,
+      "loss": 1.3577,
       "step": 1155
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 7.719921104536489e-05,
-      "loss": 1.6069,
+      "epoch": 5.16,
+      "learning_rate": 8.967857142857143e-05,
+      "loss": 1.3456,
       "step": 1156
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 7.717948717948718e-05,
-      "loss": 1.7337,
+      "epoch": 5.17,
+      "learning_rate": 8.966964285714287e-05,
+      "loss": 1.4049,
       "step": 1157
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 7.715976331360946e-05,
-      "loss": 1.6981,
+      "epoch": 5.17,
+      "learning_rate": 8.966071428571428e-05,
+      "loss": 1.3713,
       "step": 1158
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 7.714003944773177e-05,
-      "loss": 1.777,
+      "epoch": 5.17,
+      "learning_rate": 8.965178571428572e-05,
+      "loss": 1.3356,
       "step": 1159
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 7.712031558185406e-05,
-      "loss": 1.8702,
+      "epoch": 5.18,
+      "learning_rate": 8.964285714285715e-05,
+      "loss": 1.2692,
       "step": 1160
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 7.710059171597634e-05,
-      "loss": 1.8264,
+      "epoch": 5.18,
+      "learning_rate": 8.963392857142857e-05,
+      "loss": 1.3269,
       "step": 1161
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 7.708086785009863e-05,
-      "loss": 1.7243,
+      "epoch": 5.19,
+      "learning_rate": 8.962500000000001e-05,
+      "loss": 1.2851,
       "step": 1162
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 7.706114398422091e-05,
-      "loss": 1.7814,
+      "epoch": 5.19,
+      "learning_rate": 8.961607142857144e-05,
+      "loss": 1.3317,
       "step": 1163
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 7.70414201183432e-05,
-      "loss": 1.7619,
+      "epoch": 5.2,
+      "learning_rate": 8.960714285714286e-05,
+      "loss": 1.3696,
       "step": 1164
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 7.702169625246549e-05,
-      "loss": 1.7174,
+      "epoch": 5.2,
+      "learning_rate": 8.959821428571429e-05,
+      "loss": 1.3093,
       "step": 1165
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 7.700197238658777e-05,
-      "loss": 1.7622,
+      "epoch": 5.21,
+      "learning_rate": 8.958928571428573e-05,
+      "loss": 1.4071,
       "step": 1166
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 7.698224852071006e-05,
-      "loss": 1.8018,
+      "epoch": 5.21,
+      "learning_rate": 8.958035714285715e-05,
+      "loss": 1.3208,
       "step": 1167
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 7.696252465483234e-05,
-      "loss": 1.7336,
+      "epoch": 5.21,
+      "learning_rate": 8.957142857142858e-05,
+      "loss": 1.3614,
       "step": 1168
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 7.694280078895463e-05,
-      "loss": 1.7565,
+      "epoch": 5.22,
+      "learning_rate": 8.95625e-05,
+      "loss": 1.2751,
       "step": 1169
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 7.692307692307693e-05,
-      "loss": 1.8548,
+      "epoch": 5.22,
+      "learning_rate": 8.955357142857144e-05,
+      "loss": 1.3542,
       "step": 1170
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 7.690335305719922e-05,
-      "loss": 1.6861,
+      "epoch": 5.23,
+      "learning_rate": 8.954464285714285e-05,
+      "loss": 1.3391,
       "step": 1171
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 7.688362919132151e-05,
-      "loss": 1.6818,
+      "epoch": 5.23,
+      "learning_rate": 8.953571428571429e-05,
+      "loss": 1.2806,
       "step": 1172
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 7.686390532544379e-05,
-      "loss": 1.6572,
+      "epoch": 5.24,
+      "learning_rate": 8.952678571428572e-05,
+      "loss": 1.4267,
       "step": 1173
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 7.684418145956608e-05,
-      "loss": 1.8161,
+      "epoch": 5.24,
+      "learning_rate": 8.951785714285714e-05,
+      "loss": 1.3427,
       "step": 1174
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 7.682445759368837e-05,
-      "loss": 1.6673,
+      "epoch": 5.25,
+      "learning_rate": 8.950892857142857e-05,
+      "loss": 1.3081,
       "step": 1175
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 7.680473372781065e-05,
-      "loss": 1.7914,
+      "epoch": 5.25,
+      "learning_rate": 8.950000000000001e-05,
+      "loss": 1.4305,
       "step": 1176
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 7.678500986193294e-05,
-      "loss": 1.7063,
+      "epoch": 5.25,
+      "learning_rate": 8.949107142857143e-05,
+      "loss": 1.396,
       "step": 1177
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 7.676528599605522e-05,
-      "loss": 1.7158,
+      "epoch": 5.26,
+      "learning_rate": 8.948214285714286e-05,
+      "loss": 1.3326,
       "step": 1178
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 7.674556213017751e-05,
-      "loss": 1.743,
+      "epoch": 5.26,
+      "learning_rate": 8.94732142857143e-05,
+      "loss": 1.3219,
       "step": 1179
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 7.67258382642998e-05,
-      "loss": 1.8005,
+      "epoch": 5.27,
+      "learning_rate": 8.946428571428572e-05,
+      "loss": 1.2972,
       "step": 1180
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 7.670611439842208e-05,
-      "loss": 1.7206,
+      "epoch": 5.27,
+      "learning_rate": 8.945535714285715e-05,
+      "loss": 1.4189,
       "step": 1181
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 7.668639053254438e-05,
-      "loss": 1.7477,
+      "epoch": 5.28,
+      "learning_rate": 8.944642857142857e-05,
+      "loss": 1.3569,
       "step": 1182
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 7.666666666666667e-05,
-      "loss": 1.8617,
+      "epoch": 5.28,
+      "learning_rate": 8.943750000000001e-05,
+      "loss": 1.2922,
       "step": 1183
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 7.664694280078896e-05,
-      "loss": 1.7711,
+      "epoch": 5.29,
+      "learning_rate": 8.942857142857142e-05,
+      "loss": 1.3403,
       "step": 1184
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 7.662721893491125e-05,
-      "loss": 1.7947,
+      "epoch": 5.29,
+      "learning_rate": 8.941964285714286e-05,
+      "loss": 1.3067,
       "step": 1185
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 7.660749506903353e-05,
-      "loss": 1.6845,
+      "epoch": 5.29,
+      "learning_rate": 8.941071428571429e-05,
+      "loss": 1.2642,
       "step": 1186
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 7.658777120315582e-05,
-      "loss": 1.7788,
+      "epoch": 5.3,
+      "learning_rate": 8.940178571428573e-05,
+      "loss": 1.2583,
       "step": 1187
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 7.656804733727812e-05,
-      "loss": 1.7097,
+      "epoch": 5.3,
+      "learning_rate": 8.939285714285714e-05,
+      "loss": 1.3495,
       "step": 1188
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 7.65483234714004e-05,
-      "loss": 1.6946,
+      "epoch": 5.31,
+      "learning_rate": 8.938392857142858e-05,
+      "loss": 1.3563,
       "step": 1189
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 7.652859960552269e-05,
-      "loss": 1.7359,
+      "epoch": 5.31,
+      "learning_rate": 8.9375e-05,
+      "loss": 1.3273,
       "step": 1190
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 7.650887573964497e-05,
-      "loss": 1.841,
+      "epoch": 5.32,
+      "learning_rate": 8.936607142857143e-05,
+      "loss": 1.3176,
       "step": 1191
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 7.648915187376726e-05,
-      "loss": 1.6834,
+      "epoch": 5.32,
+      "learning_rate": 8.935714285714285e-05,
+      "loss": 1.2764,
       "step": 1192
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 7.646942800788955e-05,
-      "loss": 1.8451,
+      "epoch": 5.33,
+      "learning_rate": 8.93482142857143e-05,
+      "loss": 1.4061,
       "step": 1193
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 7.644970414201184e-05,
-      "loss": 1.6765,
+      "epoch": 5.33,
+      "learning_rate": 8.933928571428572e-05,
+      "loss": 1.2528,
       "step": 1194
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 7.642998027613413e-05,
-      "loss": 1.6993,
+      "epoch": 5.33,
+      "learning_rate": 8.933035714285714e-05,
+      "loss": 1.2912,
       "step": 1195
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 7.641025641025641e-05,
-      "loss": 1.7287,
+      "epoch": 5.34,
+      "learning_rate": 8.932142857142858e-05,
+      "loss": 1.3532,
       "step": 1196
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 7.63905325443787e-05,
-      "loss": 1.6839,
+      "epoch": 5.34,
+      "learning_rate": 8.93125e-05,
+      "loss": 1.2985,
       "step": 1197
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 7.6370808678501e-05,
-      "loss": 1.6964,
+      "epoch": 5.35,
+      "learning_rate": 8.930357142857143e-05,
+      "loss": 1.3191,
       "step": 1198
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 7.635108481262328e-05,
-      "loss": 1.6909,
+      "epoch": 5.35,
+      "learning_rate": 8.929464285714286e-05,
+      "loss": 1.3183,
       "step": 1199
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 7.633136094674557e-05,
-      "loss": 1.7666,
+      "epoch": 5.36,
+      "learning_rate": 8.92857142857143e-05,
+      "loss": 1.3221,
       "step": 1200
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 7.631163708086785e-05,
-      "loss": 1.807,
+      "epoch": 5.36,
+      "eval_loss": 1.2428117990493774,
+      "eval_runtime": 47.0864,
+      "eval_samples_per_second": 278.085,
+      "eval_steps_per_second": 1.465,
+      "step": 1200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.927678571428571e-05,
+      "loss": 1.3079,
       "step": 1201
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 7.629191321499014e-05,
-      "loss": 1.6543,
+      "epoch": 5.37,
+      "learning_rate": 8.926785714285715e-05,
+      "loss": 1.4003,
       "step": 1202
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 7.627218934911243e-05,
-      "loss": 1.7786,
+      "epoch": 5.37,
+      "learning_rate": 8.925892857142858e-05,
+      "loss": 1.3126,
       "step": 1203
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 7.625246548323471e-05,
-      "loss": 1.6372,
+      "epoch": 5.38,
+      "learning_rate": 8.925e-05,
+      "loss": 1.2915,
       "step": 1204
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 7.6232741617357e-05,
-      "loss": 1.733,
+      "epoch": 5.38,
+      "learning_rate": 8.924107142857143e-05,
+      "loss": 1.3174,
       "step": 1205
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 7.621301775147929e-05,
-      "loss": 1.7511,
+      "epoch": 5.38,
+      "learning_rate": 8.923214285714287e-05,
+      "loss": 1.4423,
       "step": 1206
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 7.619329388560158e-05,
-      "loss": 1.7956,
+      "epoch": 5.39,
+      "learning_rate": 8.922321428571429e-05,
+      "loss": 1.3327,
       "step": 1207
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 7.617357001972388e-05,
-      "loss": 1.7664,
+      "epoch": 5.39,
+      "learning_rate": 8.921428571428572e-05,
+      "loss": 1.2859,
       "step": 1208
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 7.615384615384616e-05,
-      "loss": 1.7942,
+      "epoch": 5.4,
+      "learning_rate": 8.920535714285716e-05,
+      "loss": 1.2904,
       "step": 1209
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 7.613412228796845e-05,
-      "loss": 1.6379,
+      "epoch": 5.4,
+      "learning_rate": 8.919642857142858e-05,
+      "loss": 1.2771,
       "step": 1210
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 7.611439842209074e-05,
-      "loss": 1.7422,
+      "epoch": 5.41,
+      "learning_rate": 8.91875e-05,
+      "loss": 1.2925,
       "step": 1211
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 7.609467455621302e-05,
-      "loss": 1.6734,
+      "epoch": 5.41,
+      "learning_rate": 8.917857142857143e-05,
+      "loss": 1.2105,
       "step": 1212
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 7.607495069033531e-05,
-      "loss": 1.762,
+      "epoch": 5.42,
+      "learning_rate": 8.916964285714287e-05,
+      "loss": 1.3622,
       "step": 1213
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 7.605522682445759e-05,
-      "loss": 1.7996,
+      "epoch": 5.42,
+      "learning_rate": 8.916071428571428e-05,
+      "loss": 1.3792,
       "step": 1214
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 7.603550295857988e-05,
-      "loss": 1.6864,
+      "epoch": 5.42,
+      "learning_rate": 8.915178571428572e-05,
+      "loss": 1.3658,
       "step": 1215
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 7.601577909270217e-05,
-      "loss": 1.8164,
+      "epoch": 5.43,
+      "learning_rate": 8.914285714285715e-05,
+      "loss": 1.2954,
       "step": 1216
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 7.599605522682447e-05,
-      "loss": 1.7541,
+      "epoch": 5.43,
+      "learning_rate": 8.913392857142857e-05,
+      "loss": 1.2841,
       "step": 1217
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 7.597633136094676e-05,
-      "loss": 1.7381,
+      "epoch": 5.44,
+      "learning_rate": 8.9125e-05,
+      "loss": 1.2946,
       "step": 1218
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 7.595660749506904e-05,
-      "loss": 1.6901,
+      "epoch": 5.44,
+      "learning_rate": 8.911607142857144e-05,
+      "loss": 1.2697,
       "step": 1219
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 7.593688362919133e-05,
-      "loss": 1.7559,
+      "epoch": 5.45,
+      "learning_rate": 8.910714285714286e-05,
+      "loss": 1.3711,
       "step": 1220
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 7.591715976331362e-05,
-      "loss": 1.7055,
+      "epoch": 5.45,
+      "learning_rate": 8.909821428571429e-05,
+      "loss": 1.2988,
       "step": 1221
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 7.58974358974359e-05,
-      "loss": 1.7109,
+      "epoch": 5.46,
+      "learning_rate": 8.908928571428571e-05,
+      "loss": 1.3764,
       "step": 1222
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 7.587771203155819e-05,
-      "loss": 1.7121,
+      "epoch": 5.46,
+      "learning_rate": 8.908035714285715e-05,
+      "loss": 1.236,
       "step": 1223
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 7.585798816568048e-05,
-      "loss": 1.6606,
+      "epoch": 5.46,
+      "learning_rate": 8.907142857142858e-05,
+      "loss": 1.3678,
       "step": 1224
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 7.583826429980276e-05,
-      "loss": 1.6536,
+      "epoch": 5.47,
+      "learning_rate": 8.90625e-05,
+      "loss": 1.3443,
       "step": 1225
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 7.581854043392505e-05,
-      "loss": 1.7262,
+      "epoch": 5.47,
+      "learning_rate": 8.905357142857144e-05,
+      "loss": 1.3585,
       "step": 1226
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 7.579881656804733e-05,
-      "loss": 1.7705,
+      "epoch": 5.48,
+      "learning_rate": 8.904464285714285e-05,
+      "loss": 1.4031,
       "step": 1227
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 7.577909270216962e-05,
-      "loss": 1.7405,
+      "epoch": 5.48,
+      "learning_rate": 8.903571428571429e-05,
+      "loss": 1.3535,
       "step": 1228
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 7.575936883629192e-05,
-      "loss": 1.6101,
+      "epoch": 5.49,
+      "learning_rate": 8.902678571428572e-05,
+      "loss": 1.353,
       "step": 1229
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 7.573964497041421e-05,
-      "loss": 1.6613,
+      "epoch": 5.49,
+      "learning_rate": 8.901785714285714e-05,
+      "loss": 1.3122,
       "step": 1230
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 7.57199211045365e-05,
-      "loss": 1.8146,
+      "epoch": 5.5,
+      "learning_rate": 8.900892857142857e-05,
+      "loss": 1.2815,
       "step": 1231
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 7.570019723865878e-05,
-      "loss": 1.8195,
+      "epoch": 5.5,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 1.2997,
       "step": 1232
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 7.568047337278107e-05,
-      "loss": 1.6891,
+      "epoch": 5.5,
+      "learning_rate": 8.899107142857143e-05,
+      "loss": 1.2069,
       "step": 1233
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 7.566074950690336e-05,
-      "loss": 1.7344,
+      "epoch": 5.51,
+      "learning_rate": 8.898214285714286e-05,
+      "loss": 1.4656,
       "step": 1234
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 7.564102564102564e-05,
-      "loss": 1.768,
+      "epoch": 5.51,
+      "learning_rate": 8.897321428571428e-05,
+      "loss": 1.3941,
       "step": 1235
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 7.562130177514793e-05,
-      "loss": 1.7537,
+      "epoch": 5.52,
+      "learning_rate": 8.896428571428572e-05,
+      "loss": 1.3433,
       "step": 1236
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 7.560157790927021e-05,
-      "loss": 1.7904,
+      "epoch": 5.52,
+      "learning_rate": 8.895535714285715e-05,
+      "loss": 1.3834,
       "step": 1237
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 7.55818540433925e-05,
-      "loss": 1.6626,
+      "epoch": 5.53,
+      "learning_rate": 8.894642857142857e-05,
+      "loss": 1.3224,
       "step": 1238
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 7.55621301775148e-05,
-      "loss": 1.6876,
+      "epoch": 5.53,
+      "learning_rate": 8.89375e-05,
+      "loss": 1.3678,
       "step": 1239
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 7.554240631163708e-05,
-      "loss": 1.7696,
+      "epoch": 5.54,
+      "learning_rate": 8.892857142857143e-05,
+      "loss": 1.4096,
       "step": 1240
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 7.552268244575937e-05,
-      "loss": 1.7005,
+      "epoch": 5.54,
+      "learning_rate": 8.891964285714286e-05,
+      "loss": 1.3703,
       "step": 1241
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 7.550295857988166e-05,
-      "loss": 1.6214,
+      "epoch": 5.54,
+      "learning_rate": 8.891071428571429e-05,
+      "loss": 1.3608,
       "step": 1242
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 7.548323471400395e-05,
-      "loss": 1.8126,
+      "epoch": 5.55,
+      "learning_rate": 8.890178571428573e-05,
+      "loss": 1.3392,
       "step": 1243
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 7.546351084812624e-05,
-      "loss": 1.7336,
+      "epoch": 5.55,
+      "learning_rate": 8.889285714285714e-05,
+      "loss": 1.3858,
       "step": 1244
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 7.544378698224852e-05,
-      "loss": 1.7196,
+      "epoch": 5.56,
+      "learning_rate": 8.888392857142858e-05,
+      "loss": 1.4167,
       "step": 1245
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 7.542406311637081e-05,
-      "loss": 1.6762,
+      "epoch": 5.56,
+      "learning_rate": 8.8875e-05,
+      "loss": 1.3517,
       "step": 1246
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 7.54043392504931e-05,
-      "loss": 1.7382,
+      "epoch": 5.57,
+      "learning_rate": 8.886607142857143e-05,
+      "loss": 1.3692,
       "step": 1247
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 7.538461538461539e-05,
-      "loss": 1.8384,
+      "epoch": 5.57,
+      "learning_rate": 8.885714285714286e-05,
+      "loss": 1.2646,
       "step": 1248
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 7.536489151873768e-05,
-      "loss": 1.6117,
+      "epoch": 5.58,
+      "learning_rate": 8.88482142857143e-05,
+      "loss": 1.3358,
       "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,
+      "epoch": 5.58,
+      "learning_rate": 8.883928571428571e-05,
+      "loss": 1.2838,
       "step": 1250
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 7.532544378698225e-05,
-      "loss": 1.7283,
+      "epoch": 5.58,
+      "learning_rate": 8.883035714285715e-05,
+      "loss": 1.2873,
       "step": 1251
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 7.530571992110454e-05,
-      "loss": 1.7611,
+      "epoch": 5.59,
+      "learning_rate": 8.882142857142857e-05,
+      "loss": 1.2486,
       "step": 1252
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 7.528599605522683e-05,
-      "loss": 1.7252,
+      "epoch": 5.59,
+      "learning_rate": 8.881250000000001e-05,
+      "loss": 1.3283,
       "step": 1253
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 7.526627218934912e-05,
-      "loss": 1.7686,
+      "epoch": 5.6,
+      "learning_rate": 8.880357142857144e-05,
+      "loss": 1.3636,
       "step": 1254
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 7.52465483234714e-05,
-      "loss": 1.8094,
+      "epoch": 5.6,
+      "learning_rate": 8.879464285714286e-05,
+      "loss": 1.3071,
       "step": 1255
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 7.52268244575937e-05,
-      "loss": 1.7976,
+      "epoch": 5.61,
+      "learning_rate": 8.87857142857143e-05,
+      "loss": 1.3719,
       "step": 1256
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 7.520710059171599e-05,
-      "loss": 1.7618,
+      "epoch": 5.61,
+      "learning_rate": 8.877678571428571e-05,
+      "loss": 1.3633,
       "step": 1257
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 7.518737672583827e-05,
-      "loss": 1.8102,
+      "epoch": 5.62,
+      "learning_rate": 8.876785714285715e-05,
+      "loss": 1.3057,
       "step": 1258
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 7.516765285996056e-05,
-      "loss": 1.7747,
+      "epoch": 5.62,
+      "learning_rate": 8.875892857142858e-05,
+      "loss": 1.3251,
       "step": 1259
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 7.514792899408284e-05,
-      "loss": 1.6812,
+      "epoch": 5.62,
+      "learning_rate": 8.875e-05,
+      "loss": 1.3707,
       "step": 1260
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 7.512820512820513e-05,
-      "loss": 1.712,
+      "epoch": 5.63,
+      "learning_rate": 8.874107142857143e-05,
+      "loss": 1.3529,
       "step": 1261
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 7.510848126232742e-05,
-      "loss": 1.7454,
+      "epoch": 5.63,
+      "learning_rate": 8.873214285714287e-05,
+      "loss": 1.3511,
       "step": 1262
     },
     {
-      "epoch": 3.74,
-      "learning_rate": 7.50887573964497e-05,
-      "loss": 1.6957,
+      "epoch": 5.64,
+      "learning_rate": 8.872321428571428e-05,
+      "loss": 1.368,
       "step": 1263
     },
     {
-      "epoch": 3.74,
-      "learning_rate": 7.506903353057199e-05,
-      "loss": 1.6991,
+      "epoch": 5.64,
+      "learning_rate": 8.871428571428572e-05,
+      "loss": 1.3895,
       "step": 1264
     },
     {
-      "epoch": 3.74,
-      "learning_rate": 7.504930966469428e-05,
-      "loss": 1.8196,
+      "epoch": 5.65,
+      "learning_rate": 8.870535714285714e-05,
+      "loss": 1.2695,
       "step": 1265
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 7.502958579881658e-05,
-      "loss": 1.6657,
+      "epoch": 5.65,
+      "learning_rate": 8.869642857142858e-05,
+      "loss": 1.3141,
       "step": 1266
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 7.500986193293887e-05,
-      "loss": 1.7202,
+      "epoch": 5.66,
+      "learning_rate": 8.868750000000001e-05,
+      "loss": 1.3022,
       "step": 1267
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 7.499013806706115e-05,
-      "loss": 1.7254,
+      "epoch": 5.66,
+      "learning_rate": 8.867857142857143e-05,
+      "loss": 1.3306,
       "step": 1268
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 7.497041420118344e-05,
-      "loss": 1.6749,
+      "epoch": 5.67,
+      "learning_rate": 8.866964285714286e-05,
+      "loss": 1.3946,
       "step": 1269
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 7.495069033530573e-05,
-      "loss": 1.7471,
+      "epoch": 5.67,
+      "learning_rate": 8.866071428571428e-05,
+      "loss": 1.2369,
       "step": 1270
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 7.493096646942801e-05,
-      "loss": 1.7241,
+      "epoch": 5.67,
+      "learning_rate": 8.865178571428572e-05,
+      "loss": 1.3087,
       "step": 1271
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 7.49112426035503e-05,
-      "loss": 1.6611,
+      "epoch": 5.68,
+      "learning_rate": 8.864285714285715e-05,
+      "loss": 1.2523,
       "step": 1272
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 7.489151873767258e-05,
-      "loss": 1.7788,
+      "epoch": 5.68,
+      "learning_rate": 8.863392857142857e-05,
+      "loss": 1.3203,
       "step": 1273
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 7.487179487179487e-05,
-      "loss": 1.8294,
+      "epoch": 5.69,
+      "learning_rate": 8.8625e-05,
+      "loss": 1.3965,
       "step": 1274
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 7.485207100591716e-05,
-      "loss": 1.6911,
+      "epoch": 5.69,
+      "learning_rate": 8.861607142857144e-05,
+      "loss": 1.2964,
       "step": 1275
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 7.483234714003946e-05,
-      "loss": 1.8059,
+      "epoch": 5.7,
+      "learning_rate": 8.860714285714286e-05,
+      "loss": 1.3292,
       "step": 1276
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 7.481262327416175e-05,
-      "loss": 1.739,
+      "epoch": 5.7,
+      "learning_rate": 8.859821428571429e-05,
+      "loss": 1.3425,
       "step": 1277
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 7.479289940828403e-05,
-      "loss": 1.6999,
+      "epoch": 5.71,
+      "learning_rate": 8.858928571428571e-05,
+      "loss": 1.3862,
       "step": 1278
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 7.477317554240632e-05,
-      "loss": 1.7439,
+      "epoch": 5.71,
+      "learning_rate": 8.858035714285715e-05,
+      "loss": 1.2401,
       "step": 1279
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 7.475345167652861e-05,
-      "loss": 1.7018,
+      "epoch": 5.71,
+      "learning_rate": 8.857142857142857e-05,
+      "loss": 1.39,
       "step": 1280
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 7.473372781065089e-05,
-      "loss": 1.7103,
+      "epoch": 5.72,
+      "learning_rate": 8.85625e-05,
+      "loss": 1.3249,
       "step": 1281
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 7.471400394477318e-05,
-      "loss": 1.7542,
+      "epoch": 5.72,
+      "learning_rate": 8.855357142857143e-05,
+      "loss": 1.312,
       "step": 1282
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 7.469428007889546e-05,
-      "loss": 1.7068,
+      "epoch": 5.73,
+      "learning_rate": 8.854464285714286e-05,
+      "loss": 1.3731,
       "step": 1283
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 7.467455621301775e-05,
-      "loss": 1.7451,
+      "epoch": 5.73,
+      "learning_rate": 8.85357142857143e-05,
+      "loss": 1.3351,
       "step": 1284
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 7.465483234714004e-05,
-      "loss": 1.7675,
+      "epoch": 5.74,
+      "learning_rate": 8.852678571428572e-05,
+      "loss": 1.3187,
       "step": 1285
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 7.463510848126232e-05,
-      "loss": 1.597,
+      "epoch": 5.74,
+      "learning_rate": 8.851785714285716e-05,
+      "loss": 1.3433,
       "step": 1286
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 7.461538461538462e-05,
-      "loss": 1.7752,
+      "epoch": 5.75,
+      "learning_rate": 8.850892857142857e-05,
+      "loss": 1.3844,
       "step": 1287
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 7.459566074950691e-05,
-      "loss": 1.8056,
+      "epoch": 5.75,
+      "learning_rate": 8.850000000000001e-05,
+      "loss": 1.3732,
       "step": 1288
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 7.45759368836292e-05,
-      "loss": 1.7452,
+      "epoch": 5.75,
+      "learning_rate": 8.849107142857144e-05,
+      "loss": 1.2976,
       "step": 1289
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 7.455621301775149e-05,
-      "loss": 1.7858,
+      "epoch": 5.76,
+      "learning_rate": 8.848214285714286e-05,
+      "loss": 1.326,
       "step": 1290
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 7.453648915187377e-05,
-      "loss": 1.772,
+      "epoch": 5.76,
+      "learning_rate": 8.847321428571429e-05,
+      "loss": 1.3656,
       "step": 1291
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 7.451676528599606e-05,
-      "loss": 1.7693,
+      "epoch": 5.77,
+      "learning_rate": 8.846428571428573e-05,
+      "loss": 1.3712,
       "step": 1292
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 7.449704142011835e-05,
-      "loss": 1.7307,
+      "epoch": 5.77,
+      "learning_rate": 8.845535714285714e-05,
+      "loss": 1.3667,
       "step": 1293
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 7.447731755424063e-05,
-      "loss": 1.6464,
+      "epoch": 5.78,
+      "learning_rate": 8.844642857142858e-05,
+      "loss": 1.4675,
       "step": 1294
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 7.445759368836292e-05,
-      "loss": 1.6068,
+      "epoch": 5.78,
+      "learning_rate": 8.84375e-05,
+      "loss": 1.3054,
       "step": 1295
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 7.44378698224852e-05,
-      "loss": 1.7304,
+      "epoch": 5.79,
+      "learning_rate": 8.842857142857143e-05,
+      "loss": 1.2681,
       "step": 1296
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 7.44181459566075e-05,
-      "loss": 1.6769,
+      "epoch": 5.79,
+      "learning_rate": 8.841964285714285e-05,
+      "loss": 1.3848,
       "step": 1297
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 7.439842209072979e-05,
-      "loss": 1.675,
+      "epoch": 5.79,
+      "learning_rate": 8.841071428571429e-05,
+      "loss": 1.3673,
       "step": 1298
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 7.437869822485207e-05,
-      "loss": 1.6344,
+      "epoch": 5.8,
+      "learning_rate": 8.840178571428572e-05,
+      "loss": 1.3014,
       "step": 1299
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 7.435897435897436e-05,
-      "loss": 1.6211,
+      "epoch": 5.8,
+      "learning_rate": 8.839285714285714e-05,
+      "loss": 1.3327,
       "step": 1300
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 7.433925049309665e-05,
-      "loss": 1.6819,
+      "epoch": 5.81,
+      "learning_rate": 8.838392857142858e-05,
+      "loss": 1.3035,
       "step": 1301
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 7.431952662721894e-05,
-      "loss": 1.7593,
+      "epoch": 5.81,
+      "learning_rate": 8.837500000000001e-05,
+      "loss": 1.3575,
       "step": 1302
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 7.429980276134123e-05,
-      "loss": 1.7579,
+      "epoch": 5.82,
+      "learning_rate": 8.836607142857143e-05,
+      "loss": 1.2764,
       "step": 1303
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 7.428007889546351e-05,
-      "loss": 1.7311,
+      "epoch": 5.82,
+      "learning_rate": 8.835714285714286e-05,
+      "loss": 1.4501,
       "step": 1304
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 7.42603550295858e-05,
-      "loss": 1.6763,
+      "epoch": 5.83,
+      "learning_rate": 8.83482142857143e-05,
+      "loss": 1.3683,
       "step": 1305
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 7.424063116370808e-05,
-      "loss": 1.7589,
+      "epoch": 5.83,
+      "learning_rate": 8.833928571428571e-05,
+      "loss": 1.3292,
       "step": 1306
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 7.422090729783038e-05,
-      "loss": 1.6833,
+      "epoch": 5.83,
+      "learning_rate": 8.833035714285715e-05,
+      "loss": 1.2511,
       "step": 1307
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 7.420118343195267e-05,
-      "loss": 1.7089,
+      "epoch": 5.84,
+      "learning_rate": 8.832142857142857e-05,
+      "loss": 1.3574,
       "step": 1308
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 7.418145956607495e-05,
-      "loss": 1.7003,
+      "epoch": 5.84,
+      "learning_rate": 8.831250000000001e-05,
+      "loss": 1.3292,
       "step": 1309
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 7.416173570019724e-05,
-      "loss": 1.7385,
+      "epoch": 5.85,
+      "learning_rate": 8.830357142857142e-05,
+      "loss": 1.2956,
       "step": 1310
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 7.414201183431953e-05,
-      "loss": 1.6371,
+      "epoch": 5.85,
+      "learning_rate": 8.829464285714286e-05,
+      "loss": 1.3009,
       "step": 1311
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 7.412228796844182e-05,
-      "loss": 1.8231,
+      "epoch": 5.86,
+      "learning_rate": 8.828571428571429e-05,
+      "loss": 1.3387,
       "step": 1312
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 7.410256410256412e-05,
-      "loss": 1.7619,
+      "epoch": 5.86,
+      "learning_rate": 8.827678571428571e-05,
+      "loss": 1.3397,
       "step": 1313
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 7.40828402366864e-05,
-      "loss": 1.8075,
+      "epoch": 5.87,
+      "learning_rate": 8.826785714285715e-05,
+      "loss": 1.3005,
       "step": 1314
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 7.406311637080869e-05,
-      "loss": 1.6752,
+      "epoch": 5.87,
+      "learning_rate": 8.825892857142858e-05,
+      "loss": 1.3181,
       "step": 1315
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 7.404339250493098e-05,
-      "loss": 1.8428,
+      "epoch": 5.88,
+      "learning_rate": 8.825e-05,
+      "loss": 1.3058,
       "step": 1316
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 7.402366863905326e-05,
-      "loss": 1.7997,
+      "epoch": 5.88,
+      "learning_rate": 8.824107142857143e-05,
+      "loss": 1.3646,
       "step": 1317
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 7.400394477317555e-05,
-      "loss": 1.644,
+      "epoch": 5.88,
+      "learning_rate": 8.823214285714287e-05,
+      "loss": 1.3002,
       "step": 1318
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 7.398422090729783e-05,
-      "loss": 1.6914,
+      "epoch": 5.89,
+      "learning_rate": 8.82232142857143e-05,
+      "loss": 1.3388,
       "step": 1319
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 7.396449704142012e-05,
-      "loss": 1.7578,
+      "epoch": 5.89,
+      "learning_rate": 8.821428571428572e-05,
+      "loss": 1.2696,
       "step": 1320
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 7.394477317554241e-05,
-      "loss": 1.7929,
+      "epoch": 5.9,
+      "learning_rate": 8.820535714285714e-05,
+      "loss": 1.3874,
       "step": 1321
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 7.392504930966469e-05,
-      "loss": 1.699,
+      "epoch": 5.9,
+      "learning_rate": 8.819642857142858e-05,
+      "loss": 1.2757,
       "step": 1322
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 7.390532544378698e-05,
-      "loss": 1.6683,
+      "epoch": 5.91,
+      "learning_rate": 8.81875e-05,
+      "loss": 1.3629,
       "step": 1323
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 7.388560157790927e-05,
-      "loss": 1.7566,
+      "epoch": 5.91,
+      "learning_rate": 8.817857142857143e-05,
+      "loss": 1.3221,
       "step": 1324
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 7.386587771203157e-05,
-      "loss": 1.7643,
+      "epoch": 5.92,
+      "learning_rate": 8.816964285714286e-05,
+      "loss": 1.3916,
       "step": 1325
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 7.384615384615386e-05,
-      "loss": 1.6878,
+      "epoch": 5.92,
+      "learning_rate": 8.816071428571429e-05,
+      "loss": 1.3579,
       "step": 1326
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 7.382642998027614e-05,
-      "loss": 1.7265,
+      "epoch": 5.92,
+      "learning_rate": 8.815178571428571e-05,
+      "loss": 1.4067,
       "step": 1327
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 7.380670611439843e-05,
-      "loss": 1.6404,
+      "epoch": 5.93,
+      "learning_rate": 8.814285714285715e-05,
+      "loss": 1.3272,
       "step": 1328
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 7.378698224852071e-05,
-      "loss": 1.6106,
+      "epoch": 5.93,
+      "learning_rate": 8.813392857142858e-05,
+      "loss": 1.4015,
       "step": 1329
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 7.3767258382643e-05,
-      "loss": 1.6446,
+      "epoch": 5.94,
+      "learning_rate": 8.8125e-05,
+      "loss": 1.2972,
       "step": 1330
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 7.374753451676529e-05,
-      "loss": 1.6853,
+      "epoch": 5.94,
+      "learning_rate": 8.811607142857144e-05,
+      "loss": 1.3409,
       "step": 1331
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 7.372781065088757e-05,
-      "loss": 1.6794,
+      "epoch": 5.95,
+      "learning_rate": 8.810714285714287e-05,
+      "loss": 1.3372,
       "step": 1332
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 7.370808678500986e-05,
-      "loss": 1.7357,
+      "epoch": 5.95,
+      "learning_rate": 8.809821428571429e-05,
+      "loss": 1.3754,
       "step": 1333
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 7.368836291913215e-05,
-      "loss": 1.7443,
+      "epoch": 5.96,
+      "learning_rate": 8.808928571428572e-05,
+      "loss": 1.3625,
       "step": 1334
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 7.366863905325445e-05,
-      "loss": 1.702,
+      "epoch": 5.96,
+      "learning_rate": 8.808035714285716e-05,
+      "loss": 1.3902,
       "step": 1335
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 7.364891518737674e-05,
-      "loss": 1.6853,
+      "epoch": 5.96,
+      "learning_rate": 8.807142857142857e-05,
+      "loss": 1.2713,
       "step": 1336
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 7.362919132149902e-05,
-      "loss": 1.6812,
+      "epoch": 5.97,
+      "learning_rate": 8.80625e-05,
+      "loss": 1.333,
       "step": 1337
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 7.360946745562131e-05,
-      "loss": 1.6566,
+      "epoch": 5.97,
+      "learning_rate": 8.805357142857143e-05,
+      "loss": 1.3449,
       "step": 1338
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 7.35897435897436e-05,
-      "loss": 1.7234,
+      "epoch": 5.98,
+      "learning_rate": 8.804464285714286e-05,
+      "loss": 1.3223,
       "step": 1339
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 7.357001972386588e-05,
-      "loss": 1.6996,
+      "epoch": 5.98,
+      "learning_rate": 8.803571428571428e-05,
+      "loss": 1.318,
       "step": 1340
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 7.355029585798817e-05,
-      "loss": 1.7073,
+      "epoch": 5.99,
+      "learning_rate": 8.802678571428572e-05,
+      "loss": 1.2575,
       "step": 1341
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 7.353057199211045e-05,
-      "loss": 1.692,
+      "epoch": 5.99,
+      "learning_rate": 8.801785714285715e-05,
+      "loss": 1.3253,
       "step": 1342
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 7.351084812623274e-05,
-      "loss": 1.6757,
+      "epoch": 6.0,
+      "learning_rate": 8.800892857142857e-05,
+      "loss": 1.3112,
       "step": 1343
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 7.349112426035504e-05,
-      "loss": 1.67,
+      "epoch": 6.0,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.3623,
       "step": 1344
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 7.347140039447731e-05,
-      "loss": 1.7981,
+      "epoch": 6.0,
+      "learning_rate": 8.799107142857144e-05,
+      "loss": 1.2699,
       "step": 1345
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 7.34516765285996e-05,
-      "loss": 1.7532,
+      "epoch": 6.01,
+      "learning_rate": 8.798214285714286e-05,
+      "loss": 1.3158,
       "step": 1346
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 7.34319526627219e-05,
-      "loss": 1.8358,
+      "epoch": 6.01,
+      "learning_rate": 8.797321428571429e-05,
+      "loss": 1.3174,
       "step": 1347
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 7.341222879684419e-05,
-      "loss": 1.6859,
+      "epoch": 6.02,
+      "learning_rate": 8.796428571428573e-05,
+      "loss": 1.3546,
       "step": 1348
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 7.339250493096648e-05,
-      "loss": 1.7356,
+      "epoch": 6.02,
+      "learning_rate": 8.795535714285714e-05,
+      "loss": 1.3079,
       "step": 1349
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 7.337278106508876e-05,
-      "loss": 1.7954,
+      "epoch": 6.03,
+      "learning_rate": 8.794642857142858e-05,
+      "loss": 1.3251,
       "step": 1350
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 7.335305719921105e-05,
-      "loss": 1.793,
+      "epoch": 6.03,
+      "learning_rate": 8.79375e-05,
+      "loss": 1.3366,
       "step": 1351
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 7.333333333333333e-05,
-      "loss": 1.6172,
+      "epoch": 6.04,
+      "learning_rate": 8.792857142857144e-05,
+      "loss": 1.3214,
       "step": 1352
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 7.331360946745562e-05,
-      "loss": 1.6901,
+      "epoch": 6.04,
+      "learning_rate": 8.791964285714285e-05,
+      "loss": 1.2826,
       "step": 1353
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.329388560157792e-05,
-      "loss": 1.698,
+      "epoch": 6.04,
+      "learning_rate": 8.791071428571429e-05,
+      "loss": 1.3233,
       "step": 1354
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.32741617357002e-05,
-      "loss": 1.7814,
+      "epoch": 6.05,
+      "learning_rate": 8.790178571428572e-05,
+      "loss": 1.2938,
       "step": 1355
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.325443786982249e-05,
-      "loss": 1.6698,
+      "epoch": 6.05,
+      "learning_rate": 8.789285714285714e-05,
+      "loss": 1.2921,
       "step": 1356
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.323471400394478e-05,
-      "loss": 1.6042,
+      "epoch": 6.06,
+      "learning_rate": 8.788392857142857e-05,
+      "loss": 1.2866,
       "step": 1357
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 7.321499013806706e-05,
-      "loss": 1.6576,
+      "epoch": 6.06,
+      "learning_rate": 8.787500000000001e-05,
+      "loss": 1.2465,
       "step": 1358
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 7.319526627218935e-05,
-      "loss": 1.8264,
+      "epoch": 6.07,
+      "learning_rate": 8.786607142857143e-05,
+      "loss": 1.2143,
       "step": 1359
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 7.317554240631164e-05,
-      "loss": 1.7129,
+      "epoch": 6.07,
+      "learning_rate": 8.785714285714286e-05,
+      "loss": 1.3789,
       "step": 1360
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 7.315581854043393e-05,
-      "loss": 1.7531,
+      "epoch": 6.08,
+      "learning_rate": 8.78482142857143e-05,
+      "loss": 1.3393,
       "step": 1361
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 7.313609467455623e-05,
-      "loss": 1.6028,
+      "epoch": 6.08,
+      "learning_rate": 8.783928571428572e-05,
+      "loss": 1.2385,
       "step": 1362
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 7.31163708086785e-05,
-      "loss": 1.7758,
+      "epoch": 6.08,
+      "learning_rate": 8.783035714285715e-05,
+      "loss": 1.2872,
       "step": 1363
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.30966469428008e-05,
-      "loss": 1.6812,
+      "epoch": 6.09,
+      "learning_rate": 8.782142857142857e-05,
+      "loss": 1.2677,
       "step": 1364
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.307692307692307e-05,
-      "loss": 1.7393,
+      "epoch": 6.09,
+      "learning_rate": 8.781250000000001e-05,
+      "loss": 1.429,
       "step": 1365
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.305719921104537e-05,
-      "loss": 1.6751,
+      "epoch": 6.1,
+      "learning_rate": 8.780357142857143e-05,
+      "loss": 1.2717,
       "step": 1366
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.303747534516766e-05,
-      "loss": 1.7712,
+      "epoch": 6.1,
+      "learning_rate": 8.779464285714286e-05,
+      "loss": 1.2881,
       "step": 1367
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 7.301775147928994e-05,
-      "loss": 1.7143,
+      "epoch": 6.11,
+      "learning_rate": 8.778571428571429e-05,
+      "loss": 1.3319,
       "step": 1368
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 7.299802761341223e-05,
-      "loss": 1.7282,
+      "epoch": 6.11,
+      "learning_rate": 8.777678571428572e-05,
+      "loss": 1.2851,
       "step": 1369
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 7.297830374753452e-05,
-      "loss": 1.7009,
+      "epoch": 6.12,
+      "learning_rate": 8.776785714285714e-05,
+      "loss": 1.3584,
       "step": 1370
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 7.295857988165681e-05,
-      "loss": 1.7833,
+      "epoch": 6.12,
+      "learning_rate": 8.775892857142858e-05,
+      "loss": 1.292,
       "step": 1371
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 7.29388560157791e-05,
-      "loss": 1.7464,
+      "epoch": 6.12,
+      "learning_rate": 8.775e-05,
+      "loss": 1.402,
       "step": 1372
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 7.291913214990138e-05,
-      "loss": 1.8077,
+      "epoch": 6.13,
+      "learning_rate": 8.774107142857143e-05,
+      "loss": 1.3687,
       "step": 1373
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 7.289940828402368e-05,
-      "loss": 1.6965,
+      "epoch": 6.13,
+      "learning_rate": 8.773214285714286e-05,
+      "loss": 1.3625,
       "step": 1374
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 7.287968441814596e-05,
-      "loss": 1.6784,
+      "epoch": 6.14,
+      "learning_rate": 8.77232142857143e-05,
+      "loss": 1.2168,
       "step": 1375
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 7.285996055226825e-05,
-      "loss": 1.6507,
+      "epoch": 6.14,
+      "learning_rate": 8.771428571428572e-05,
+      "loss": 1.2671,
       "step": 1376
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 7.284023668639054e-05,
-      "loss": 1.6128,
+      "epoch": 6.15,
+      "learning_rate": 8.770535714285715e-05,
+      "loss": 1.2899,
       "step": 1377
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 7.282051282051282e-05,
-      "loss": 1.7519,
+      "epoch": 6.15,
+      "learning_rate": 8.769642857142858e-05,
+      "loss": 1.3313,
       "step": 1378
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 7.280078895463511e-05,
-      "loss": 1.5589,
+      "epoch": 6.16,
+      "learning_rate": 8.76875e-05,
+      "loss": 1.33,
       "step": 1379
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 7.27810650887574e-05,
-      "loss": 1.7534,
+      "epoch": 6.16,
+      "learning_rate": 8.767857142857144e-05,
+      "loss": 1.3974,
       "step": 1380
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 7.276134122287968e-05,
-      "loss": 1.658,
+      "epoch": 6.17,
+      "learning_rate": 8.766964285714286e-05,
+      "loss": 1.3188,
       "step": 1381
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 7.274161735700197e-05,
-      "loss": 1.7008,
+      "epoch": 6.17,
+      "learning_rate": 8.766071428571429e-05,
+      "loss": 1.3082,
       "step": 1382
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 7.272189349112426e-05,
-      "loss": 1.647,
+      "epoch": 6.17,
+      "learning_rate": 8.765178571428571e-05,
+      "loss": 1.3057,
       "step": 1383
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 7.270216962524656e-05,
-      "loss": 1.6741,
+      "epoch": 6.18,
+      "learning_rate": 8.764285714285715e-05,
+      "loss": 1.2324,
       "step": 1384
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 7.268244575936885e-05,
-      "loss": 1.6591,
+      "epoch": 6.18,
+      "learning_rate": 8.763392857142858e-05,
+      "loss": 1.2818,
       "step": 1385
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 7.266272189349113e-05,
-      "loss": 1.6904,
+      "epoch": 6.19,
+      "learning_rate": 8.7625e-05,
+      "loss": 1.2314,
       "step": 1386
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 7.264299802761342e-05,
-      "loss": 1.7178,
+      "epoch": 6.19,
+      "learning_rate": 8.761607142857143e-05,
+      "loss": 1.2995,
       "step": 1387
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 7.26232741617357e-05,
-      "loss": 1.6854,
+      "epoch": 6.2,
+      "learning_rate": 8.760714285714287e-05,
+      "loss": 1.3208,
       "step": 1388
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 7.260355029585799e-05,
-      "loss": 1.7031,
+      "epoch": 6.2,
+      "learning_rate": 8.759821428571429e-05,
+      "loss": 1.2755,
       "step": 1389
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 7.258382642998028e-05,
-      "loss": 1.6972,
+      "epoch": 6.21,
+      "learning_rate": 8.758928571428572e-05,
+      "loss": 1.4128,
       "step": 1390
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 7.256410256410256e-05,
-      "loss": 1.5943,
+      "epoch": 6.21,
+      "learning_rate": 8.758035714285716e-05,
+      "loss": 1.3482,
       "step": 1391
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 7.254437869822485e-05,
-      "loss": 1.6556,
+      "epoch": 6.21,
+      "learning_rate": 8.757142857142857e-05,
+      "loss": 1.2618,
       "step": 1392
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 7.252465483234713e-05,
-      "loss": 1.6773,
+      "epoch": 6.22,
+      "learning_rate": 8.756250000000001e-05,
+      "loss": 1.2711,
       "step": 1393
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 7.250493096646944e-05,
-      "loss": 1.7108,
+      "epoch": 6.22,
+      "learning_rate": 8.755357142857143e-05,
+      "loss": 1.3158,
       "step": 1394
     },
     {
-      "epoch": 4.13,
-      "learning_rate": 7.248520710059173e-05,
-      "loss": 1.6129,
+      "epoch": 6.23,
+      "learning_rate": 8.754464285714287e-05,
+      "loss": 1.2439,
       "step": 1395
     },
     {
-      "epoch": 4.13,
-      "learning_rate": 7.246548323471401e-05,
-      "loss": 1.725,
+      "epoch": 6.23,
+      "learning_rate": 8.753571428571428e-05,
+      "loss": 1.2593,
       "step": 1396
     },
     {
-      "epoch": 4.13,
-      "learning_rate": 7.24457593688363e-05,
-      "loss": 1.67,
+      "epoch": 6.24,
+      "learning_rate": 8.752678571428572e-05,
+      "loss": 1.3472,
       "step": 1397
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 7.242603550295858e-05,
-      "loss": 1.665,
+      "epoch": 6.24,
+      "learning_rate": 8.751785714285715e-05,
+      "loss": 1.3512,
       "step": 1398
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 7.240631163708087e-05,
-      "loss": 1.7091,
+      "epoch": 6.25,
+      "learning_rate": 8.750892857142857e-05,
+      "loss": 1.3395,
       "step": 1399
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 7.238658777120316e-05,
-      "loss": 1.703,
+      "epoch": 6.25,
+      "learning_rate": 8.75e-05,
+      "loss": 1.3558,
       "step": 1400
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 7.236686390532544e-05,
-      "loss": 1.7505,
+      "epoch": 6.25,
+      "learning_rate": 8.749107142857144e-05,
+      "loss": 1.3671,
       "step": 1401
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 7.234714003944773e-05,
-      "loss": 1.749,
+      "epoch": 6.26,
+      "learning_rate": 8.748214285714286e-05,
+      "loss": 1.3336,
       "step": 1402
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 7.232741617357003e-05,
-      "loss": 1.6112,
+      "epoch": 6.26,
+      "learning_rate": 8.747321428571429e-05,
+      "loss": 1.2835,
       "step": 1403
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 7.23076923076923e-05,
-      "loss": 1.6849,
+      "epoch": 6.27,
+      "learning_rate": 8.746428571428571e-05,
+      "loss": 1.3185,
       "step": 1404
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 7.22879684418146e-05,
-      "loss": 1.6101,
+      "epoch": 6.27,
+      "learning_rate": 8.745535714285714e-05,
+      "loss": 1.3325,
       "step": 1405
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 7.226824457593689e-05,
-      "loss": 1.7052,
+      "epoch": 6.28,
+      "learning_rate": 8.744642857142858e-05,
+      "loss": 1.3167,
       "step": 1406
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 7.224852071005918e-05,
-      "loss": 1.693,
+      "epoch": 6.28,
+      "learning_rate": 8.74375e-05,
+      "loss": 1.3963,
       "step": 1407
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 7.222879684418147e-05,
-      "loss": 1.7472,
+      "epoch": 6.29,
+      "learning_rate": 8.742857142857144e-05,
+      "loss": 1.3937,
       "step": 1408
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 7.220907297830375e-05,
-      "loss": 1.5884,
+      "epoch": 6.29,
+      "learning_rate": 8.741964285714286e-05,
+      "loss": 1.4223,
       "step": 1409
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 7.218934911242604e-05,
-      "loss": 1.6703,
+      "epoch": 6.29,
+      "learning_rate": 8.74107142857143e-05,
+      "loss": 1.2583,
       "step": 1410
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 7.216962524654832e-05,
-      "loss": 1.7915,
+      "epoch": 6.3,
+      "learning_rate": 8.740178571428572e-05,
+      "loss": 1.2633,
       "step": 1411
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 7.214990138067061e-05,
-      "loss": 1.7117,
+      "epoch": 6.3,
+      "learning_rate": 8.739285714285715e-05,
+      "loss": 1.3449,
       "step": 1412
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 7.21301775147929e-05,
-      "loss": 1.7337,
+      "epoch": 6.31,
+      "learning_rate": 8.738392857142857e-05,
+      "loss": 1.3189,
       "step": 1413
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 7.211045364891518e-05,
-      "loss": 1.6531,
+      "epoch": 6.31,
+      "learning_rate": 8.737500000000001e-05,
+      "loss": 1.3123,
       "step": 1414
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 7.209072978303748e-05,
-      "loss": 1.6375,
+      "epoch": 6.32,
+      "learning_rate": 8.736607142857142e-05,
+      "loss": 1.3182,
       "step": 1415
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 7.207100591715976e-05,
-      "loss": 1.8062,
+      "epoch": 6.32,
+      "learning_rate": 8.735714285714286e-05,
+      "loss": 1.3627,
       "step": 1416
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 7.205128205128205e-05,
-      "loss": 1.71,
+      "epoch": 6.33,
+      "learning_rate": 8.734821428571429e-05,
+      "loss": 1.279,
       "step": 1417
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 7.203155818540434e-05,
-      "loss": 1.7602,
+      "epoch": 6.33,
+      "learning_rate": 8.733928571428572e-05,
+      "loss": 1.2018,
       "step": 1418
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 7.201183431952663e-05,
-      "loss": 1.6197,
+      "epoch": 6.33,
+      "learning_rate": 8.733035714285715e-05,
+      "loss": 1.3105,
       "step": 1419
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 7.199211045364892e-05,
-      "loss": 1.7183,
+      "epoch": 6.34,
+      "learning_rate": 8.732142857142858e-05,
+      "loss": 1.2725,
       "step": 1420
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 7.19723865877712e-05,
-      "loss": 1.7505,
+      "epoch": 6.34,
+      "learning_rate": 8.731250000000001e-05,
+      "loss": 1.3197,
       "step": 1421
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 7.19526627218935e-05,
-      "loss": 1.6458,
+      "epoch": 6.35,
+      "learning_rate": 8.730357142857143e-05,
+      "loss": 1.2812,
       "step": 1422
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 7.193293885601579e-05,
-      "loss": 1.7194,
+      "epoch": 6.35,
+      "learning_rate": 8.729464285714287e-05,
+      "loss": 1.3344,
       "step": 1423
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 7.191321499013807e-05,
-      "loss": 1.7151,
+      "epoch": 6.36,
+      "learning_rate": 8.728571428571429e-05,
+      "loss": 1.3457,
       "step": 1424
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 7.189349112426036e-05,
-      "loss": 1.607,
+      "epoch": 6.36,
+      "learning_rate": 8.727678571428572e-05,
+      "loss": 1.3252,
       "step": 1425
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 7.187376725838265e-05,
-      "loss": 1.6577,
+      "epoch": 6.37,
+      "learning_rate": 8.726785714285714e-05,
+      "loss": 1.3681,
       "step": 1426
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 7.185404339250493e-05,
-      "loss": 1.7345,
+      "epoch": 6.37,
+      "learning_rate": 8.725892857142858e-05,
+      "loss": 1.3765,
       "step": 1427
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 7.183431952662722e-05,
-      "loss": 1.643,
+      "epoch": 6.38,
+      "learning_rate": 8.725e-05,
+      "loss": 1.3452,
       "step": 1428
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 7.181459566074951e-05,
-      "loss": 1.6539,
+      "epoch": 6.38,
+      "learning_rate": 8.724107142857143e-05,
+      "loss": 1.3163,
       "step": 1429
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 7.17948717948718e-05,
-      "loss": 1.6924,
+      "epoch": 6.38,
+      "learning_rate": 8.723214285714286e-05,
+      "loss": 1.2611,
       "step": 1430
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 7.17751479289941e-05,
-      "loss": 1.774,
+      "epoch": 6.39,
+      "learning_rate": 8.72232142857143e-05,
+      "loss": 1.3233,
       "step": 1431
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 7.175542406311638e-05,
-      "loss": 1.6964,
+      "epoch": 6.39,
+      "learning_rate": 8.721428571428571e-05,
+      "loss": 1.3277,
       "step": 1432
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 7.173570019723867e-05,
-      "loss": 1.706,
+      "epoch": 6.4,
+      "learning_rate": 8.720535714285715e-05,
+      "loss": 1.2523,
       "step": 1433
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 7.171597633136095e-05,
-      "loss": 1.7393,
+      "epoch": 6.4,
+      "learning_rate": 8.719642857142857e-05,
+      "loss": 1.1878,
       "step": 1434
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 7.169625246548324e-05,
-      "loss": 1.7168,
+      "epoch": 6.41,
+      "learning_rate": 8.71875e-05,
+      "loss": 1.3681,
       "step": 1435
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 7.167652859960553e-05,
-      "loss": 1.6651,
+      "epoch": 6.41,
+      "learning_rate": 8.717857142857144e-05,
+      "loss": 1.3138,
       "step": 1436
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 7.165680473372781e-05,
-      "loss": 1.6461,
+      "epoch": 6.42,
+      "learning_rate": 8.716964285714286e-05,
+      "loss": 1.3392,
       "step": 1437
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 7.16370808678501e-05,
-      "loss": 1.612,
+      "epoch": 6.42,
+      "learning_rate": 8.716071428571429e-05,
+      "loss": 1.2726,
       "step": 1438
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 7.161735700197238e-05,
-      "loss": 1.7587,
+      "epoch": 6.42,
+      "learning_rate": 8.715178571428571e-05,
+      "loss": 1.3008,
       "step": 1439
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 7.159763313609467e-05,
-      "loss": 1.6489,
+      "epoch": 6.43,
+      "learning_rate": 8.714285714285715e-05,
+      "loss": 1.4073,
       "step": 1440
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 7.157790927021696e-05,
-      "loss": 1.817,
+      "epoch": 6.43,
+      "learning_rate": 8.713392857142858e-05,
+      "loss": 1.2649,
       "step": 1441
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 7.155818540433926e-05,
-      "loss": 1.6523,
+      "epoch": 6.44,
+      "learning_rate": 8.7125e-05,
+      "loss": 1.2958,
       "step": 1442
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 7.153846153846155e-05,
-      "loss": 1.6976,
+      "epoch": 6.44,
+      "learning_rate": 8.711607142857143e-05,
+      "loss": 1.3022,
       "step": 1443
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 7.151873767258384e-05,
-      "loss": 1.609,
+      "epoch": 6.45,
+      "learning_rate": 8.710714285714287e-05,
+      "loss": 1.3126,
       "step": 1444
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 7.149901380670612e-05,
-      "loss": 1.7544,
+      "epoch": 6.45,
+      "learning_rate": 8.709821428571428e-05,
+      "loss": 1.3746,
       "step": 1445
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 7.147928994082841e-05,
-      "loss": 1.7576,
+      "epoch": 6.46,
+      "learning_rate": 8.708928571428572e-05,
+      "loss": 1.3387,
       "step": 1446
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 7.145956607495069e-05,
-      "loss": 1.6113,
+      "epoch": 6.46,
+      "learning_rate": 8.708035714285714e-05,
+      "loss": 1.285,
       "step": 1447
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 7.143984220907298e-05,
-      "loss": 1.7401,
+      "epoch": 6.46,
+      "learning_rate": 8.707142857142857e-05,
+      "loss": 1.2951,
       "step": 1448
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 7.142011834319527e-05,
-      "loss": 1.6532,
+      "epoch": 6.47,
+      "learning_rate": 8.706250000000001e-05,
+      "loss": 1.3007,
       "step": 1449
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 7.140039447731755e-05,
-      "loss": 1.7223,
+      "epoch": 6.47,
+      "learning_rate": 8.705357142857143e-05,
+      "loss": 1.3205,
       "step": 1450
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 7.138067061143984e-05,
-      "loss": 1.7033,
+      "epoch": 6.48,
+      "learning_rate": 8.704464285714286e-05,
+      "loss": 1.2407,
       "step": 1451
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 7.136094674556214e-05,
-      "loss": 1.785,
+      "epoch": 6.48,
+      "learning_rate": 8.703571428571429e-05,
+      "loss": 1.3696,
       "step": 1452
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 7.134122287968443e-05,
-      "loss": 1.6098,
+      "epoch": 6.49,
+      "learning_rate": 8.702678571428572e-05,
+      "loss": 1.3254,
       "step": 1453
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 7.132149901380672e-05,
-      "loss": 1.6419,
+      "epoch": 6.49,
+      "learning_rate": 8.701785714285715e-05,
+      "loss": 1.2844,
       "step": 1454
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 7.1301775147929e-05,
-      "loss": 1.6428,
+      "epoch": 6.5,
+      "learning_rate": 8.700892857142858e-05,
+      "loss": 1.3054,
       "step": 1455
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 7.128205128205129e-05,
-      "loss": 1.7238,
+      "epoch": 6.5,
+      "learning_rate": 8.7e-05,
+      "loss": 1.3074,
       "step": 1456
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 7.126232741617357e-05,
-      "loss": 1.7275,
+      "epoch": 6.5,
+      "learning_rate": 8.699107142857144e-05,
+      "loss": 1.2503,
       "step": 1457
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 7.124260355029586e-05,
-      "loss": 1.656,
+      "epoch": 6.51,
+      "learning_rate": 8.698214285714285e-05,
+      "loss": 1.3226,
       "step": 1458
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 7.122287968441815e-05,
-      "loss": 1.7101,
+      "epoch": 6.51,
+      "learning_rate": 8.697321428571429e-05,
+      "loss": 1.1575,
       "step": 1459
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 7.120315581854043e-05,
-      "loss": 1.6485,
+      "epoch": 6.52,
+      "learning_rate": 8.696428571428572e-05,
+      "loss": 1.2549,
       "step": 1460
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 7.118343195266272e-05,
-      "loss": 1.6507,
+      "epoch": 6.52,
+      "learning_rate": 8.695535714285715e-05,
+      "loss": 1.3328,
       "step": 1461
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 7.1163708086785e-05,
-      "loss": 1.6879,
+      "epoch": 6.53,
+      "learning_rate": 8.694642857142857e-05,
+      "loss": 1.3096,
       "step": 1462
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 7.11439842209073e-05,
-      "loss": 1.6495,
+      "epoch": 6.53,
+      "learning_rate": 8.69375e-05,
+      "loss": 1.3081,
       "step": 1463
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 7.112426035502959e-05,
-      "loss": 1.7404,
+      "epoch": 6.54,
+      "learning_rate": 8.692857142857143e-05,
+      "loss": 1.3143,
       "step": 1464
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 7.110453648915188e-05,
-      "loss": 1.643,
+      "epoch": 6.54,
+      "learning_rate": 8.691964285714286e-05,
+      "loss": 1.3216,
       "step": 1465
     },
     {
-      "epoch": 4.34,
-      "learning_rate": 7.108481262327417e-05,
-      "loss": 1.6463,
+      "epoch": 6.54,
+      "learning_rate": 8.69107142857143e-05,
+      "loss": 1.2787,
       "step": 1466
     },
     {
-      "epoch": 4.34,
-      "learning_rate": 7.106508875739646e-05,
-      "loss": 1.7981,
+      "epoch": 6.55,
+      "learning_rate": 8.690178571428572e-05,
+      "loss": 1.3515,
       "step": 1467
     },
     {
-      "epoch": 4.34,
-      "learning_rate": 7.104536489151874e-05,
-      "loss": 1.6862,
+      "epoch": 6.55,
+      "learning_rate": 8.689285714285715e-05,
+      "loss": 1.3593,
       "step": 1468
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 7.102564102564103e-05,
-      "loss": 1.6814,
+      "epoch": 6.56,
+      "learning_rate": 8.688392857142857e-05,
+      "loss": 1.2694,
       "step": 1469
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 7.100591715976331e-05,
-      "loss": 1.8018,
+      "epoch": 6.56,
+      "learning_rate": 8.687500000000001e-05,
+      "loss": 1.3177,
       "step": 1470
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 7.09861932938856e-05,
-      "loss": 1.5749,
+      "epoch": 6.57,
+      "learning_rate": 8.686607142857142e-05,
+      "loss": 1.2665,
       "step": 1471
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 7.09664694280079e-05,
-      "loss": 1.6312,
+      "epoch": 6.57,
+      "learning_rate": 8.685714285714286e-05,
+      "loss": 1.37,
       "step": 1472
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 7.094674556213018e-05,
-      "loss": 1.764,
+      "epoch": 6.58,
+      "learning_rate": 8.684821428571429e-05,
+      "loss": 1.3184,
       "step": 1473
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 7.092702169625247e-05,
-      "loss": 1.7037,
+      "epoch": 6.58,
+      "learning_rate": 8.683928571428573e-05,
+      "loss": 1.2805,
       "step": 1474
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 7.090729783037475e-05,
-      "loss": 1.5864,
+      "epoch": 6.58,
+      "learning_rate": 8.683035714285714e-05,
+      "loss": 1.2683,
       "step": 1475
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 7.088757396449704e-05,
-      "loss": 1.7195,
+      "epoch": 6.59,
+      "learning_rate": 8.682142857142858e-05,
+      "loss": 1.3647,
       "step": 1476
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 7.086785009861933e-05,
-      "loss": 1.6965,
+      "epoch": 6.59,
+      "learning_rate": 8.68125e-05,
+      "loss": 1.3388,
       "step": 1477
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 7.084812623274162e-05,
-      "loss": 1.6735,
+      "epoch": 6.6,
+      "learning_rate": 8.680357142857143e-05,
+      "loss": 1.265,
       "step": 1478
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 7.082840236686391e-05,
-      "loss": 1.7213,
+      "epoch": 6.6,
+      "learning_rate": 8.679464285714287e-05,
+      "loss": 1.2461,
       "step": 1479
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 7.08086785009862e-05,
-      "loss": 1.6963,
+      "epoch": 6.61,
+      "learning_rate": 8.678571428571429e-05,
+      "loss": 1.2951,
       "step": 1480
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 7.078895463510849e-05,
-      "loss": 1.7314,
+      "epoch": 6.61,
+      "learning_rate": 8.677678571428572e-05,
+      "loss": 1.2864,
       "step": 1481
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 7.076923076923078e-05,
-      "loss": 1.5856,
+      "epoch": 6.62,
+      "learning_rate": 8.676785714285714e-05,
+      "loss": 1.2511,
       "step": 1482
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 7.074950690335306e-05,
-      "loss": 1.7434,
+      "epoch": 6.62,
+      "learning_rate": 8.675892857142858e-05,
+      "loss": 1.3122,
       "step": 1483
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 7.072978303747535e-05,
-      "loss": 1.5951,
+      "epoch": 6.62,
+      "learning_rate": 8.675000000000001e-05,
+      "loss": 1.3223,
       "step": 1484
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 7.071005917159763e-05,
-      "loss": 1.6218,
+      "epoch": 6.63,
+      "learning_rate": 8.674107142857143e-05,
+      "loss": 1.2915,
       "step": 1485
     },
     {
-      "epoch": 4.4,
-      "learning_rate": 7.069033530571992e-05,
-      "loss": 1.6218,
+      "epoch": 6.63,
+      "learning_rate": 8.673214285714286e-05,
+      "loss": 1.4255,
       "step": 1486
     },
     {
-      "epoch": 4.4,
-      "learning_rate": 7.067061143984221e-05,
-      "loss": 1.6577,
+      "epoch": 6.64,
+      "learning_rate": 8.67232142857143e-05,
+      "loss": 1.3046,
       "step": 1487
     },
     {
-      "epoch": 4.4,
-      "learning_rate": 7.06508875739645e-05,
-      "loss": 1.6189,
+      "epoch": 6.64,
+      "learning_rate": 8.671428571428571e-05,
+      "loss": 1.3412,
       "step": 1488
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 7.06311637080868e-05,
-      "loss": 1.7413,
+      "epoch": 6.65,
+      "learning_rate": 8.670535714285715e-05,
+      "loss": 1.3064,
       "step": 1489
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 7.061143984220909e-05,
-      "loss": 1.6668,
+      "epoch": 6.65,
+      "learning_rate": 8.669642857142857e-05,
+      "loss": 1.28,
       "step": 1490
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 7.059171597633137e-05,
-      "loss": 1.6773,
+      "epoch": 6.66,
+      "learning_rate": 8.66875e-05,
+      "loss": 1.333,
       "step": 1491
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 7.057199211045366e-05,
-      "loss": 1.675,
+      "epoch": 6.66,
+      "learning_rate": 8.667857142857143e-05,
+      "loss": 1.3476,
       "step": 1492
     },
     {
-      "epoch": 4.42,
-      "learning_rate": 7.055226824457594e-05,
-      "loss": 1.7214,
+      "epoch": 6.67,
+      "learning_rate": 8.666964285714286e-05,
+      "loss": 1.2861,
       "step": 1493
     },
     {
-      "epoch": 4.42,
-      "learning_rate": 7.053254437869823e-05,
-      "loss": 1.6433,
+      "epoch": 6.67,
+      "learning_rate": 8.666071428571429e-05,
+      "loss": 1.2717,
       "step": 1494
     },
     {
-      "epoch": 4.42,
-      "learning_rate": 7.051282051282052e-05,
-      "loss": 1.6909,
+      "epoch": 6.67,
+      "learning_rate": 8.665178571428572e-05,
+      "loss": 1.3296,
       "step": 1495
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 7.04930966469428e-05,
-      "loss": 1.6688,
+      "epoch": 6.68,
+      "learning_rate": 8.664285714285715e-05,
+      "loss": 1.3088,
       "step": 1496
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 7.047337278106509e-05,
-      "loss": 1.6395,
+      "epoch": 6.68,
+      "learning_rate": 8.663392857142858e-05,
+      "loss": 1.2687,
       "step": 1497
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 7.045364891518737e-05,
-      "loss": 1.6815,
+      "epoch": 6.69,
+      "learning_rate": 8.6625e-05,
+      "loss": 1.267,
       "step": 1498
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 7.043392504930966e-05,
-      "loss": 1.6835,
+      "epoch": 6.69,
+      "learning_rate": 8.661607142857143e-05,
+      "loss": 1.2782,
       "step": 1499
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 7.041420118343195e-05,
-      "loss": 1.6929,
+      "epoch": 6.7,
+      "learning_rate": 8.660714285714287e-05,
+      "loss": 1.3116,
       "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,
+      "epoch": 6.7,
+      "eval_loss": 1.2258427143096924,
+      "eval_runtime": 45.572,
+      "eval_samples_per_second": 287.326,
+      "eval_steps_per_second": 1.514,
       "step": 1500
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 7.039447731755425e-05,
-      "loss": 1.695,
+      "epoch": 6.7,
+      "learning_rate": 8.659821428571428e-05,
+      "loss": 1.2908,
       "step": 1501
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 7.037475345167654e-05,
-      "loss": 1.7585,
+      "epoch": 6.71,
+      "learning_rate": 8.658928571428572e-05,
+      "loss": 1.3291,
       "step": 1502
     },
     {
-      "epoch": 4.45,
-      "learning_rate": 7.035502958579882e-05,
-      "loss": 1.5668,
+      "epoch": 6.71,
+      "learning_rate": 8.658035714285715e-05,
+      "loss": 1.3327,
       "step": 1503
     },
     {
-      "epoch": 4.45,
-      "learning_rate": 7.033530571992111e-05,
-      "loss": 1.7283,
+      "epoch": 6.71,
+      "learning_rate": 8.657142857142858e-05,
+      "loss": 1.3323,
       "step": 1504
     },
     {
-      "epoch": 4.45,
-      "learning_rate": 7.03155818540434e-05,
-      "loss": 1.7085,
+      "epoch": 6.72,
+      "learning_rate": 8.65625e-05,
+      "loss": 1.3009,
       "step": 1505
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 7.029585798816568e-05,
-      "loss": 1.8013,
+      "epoch": 6.72,
+      "learning_rate": 8.655357142857144e-05,
+      "loss": 1.2813,
       "step": 1506
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 7.027613412228797e-05,
-      "loss": 1.6614,
+      "epoch": 6.73,
+      "learning_rate": 8.654464285714286e-05,
+      "loss": 1.2602,
       "step": 1507
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 7.025641025641025e-05,
-      "loss": 1.7083,
+      "epoch": 6.73,
+      "learning_rate": 8.653571428571429e-05,
+      "loss": 1.3479,
       "step": 1508
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 7.023668639053254e-05,
-      "loss": 1.6591,
+      "epoch": 6.74,
+      "learning_rate": 8.652678571428571e-05,
+      "loss": 1.3255,
       "step": 1509
     },
     {
-      "epoch": 4.47,
-      "learning_rate": 7.021696252465483e-05,
-      "loss": 1.6939,
+      "epoch": 6.74,
+      "learning_rate": 8.651785714285715e-05,
+      "loss": 1.3823,
       "step": 1510
     },
     {
-      "epoch": 4.47,
-      "learning_rate": 7.019723865877713e-05,
-      "loss": 1.6936,
+      "epoch": 6.75,
+      "learning_rate": 8.650892857142858e-05,
+      "loss": 1.305,
       "step": 1511
     },
     {
-      "epoch": 4.47,
-      "learning_rate": 7.017751479289942e-05,
-      "loss": 1.6156,
+      "epoch": 6.75,
+      "learning_rate": 8.65e-05,
+      "loss": 1.2929,
       "step": 1512
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 7.015779092702171e-05,
-      "loss": 1.6188,
+      "epoch": 6.75,
+      "learning_rate": 8.649107142857144e-05,
+      "loss": 1.2897,
       "step": 1513
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 7.013806706114399e-05,
-      "loss": 1.7037,
+      "epoch": 6.76,
+      "learning_rate": 8.648214285714285e-05,
+      "loss": 1.3458,
       "step": 1514
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 7.011834319526628e-05,
-      "loss": 1.7514,
+      "epoch": 6.76,
+      "learning_rate": 8.647321428571429e-05,
+      "loss": 1.3233,
       "step": 1515
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 7.009861932938856e-05,
-      "loss": 1.6374,
+      "epoch": 6.77,
+      "learning_rate": 8.646428571428572e-05,
+      "loss": 1.2527,
       "step": 1516
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 7.007889546351085e-05,
-      "loss": 1.7444,
+      "epoch": 6.77,
+      "learning_rate": 8.645535714285716e-05,
+      "loss": 1.2963,
       "step": 1517
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 7.005917159763314e-05,
-      "loss": 1.6806,
+      "epoch": 6.78,
+      "learning_rate": 8.644642857142857e-05,
+      "loss": 1.2763,
       "step": 1518
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 7.003944773175542e-05,
-      "loss": 1.7,
+      "epoch": 6.78,
+      "learning_rate": 8.643750000000001e-05,
+      "loss": 1.2669,
       "step": 1519
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 7.001972386587772e-05,
-      "loss": 1.6271,
+      "epoch": 6.79,
+      "learning_rate": 8.642857142857143e-05,
+      "loss": 1.3188,
       "step": 1520
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 7e-05,
-      "loss": 1.6974,
+      "epoch": 6.79,
+      "learning_rate": 8.641964285714286e-05,
+      "loss": 1.2829,
       "step": 1521
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 6.998027613412229e-05,
-      "loss": 1.655,
+      "epoch": 6.79,
+      "learning_rate": 8.641071428571428e-05,
+      "loss": 1.3218,
       "step": 1522
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 6.996055226824458e-05,
-      "loss": 1.602,
+      "epoch": 6.8,
+      "learning_rate": 8.640178571428572e-05,
+      "loss": 1.2814,
       "step": 1523
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 6.994082840236687e-05,
-      "loss": 1.7667,
+      "epoch": 6.8,
+      "learning_rate": 8.639285714285715e-05,
+      "loss": 1.2322,
       "step": 1524
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 6.992110453648916e-05,
-      "loss": 1.7123,
+      "epoch": 6.81,
+      "learning_rate": 8.638392857142857e-05,
+      "loss": 1.2861,
       "step": 1525
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 6.990138067061144e-05,
-      "loss": 1.8451,
+      "epoch": 6.81,
+      "learning_rate": 8.637500000000001e-05,
+      "loss": 1.2772,
       "step": 1526
     },
     {
-      "epoch": 4.52,
-      "learning_rate": 6.988165680473373e-05,
-      "loss": 1.6255,
+      "epoch": 6.82,
+      "learning_rate": 8.636607142857144e-05,
+      "loss": 1.2536,
       "step": 1527
     },
     {
-      "epoch": 4.52,
-      "learning_rate": 6.986193293885603e-05,
-      "loss": 1.7041,
+      "epoch": 6.82,
+      "learning_rate": 8.635714285714286e-05,
+      "loss": 1.2774,
       "step": 1528
     },
     {
-      "epoch": 4.52,
-      "learning_rate": 6.98422090729783e-05,
-      "loss": 1.7547,
+      "epoch": 6.83,
+      "learning_rate": 8.634821428571429e-05,
+      "loss": 1.3386,
       "step": 1529
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 6.98224852071006e-05,
-      "loss": 1.604,
+      "epoch": 6.83,
+      "learning_rate": 8.633928571428573e-05,
+      "loss": 1.3159,
       "step": 1530
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 6.980276134122287e-05,
-      "loss": 1.6659,
+      "epoch": 6.83,
+      "learning_rate": 8.633035714285714e-05,
+      "loss": 1.2939,
       "step": 1531
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 6.978303747534517e-05,
-      "loss": 1.7572,
+      "epoch": 6.84,
+      "learning_rate": 8.632142857142858e-05,
+      "loss": 1.2412,
       "step": 1532
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 6.976331360946746e-05,
-      "loss": 1.7103,
+      "epoch": 6.84,
+      "learning_rate": 8.63125e-05,
+      "loss": 1.3481,
       "step": 1533
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 6.974358974358974e-05,
-      "loss": 1.6534,
+      "epoch": 6.85,
+      "learning_rate": 8.630357142857143e-05,
+      "loss": 1.2972,
       "step": 1534
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 6.972386587771203e-05,
-      "loss": 1.6458,
+      "epoch": 6.85,
+      "learning_rate": 8.629464285714286e-05,
+      "loss": 1.2298,
       "step": 1535
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 6.970414201183432e-05,
-      "loss": 1.6916,
+      "epoch": 6.86,
+      "learning_rate": 8.62857142857143e-05,
+      "loss": 1.2645,
       "step": 1536
     },
     {
-      "epoch": 4.55,
-      "learning_rate": 6.968441814595661e-05,
-      "loss": 1.6783,
+      "epoch": 6.86,
+      "learning_rate": 8.627678571428572e-05,
+      "loss": 1.2877,
       "step": 1537
     },
     {
-      "epoch": 4.55,
-      "learning_rate": 6.96646942800789e-05,
-      "loss": 1.7412,
+      "epoch": 6.87,
+      "learning_rate": 8.626785714285714e-05,
+      "loss": 1.3413,
       "step": 1538
     },
     {
-      "epoch": 4.55,
-      "learning_rate": 6.964497041420118e-05,
-      "loss": 1.7834,
+      "epoch": 6.87,
+      "learning_rate": 8.625892857142857e-05,
+      "loss": 1.2504,
       "step": 1539
     },
     {
-      "epoch": 4.56,
-      "learning_rate": 6.962524654832348e-05,
-      "loss": 1.7327,
+      "epoch": 6.88,
+      "learning_rate": 8.625000000000001e-05,
+      "loss": 1.241,
       "step": 1540
     },
     {
-      "epoch": 4.56,
-      "learning_rate": 6.960552268244577e-05,
-      "loss": 1.6625,
+      "epoch": 6.88,
+      "learning_rate": 8.624107142857143e-05,
+      "loss": 1.3422,
       "step": 1541
     },
     {
-      "epoch": 4.56,
-      "learning_rate": 6.958579881656805e-05,
-      "loss": 1.6921,
+      "epoch": 6.88,
+      "learning_rate": 8.623214285714286e-05,
+      "loss": 1.2118,
       "step": 1542
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 6.956607495069034e-05,
-      "loss": 1.6634,
+      "epoch": 6.89,
+      "learning_rate": 8.62232142857143e-05,
+      "loss": 1.2568,
       "step": 1543
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 6.954635108481262e-05,
-      "loss": 1.6871,
+      "epoch": 6.89,
+      "learning_rate": 8.621428571428571e-05,
+      "loss": 1.2549,
       "step": 1544
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 6.952662721893491e-05,
-      "loss": 1.706,
+      "epoch": 6.9,
+      "learning_rate": 8.620535714285715e-05,
+      "loss": 1.2907,
       "step": 1545
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 6.95069033530572e-05,
-      "loss": 1.7756,
+      "epoch": 6.9,
+      "learning_rate": 8.619642857142858e-05,
+      "loss": 1.3238,
       "step": 1546
     },
     {
-      "epoch": 4.58,
-      "learning_rate": 6.94871794871795e-05,
-      "loss": 1.7524,
+      "epoch": 6.91,
+      "learning_rate": 8.61875e-05,
+      "loss": 1.2675,
       "step": 1547
     },
     {
-      "epoch": 4.58,
-      "learning_rate": 6.946745562130179e-05,
-      "loss": 1.6669,
+      "epoch": 6.91,
+      "learning_rate": 8.617857142857143e-05,
+      "loss": 1.3036,
       "step": 1548
     },
     {
-      "epoch": 4.58,
-      "learning_rate": 6.944773175542406e-05,
-      "loss": 1.7623,
+      "epoch": 6.92,
+      "learning_rate": 8.616964285714287e-05,
+      "loss": 1.2667,
       "step": 1549
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 6.942800788954636e-05,
-      "loss": 1.6281,
+      "epoch": 6.92,
+      "learning_rate": 8.616071428571429e-05,
+      "loss": 1.2508,
       "step": 1550
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 6.940828402366865e-05,
-      "loss": 1.7002,
+      "epoch": 6.92,
+      "learning_rate": 8.615178571428572e-05,
+      "loss": 1.3513,
       "step": 1551
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 6.938856015779093e-05,
-      "loss": 1.6835,
+      "epoch": 6.93,
+      "learning_rate": 8.614285714285714e-05,
+      "loss": 1.2382,
       "step": 1552
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 6.936883629191322e-05,
-      "loss": 1.7039,
+      "epoch": 6.93,
+      "learning_rate": 8.613392857142858e-05,
+      "loss": 1.2715,
       "step": 1553
     },
     {
-      "epoch": 4.6,
-      "learning_rate": 6.93491124260355e-05,
-      "loss": 1.6842,
+      "epoch": 6.94,
+      "learning_rate": 8.6125e-05,
+      "loss": 1.2074,
       "step": 1554
     },
     {
-      "epoch": 4.6,
-      "learning_rate": 6.932938856015779e-05,
-      "loss": 1.617,
+      "epoch": 6.94,
+      "learning_rate": 8.611607142857143e-05,
+      "loss": 1.2573,
       "step": 1555
     },
     {
-      "epoch": 4.6,
-      "learning_rate": 6.930966469428008e-05,
-      "loss": 1.5942,
+      "epoch": 6.95,
+      "learning_rate": 8.610714285714286e-05,
+      "loss": 1.3434,
       "step": 1556
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 6.928994082840236e-05,
-      "loss": 1.6847,
+      "epoch": 6.95,
+      "learning_rate": 8.609821428571428e-05,
+      "loss": 1.2414,
       "step": 1557
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 6.927021696252465e-05,
-      "loss": 1.6843,
+      "epoch": 6.96,
+      "learning_rate": 8.608928571428572e-05,
+      "loss": 1.3191,
       "step": 1558
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 6.925049309664695e-05,
-      "loss": 1.6311,
+      "epoch": 6.96,
+      "learning_rate": 8.608035714285715e-05,
+      "loss": 1.309,
       "step": 1559
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 6.923076923076924e-05,
-      "loss": 1.7402,
+      "epoch": 6.96,
+      "learning_rate": 8.607142857142859e-05,
+      "loss": 1.3664,
       "step": 1560
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 6.921104536489153e-05,
-      "loss": 1.8036,
+      "epoch": 6.97,
+      "learning_rate": 8.60625e-05,
+      "loss": 1.3968,
       "step": 1561
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 6.919132149901381e-05,
-      "loss": 1.6092,
+      "epoch": 6.97,
+      "learning_rate": 8.605357142857144e-05,
+      "loss": 1.2849,
       "step": 1562
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 6.91715976331361e-05,
-      "loss": 1.634,
+      "epoch": 6.98,
+      "learning_rate": 8.604464285714286e-05,
+      "loss": 1.2634,
       "step": 1563
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 6.915187376725839e-05,
-      "loss": 1.7229,
+      "epoch": 6.98,
+      "learning_rate": 8.603571428571429e-05,
+      "loss": 1.2989,
       "step": 1564
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 6.913214990138067e-05,
-      "loss": 1.6745,
+      "epoch": 6.99,
+      "learning_rate": 8.602678571428571e-05,
+      "loss": 1.3318,
       "step": 1565
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 6.911242603550296e-05,
-      "loss": 1.6808,
+      "epoch": 6.99,
+      "learning_rate": 8.601785714285715e-05,
+      "loss": 1.2379,
       "step": 1566
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 6.909270216962524e-05,
-      "loss": 1.6811,
+      "epoch": 7.0,
+      "learning_rate": 8.600892857142856e-05,
+      "loss": 1.3214,
       "step": 1567
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 6.907297830374753e-05,
-      "loss": 1.6214,
+      "epoch": 7.0,
+      "learning_rate": 8.6e-05,
+      "loss": 1.4629,
       "step": 1568
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 6.905325443786983e-05,
-      "loss": 1.6334,
+      "epoch": 7.0,
+      "learning_rate": 8.599107142857143e-05,
+      "loss": 1.22,
       "step": 1569
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 6.903353057199212e-05,
-      "loss": 1.7347,
+      "epoch": 7.01,
+      "learning_rate": 8.598214285714287e-05,
+      "loss": 1.274,
       "step": 1570
     },
     {
-      "epoch": 4.65,
-      "learning_rate": 6.901380670611441e-05,
-      "loss": 1.586,
+      "epoch": 7.01,
+      "learning_rate": 8.597321428571429e-05,
+      "loss": 1.2594,
       "step": 1571
     },
     {
-      "epoch": 4.65,
-      "learning_rate": 6.899408284023669e-05,
-      "loss": 1.6061,
+      "epoch": 7.02,
+      "learning_rate": 8.596428571428572e-05,
+      "loss": 1.2579,
       "step": 1572
     },
     {
-      "epoch": 4.65,
-      "learning_rate": 6.897435897435898e-05,
-      "loss": 1.6794,
+      "epoch": 7.02,
+      "learning_rate": 8.595535714285716e-05,
+      "loss": 1.2271,
       "step": 1573
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 6.895463510848127e-05,
-      "loss": 1.6595,
+      "epoch": 7.03,
+      "learning_rate": 8.594642857142857e-05,
+      "loss": 1.2844,
       "step": 1574
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 6.893491124260355e-05,
-      "loss": 1.8368,
+      "epoch": 7.03,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 1.3427,
       "step": 1575
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 6.891518737672584e-05,
-      "loss": 1.5694,
+      "epoch": 7.04,
+      "learning_rate": 8.592857142857143e-05,
+      "loss": 1.2032,
       "step": 1576
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 6.889546351084814e-05,
-      "loss": 1.7532,
+      "epoch": 7.04,
+      "learning_rate": 8.591964285714286e-05,
+      "loss": 1.3029,
       "step": 1577
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 6.887573964497041e-05,
-      "loss": 1.7016,
+      "epoch": 7.04,
+      "learning_rate": 8.591071428571428e-05,
+      "loss": 1.2335,
       "step": 1578
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 6.88560157790927e-05,
-      "loss": 1.6331,
+      "epoch": 7.05,
+      "learning_rate": 8.590178571428572e-05,
+      "loss": 1.3649,
       "step": 1579
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 6.883629191321498e-05,
-      "loss": 1.7002,
+      "epoch": 7.05,
+      "learning_rate": 8.589285714285714e-05,
+      "loss": 1.2817,
       "step": 1580
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 6.881656804733728e-05,
-      "loss": 1.7429,
+      "epoch": 7.06,
+      "learning_rate": 8.588392857142857e-05,
+      "loss": 1.2112,
       "step": 1581
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 6.879684418145957e-05,
-      "loss": 1.5845,
+      "epoch": 7.06,
+      "learning_rate": 8.5875e-05,
+      "loss": 1.2502,
       "step": 1582
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 6.877712031558186e-05,
-      "loss": 1.6967,
+      "epoch": 7.07,
+      "learning_rate": 8.586607142857144e-05,
+      "loss": 1.376,
       "step": 1583
     },
     {
-      "epoch": 4.69,
-      "learning_rate": 6.875739644970415e-05,
-      "loss": 1.671,
+      "epoch": 7.07,
+      "learning_rate": 8.585714285714286e-05,
+      "loss": 1.2492,
       "step": 1584
     },
     {
-      "epoch": 4.69,
-      "learning_rate": 6.873767258382643e-05,
-      "loss": 1.6277,
+      "epoch": 7.08,
+      "learning_rate": 8.584821428571429e-05,
+      "loss": 1.2415,
       "step": 1585
     },
     {
-      "epoch": 4.69,
-      "learning_rate": 6.871794871794872e-05,
-      "loss": 1.7199,
+      "epoch": 7.08,
+      "learning_rate": 8.583928571428572e-05,
+      "loss": 1.4028,
       "step": 1586
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 6.869822485207102e-05,
-      "loss": 1.7113,
+      "epoch": 7.08,
+      "learning_rate": 8.583035714285714e-05,
+      "loss": 1.2476,
       "step": 1587
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 6.86785009861933e-05,
-      "loss": 1.5845,
+      "epoch": 7.09,
+      "learning_rate": 8.582142857142858e-05,
+      "loss": 1.2275,
       "step": 1588
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 6.865877712031559e-05,
-      "loss": 1.6749,
+      "epoch": 7.09,
+      "learning_rate": 8.58125e-05,
+      "loss": 1.294,
       "step": 1589
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 6.863905325443787e-05,
-      "loss": 1.5795,
+      "epoch": 7.1,
+      "learning_rate": 8.580357142857143e-05,
+      "loss": 1.2366,
       "step": 1590
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 6.861932938856016e-05,
-      "loss": 1.5712,
+      "epoch": 7.1,
+      "learning_rate": 8.579464285714286e-05,
+      "loss": 1.3436,
       "step": 1591
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 6.859960552268245e-05,
-      "loss": 1.7565,
+      "epoch": 7.11,
+      "learning_rate": 8.57857142857143e-05,
+      "loss": 1.279,
       "step": 1592
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 6.857988165680473e-05,
-      "loss": 1.6717,
+      "epoch": 7.11,
+      "learning_rate": 8.577678571428572e-05,
+      "loss": 1.2981,
       "step": 1593
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 6.856015779092702e-05,
-      "loss": 1.6726,
+      "epoch": 7.12,
+      "learning_rate": 8.576785714285715e-05,
+      "loss": 1.3042,
       "step": 1594
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 6.854043392504931e-05,
-      "loss": 1.6378,
+      "epoch": 7.12,
+      "learning_rate": 8.575892857142857e-05,
+      "loss": 1.3182,
       "step": 1595
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 6.85207100591716e-05,
-      "loss": 1.6823,
+      "epoch": 7.12,
+      "learning_rate": 8.575000000000001e-05,
+      "loss": 1.2732,
       "step": 1596
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 6.85009861932939e-05,
-      "loss": 1.7148,
+      "epoch": 7.13,
+      "learning_rate": 8.574107142857142e-05,
+      "loss": 1.3271,
       "step": 1597
     },
     {
-      "epoch": 4.73,
-      "learning_rate": 6.848126232741617e-05,
-      "loss": 1.745,
+      "epoch": 7.13,
+      "learning_rate": 8.573214285714286e-05,
+      "loss": 1.2495,
       "step": 1598
     },
     {
-      "epoch": 4.73,
-      "learning_rate": 6.846153846153847e-05,
-      "loss": 1.629,
+      "epoch": 7.14,
+      "learning_rate": 8.572321428571429e-05,
+      "loss": 1.2367,
       "step": 1599
     },
     {
-      "epoch": 4.73,
-      "learning_rate": 6.844181459566076e-05,
-      "loss": 1.6951,
+      "epoch": 7.14,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.2849,
       "step": 1600
     },
     {
-      "epoch": 4.74,
-      "learning_rate": 6.842209072978304e-05,
-      "loss": 1.6388,
+      "epoch": 7.15,
+      "learning_rate": 8.570535714285715e-05,
+      "loss": 1.2964,
       "step": 1601
     },
     {
-      "epoch": 4.74,
-      "learning_rate": 6.840236686390533e-05,
-      "loss": 1.6383,
+      "epoch": 7.15,
+      "learning_rate": 8.569642857142858e-05,
+      "loss": 1.2721,
       "step": 1602
     },
     {
-      "epoch": 4.74,
-      "learning_rate": 6.838264299802761e-05,
-      "loss": 1.5902,
+      "epoch": 7.16,
+      "learning_rate": 8.568750000000002e-05,
+      "loss": 1.3283,
       "step": 1603
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 6.83629191321499e-05,
-      "loss": 1.6663,
+      "epoch": 7.16,
+      "learning_rate": 8.567857142857143e-05,
+      "loss": 1.2236,
       "step": 1604
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 6.834319526627219e-05,
-      "loss": 1.7243,
+      "epoch": 7.17,
+      "learning_rate": 8.566964285714287e-05,
+      "loss": 1.3142,
       "step": 1605
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 6.832347140039448e-05,
-      "loss": 1.5873,
+      "epoch": 7.17,
+      "learning_rate": 8.566071428571429e-05,
+      "loss": 1.2508,
       "step": 1606
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 6.830374753451678e-05,
-      "loss": 1.6889,
+      "epoch": 7.17,
+      "learning_rate": 8.565178571428572e-05,
+      "loss": 1.28,
       "step": 1607
     },
     {
-      "epoch": 4.76,
-      "learning_rate": 6.828402366863906e-05,
-      "loss": 1.5907,
+      "epoch": 7.18,
+      "learning_rate": 8.564285714285714e-05,
+      "loss": 1.2666,
       "step": 1608
     },
     {
-      "epoch": 4.76,
-      "learning_rate": 6.826429980276135e-05,
-      "loss": 1.6725,
+      "epoch": 7.18,
+      "learning_rate": 8.563392857142858e-05,
+      "loss": 1.268,
       "step": 1609
     },
     {
-      "epoch": 4.76,
-      "learning_rate": 6.824457593688364e-05,
-      "loss": 1.712,
+      "epoch": 7.19,
+      "learning_rate": 8.5625e-05,
+      "loss": 1.2804,
       "step": 1610
     },
     {
-      "epoch": 4.77,
-      "learning_rate": 6.822485207100592e-05,
-      "loss": 1.7375,
+      "epoch": 7.19,
+      "learning_rate": 8.561607142857143e-05,
+      "loss": 1.3133,
       "step": 1611
     },
     {
-      "epoch": 4.77,
-      "learning_rate": 6.820512820512821e-05,
-      "loss": 1.6592,
+      "epoch": 7.2,
+      "learning_rate": 8.560714285714286e-05,
+      "loss": 1.254,
       "step": 1612
     },
     {
-      "epoch": 4.77,
-      "learning_rate": 6.818540433925049e-05,
-      "loss": 1.7399,
+      "epoch": 7.2,
+      "learning_rate": 8.559821428571428e-05,
+      "loss": 1.257,
       "step": 1613
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 6.816568047337278e-05,
-      "loss": 1.7114,
+      "epoch": 7.21,
+      "learning_rate": 8.558928571428571e-05,
+      "loss": 1.2497,
       "step": 1614
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 6.814595660749507e-05,
-      "loss": 1.7714,
+      "epoch": 7.21,
+      "learning_rate": 8.558035714285715e-05,
+      "loss": 1.2807,
       "step": 1615
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 6.812623274161735e-05,
-      "loss": 1.6295,
+      "epoch": 7.21,
+      "learning_rate": 8.557142857142857e-05,
+      "loss": 1.368,
       "step": 1616
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 6.810650887573964e-05,
-      "loss": 1.6547,
+      "epoch": 7.22,
+      "learning_rate": 8.55625e-05,
+      "loss": 1.265,
       "step": 1617
     },
     {
-      "epoch": 4.79,
-      "learning_rate": 6.808678500986194e-05,
-      "loss": 1.647,
+      "epoch": 7.22,
+      "learning_rate": 8.555357142857144e-05,
+      "loss": 1.3338,
       "step": 1618
     },
     {
-      "epoch": 4.79,
-      "learning_rate": 6.806706114398423e-05,
-      "loss": 1.7809,
+      "epoch": 7.23,
+      "learning_rate": 8.554464285714286e-05,
+      "loss": 1.2262,
       "step": 1619
     },
     {
-      "epoch": 4.79,
-      "learning_rate": 6.804733727810652e-05,
-      "loss": 1.5747,
+      "epoch": 7.23,
+      "learning_rate": 8.553571428571429e-05,
+      "loss": 1.2791,
       "step": 1620
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 6.80276134122288e-05,
-      "loss": 1.6598,
+      "epoch": 7.24,
+      "learning_rate": 8.552678571428571e-05,
+      "loss": 1.3362,
       "step": 1621
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 6.800788954635109e-05,
-      "loss": 1.6836,
+      "epoch": 7.24,
+      "learning_rate": 8.551785714285715e-05,
+      "loss": 1.3548,
       "step": 1622
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 6.798816568047338e-05,
-      "loss": 1.6107,
+      "epoch": 7.25,
+      "learning_rate": 8.550892857142857e-05,
+      "loss": 1.2626,
       "step": 1623
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 6.796844181459566e-05,
-      "loss": 1.7016,
+      "epoch": 7.25,
+      "learning_rate": 8.55e-05,
+      "loss": 1.2393,
       "step": 1624
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 6.794871794871795e-05,
-      "loss": 1.6553,
+      "epoch": 7.25,
+      "learning_rate": 8.549107142857143e-05,
+      "loss": 1.2409,
       "step": 1625
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 6.792899408284023e-05,
-      "loss": 1.7569,
+      "epoch": 7.26,
+      "learning_rate": 8.548214285714287e-05,
+      "loss": 1.2629,
       "step": 1626
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 6.790927021696252e-05,
-      "loss": 1.6879,
+      "epoch": 7.26,
+      "learning_rate": 8.547321428571428e-05,
+      "loss": 1.3937,
       "step": 1627
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 6.788954635108482e-05,
-      "loss": 1.6139,
+      "epoch": 7.27,
+      "learning_rate": 8.546428571428572e-05,
+      "loss": 1.1976,
       "step": 1628
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 6.786982248520711e-05,
-      "loss": 1.6525,
+      "epoch": 7.27,
+      "learning_rate": 8.545535714285715e-05,
+      "loss": 1.3664,
       "step": 1629
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 6.78500986193294e-05,
-      "loss": 1.7435,
+      "epoch": 7.28,
+      "learning_rate": 8.544642857142857e-05,
+      "loss": 1.3014,
       "step": 1630
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 6.783037475345168e-05,
-      "loss": 1.6012,
+      "epoch": 7.28,
+      "learning_rate": 8.543750000000001e-05,
+      "loss": 1.3267,
       "step": 1631
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 6.781065088757397e-05,
-      "loss": 1.6758,
+      "epoch": 7.29,
+      "learning_rate": 8.542857142857144e-05,
+      "loss": 1.2736,
       "step": 1632
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 6.779092702169626e-05,
-      "loss": 1.6789,
+      "epoch": 7.29,
+      "learning_rate": 8.541964285714286e-05,
+      "loss": 1.3544,
       "step": 1633
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 6.777120315581854e-05,
-      "loss": 1.6851,
+      "epoch": 7.29,
+      "learning_rate": 8.541071428571429e-05,
+      "loss": 1.3056,
       "step": 1634
     },
     {
-      "epoch": 4.84,
-      "learning_rate": 6.775147928994083e-05,
-      "loss": 1.6771,
+      "epoch": 7.3,
+      "learning_rate": 8.540178571428573e-05,
+      "loss": 1.3222,
       "step": 1635
     },
     {
-      "epoch": 4.84,
-      "learning_rate": 6.773175542406311e-05,
-      "loss": 1.7053,
+      "epoch": 7.3,
+      "learning_rate": 8.539285714285715e-05,
+      "loss": 1.2309,
       "step": 1636
     },
     {
-      "epoch": 4.84,
-      "learning_rate": 6.77120315581854e-05,
-      "loss": 1.7031,
+      "epoch": 7.31,
+      "learning_rate": 8.538392857142858e-05,
+      "loss": 1.2717,
       "step": 1637
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 6.76923076923077e-05,
-      "loss": 1.6993,
+      "epoch": 7.31,
+      "learning_rate": 8.5375e-05,
+      "loss": 1.2966,
       "step": 1638
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 6.767258382642998e-05,
-      "loss": 1.7541,
+      "epoch": 7.32,
+      "learning_rate": 8.536607142857144e-05,
+      "loss": 1.2558,
       "step": 1639
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 6.765285996055227e-05,
-      "loss": 1.7491,
+      "epoch": 7.32,
+      "learning_rate": 8.535714285714285e-05,
+      "loss": 1.2759,
       "step": 1640
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 6.763313609467456e-05,
-      "loss": 1.7476,
+      "epoch": 7.33,
+      "learning_rate": 8.534821428571429e-05,
+      "loss": 1.2623,
       "step": 1641
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 6.761341222879685e-05,
-      "loss": 1.7134,
+      "epoch": 7.33,
+      "learning_rate": 8.533928571428572e-05,
+      "loss": 1.1874,
       "step": 1642
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 6.759368836291914e-05,
-      "loss": 1.671,
+      "epoch": 7.33,
+      "learning_rate": 8.533035714285714e-05,
+      "loss": 1.1897,
       "step": 1643
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 6.757396449704142e-05,
-      "loss": 1.6314,
+      "epoch": 7.34,
+      "learning_rate": 8.532142857142857e-05,
+      "loss": 1.3311,
       "step": 1644
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 6.755424063116371e-05,
-      "loss": 1.7459,
+      "epoch": 7.34,
+      "learning_rate": 8.531250000000001e-05,
+      "loss": 1.2969,
       "step": 1645
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 6.7534516765286e-05,
-      "loss": 1.6199,
+      "epoch": 7.35,
+      "learning_rate": 8.530357142857143e-05,
+      "loss": 1.3205,
       "step": 1646
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 6.751479289940829e-05,
-      "loss": 1.7738,
+      "epoch": 7.35,
+      "learning_rate": 8.529464285714286e-05,
+      "loss": 1.2821,
       "step": 1647
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 6.749506903353058e-05,
-      "loss": 1.5952,
+      "epoch": 7.36,
+      "learning_rate": 8.52857142857143e-05,
+      "loss": 1.219,
       "step": 1648
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 6.747534516765286e-05,
-      "loss": 1.5989,
+      "epoch": 7.36,
+      "learning_rate": 8.527678571428572e-05,
+      "loss": 1.2912,
       "step": 1649
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 6.745562130177515e-05,
-      "loss": 1.649,
+      "epoch": 7.37,
+      "learning_rate": 8.526785714285715e-05,
+      "loss": 1.3554,
       "step": 1650
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 6.743589743589744e-05,
-      "loss": 1.6976,
+      "epoch": 7.37,
+      "learning_rate": 8.525892857142857e-05,
+      "loss": 1.2973,
       "step": 1651
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 6.741617357001972e-05,
-      "loss": 1.7155,
+      "epoch": 7.38,
+      "learning_rate": 8.525000000000001e-05,
+      "loss": 1.2733,
       "step": 1652
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 6.739644970414201e-05,
-      "loss": 1.6374,
+      "epoch": 7.38,
+      "learning_rate": 8.524107142857142e-05,
+      "loss": 1.2322,
       "step": 1653
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 6.73767258382643e-05,
-      "loss": 1.6676,
+      "epoch": 7.38,
+      "learning_rate": 8.523214285714286e-05,
+      "loss": 1.2651,
       "step": 1654
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 6.73570019723866e-05,
-      "loss": 1.7835,
+      "epoch": 7.39,
+      "learning_rate": 8.522321428571429e-05,
+      "loss": 1.2752,
       "step": 1655
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 6.733727810650889e-05,
-      "loss": 1.6443,
+      "epoch": 7.39,
+      "learning_rate": 8.521428571428571e-05,
+      "loss": 1.2903,
       "step": 1656
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 6.731755424063117e-05,
-      "loss": 1.5772,
+      "epoch": 7.4,
+      "learning_rate": 8.520535714285714e-05,
+      "loss": 1.429,
       "step": 1657
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 6.729783037475346e-05,
-      "loss": 1.6758,
+      "epoch": 7.4,
+      "learning_rate": 8.519642857142858e-05,
+      "loss": 1.2588,
       "step": 1658
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 6.727810650887574e-05,
-      "loss": 1.6709,
+      "epoch": 7.41,
+      "learning_rate": 8.51875e-05,
+      "loss": 1.3591,
       "step": 1659
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 6.725838264299803e-05,
-      "loss": 1.651,
+      "epoch": 7.41,
+      "learning_rate": 8.517857142857143e-05,
+      "loss": 1.2115,
       "step": 1660
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 6.723865877712032e-05,
-      "loss": 1.755,
+      "epoch": 7.42,
+      "learning_rate": 8.516964285714287e-05,
+      "loss": 1.2782,
       "step": 1661
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 6.72189349112426e-05,
-      "loss": 1.6905,
+      "epoch": 7.42,
+      "learning_rate": 8.51607142857143e-05,
+      "loss": 1.1953,
       "step": 1662
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 6.719921104536489e-05,
-      "loss": 1.6215,
+      "epoch": 7.42,
+      "learning_rate": 8.515178571428572e-05,
+      "loss": 1.3413,
       "step": 1663
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 6.717948717948718e-05,
-      "loss": 1.6125,
+      "epoch": 7.43,
+      "learning_rate": 8.514285714285714e-05,
+      "loss": 1.2717,
       "step": 1664
     },
     {
-      "epoch": 4.93,
-      "learning_rate": 6.715976331360948e-05,
-      "loss": 1.6466,
+      "epoch": 7.43,
+      "learning_rate": 8.513392857142858e-05,
+      "loss": 1.3349,
       "step": 1665
     },
     {
-      "epoch": 4.93,
-      "learning_rate": 6.714003944773177e-05,
-      "loss": 1.7112,
+      "epoch": 7.44,
+      "learning_rate": 8.5125e-05,
+      "loss": 1.3566,
       "step": 1666
     },
     {
-      "epoch": 4.93,
-      "learning_rate": 6.712031558185405e-05,
-      "loss": 1.7251,
+      "epoch": 7.44,
+      "learning_rate": 8.511607142857143e-05,
+      "loss": 1.2318,
       "step": 1667
     },
     {
-      "epoch": 4.93,
-      "learning_rate": 6.710059171597634e-05,
-      "loss": 1.7104,
+      "epoch": 7.45,
+      "learning_rate": 8.510714285714286e-05,
+      "loss": 1.293,
       "step": 1668
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 6.708086785009863e-05,
-      "loss": 1.691,
+      "epoch": 7.45,
+      "learning_rate": 8.50982142857143e-05,
+      "loss": 1.2298,
       "step": 1669
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 6.706114398422091e-05,
-      "loss": 1.7012,
+      "epoch": 7.46,
+      "learning_rate": 8.508928571428571e-05,
+      "loss": 1.3032,
       "step": 1670
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 6.70414201183432e-05,
-      "loss": 1.6405,
+      "epoch": 7.46,
+      "learning_rate": 8.508035714285715e-05,
+      "loss": 1.292,
       "step": 1671
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 6.702169625246548e-05,
-      "loss": 1.7031,
+      "epoch": 7.46,
+      "learning_rate": 8.507142857142858e-05,
+      "loss": 1.3088,
       "step": 1672
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 6.700197238658777e-05,
-      "loss": 1.7001,
+      "epoch": 7.47,
+      "learning_rate": 8.50625e-05,
+      "loss": 1.2567,
       "step": 1673
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 6.698224852071006e-05,
-      "loss": 1.6615,
+      "epoch": 7.47,
+      "learning_rate": 8.505357142857143e-05,
+      "loss": 1.2931,
       "step": 1674
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 6.696252465483234e-05,
-      "loss": 1.6866,
+      "epoch": 7.48,
+      "learning_rate": 8.504464285714287e-05,
+      "loss": 1.2408,
       "step": 1675
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 6.694280078895463e-05,
-      "loss": 1.6486,
+      "epoch": 7.48,
+      "learning_rate": 8.503571428571429e-05,
+      "loss": 1.2654,
       "step": 1676
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 6.692307692307693e-05,
-      "loss": 1.7297,
+      "epoch": 7.49,
+      "learning_rate": 8.502678571428572e-05,
+      "loss": 1.2985,
       "step": 1677
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 6.690335305719922e-05,
-      "loss": 1.6228,
+      "epoch": 7.49,
+      "learning_rate": 8.501785714285716e-05,
+      "loss": 1.3217,
       "step": 1678
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 6.688362919132151e-05,
-      "loss": 1.6862,
+      "epoch": 7.5,
+      "learning_rate": 8.500892857142858e-05,
+      "loss": 1.3957,
       "step": 1679
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 6.686390532544379e-05,
-      "loss": 1.6468,
+      "epoch": 7.5,
+      "learning_rate": 8.5e-05,
+      "loss": 1.2568,
       "step": 1680
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 6.684418145956608e-05,
-      "loss": 1.652,
+      "epoch": 7.5,
+      "learning_rate": 8.499107142857143e-05,
+      "loss": 1.367,
       "step": 1681
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 6.682445759368836e-05,
-      "loss": 1.7169,
+      "epoch": 7.51,
+      "learning_rate": 8.498214285714287e-05,
+      "loss": 1.2576,
       "step": 1682
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 6.680473372781065e-05,
-      "loss": 1.7838,
+      "epoch": 7.51,
+      "learning_rate": 8.497321428571428e-05,
+      "loss": 1.215,
       "step": 1683
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 6.678500986193294e-05,
-      "loss": 1.6955,
+      "epoch": 7.52,
+      "learning_rate": 8.496428571428572e-05,
+      "loss": 1.3092,
       "step": 1684
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.676528599605522e-05,
-      "loss": 1.712,
+      "epoch": 7.52,
+      "learning_rate": 8.495535714285715e-05,
+      "loss": 1.2549,
       "step": 1685
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.674556213017751e-05,
-      "loss": 1.6439,
+      "epoch": 7.53,
+      "learning_rate": 8.494642857142857e-05,
+      "loss": 1.282,
       "step": 1686
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.672583826429981e-05,
-      "loss": 1.6041,
+      "epoch": 7.53,
+      "learning_rate": 8.49375e-05,
+      "loss": 1.1735,
       "step": 1687
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.67061143984221e-05,
-      "loss": 1.75,
+      "epoch": 7.54,
+      "learning_rate": 8.492857142857144e-05,
+      "loss": 1.2821,
       "step": 1688
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 6.668639053254439e-05,
-      "loss": 1.6455,
+      "epoch": 7.54,
+      "learning_rate": 8.491964285714286e-05,
+      "loss": 1.2471,
       "step": 1689
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 6.666666666666667e-05,
-      "loss": 1.5692,
+      "epoch": 7.54,
+      "learning_rate": 8.491071428571429e-05,
+      "loss": 1.3194,
       "step": 1690
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 6.664694280078896e-05,
-      "loss": 1.6062,
+      "epoch": 7.55,
+      "learning_rate": 8.490178571428571e-05,
+      "loss": 1.2688,
       "step": 1691
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.662721893491125e-05,
-      "loss": 1.6053,
+      "epoch": 7.55,
+      "learning_rate": 8.489285714285715e-05,
+      "loss": 1.2669,
       "step": 1692
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.660749506903353e-05,
-      "loss": 1.6259,
+      "epoch": 7.56,
+      "learning_rate": 8.488392857142858e-05,
+      "loss": 1.2303,
       "step": 1693
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.658777120315582e-05,
-      "loss": 1.6173,
+      "epoch": 7.56,
+      "learning_rate": 8.4875e-05,
+      "loss": 1.2825,
       "step": 1694
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.65680473372781e-05,
-      "loss": 1.6866,
+      "epoch": 7.57,
+      "learning_rate": 8.486607142857144e-05,
+      "loss": 1.2656,
       "step": 1695
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 6.65483234714004e-05,
-      "loss": 1.7775,
+      "epoch": 7.57,
+      "learning_rate": 8.485714285714285e-05,
+      "loss": 1.3312,
       "step": 1696
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 6.652859960552269e-05,
-      "loss": 1.6216,
+      "epoch": 7.58,
+      "learning_rate": 8.484821428571429e-05,
+      "loss": 1.1707,
       "step": 1697
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 6.650887573964497e-05,
-      "loss": 1.7687,
+      "epoch": 7.58,
+      "learning_rate": 8.483928571428572e-05,
+      "loss": 1.2215,
       "step": 1698
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 6.648915187376726e-05,
-      "loss": 1.6166,
+      "epoch": 7.58,
+      "learning_rate": 8.483035714285714e-05,
+      "loss": 1.2425,
       "step": 1699
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 6.646942800788955e-05,
-      "loss": 1.6359,
+      "epoch": 7.59,
+      "learning_rate": 8.482142857142857e-05,
+      "loss": 1.2486,
       "step": 1700
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 6.644970414201184e-05,
-      "loss": 1.6395,
+      "epoch": 7.59,
+      "learning_rate": 8.481250000000001e-05,
+      "loss": 1.2982,
       "step": 1701
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.642998027613413e-05,
-      "loss": 1.5865,
+      "epoch": 7.6,
+      "learning_rate": 8.480357142857143e-05,
+      "loss": 1.2993,
       "step": 1702
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.641025641025641e-05,
-      "loss": 1.6524,
+      "epoch": 7.6,
+      "learning_rate": 8.479464285714286e-05,
+      "loss": 1.2606,
       "step": 1703
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.63905325443787e-05,
-      "loss": 1.743,
+      "epoch": 7.61,
+      "learning_rate": 8.478571428571428e-05,
+      "loss": 1.2685,
       "step": 1704
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.637080867850098e-05,
-      "loss": 1.5959,
+      "epoch": 7.61,
+      "learning_rate": 8.477678571428572e-05,
+      "loss": 1.281,
       "step": 1705
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 6.635108481262328e-05,
-      "loss": 1.6975,
+      "epoch": 7.62,
+      "learning_rate": 8.476785714285715e-05,
+      "loss": 1.2711,
       "step": 1706
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 6.633136094674557e-05,
-      "loss": 1.5926,
+      "epoch": 7.62,
+      "learning_rate": 8.475892857142857e-05,
+      "loss": 1.2798,
       "step": 1707
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 6.631163708086785e-05,
-      "loss": 1.6271,
+      "epoch": 7.62,
+      "learning_rate": 8.475000000000001e-05,
+      "loss": 1.1871,
       "step": 1708
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 6.629191321499014e-05,
-      "loss": 1.6931,
+      "epoch": 7.63,
+      "learning_rate": 8.474107142857143e-05,
+      "loss": 1.3282,
       "step": 1709
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 6.627218934911243e-05,
-      "loss": 1.703,
+      "epoch": 7.63,
+      "learning_rate": 8.473214285714286e-05,
+      "loss": 1.1779,
       "step": 1710
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 6.625246548323471e-05,
-      "loss": 1.6567,
+      "epoch": 7.64,
+      "learning_rate": 8.472321428571429e-05,
+      "loss": 1.2966,
       "step": 1711
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 6.6232741617357e-05,
-      "loss": 1.6859,
+      "epoch": 7.64,
+      "learning_rate": 8.471428571428573e-05,
+      "loss": 1.2751,
       "step": 1712
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 6.62130177514793e-05,
-      "loss": 1.7783,
+      "epoch": 7.65,
+      "learning_rate": 8.470535714285714e-05,
+      "loss": 1.2487,
       "step": 1713
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 6.619329388560159e-05,
-      "loss": 1.5862,
+      "epoch": 7.65,
+      "learning_rate": 8.469642857142858e-05,
+      "loss": 1.1819,
       "step": 1714
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 6.617357001972388e-05,
-      "loss": 1.6993,
+      "epoch": 7.66,
+      "learning_rate": 8.46875e-05,
+      "loss": 1.2998,
       "step": 1715
     },
     {
-      "epoch": 5.08,
-      "learning_rate": 6.615384615384616e-05,
-      "loss": 1.6873,
+      "epoch": 7.66,
+      "learning_rate": 8.467857142857143e-05,
+      "loss": 1.2336,
       "step": 1716
     },
     {
-      "epoch": 5.08,
-      "learning_rate": 6.613412228796845e-05,
-      "loss": 1.661,
+      "epoch": 7.67,
+      "learning_rate": 8.466964285714286e-05,
+      "loss": 1.2699,
       "step": 1717
     },
     {
-      "epoch": 5.08,
-      "learning_rate": 6.611439842209073e-05,
-      "loss": 1.5867,
+      "epoch": 7.67,
+      "learning_rate": 8.46607142857143e-05,
+      "loss": 1.2653,
       "step": 1718
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 6.609467455621302e-05,
-      "loss": 1.6718,
+      "epoch": 7.67,
+      "learning_rate": 8.465178571428572e-05,
+      "loss": 1.2649,
       "step": 1719
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 6.607495069033531e-05,
-      "loss": 1.7067,
+      "epoch": 7.68,
+      "learning_rate": 8.464285714285715e-05,
+      "loss": 1.2119,
       "step": 1720
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 6.605522682445759e-05,
-      "loss": 1.7595,
+      "epoch": 7.68,
+      "learning_rate": 8.463392857142857e-05,
+      "loss": 1.2991,
       "step": 1721
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 6.603550295857988e-05,
-      "loss": 1.7694,
+      "epoch": 7.69,
+      "learning_rate": 8.4625e-05,
+      "loss": 1.2583,
       "step": 1722
     },
     {
-      "epoch": 5.1,
-      "learning_rate": 6.601577909270217e-05,
-      "loss": 1.585,
+      "epoch": 7.69,
+      "learning_rate": 8.461607142857144e-05,
+      "loss": 1.2953,
       "step": 1723
     },
     {
-      "epoch": 5.1,
-      "learning_rate": 6.599605522682447e-05,
-      "loss": 1.5637,
+      "epoch": 7.7,
+      "learning_rate": 8.460714285714286e-05,
+      "loss": 1.278,
       "step": 1724
     },
     {
-      "epoch": 5.1,
-      "learning_rate": 6.597633136094676e-05,
-      "loss": 1.6366,
+      "epoch": 7.7,
+      "learning_rate": 8.45982142857143e-05,
+      "loss": 1.3429,
       "step": 1725
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 6.595660749506904e-05,
-      "loss": 1.6975,
+      "epoch": 7.71,
+      "learning_rate": 8.458928571428571e-05,
+      "loss": 1.1828,
       "step": 1726
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 6.593688362919133e-05,
-      "loss": 1.6301,
+      "epoch": 7.71,
+      "learning_rate": 8.458035714285715e-05,
+      "loss": 1.2875,
       "step": 1727
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 6.591715976331361e-05,
-      "loss": 1.6111,
+      "epoch": 7.71,
+      "learning_rate": 8.457142857142858e-05,
+      "loss": 1.3019,
       "step": 1728
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 6.58974358974359e-05,
-      "loss": 1.6016,
+      "epoch": 7.72,
+      "learning_rate": 8.45625e-05,
+      "loss": 1.2747,
       "step": 1729
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 6.587771203155819e-05,
-      "loss": 1.5586,
+      "epoch": 7.72,
+      "learning_rate": 8.455357142857143e-05,
+      "loss": 1.3318,
       "step": 1730
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 6.585798816568047e-05,
-      "loss": 1.7331,
+      "epoch": 7.73,
+      "learning_rate": 8.454464285714287e-05,
+      "loss": 1.3093,
       "step": 1731
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 6.583826429980276e-05,
-      "loss": 1.6426,
+      "epoch": 7.73,
+      "learning_rate": 8.453571428571428e-05,
+      "loss": 1.2928,
       "step": 1732
     },
     {
-      "epoch": 5.13,
-      "learning_rate": 6.581854043392505e-05,
-      "loss": 1.6112,
+      "epoch": 7.74,
+      "learning_rate": 8.452678571428572e-05,
+      "loss": 1.2545,
       "step": 1733
     },
     {
-      "epoch": 5.13,
-      "learning_rate": 6.579881656804733e-05,
-      "loss": 1.6679,
+      "epoch": 7.74,
+      "learning_rate": 8.451785714285714e-05,
+      "loss": 1.4179,
       "step": 1734
     },
     {
-      "epoch": 5.13,
-      "learning_rate": 6.577909270216963e-05,
-      "loss": 1.6295,
+      "epoch": 7.75,
+      "learning_rate": 8.450892857142858e-05,
+      "loss": 1.3877,
       "step": 1735
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 6.575936883629192e-05,
-      "loss": 1.7395,
+      "epoch": 7.75,
+      "learning_rate": 8.450000000000001e-05,
+      "loss": 1.3207,
       "step": 1736
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 6.573964497041421e-05,
-      "loss": 1.7692,
+      "epoch": 7.75,
+      "learning_rate": 8.449107142857143e-05,
+      "loss": 1.2022,
       "step": 1737
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 6.57199211045365e-05,
-      "loss": 1.6747,
+      "epoch": 7.76,
+      "learning_rate": 8.448214285714287e-05,
+      "loss": 1.2808,
       "step": 1738
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 6.570019723865878e-05,
-      "loss": 1.6445,
+      "epoch": 7.76,
+      "learning_rate": 8.447321428571428e-05,
+      "loss": 1.2986,
       "step": 1739
     },
     {
-      "epoch": 5.15,
-      "learning_rate": 6.568047337278107e-05,
-      "loss": 1.7905,
+      "epoch": 7.77,
+      "learning_rate": 8.446428571428572e-05,
+      "loss": 1.2739,
       "step": 1740
     },
     {
-      "epoch": 5.15,
-      "learning_rate": 6.566074950690335e-05,
-      "loss": 1.7085,
+      "epoch": 7.77,
+      "learning_rate": 8.445535714285715e-05,
+      "loss": 1.3019,
       "step": 1741
     },
     {
-      "epoch": 5.15,
-      "learning_rate": 6.564102564102564e-05,
-      "loss": 1.7785,
+      "epoch": 7.78,
+      "learning_rate": 8.444642857142857e-05,
+      "loss": 1.2792,
       "step": 1742
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 6.562130177514794e-05,
-      "loss": 1.6156,
+      "epoch": 7.78,
+      "learning_rate": 8.44375e-05,
+      "loss": 1.254,
       "step": 1743
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 6.560157790927021e-05,
-      "loss": 1.7902,
+      "epoch": 7.79,
+      "learning_rate": 8.442857142857144e-05,
+      "loss": 1.4188,
       "step": 1744
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 6.55818540433925e-05,
-      "loss": 1.5176,
+      "epoch": 7.79,
+      "learning_rate": 8.441964285714286e-05,
+      "loss": 1.2245,
       "step": 1745
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 6.55621301775148e-05,
-      "loss": 1.5902,
+      "epoch": 7.79,
+      "learning_rate": 8.441071428571429e-05,
+      "loss": 1.1509,
       "step": 1746
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 6.554240631163709e-05,
-      "loss": 1.584,
+      "epoch": 7.8,
+      "learning_rate": 8.440178571428571e-05,
+      "loss": 1.3086,
       "step": 1747
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 6.552268244575938e-05,
-      "loss": 1.664,
+      "epoch": 7.8,
+      "learning_rate": 8.439285714285715e-05,
+      "loss": 1.2312,
       "step": 1748
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 6.550295857988166e-05,
-      "loss": 1.6822,
+      "epoch": 7.81,
+      "learning_rate": 8.438392857142857e-05,
+      "loss": 1.2703,
       "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,
+      "epoch": 7.81,
+      "learning_rate": 8.4375e-05,
+      "loss": 1.369,
       "step": 1750
     },
     {
-      "epoch": 5.18,
-      "learning_rate": 6.546351084812623e-05,
-      "loss": 1.6999,
+      "epoch": 7.82,
+      "learning_rate": 8.436607142857143e-05,
+      "loss": 1.3104,
       "step": 1751
     },
     {
-      "epoch": 5.18,
-      "learning_rate": 6.544378698224852e-05,
-      "loss": 1.6479,
+      "epoch": 7.82,
+      "learning_rate": 8.435714285714286e-05,
+      "loss": 1.2442,
       "step": 1752
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 6.542406311637082e-05,
-      "loss": 1.6849,
+      "epoch": 7.83,
+      "learning_rate": 8.43482142857143e-05,
+      "loss": 1.2945,
       "step": 1753
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 6.54043392504931e-05,
-      "loss": 1.6577,
+      "epoch": 7.83,
+      "learning_rate": 8.433928571428572e-05,
+      "loss": 1.2897,
       "step": 1754
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 6.538461538461539e-05,
-      "loss": 1.697,
+      "epoch": 7.83,
+      "learning_rate": 8.433035714285715e-05,
+      "loss": 1.2955,
       "step": 1755
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 6.536489151873768e-05,
-      "loss": 1.6922,
+      "epoch": 7.84,
+      "learning_rate": 8.432142857142857e-05,
+      "loss": 1.3474,
       "step": 1756
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 6.534516765285996e-05,
-      "loss": 1.6819,
+      "epoch": 7.84,
+      "learning_rate": 8.431250000000001e-05,
+      "loss": 1.2467,
       "step": 1757
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 6.532544378698225e-05,
-      "loss": 1.6994,
+      "epoch": 7.85,
+      "learning_rate": 8.430357142857144e-05,
+      "loss": 1.31,
       "step": 1758
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 6.530571992110454e-05,
-      "loss": 1.5715,
+      "epoch": 7.85,
+      "learning_rate": 8.429464285714286e-05,
+      "loss": 1.3254,
       "step": 1759
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 6.528599605522683e-05,
-      "loss": 1.7086,
+      "epoch": 7.86,
+      "learning_rate": 8.428571428571429e-05,
+      "loss": 1.245,
       "step": 1760
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 6.526627218934913e-05,
-      "loss": 1.6502,
+      "epoch": 7.86,
+      "learning_rate": 8.427678571428573e-05,
+      "loss": 1.119,
       "step": 1761
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 6.52465483234714e-05,
-      "loss": 1.6834,
+      "epoch": 7.87,
+      "learning_rate": 8.426785714285714e-05,
+      "loss": 1.2399,
       "step": 1762
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 6.52268244575937e-05,
-      "loss": 1.6329,
+      "epoch": 7.87,
+      "learning_rate": 8.425892857142858e-05,
+      "loss": 1.3052,
       "step": 1763
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 6.520710059171597e-05,
-      "loss": 1.6813,
+      "epoch": 7.88,
+      "learning_rate": 8.425e-05,
+      "loss": 1.1807,
       "step": 1764
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 6.518737672583827e-05,
-      "loss": 1.5854,
+      "epoch": 7.88,
+      "learning_rate": 8.424107142857143e-05,
+      "loss": 1.1975,
       "step": 1765
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 6.516765285996056e-05,
-      "loss": 1.6623,
+      "epoch": 7.88,
+      "learning_rate": 8.423214285714287e-05,
+      "loss": 1.308,
       "step": 1766
     },
     {
-      "epoch": 5.23,
-      "learning_rate": 6.514792899408284e-05,
-      "loss": 1.7244,
+      "epoch": 7.89,
+      "learning_rate": 8.422321428571429e-05,
+      "loss": 1.2928,
       "step": 1767
     },
     {
-      "epoch": 5.23,
-      "learning_rate": 6.512820512820513e-05,
-      "loss": 1.7064,
+      "epoch": 7.89,
+      "learning_rate": 8.421428571428572e-05,
+      "loss": 1.2857,
       "step": 1768
     },
     {
-      "epoch": 5.23,
-      "learning_rate": 6.510848126232741e-05,
-      "loss": 1.734,
+      "epoch": 7.9,
+      "learning_rate": 8.420535714285714e-05,
+      "loss": 1.2686,
       "step": 1769
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 6.50887573964497e-05,
-      "loss": 1.6468,
+      "epoch": 7.9,
+      "learning_rate": 8.419642857142858e-05,
+      "loss": 1.3517,
       "step": 1770
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 6.506903353057199e-05,
-      "loss": 1.6086,
+      "epoch": 7.91,
+      "learning_rate": 8.41875e-05,
+      "loss": 1.2465,
       "step": 1771
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 6.504930966469428e-05,
-      "loss": 1.6016,
+      "epoch": 7.91,
+      "learning_rate": 8.417857142857143e-05,
+      "loss": 1.2231,
       "step": 1772
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 6.502958579881658e-05,
-      "loss": 1.6209,
+      "epoch": 7.92,
+      "learning_rate": 8.416964285714286e-05,
+      "loss": 1.3109,
       "step": 1773
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 6.500986193293886e-05,
-      "loss": 1.6177,
+      "epoch": 7.92,
+      "learning_rate": 8.41607142857143e-05,
+      "loss": 1.2183,
       "step": 1774
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 6.499013806706115e-05,
-      "loss": 1.5701,
+      "epoch": 7.92,
+      "learning_rate": 8.415178571428571e-05,
+      "loss": 1.3258,
       "step": 1775
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 6.497041420118344e-05,
-      "loss": 1.6554,
+      "epoch": 7.93,
+      "learning_rate": 8.414285714285715e-05,
+      "loss": 1.2524,
       "step": 1776
     },
     {
-      "epoch": 5.26,
-      "learning_rate": 6.495069033530572e-05,
-      "loss": 1.6647,
+      "epoch": 7.93,
+      "learning_rate": 8.413392857142857e-05,
+      "loss": 1.2449,
       "step": 1777
     },
     {
-      "epoch": 5.26,
-      "learning_rate": 6.493096646942801e-05,
-      "loss": 1.5974,
+      "epoch": 7.94,
+      "learning_rate": 8.412500000000001e-05,
+      "loss": 1.2324,
       "step": 1778
     },
     {
-      "epoch": 5.26,
-      "learning_rate": 6.49112426035503e-05,
-      "loss": 1.6408,
+      "epoch": 7.94,
+      "learning_rate": 8.411607142857142e-05,
+      "loss": 1.301,
       "step": 1779
     },
     {
-      "epoch": 5.27,
-      "learning_rate": 6.489151873767258e-05,
-      "loss": 1.6932,
+      "epoch": 7.95,
+      "learning_rate": 8.410714285714286e-05,
+      "loss": 1.2829,
       "step": 1780
     },
     {
-      "epoch": 5.27,
-      "learning_rate": 6.487179487179487e-05,
-      "loss": 1.6165,
+      "epoch": 7.95,
+      "learning_rate": 8.409821428571429e-05,
+      "loss": 1.2086,
       "step": 1781
     },
     {
-      "epoch": 5.27,
-      "learning_rate": 6.485207100591716e-05,
-      "loss": 1.6844,
+      "epoch": 7.96,
+      "learning_rate": 8.408928571428571e-05,
+      "loss": 1.194,
       "step": 1782
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 6.483234714003946e-05,
-      "loss": 1.664,
+      "epoch": 7.96,
+      "learning_rate": 8.408035714285715e-05,
+      "loss": 1.1942,
       "step": 1783
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 6.481262327416175e-05,
-      "loss": 1.7908,
+      "epoch": 7.96,
+      "learning_rate": 8.407142857142858e-05,
+      "loss": 1.1858,
       "step": 1784
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 6.479289940828403e-05,
-      "loss": 1.7524,
+      "epoch": 7.97,
+      "learning_rate": 8.40625e-05,
+      "loss": 1.2691,
       "step": 1785
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 6.477317554240632e-05,
-      "loss": 1.649,
+      "epoch": 7.97,
+      "learning_rate": 8.405357142857143e-05,
+      "loss": 1.2615,
       "step": 1786
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 6.47534516765286e-05,
-      "loss": 1.5624,
+      "epoch": 7.98,
+      "learning_rate": 8.404464285714287e-05,
+      "loss": 1.3228,
       "step": 1787
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 6.473372781065089e-05,
-      "loss": 1.6394,
+      "epoch": 7.98,
+      "learning_rate": 8.403571428571428e-05,
+      "loss": 1.3383,
       "step": 1788
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 6.471400394477318e-05,
-      "loss": 1.6399,
+      "epoch": 7.99,
+      "learning_rate": 8.402678571428572e-05,
+      "loss": 1.2244,
       "step": 1789
     },
     {
-      "epoch": 5.3,
-      "learning_rate": 6.469428007889546e-05,
-      "loss": 1.5973,
+      "epoch": 7.99,
+      "learning_rate": 8.401785714285714e-05,
+      "loss": 1.2694,
       "step": 1790
     },
     {
-      "epoch": 5.3,
-      "learning_rate": 6.467455621301775e-05,
-      "loss": 1.6413,
+      "epoch": 8.0,
+      "learning_rate": 8.400892857142858e-05,
+      "loss": 1.3087,
       "step": 1791
     },
     {
-      "epoch": 5.3,
-      "learning_rate": 6.465483234714003e-05,
-      "loss": 1.6051,
+      "epoch": 8.0,
+      "learning_rate": 8.4e-05,
+      "loss": 1.318,
       "step": 1792
     },
     {
-      "epoch": 5.3,
-      "learning_rate": 6.463510848126232e-05,
-      "loss": 1.7984,
+      "epoch": 8.0,
+      "learning_rate": 8.399107142857143e-05,
+      "loss": 1.205,
       "step": 1793
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 6.461538461538462e-05,
-      "loss": 1.703,
+      "epoch": 8.01,
+      "learning_rate": 8.398214285714286e-05,
+      "loss": 1.1739,
       "step": 1794
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 6.459566074950691e-05,
-      "loss": 1.6493,
+      "epoch": 8.01,
+      "learning_rate": 8.397321428571429e-05,
+      "loss": 1.229,
       "step": 1795
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 6.45759368836292e-05,
-      "loss": 1.6355,
+      "epoch": 8.02,
+      "learning_rate": 8.396428571428572e-05,
+      "loss": 1.2003,
       "step": 1796
     },
     {
-      "epoch": 5.32,
-      "learning_rate": 6.455621301775149e-05,
-      "loss": 1.7316,
+      "epoch": 8.02,
+      "learning_rate": 8.395535714285715e-05,
+      "loss": 1.2641,
       "step": 1797
     },
     {
-      "epoch": 5.32,
-      "learning_rate": 6.453648915187377e-05,
-      "loss": 1.5653,
+      "epoch": 8.03,
+      "learning_rate": 8.394642857142858e-05,
+      "loss": 1.2996,
       "step": 1798
     },
     {
-      "epoch": 5.32,
-      "learning_rate": 6.451676528599606e-05,
-      "loss": 1.5851,
+      "epoch": 8.03,
+      "learning_rate": 8.39375e-05,
+      "loss": 1.1516,
       "step": 1799
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 6.449704142011834e-05,
-      "loss": 1.7155,
+      "epoch": 8.04,
+      "learning_rate": 8.392857142857144e-05,
+      "loss": 1.2768,
       "step": 1800
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 6.447731755424063e-05,
-      "loss": 1.5956,
+      "epoch": 8.04,
+      "eval_loss": 1.2090328931808472,
+      "eval_runtime": 46.031,
+      "eval_samples_per_second": 284.461,
+      "eval_steps_per_second": 1.499,
+      "step": 1800
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 8.391964285714287e-05,
+      "loss": 1.183,
       "step": 1801
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 6.445759368836293e-05,
-      "loss": 1.7257,
+      "epoch": 8.04,
+      "learning_rate": 8.391071428571429e-05,
+      "loss": 1.2355,
       "step": 1802
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 6.44378698224852e-05,
-      "loss": 1.7179,
+      "epoch": 8.05,
+      "learning_rate": 8.390178571428572e-05,
+      "loss": 1.2936,
       "step": 1803
     },
     {
-      "epoch": 5.34,
-      "learning_rate": 6.44181459566075e-05,
-      "loss": 1.7321,
+      "epoch": 8.05,
+      "learning_rate": 8.389285714285715e-05,
+      "loss": 1.2972,
       "step": 1804
     },
     {
-      "epoch": 5.34,
-      "learning_rate": 6.439842209072979e-05,
-      "loss": 1.5654,
+      "epoch": 8.06,
+      "learning_rate": 8.388392857142857e-05,
+      "loss": 1.287,
       "step": 1805
     },
     {
-      "epoch": 5.34,
-      "learning_rate": 6.437869822485208e-05,
-      "loss": 1.6664,
+      "epoch": 8.06,
+      "learning_rate": 8.3875e-05,
+      "loss": 1.2252,
       "step": 1806
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 6.435897435897437e-05,
-      "loss": 1.5924,
+      "epoch": 8.07,
+      "learning_rate": 8.386607142857143e-05,
+      "loss": 1.2254,
       "step": 1807
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 6.433925049309665e-05,
-      "loss": 1.5771,
+      "epoch": 8.07,
+      "learning_rate": 8.385714285714286e-05,
+      "loss": 1.2293,
       "step": 1808
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 6.431952662721894e-05,
-      "loss": 1.6284,
+      "epoch": 8.08,
+      "learning_rate": 8.384821428571428e-05,
+      "loss": 1.2946,
       "step": 1809
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 6.429980276134122e-05,
-      "loss": 1.633,
+      "epoch": 8.08,
+      "learning_rate": 8.383928571428572e-05,
+      "loss": 1.2595,
       "step": 1810
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 6.428007889546351e-05,
-      "loss": 1.6503,
+      "epoch": 8.08,
+      "learning_rate": 8.383035714285715e-05,
+      "loss": 1.2687,
       "step": 1811
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 6.42603550295858e-05,
-      "loss": 1.5963,
+      "epoch": 8.09,
+      "learning_rate": 8.382142857142857e-05,
+      "loss": 1.2218,
       "step": 1812
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 6.424063116370808e-05,
-      "loss": 1.6556,
+      "epoch": 8.09,
+      "learning_rate": 8.381250000000001e-05,
+      "loss": 1.2311,
       "step": 1813
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 6.422090729783038e-05,
-      "loss": 1.6577,
+      "epoch": 8.1,
+      "learning_rate": 8.380357142857144e-05,
+      "loss": 1.175,
       "step": 1814
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 6.420118343195266e-05,
-      "loss": 1.6834,
+      "epoch": 8.1,
+      "learning_rate": 8.379464285714286e-05,
+      "loss": 1.2618,
       "step": 1815
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 6.418145956607495e-05,
-      "loss": 1.6147,
+      "epoch": 8.11,
+      "learning_rate": 8.378571428571429e-05,
+      "loss": 1.3516,
       "step": 1816
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 6.416173570019724e-05,
-      "loss": 1.7063,
+      "epoch": 8.11,
+      "learning_rate": 8.377678571428573e-05,
+      "loss": 1.2579,
       "step": 1817
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 6.414201183431953e-05,
-      "loss": 1.6561,
+      "epoch": 8.12,
+      "learning_rate": 8.376785714285714e-05,
+      "loss": 1.2623,
       "step": 1818
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 6.412228796844182e-05,
-      "loss": 1.6494,
+      "epoch": 8.12,
+      "learning_rate": 8.375892857142858e-05,
+      "loss": 1.2031,
       "step": 1819
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 6.410256410256412e-05,
-      "loss": 1.6226,
+      "epoch": 8.12,
+      "learning_rate": 8.375e-05,
+      "loss": 1.2312,
       "step": 1820
     },
     {
-      "epoch": 5.39,
-      "learning_rate": 6.40828402366864e-05,
-      "loss": 1.7249,
+      "epoch": 8.13,
+      "learning_rate": 8.374107142857143e-05,
+      "loss": 1.2918,
       "step": 1821
     },
     {
-      "epoch": 5.39,
-      "learning_rate": 6.406311637080869e-05,
-      "loss": 1.6362,
+      "epoch": 8.13,
+      "learning_rate": 8.373214285714285e-05,
+      "loss": 1.254,
       "step": 1822
     },
     {
-      "epoch": 5.39,
-      "learning_rate": 6.404339250493097e-05,
-      "loss": 1.6458,
+      "epoch": 8.14,
+      "learning_rate": 8.372321428571429e-05,
+      "loss": 1.2501,
       "step": 1823
     },
     {
-      "epoch": 5.4,
-      "learning_rate": 6.402366863905326e-05,
-      "loss": 1.5809,
+      "epoch": 8.14,
+      "learning_rate": 8.371428571428572e-05,
+      "loss": 1.2603,
       "step": 1824
     },
     {
-      "epoch": 5.4,
-      "learning_rate": 6.400394477317555e-05,
-      "loss": 1.6508,
+      "epoch": 8.15,
+      "learning_rate": 8.370535714285714e-05,
+      "loss": 1.2243,
       "step": 1825
     },
     {
-      "epoch": 5.4,
-      "learning_rate": 6.398422090729783e-05,
-      "loss": 1.6044,
+      "epoch": 8.15,
+      "learning_rate": 8.369642857142857e-05,
+      "loss": 1.2198,
       "step": 1826
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 6.396449704142012e-05,
-      "loss": 1.5575,
+      "epoch": 8.16,
+      "learning_rate": 8.368750000000001e-05,
+      "loss": 1.281,
       "step": 1827
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 6.39447731755424e-05,
-      "loss": 1.7022,
+      "epoch": 8.16,
+      "learning_rate": 8.367857142857143e-05,
+      "loss": 1.0902,
       "step": 1828
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 6.392504930966469e-05,
-      "loss": 1.6891,
+      "epoch": 8.17,
+      "learning_rate": 8.366964285714286e-05,
+      "loss": 1.3156,
       "step": 1829
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 6.390532544378698e-05,
-      "loss": 1.6565,
+      "epoch": 8.17,
+      "learning_rate": 8.36607142857143e-05,
+      "loss": 1.2207,
       "step": 1830
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 6.388560157790928e-05,
-      "loss": 1.6505,
+      "epoch": 8.17,
+      "learning_rate": 8.365178571428571e-05,
+      "loss": 1.2493,
       "step": 1831
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 6.386587771203157e-05,
-      "loss": 1.602,
+      "epoch": 8.18,
+      "learning_rate": 8.364285714285715e-05,
+      "loss": 1.1706,
       "step": 1832
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 6.384615384615385e-05,
-      "loss": 1.6895,
+      "epoch": 8.18,
+      "learning_rate": 8.363392857142857e-05,
+      "loss": 1.3344,
       "step": 1833
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 6.382642998027614e-05,
-      "loss": 1.6774,
+      "epoch": 8.19,
+      "learning_rate": 8.362500000000001e-05,
+      "loss": 1.2254,
       "step": 1834
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 6.380670611439843e-05,
-      "loss": 1.7063,
+      "epoch": 8.19,
+      "learning_rate": 8.361607142857143e-05,
+      "loss": 1.2493,
       "step": 1835
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 6.378698224852071e-05,
-      "loss": 1.6098,
+      "epoch": 8.2,
+      "learning_rate": 8.360714285714286e-05,
+      "loss": 1.2908,
       "step": 1836
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 6.3767258382643e-05,
-      "loss": 1.5493,
+      "epoch": 8.2,
+      "learning_rate": 8.359821428571429e-05,
+      "loss": 1.2766,
       "step": 1837
     },
     {
-      "epoch": 5.44,
-      "learning_rate": 6.374753451676528e-05,
-      "loss": 1.7638,
+      "epoch": 8.21,
+      "learning_rate": 8.358928571428572e-05,
+      "loss": 1.3055,
       "step": 1838
     },
     {
-      "epoch": 5.44,
-      "learning_rate": 6.372781065088757e-05,
-      "loss": 1.5486,
+      "epoch": 8.21,
+      "learning_rate": 8.358035714285714e-05,
+      "loss": 1.2454,
       "step": 1839
     },
     {
-      "epoch": 5.44,
-      "learning_rate": 6.370808678500986e-05,
-      "loss": 1.5467,
+      "epoch": 8.21,
+      "learning_rate": 8.357142857142858e-05,
+      "loss": 1.2931,
       "step": 1840
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 6.368836291913216e-05,
-      "loss": 1.7039,
+      "epoch": 8.22,
+      "learning_rate": 8.35625e-05,
+      "loss": 1.1867,
       "step": 1841
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 6.366863905325445e-05,
-      "loss": 1.6133,
+      "epoch": 8.22,
+      "learning_rate": 8.355357142857143e-05,
+      "loss": 1.2695,
       "step": 1842
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 6.364891518737674e-05,
-      "loss": 1.6649,
+      "epoch": 8.23,
+      "learning_rate": 8.354464285714287e-05,
+      "loss": 1.2852,
       "step": 1843
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 6.362919132149902e-05,
-      "loss": 1.619,
+      "epoch": 8.23,
+      "learning_rate": 8.35357142857143e-05,
+      "loss": 1.1867,
       "step": 1844
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 6.360946745562131e-05,
-      "loss": 1.6084,
+      "epoch": 8.24,
+      "learning_rate": 8.352678571428572e-05,
+      "loss": 1.2355,
       "step": 1845
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 6.358974358974359e-05,
-      "loss": 1.778,
+      "epoch": 8.24,
+      "learning_rate": 8.351785714285715e-05,
+      "loss": 1.3032,
       "step": 1846
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 6.357001972386588e-05,
-      "loss": 1.5867,
+      "epoch": 8.25,
+      "learning_rate": 8.350892857142858e-05,
+      "loss": 1.1947,
       "step": 1847
     },
     {
-      "epoch": 5.47,
-      "learning_rate": 6.355029585798817e-05,
-      "loss": 1.6142,
+      "epoch": 8.25,
+      "learning_rate": 8.35e-05,
+      "loss": 1.2797,
       "step": 1848
     },
     {
-      "epoch": 5.47,
-      "learning_rate": 6.353057199211045e-05,
-      "loss": 1.5622,
+      "epoch": 8.25,
+      "learning_rate": 8.349107142857144e-05,
+      "loss": 1.236,
       "step": 1849
     },
     {
-      "epoch": 5.47,
-      "learning_rate": 6.351084812623274e-05,
-      "loss": 1.6251,
+      "epoch": 8.26,
+      "learning_rate": 8.348214285714286e-05,
+      "loss": 1.3526,
       "step": 1850
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 6.349112426035502e-05,
-      "loss": 1.6941,
+      "epoch": 8.26,
+      "learning_rate": 8.347321428571429e-05,
+      "loss": 1.2548,
       "step": 1851
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 6.347140039447731e-05,
-      "loss": 1.7277,
+      "epoch": 8.27,
+      "learning_rate": 8.346428571428571e-05,
+      "loss": 1.2959,
       "step": 1852
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 6.34516765285996e-05,
-      "loss": 1.6248,
+      "epoch": 8.27,
+      "learning_rate": 8.345535714285715e-05,
+      "loss": 1.3169,
       "step": 1853
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 6.34319526627219e-05,
-      "loss": 1.6421,
+      "epoch": 8.28,
+      "learning_rate": 8.344642857142858e-05,
+      "loss": 1.2302,
       "step": 1854
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 6.341222879684419e-05,
-      "loss": 1.6356,
+      "epoch": 8.28,
+      "learning_rate": 8.34375e-05,
+      "loss": 1.1605,
       "step": 1855
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 6.339250493096647e-05,
-      "loss": 1.6892,
+      "epoch": 8.29,
+      "learning_rate": 8.342857142857143e-05,
+      "loss": 1.2609,
       "step": 1856
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 6.337278106508876e-05,
-      "loss": 1.6854,
+      "epoch": 8.29,
+      "learning_rate": 8.341964285714287e-05,
+      "loss": 1.3158,
       "step": 1857
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 6.335305719921105e-05,
-      "loss": 1.6351,
+      "epoch": 8.29,
+      "learning_rate": 8.341071428571429e-05,
+      "loss": 1.3449,
       "step": 1858
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 6.333333333333333e-05,
-      "loss": 1.6677,
+      "epoch": 8.3,
+      "learning_rate": 8.340178571428572e-05,
+      "loss": 1.1981,
       "step": 1859
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 6.331360946745562e-05,
-      "loss": 1.748,
+      "epoch": 8.3,
+      "learning_rate": 8.339285714285716e-05,
+      "loss": 1.2334,
       "step": 1860
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 6.32938856015779e-05,
-      "loss": 1.6874,
+      "epoch": 8.31,
+      "learning_rate": 8.338392857142857e-05,
+      "loss": 1.2315,
       "step": 1861
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 6.32741617357002e-05,
-      "loss": 1.5966,
+      "epoch": 8.31,
+      "learning_rate": 8.337500000000001e-05,
+      "loss": 1.1864,
       "step": 1862
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 6.325443786982249e-05,
-      "loss": 1.6354,
+      "epoch": 8.32,
+      "learning_rate": 8.336607142857143e-05,
+      "loss": 1.2465,
       "step": 1863
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 6.323471400394478e-05,
-      "loss": 1.6946,
+      "epoch": 8.32,
+      "learning_rate": 8.335714285714286e-05,
+      "loss": 1.2088,
       "step": 1864
     },
     {
-      "epoch": 5.52,
-      "learning_rate": 6.321499013806707e-05,
-      "loss": 1.7417,
+      "epoch": 8.33,
+      "learning_rate": 8.334821428571428e-05,
+      "loss": 1.3303,
       "step": 1865
     },
     {
-      "epoch": 5.52,
-      "learning_rate": 6.319526627218936e-05,
-      "loss": 1.6535,
+      "epoch": 8.33,
+      "learning_rate": 8.333928571428572e-05,
+      "loss": 1.22,
       "step": 1866
     },
     {
-      "epoch": 5.52,
-      "learning_rate": 6.317554240631164e-05,
-      "loss": 1.7221,
+      "epoch": 8.33,
+      "learning_rate": 8.333035714285715e-05,
+      "loss": 1.3273,
       "step": 1867
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 6.315581854043393e-05,
-      "loss": 1.7401,
+      "epoch": 8.34,
+      "learning_rate": 8.332142857142857e-05,
+      "loss": 1.2877,
       "step": 1868
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 6.313609467455621e-05,
-      "loss": 1.5997,
+      "epoch": 8.34,
+      "learning_rate": 8.33125e-05,
+      "loss": 1.2183,
       "step": 1869
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 6.31163708086785e-05,
-      "loss": 1.6496,
+      "epoch": 8.35,
+      "learning_rate": 8.330357142857144e-05,
+      "loss": 1.3573,
       "step": 1870
     },
     {
-      "epoch": 5.54,
-      "learning_rate": 6.30966469428008e-05,
-      "loss": 1.6385,
+      "epoch": 8.35,
+      "learning_rate": 8.329464285714286e-05,
+      "loss": 1.2154,
       "step": 1871
     },
     {
-      "epoch": 5.54,
-      "learning_rate": 6.307692307692308e-05,
-      "loss": 1.5846,
+      "epoch": 8.36,
+      "learning_rate": 8.328571428571429e-05,
+      "loss": 1.1974,
       "step": 1872
     },
     {
-      "epoch": 5.54,
-      "learning_rate": 6.305719921104537e-05,
-      "loss": 1.5461,
+      "epoch": 8.36,
+      "learning_rate": 8.327678571428571e-05,
+      "loss": 1.1725,
       "step": 1873
     },
     {
-      "epoch": 5.54,
-      "learning_rate": 6.303747534516765e-05,
-      "loss": 1.7008,
+      "epoch": 8.37,
+      "learning_rate": 8.326785714285714e-05,
+      "loss": 1.1916,
       "step": 1874
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 6.301775147928994e-05,
-      "loss": 1.6542,
+      "epoch": 8.37,
+      "learning_rate": 8.325892857142858e-05,
+      "loss": 1.2527,
       "step": 1875
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 6.299802761341223e-05,
-      "loss": 1.6809,
+      "epoch": 8.38,
+      "learning_rate": 8.325e-05,
+      "loss": 1.2824,
       "step": 1876
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 6.297830374753452e-05,
-      "loss": 1.6583,
+      "epoch": 8.38,
+      "learning_rate": 8.324107142857144e-05,
+      "loss": 1.3335,
       "step": 1877
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 6.295857988165681e-05,
-      "loss": 1.6528,
+      "epoch": 8.38,
+      "learning_rate": 8.323214285714286e-05,
+      "loss": 1.2082,
       "step": 1878
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 6.29388560157791e-05,
-      "loss": 1.632,
+      "epoch": 8.39,
+      "learning_rate": 8.32232142857143e-05,
+      "loss": 1.2788,
       "step": 1879
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 6.291913214990139e-05,
-      "loss": 1.6759,
+      "epoch": 8.39,
+      "learning_rate": 8.321428571428572e-05,
+      "loss": 1.2743,
       "step": 1880
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 6.289940828402368e-05,
-      "loss": 1.6115,
+      "epoch": 8.4,
+      "learning_rate": 8.320535714285715e-05,
+      "loss": 1.2922,
       "step": 1881
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 6.287968441814596e-05,
-      "loss": 1.6452,
+      "epoch": 8.4,
+      "learning_rate": 8.319642857142857e-05,
+      "loss": 1.2437,
       "step": 1882
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 6.285996055226825e-05,
-      "loss": 1.6374,
+      "epoch": 8.41,
+      "learning_rate": 8.318750000000001e-05,
+      "loss": 1.2871,
       "step": 1883
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 6.284023668639053e-05,
-      "loss": 1.6466,
+      "epoch": 8.41,
+      "learning_rate": 8.317857142857142e-05,
+      "loss": 1.2196,
       "step": 1884
     },
     {
-      "epoch": 5.58,
-      "learning_rate": 6.282051282051282e-05,
-      "loss": 1.6743,
+      "epoch": 8.42,
+      "learning_rate": 8.316964285714286e-05,
+      "loss": 1.1609,
       "step": 1885
     },
     {
-      "epoch": 5.58,
-      "learning_rate": 6.280078895463511e-05,
-      "loss": 1.7694,
+      "epoch": 8.42,
+      "learning_rate": 8.316071428571429e-05,
+      "loss": 1.3312,
       "step": 1886
     },
     {
-      "epoch": 5.58,
-      "learning_rate": 6.278106508875739e-05,
-      "loss": 1.6999,
+      "epoch": 8.42,
+      "learning_rate": 8.315178571428572e-05,
+      "loss": 1.2212,
       "step": 1887
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 6.276134122287968e-05,
-      "loss": 1.5366,
+      "epoch": 8.43,
+      "learning_rate": 8.314285714285715e-05,
+      "loss": 1.2951,
       "step": 1888
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 6.274161735700199e-05,
-      "loss": 1.6058,
+      "epoch": 8.43,
+      "learning_rate": 8.313392857142858e-05,
+      "loss": 1.3359,
       "step": 1889
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 6.272189349112427e-05,
-      "loss": 1.5548,
+      "epoch": 8.44,
+      "learning_rate": 8.312500000000001e-05,
+      "loss": 1.2518,
       "step": 1890
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 6.270216962524656e-05,
-      "loss": 1.574,
+      "epoch": 8.44,
+      "learning_rate": 8.311607142857143e-05,
+      "loss": 1.2133,
       "step": 1891
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 6.268244575936884e-05,
-      "loss": 1.6095,
+      "epoch": 8.45,
+      "learning_rate": 8.310714285714287e-05,
+      "loss": 1.2063,
       "step": 1892
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 6.266272189349113e-05,
-      "loss": 1.6208,
+      "epoch": 8.45,
+      "learning_rate": 8.309821428571429e-05,
+      "loss": 1.2638,
       "step": 1893
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 6.264299802761342e-05,
-      "loss": 1.6171,
+      "epoch": 8.46,
+      "learning_rate": 8.308928571428572e-05,
+      "loss": 1.2617,
       "step": 1894
     },
     {
-      "epoch": 5.61,
-      "learning_rate": 6.26232741617357e-05,
-      "loss": 1.5873,
+      "epoch": 8.46,
+      "learning_rate": 8.308035714285714e-05,
+      "loss": 1.2615,
       "step": 1895
     },
     {
-      "epoch": 5.61,
-      "learning_rate": 6.260355029585799e-05,
-      "loss": 1.7138,
+      "epoch": 8.46,
+      "learning_rate": 8.307142857142858e-05,
+      "loss": 1.1942,
       "step": 1896
     },
     {
-      "epoch": 5.61,
-      "learning_rate": 6.258382642998027e-05,
-      "loss": 1.6191,
+      "epoch": 8.47,
+      "learning_rate": 8.306249999999999e-05,
+      "loss": 1.2962,
       "step": 1897
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 6.256410256410256e-05,
-      "loss": 1.6254,
+      "epoch": 8.47,
+      "learning_rate": 8.305357142857143e-05,
+      "loss": 1.2604,
       "step": 1898
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 6.254437869822485e-05,
-      "loss": 1.6237,
+      "epoch": 8.48,
+      "learning_rate": 8.304464285714286e-05,
+      "loss": 1.3647,
       "step": 1899
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 6.252465483234715e-05,
-      "loss": 1.6244,
+      "epoch": 8.48,
+      "learning_rate": 8.30357142857143e-05,
+      "loss": 1.2209,
       "step": 1900
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 6.250493096646944e-05,
-      "loss": 1.5734,
+      "epoch": 8.49,
+      "learning_rate": 8.302678571428572e-05,
+      "loss": 1.3072,
       "step": 1901
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 6.248520710059172e-05,
-      "loss": 1.7106,
+      "epoch": 8.49,
+      "learning_rate": 8.301785714285715e-05,
+      "loss": 1.2162,
       "step": 1902
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 6.246548323471401e-05,
-      "loss": 1.6269,
+      "epoch": 8.5,
+      "learning_rate": 8.300892857142857e-05,
+      "loss": 1.2375,
       "step": 1903
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 6.24457593688363e-05,
-      "loss": 1.584,
+      "epoch": 8.5,
+      "learning_rate": 8.3e-05,
+      "loss": 1.1987,
       "step": 1904
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 6.242603550295858e-05,
-      "loss": 1.6794,
+      "epoch": 8.5,
+      "learning_rate": 8.299107142857144e-05,
+      "loss": 1.2244,
       "step": 1905
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 6.240631163708087e-05,
-      "loss": 1.7075,
+      "epoch": 8.51,
+      "learning_rate": 8.298214285714286e-05,
+      "loss": 1.2497,
       "step": 1906
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 6.238658777120316e-05,
-      "loss": 1.7441,
+      "epoch": 8.51,
+      "learning_rate": 8.297321428571429e-05,
+      "loss": 1.2988,
       "step": 1907
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 6.236686390532544e-05,
-      "loss": 1.56,
+      "epoch": 8.52,
+      "learning_rate": 8.296428571428571e-05,
+      "loss": 1.2629,
       "step": 1908
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 6.234714003944773e-05,
-      "loss": 1.6514,
+      "epoch": 8.52,
+      "learning_rate": 8.295535714285715e-05,
+      "loss": 1.2108,
       "step": 1909
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 6.232741617357001e-05,
-      "loss": 1.6301,
+      "epoch": 8.53,
+      "learning_rate": 8.294642857142858e-05,
+      "loss": 1.2661,
       "step": 1910
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 6.23076923076923e-05,
-      "loss": 1.6797,
+      "epoch": 8.53,
+      "learning_rate": 8.29375e-05,
+      "loss": 1.2464,
       "step": 1911
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 6.22879684418146e-05,
-      "loss": 1.7106,
+      "epoch": 8.54,
+      "learning_rate": 8.292857142857143e-05,
+      "loss": 1.2661,
       "step": 1912
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 6.226824457593689e-05,
-      "loss": 1.7865,
+      "epoch": 8.54,
+      "learning_rate": 8.291964285714287e-05,
+      "loss": 1.3357,
       "step": 1913
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 6.224852071005918e-05,
-      "loss": 1.6141,
+      "epoch": 8.54,
+      "learning_rate": 8.291071428571428e-05,
+      "loss": 1.1397,
       "step": 1914
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 6.222879684418146e-05,
-      "loss": 1.6293,
+      "epoch": 8.55,
+      "learning_rate": 8.290178571428572e-05,
+      "loss": 1.2286,
       "step": 1915
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 6.220907297830375e-05,
-      "loss": 1.5419,
+      "epoch": 8.55,
+      "learning_rate": 8.289285714285714e-05,
+      "loss": 1.2411,
       "step": 1916
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 6.218934911242604e-05,
-      "loss": 1.5938,
+      "epoch": 8.56,
+      "learning_rate": 8.288392857142857e-05,
+      "loss": 1.2529,
       "step": 1917
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 6.216962524654832e-05,
-      "loss": 1.6748,
+      "epoch": 8.56,
+      "learning_rate": 8.287500000000001e-05,
+      "loss": 1.2197,
       "step": 1918
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 6.214990138067062e-05,
-      "loss": 1.5019,
+      "epoch": 8.57,
+      "learning_rate": 8.286607142857143e-05,
+      "loss": 1.2241,
       "step": 1919
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 6.21301775147929e-05,
-      "loss": 1.5735,
+      "epoch": 8.57,
+      "learning_rate": 8.285714285714287e-05,
+      "loss": 1.2122,
       "step": 1920
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 6.211045364891519e-05,
-      "loss": 1.6608,
+      "epoch": 8.58,
+      "learning_rate": 8.284821428571429e-05,
+      "loss": 1.2238,
       "step": 1921
     },
     {
-      "epoch": 5.69,
-      "learning_rate": 6.209072978303748e-05,
-      "loss": 1.5384,
+      "epoch": 8.58,
+      "learning_rate": 8.283928571428572e-05,
+      "loss": 1.3328,
       "step": 1922
     },
     {
-      "epoch": 5.69,
-      "learning_rate": 6.207100591715977e-05,
-      "loss": 1.7461,
+      "epoch": 8.58,
+      "learning_rate": 8.283035714285715e-05,
+      "loss": 1.2468,
       "step": 1923
     },
     {
-      "epoch": 5.69,
-      "learning_rate": 6.205128205128206e-05,
-      "loss": 1.6565,
+      "epoch": 8.59,
+      "learning_rate": 8.282142857142857e-05,
+      "loss": 1.2549,
       "step": 1924
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 6.203155818540434e-05,
-      "loss": 1.7347,
+      "epoch": 8.59,
+      "learning_rate": 8.28125e-05,
+      "loss": 1.2553,
       "step": 1925
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 6.201183431952663e-05,
-      "loss": 1.5339,
+      "epoch": 8.6,
+      "learning_rate": 8.280357142857144e-05,
+      "loss": 1.2612,
       "step": 1926
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 6.199211045364892e-05,
-      "loss": 1.6116,
+      "epoch": 8.6,
+      "learning_rate": 8.279464285714285e-05,
+      "loss": 1.2776,
       "step": 1927
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 6.19723865877712e-05,
-      "loss": 1.6182,
+      "epoch": 8.61,
+      "learning_rate": 8.278571428571429e-05,
+      "loss": 1.213,
       "step": 1928
     },
     {
-      "epoch": 5.71,
-      "learning_rate": 6.19526627218935e-05,
-      "loss": 1.6928,
+      "epoch": 8.61,
+      "learning_rate": 8.277678571428572e-05,
+      "loss": 1.2754,
       "step": 1929
     },
     {
-      "epoch": 5.71,
-      "learning_rate": 6.193293885601579e-05,
-      "loss": 1.6215,
+      "epoch": 8.62,
+      "learning_rate": 8.276785714285714e-05,
+      "loss": 1.2447,
       "step": 1930
     },
     {
-      "epoch": 5.71,
-      "learning_rate": 6.191321499013807e-05,
-      "loss": 1.5391,
+      "epoch": 8.62,
+      "learning_rate": 8.275892857142857e-05,
+      "loss": 1.2677,
       "step": 1931
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 6.189349112426036e-05,
-      "loss": 1.6265,
+      "epoch": 8.62,
+      "learning_rate": 8.275e-05,
+      "loss": 1.2263,
       "step": 1932
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 6.187376725838264e-05,
-      "loss": 1.5649,
+      "epoch": 8.63,
+      "learning_rate": 8.274107142857143e-05,
+      "loss": 1.2652,
       "step": 1933
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 6.185404339250493e-05,
-      "loss": 1.6132,
+      "epoch": 8.63,
+      "learning_rate": 8.273214285714286e-05,
+      "loss": 1.2088,
       "step": 1934
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 6.183431952662722e-05,
-      "loss": 1.6076,
+      "epoch": 8.64,
+      "learning_rate": 8.27232142857143e-05,
+      "loss": 1.2302,
       "step": 1935
     },
     {
-      "epoch": 5.73,
-      "learning_rate": 6.181459566074951e-05,
-      "loss": 1.6126,
+      "epoch": 8.64,
+      "learning_rate": 8.271428571428572e-05,
+      "loss": 1.1585,
       "step": 1936
     },
     {
-      "epoch": 5.73,
-      "learning_rate": 6.17948717948718e-05,
-      "loss": 1.5696,
+      "epoch": 8.65,
+      "learning_rate": 8.270535714285715e-05,
+      "loss": 1.2152,
       "step": 1937
     },
     {
-      "epoch": 5.73,
-      "learning_rate": 6.177514792899408e-05,
-      "loss": 1.6056,
+      "epoch": 8.65,
+      "learning_rate": 8.269642857142857e-05,
+      "loss": 1.1372,
       "step": 1938
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 6.175542406311638e-05,
-      "loss": 1.497,
+      "epoch": 8.66,
+      "learning_rate": 8.268750000000001e-05,
+      "loss": 1.2147,
       "step": 1939
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 6.173570019723867e-05,
-      "loss": 1.6595,
+      "epoch": 8.66,
+      "learning_rate": 8.267857142857142e-05,
+      "loss": 1.1808,
       "step": 1940
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 6.171597633136095e-05,
-      "loss": 1.6147,
+      "epoch": 8.67,
+      "learning_rate": 8.266964285714286e-05,
+      "loss": 1.2538,
       "step": 1941
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 6.169625246548324e-05,
-      "loss": 1.6221,
+      "epoch": 8.67,
+      "learning_rate": 8.266071428571429e-05,
+      "loss": 1.2322,
       "step": 1942
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 6.167652859960552e-05,
-      "loss": 1.4354,
+      "epoch": 8.67,
+      "learning_rate": 8.265178571428573e-05,
+      "loss": 1.226,
       "step": 1943
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 6.165680473372781e-05,
-      "loss": 1.672,
+      "epoch": 8.68,
+      "learning_rate": 8.264285714285714e-05,
+      "loss": 1.3264,
       "step": 1944
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 6.16370808678501e-05,
-      "loss": 1.6458,
+      "epoch": 8.68,
+      "learning_rate": 8.263392857142858e-05,
+      "loss": 1.1731,
       "step": 1945
     },
     {
-      "epoch": 5.76,
-      "learning_rate": 6.161735700197238e-05,
-      "loss": 1.6172,
+      "epoch": 8.69,
+      "learning_rate": 8.2625e-05,
+      "loss": 1.2707,
       "step": 1946
     },
     {
-      "epoch": 5.76,
-      "learning_rate": 6.159763313609467e-05,
-      "loss": 1.6525,
+      "epoch": 8.69,
+      "learning_rate": 8.261607142857143e-05,
+      "loss": 1.2398,
       "step": 1947
     },
     {
-      "epoch": 5.76,
-      "learning_rate": 6.157790927021696e-05,
-      "loss": 1.6611,
+      "epoch": 8.7,
+      "learning_rate": 8.260714285714287e-05,
+      "loss": 1.3437,
       "step": 1948
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 6.155818540433926e-05,
-      "loss": 1.6012,
+      "epoch": 8.7,
+      "learning_rate": 8.259821428571429e-05,
+      "loss": 1.2561,
       "step": 1949
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 6.153846153846155e-05,
-      "loss": 1.688,
+      "epoch": 8.71,
+      "learning_rate": 8.258928571428572e-05,
+      "loss": 1.28,
       "step": 1950
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 6.151873767258383e-05,
-      "loss": 1.6337,
+      "epoch": 8.71,
+      "learning_rate": 8.258035714285714e-05,
+      "loss": 1.3149,
       "step": 1951
     },
     {
-      "epoch": 5.78,
-      "learning_rate": 6.149901380670612e-05,
-      "loss": 1.6499,
+      "epoch": 8.71,
+      "learning_rate": 8.257142857142858e-05,
+      "loss": 1.3176,
       "step": 1952
     },
     {
-      "epoch": 5.78,
-      "learning_rate": 6.147928994082841e-05,
-      "loss": 1.6346,
+      "epoch": 8.72,
+      "learning_rate": 8.256250000000001e-05,
+      "loss": 1.2976,
       "step": 1953
     },
     {
-      "epoch": 5.78,
-      "learning_rate": 6.145956607495069e-05,
-      "loss": 1.5665,
+      "epoch": 8.72,
+      "learning_rate": 8.255357142857143e-05,
+      "loss": 1.1668,
       "step": 1954
     },
     {
-      "epoch": 5.78,
-      "learning_rate": 6.143984220907298e-05,
-      "loss": 1.6425,
+      "epoch": 8.73,
+      "learning_rate": 8.254464285714286e-05,
+      "loss": 1.2679,
       "step": 1955
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 6.142011834319526e-05,
-      "loss": 1.5487,
+      "epoch": 8.73,
+      "learning_rate": 8.25357142857143e-05,
+      "loss": 1.2162,
       "step": 1956
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 6.140039447731755e-05,
-      "loss": 1.6378,
+      "epoch": 8.74,
+      "learning_rate": 8.252678571428571e-05,
+      "loss": 1.2689,
       "step": 1957
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 6.138067061143984e-05,
-      "loss": 1.6321,
+      "epoch": 8.74,
+      "learning_rate": 8.251785714285715e-05,
+      "loss": 1.3289,
       "step": 1958
     },
     {
-      "epoch": 5.8,
-      "learning_rate": 6.136094674556214e-05,
-      "loss": 1.6962,
+      "epoch": 8.75,
+      "learning_rate": 8.250892857142857e-05,
+      "loss": 1.1551,
       "step": 1959
     },
     {
-      "epoch": 5.8,
-      "learning_rate": 6.134122287968443e-05,
-      "loss": 1.6388,
+      "epoch": 8.75,
+      "learning_rate": 8.25e-05,
+      "loss": 1.2833,
       "step": 1960
     },
     {
-      "epoch": 5.8,
-      "learning_rate": 6.132149901380671e-05,
-      "loss": 1.6181,
+      "epoch": 8.75,
+      "learning_rate": 8.249107142857142e-05,
+      "loss": 1.283,
       "step": 1961
     },
     {
-      "epoch": 5.8,
-      "learning_rate": 6.1301775147929e-05,
-      "loss": 1.5708,
+      "epoch": 8.76,
+      "learning_rate": 8.248214285714286e-05,
+      "loss": 1.2388,
       "step": 1962
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 6.128205128205129e-05,
-      "loss": 1.6482,
+      "epoch": 8.76,
+      "learning_rate": 8.247321428571429e-05,
+      "loss": 1.3055,
       "step": 1963
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 6.126232741617357e-05,
-      "loss": 1.6112,
+      "epoch": 8.77,
+      "learning_rate": 8.246428571428571e-05,
+      "loss": 1.2121,
       "step": 1964
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 6.124260355029586e-05,
-      "loss": 1.5604,
+      "epoch": 8.77,
+      "learning_rate": 8.245535714285715e-05,
+      "loss": 1.3145,
       "step": 1965
     },
     {
-      "epoch": 5.82,
-      "learning_rate": 6.122287968441814e-05,
-      "loss": 1.6859,
+      "epoch": 8.78,
+      "learning_rate": 8.244642857142858e-05,
+      "loss": 1.3566,
       "step": 1966
     },
     {
-      "epoch": 5.82,
-      "learning_rate": 6.120315581854043e-05,
-      "loss": 1.6221,
+      "epoch": 8.78,
+      "learning_rate": 8.24375e-05,
+      "loss": 1.3505,
       "step": 1967
     },
     {
-      "epoch": 5.82,
-      "learning_rate": 6.118343195266273e-05,
-      "loss": 1.609,
+      "epoch": 8.79,
+      "learning_rate": 8.242857142857143e-05,
+      "loss": 1.2329,
       "step": 1968
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 6.1163708086785e-05,
-      "loss": 1.6889,
+      "epoch": 8.79,
+      "learning_rate": 8.241964285714287e-05,
+      "loss": 1.2904,
       "step": 1969
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 6.11439842209073e-05,
-      "loss": 1.6574,
+      "epoch": 8.79,
+      "learning_rate": 8.241071428571428e-05,
+      "loss": 1.269,
       "step": 1970
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 6.112426035502959e-05,
-      "loss": 1.698,
+      "epoch": 8.8,
+      "learning_rate": 8.240178571428572e-05,
+      "loss": 1.1612,
       "step": 1971
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 6.110453648915188e-05,
-      "loss": 1.5841,
+      "epoch": 8.8,
+      "learning_rate": 8.239285714285715e-05,
+      "loss": 1.3588,
       "step": 1972
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 6.108481262327417e-05,
-      "loss": 1.6886,
+      "epoch": 8.81,
+      "learning_rate": 8.238392857142857e-05,
+      "loss": 1.3042,
       "step": 1973
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 6.106508875739645e-05,
-      "loss": 1.7181,
+      "epoch": 8.81,
+      "learning_rate": 8.2375e-05,
+      "loss": 1.2365,
       "step": 1974
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 6.104536489151874e-05,
-      "loss": 1.6755,
+      "epoch": 8.82,
+      "learning_rate": 8.236607142857144e-05,
+      "loss": 1.2339,
       "step": 1975
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 6.1025641025641035e-05,
-      "loss": 1.5742,
+      "epoch": 8.82,
+      "learning_rate": 8.235714285714286e-05,
+      "loss": 1.2054,
       "step": 1976
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 6.1005917159763314e-05,
-      "loss": 1.5847,
+      "epoch": 8.83,
+      "learning_rate": 8.234821428571429e-05,
+      "loss": 1.2352,
       "step": 1977
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 6.0986193293885606e-05,
-      "loss": 1.5676,
+      "epoch": 8.83,
+      "learning_rate": 8.233928571428573e-05,
+      "loss": 1.3091,
       "step": 1978
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 6.096646942800789e-05,
-      "loss": 1.6134,
+      "epoch": 8.83,
+      "learning_rate": 8.233035714285715e-05,
+      "loss": 1.2572,
       "step": 1979
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 6.094674556213018e-05,
-      "loss": 1.5594,
+      "epoch": 8.84,
+      "learning_rate": 8.232142857142858e-05,
+      "loss": 1.3334,
       "step": 1980
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 6.0927021696252475e-05,
-      "loss": 1.5746,
+      "epoch": 8.84,
+      "learning_rate": 8.23125e-05,
+      "loss": 1.1912,
       "step": 1981
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 6.0907297830374754e-05,
-      "loss": 1.7408,
+      "epoch": 8.85,
+      "learning_rate": 8.230357142857144e-05,
+      "loss": 1.1274,
       "step": 1982
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 6.0887573964497046e-05,
-      "loss": 1.5906,
+      "epoch": 8.85,
+      "learning_rate": 8.229464285714285e-05,
+      "loss": 1.1794,
       "step": 1983
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 6.0867850098619325e-05,
-      "loss": 1.6684,
+      "epoch": 8.86,
+      "learning_rate": 8.228571428571429e-05,
+      "loss": 1.2509,
       "step": 1984
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 6.084812623274162e-05,
-      "loss": 1.5945,
+      "epoch": 8.86,
+      "learning_rate": 8.227678571428572e-05,
+      "loss": 1.2489,
       "step": 1985
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 6.0828402366863916e-05,
-      "loss": 1.6329,
+      "epoch": 8.87,
+      "learning_rate": 8.226785714285716e-05,
+      "loss": 1.294,
       "step": 1986
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 6.0808678500986194e-05,
-      "loss": 1.6363,
+      "epoch": 8.87,
+      "learning_rate": 8.225892857142857e-05,
+      "loss": 1.3075,
       "step": 1987
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 6.0788954635108486e-05,
-      "loss": 1.6287,
+      "epoch": 8.88,
+      "learning_rate": 8.225000000000001e-05,
+      "loss": 1.1853,
       "step": 1988
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 6.0769230769230765e-05,
-      "loss": 1.5999,
+      "epoch": 8.88,
+      "learning_rate": 8.224107142857143e-05,
+      "loss": 1.2821,
       "step": 1989
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 6.074950690335306e-05,
-      "loss": 1.7113,
+      "epoch": 8.88,
+      "learning_rate": 8.223214285714286e-05,
+      "loss": 1.2087,
       "step": 1990
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 6.072978303747535e-05,
-      "loss": 1.6081,
+      "epoch": 8.89,
+      "learning_rate": 8.222321428571428e-05,
+      "loss": 1.2271,
       "step": 1991
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 6.0710059171597635e-05,
-      "loss": 1.7062,
+      "epoch": 8.89,
+      "learning_rate": 8.221428571428572e-05,
+      "loss": 1.2942,
       "step": 1992
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 6.069033530571993e-05,
-      "loss": 1.6882,
+      "epoch": 8.9,
+      "learning_rate": 8.220535714285715e-05,
+      "loss": 1.1909,
       "step": 1993
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 6.0670611439842205e-05,
-      "loss": 1.6784,
+      "epoch": 8.9,
+      "learning_rate": 8.219642857142857e-05,
+      "loss": 1.2594,
       "step": 1994
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 6.06508875739645e-05,
-      "loss": 1.5149,
+      "epoch": 8.91,
+      "learning_rate": 8.218750000000001e-05,
+      "loss": 1.2141,
       "step": 1995
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 6.063116370808679e-05,
-      "loss": 1.6024,
+      "epoch": 8.91,
+      "learning_rate": 8.217857142857142e-05,
+      "loss": 1.2837,
       "step": 1996
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 6.0611439842209075e-05,
-      "loss": 1.5979,
+      "epoch": 8.92,
+      "learning_rate": 8.216964285714286e-05,
+      "loss": 1.2625,
       "step": 1997
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 6.059171597633137e-05,
-      "loss": 1.57,
+      "epoch": 8.92,
+      "learning_rate": 8.216071428571429e-05,
+      "loss": 1.2695,
       "step": 1998
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 6.057199211045366e-05,
-      "loss": 1.5614,
+      "epoch": 8.92,
+      "learning_rate": 8.215178571428573e-05,
+      "loss": 1.2524,
       "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,
+      "epoch": 8.93,
+      "learning_rate": 8.214285714285714e-05,
+      "loss": 1.2973,
       "step": 2000
     },
     {
-      "epoch": 5.92,
-      "learning_rate": 6.053254437869823e-05,
-      "loss": 1.7501,
+      "epoch": 8.93,
+      "learning_rate": 8.213392857142858e-05,
+      "loss": 1.3544,
       "step": 2001
     },
     {
-      "epoch": 5.92,
-      "learning_rate": 6.0512820512820515e-05,
-      "loss": 1.7029,
+      "epoch": 8.94,
+      "learning_rate": 8.2125e-05,
+      "loss": 1.3066,
       "step": 2002
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 6.049309664694281e-05,
-      "loss": 1.5721,
+      "epoch": 8.94,
+      "learning_rate": 8.211607142857143e-05,
+      "loss": 1.2681,
       "step": 2003
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 6.04733727810651e-05,
-      "loss": 1.6621,
+      "epoch": 8.95,
+      "learning_rate": 8.210714285714285e-05,
+      "loss": 1.2882,
       "step": 2004
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 6.045364891518738e-05,
-      "loss": 1.6392,
+      "epoch": 8.95,
+      "learning_rate": 8.20982142857143e-05,
+      "loss": 1.2528,
       "step": 2005
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 6.043392504930967e-05,
-      "loss": 1.715,
+      "epoch": 8.96,
+      "learning_rate": 8.208928571428572e-05,
+      "loss": 1.2726,
       "step": 2006
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 6.041420118343195e-05,
-      "loss": 1.6153,
+      "epoch": 8.96,
+      "learning_rate": 8.208035714285714e-05,
+      "loss": 1.3081,
       "step": 2007
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 6.039447731755424e-05,
-      "loss": 1.7215,
+      "epoch": 8.96,
+      "learning_rate": 8.207142857142857e-05,
+      "loss": 1.2109,
       "step": 2008
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 6.037475345167653e-05,
-      "loss": 1.7396,
+      "epoch": 8.97,
+      "learning_rate": 8.206250000000001e-05,
+      "loss": 1.2731,
       "step": 2009
     },
     {
-      "epoch": 5.95,
-      "learning_rate": 6.035502958579882e-05,
-      "loss": 1.6267,
+      "epoch": 8.97,
+      "learning_rate": 8.205357142857143e-05,
+      "loss": 1.1923,
       "step": 2010
     },
     {
-      "epoch": 5.95,
-      "learning_rate": 6.033530571992111e-05,
-      "loss": 1.5477,
+      "epoch": 8.98,
+      "learning_rate": 8.204464285714286e-05,
+      "loss": 1.2717,
       "step": 2011
     },
     {
-      "epoch": 5.95,
-      "learning_rate": 6.031558185404339e-05,
-      "loss": 1.6819,
+      "epoch": 8.98,
+      "learning_rate": 8.20357142857143e-05,
+      "loss": 1.2403,
       "step": 2012
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 6.029585798816568e-05,
-      "loss": 1.7775,
+      "epoch": 8.99,
+      "learning_rate": 8.202678571428571e-05,
+      "loss": 1.2277,
       "step": 2013
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 6.027613412228797e-05,
-      "loss": 1.6547,
+      "epoch": 8.99,
+      "learning_rate": 8.201785714285715e-05,
+      "loss": 1.2441,
       "step": 2014
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 6.025641025641026e-05,
-      "loss": 1.6192,
+      "epoch": 9.0,
+      "learning_rate": 8.200892857142858e-05,
+      "loss": 1.2113,
       "step": 2015
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 6.023668639053255e-05,
-      "loss": 1.5231,
+      "epoch": 9.0,
+      "learning_rate": 8.2e-05,
+      "loss": 1.2296,
       "step": 2016
     },
     {
-      "epoch": 5.97,
-      "learning_rate": 6.021696252465484e-05,
-      "loss": 1.6265,
+      "epoch": 9.0,
+      "learning_rate": 8.199107142857143e-05,
+      "loss": 1.2517,
       "step": 2017
     },
     {
-      "epoch": 5.97,
-      "learning_rate": 6.019723865877712e-05,
-      "loss": 1.6888,
+      "epoch": 9.01,
+      "learning_rate": 8.198214285714287e-05,
+      "loss": 1.2564,
       "step": 2018
     },
     {
-      "epoch": 5.97,
-      "learning_rate": 6.017751479289941e-05,
-      "loss": 1.5533,
+      "epoch": 9.01,
+      "learning_rate": 8.197321428571429e-05,
+      "loss": 1.2058,
       "step": 2019
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 6.01577909270217e-05,
-      "loss": 1.5677,
+      "epoch": 9.02,
+      "learning_rate": 8.196428571428572e-05,
+      "loss": 1.2162,
       "step": 2020
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 6.013806706114399e-05,
-      "loss": 1.6082,
+      "epoch": 9.02,
+      "learning_rate": 8.195535714285714e-05,
+      "loss": 1.1874,
       "step": 2021
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 6.011834319526628e-05,
-      "loss": 1.639,
+      "epoch": 9.03,
+      "learning_rate": 8.194642857142858e-05,
+      "loss": 1.2942,
       "step": 2022
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 6.009861932938856e-05,
-      "loss": 1.6443,
+      "epoch": 9.03,
+      "learning_rate": 8.19375e-05,
+      "loss": 1.2633,
       "step": 2023
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 6.0078895463510853e-05,
-      "loss": 1.6805,
+      "epoch": 9.04,
+      "learning_rate": 8.192857142857143e-05,
+      "loss": 1.1991,
       "step": 2024
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 6.005917159763313e-05,
-      "loss": 1.5888,
+      "epoch": 9.04,
+      "learning_rate": 8.191964285714287e-05,
+      "loss": 1.237,
       "step": 2025
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 6.0039447731755424e-05,
-      "loss": 1.6103,
+      "epoch": 9.04,
+      "learning_rate": 8.191071428571428e-05,
+      "loss": 1.2442,
       "step": 2026
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 6.0019723865877716e-05,
-      "loss": 1.5529,
+      "epoch": 9.05,
+      "learning_rate": 8.190178571428572e-05,
+      "loss": 1.1845,
       "step": 2027
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 6e-05,
-      "loss": 1.7657,
+      "epoch": 9.05,
+      "learning_rate": 8.189285714285715e-05,
+      "loss": 1.1808,
       "step": 2028
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 5.9980276134122294e-05,
-      "loss": 1.5298,
+      "epoch": 9.06,
+      "learning_rate": 8.188392857142859e-05,
+      "loss": 1.2564,
       "step": 2029
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.996055226824457e-05,
-      "loss": 1.6132,
+      "epoch": 9.06,
+      "learning_rate": 8.1875e-05,
+      "loss": 1.2225,
       "step": 2030
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.9940828402366864e-05,
-      "loss": 1.6396,
+      "epoch": 9.07,
+      "learning_rate": 8.186607142857144e-05,
+      "loss": 1.2227,
       "step": 2031
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.9921104536489156e-05,
-      "loss": 1.6405,
+      "epoch": 9.07,
+      "learning_rate": 8.185714285714286e-05,
+      "loss": 1.1298,
       "step": 2032
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.990138067061144e-05,
-      "loss": 1.5749,
+      "epoch": 9.08,
+      "learning_rate": 8.184821428571429e-05,
+      "loss": 1.224,
       "step": 2033
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 5.9881656804733734e-05,
-      "loss": 1.6563,
+      "epoch": 9.08,
+      "learning_rate": 8.183928571428571e-05,
+      "loss": 1.2484,
       "step": 2034
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 5.986193293885601e-05,
-      "loss": 1.6689,
+      "epoch": 9.08,
+      "learning_rate": 8.183035714285715e-05,
+      "loss": 1.2771,
       "step": 2035
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 5.9842209072978305e-05,
-      "loss": 1.683,
+      "epoch": 9.09,
+      "learning_rate": 8.182142857142858e-05,
+      "loss": 1.2399,
       "step": 2036
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 5.98224852071006e-05,
-      "loss": 1.5831,
+      "epoch": 9.09,
+      "learning_rate": 8.18125e-05,
+      "loss": 1.2028,
       "step": 2037
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 5.980276134122288e-05,
-      "loss": 1.5859,
+      "epoch": 9.1,
+      "learning_rate": 8.180357142857143e-05,
+      "loss": 1.3208,
       "step": 2038
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 5.9783037475345174e-05,
-      "loss": 1.5538,
+      "epoch": 9.1,
+      "learning_rate": 8.179464285714285e-05,
+      "loss": 1.2363,
       "step": 2039
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.9763313609467466e-05,
-      "loss": 1.6337,
+      "epoch": 9.11,
+      "learning_rate": 8.178571428571429e-05,
+      "loss": 1.3562,
       "step": 2040
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.9743589743589745e-05,
-      "loss": 1.5621,
+      "epoch": 9.11,
+      "learning_rate": 8.177678571428572e-05,
+      "loss": 1.1732,
       "step": 2041
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.972386587771204e-05,
-      "loss": 1.6455,
+      "epoch": 9.12,
+      "learning_rate": 8.176785714285716e-05,
+      "loss": 1.2414,
       "step": 2042
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.9704142011834315e-05,
-      "loss": 1.5531,
+      "epoch": 9.12,
+      "learning_rate": 8.175892857142857e-05,
+      "loss": 1.2483,
       "step": 2043
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 5.9684418145956614e-05,
-      "loss": 1.6827,
+      "epoch": 9.12,
+      "learning_rate": 8.175000000000001e-05,
+      "loss": 1.2394,
       "step": 2044
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 5.9664694280078906e-05,
-      "loss": 1.6581,
+      "epoch": 9.13,
+      "learning_rate": 8.174107142857143e-05,
+      "loss": 1.2528,
       "step": 2045
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 5.9644970414201185e-05,
-      "loss": 1.6266,
+      "epoch": 9.13,
+      "learning_rate": 8.173214285714286e-05,
+      "loss": 1.1794,
       "step": 2046
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 5.962524654832348e-05,
-      "loss": 1.5763,
+      "epoch": 9.14,
+      "learning_rate": 8.172321428571428e-05,
+      "loss": 1.2217,
       "step": 2047
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 5.9605522682445756e-05,
-      "loss": 1.6249,
+      "epoch": 9.14,
+      "learning_rate": 8.171428571428572e-05,
+      "loss": 1.2724,
       "step": 2048
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 5.958579881656805e-05,
-      "loss": 1.6313,
+      "epoch": 9.15,
+      "learning_rate": 8.170535714285714e-05,
+      "loss": 1.2131,
       "step": 2049
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 5.956607495069034e-05,
-      "loss": 1.6493,
+      "epoch": 9.15,
+      "learning_rate": 8.169642857142857e-05,
+      "loss": 1.181,
       "step": 2050
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 5.9546351084812625e-05,
-      "loss": 1.6173,
+      "epoch": 9.16,
+      "learning_rate": 8.16875e-05,
+      "loss": 1.1396,
       "step": 2051
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 5.952662721893492e-05,
-      "loss": 1.58,
+      "epoch": 9.16,
+      "learning_rate": 8.167857142857144e-05,
+      "loss": 1.2591,
       "step": 2052
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 5.9506903353057196e-05,
-      "loss": 1.7155,
+      "epoch": 9.17,
+      "learning_rate": 8.166964285714286e-05,
+      "loss": 1.2176,
       "step": 2053
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 5.948717948717949e-05,
-      "loss": 1.6468,
+      "epoch": 9.17,
+      "learning_rate": 8.166071428571429e-05,
+      "loss": 1.2274,
       "step": 2054
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 5.946745562130178e-05,
-      "loss": 1.5762,
+      "epoch": 9.17,
+      "learning_rate": 8.165178571428573e-05,
+      "loss": 1.2214,
       "step": 2055
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 5.9447731755424066e-05,
-      "loss": 1.6463,
+      "epoch": 9.18,
+      "learning_rate": 8.164285714285714e-05,
+      "loss": 1.1736,
       "step": 2056
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 5.942800788954636e-05,
-      "loss": 1.5248,
+      "epoch": 9.18,
+      "learning_rate": 8.163392857142858e-05,
+      "loss": 1.191,
       "step": 2057
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 5.9408284023668636e-05,
-      "loss": 1.662,
+      "epoch": 9.19,
+      "learning_rate": 8.1625e-05,
+      "loss": 1.1482,
       "step": 2058
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 5.938856015779093e-05,
-      "loss": 1.6276,
+      "epoch": 9.19,
+      "learning_rate": 8.161607142857143e-05,
+      "loss": 1.2075,
       "step": 2059
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 5.936883629191322e-05,
-      "loss": 1.5576,
+      "epoch": 9.2,
+      "learning_rate": 8.160714285714286e-05,
+      "loss": 1.3082,
       "step": 2060
     },
     {
-      "epoch": 6.1,
-      "learning_rate": 5.9349112426035506e-05,
-      "loss": 1.6583,
+      "epoch": 9.2,
+      "learning_rate": 8.15982142857143e-05,
+      "loss": 1.2461,
       "step": 2061
     },
     {
-      "epoch": 6.1,
-      "learning_rate": 5.93293885601578e-05,
-      "loss": 1.6066,
+      "epoch": 9.21,
+      "learning_rate": 8.158928571428572e-05,
+      "loss": 1.2073,
       "step": 2062
     },
     {
-      "epoch": 6.1,
-      "learning_rate": 5.930966469428009e-05,
-      "loss": 1.6566,
+      "epoch": 9.21,
+      "learning_rate": 8.158035714285715e-05,
+      "loss": 1.1454,
       "step": 2063
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 5.928994082840237e-05,
-      "loss": 1.5704,
+      "epoch": 9.21,
+      "learning_rate": 8.157142857142857e-05,
+      "loss": 1.1987,
       "step": 2064
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 5.927021696252466e-05,
-      "loss": 1.5443,
+      "epoch": 9.22,
+      "learning_rate": 8.156250000000001e-05,
+      "loss": 1.2748,
       "step": 2065
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 5.925049309664694e-05,
-      "loss": 1.6922,
+      "epoch": 9.22,
+      "learning_rate": 8.155357142857142e-05,
+      "loss": 1.2794,
       "step": 2066
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 5.923076923076923e-05,
-      "loss": 1.6437,
+      "epoch": 9.23,
+      "learning_rate": 8.154464285714286e-05,
+      "loss": 1.2045,
       "step": 2067
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 5.9211045364891523e-05,
-      "loss": 1.6683,
+      "epoch": 9.23,
+      "learning_rate": 8.153571428571429e-05,
+      "loss": 1.2542,
       "step": 2068
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 5.919132149901381e-05,
-      "loss": 1.606,
+      "epoch": 9.24,
+      "learning_rate": 8.152678571428571e-05,
+      "loss": 1.2378,
       "step": 2069
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 5.91715976331361e-05,
-      "loss": 1.5999,
+      "epoch": 9.24,
+      "learning_rate": 8.151785714285715e-05,
+      "loss": 1.2244,
       "step": 2070
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 5.915187376725838e-05,
-      "loss": 1.5963,
+      "epoch": 9.25,
+      "learning_rate": 8.150892857142858e-05,
+      "loss": 1.1749,
       "step": 2071
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 5.913214990138067e-05,
-      "loss": 1.6396,
+      "epoch": 9.25,
+      "learning_rate": 8.15e-05,
+      "loss": 1.2745,
       "step": 2072
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 5.9112426035502964e-05,
-      "loss": 1.7008,
+      "epoch": 9.25,
+      "learning_rate": 8.149107142857143e-05,
+      "loss": 1.2003,
       "step": 2073
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 5.909270216962525e-05,
-      "loss": 1.5412,
+      "epoch": 9.26,
+      "learning_rate": 8.148214285714287e-05,
+      "loss": 1.2185,
       "step": 2074
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 5.907297830374754e-05,
-      "loss": 1.5185,
+      "epoch": 9.26,
+      "learning_rate": 8.147321428571429e-05,
+      "loss": 1.2139,
       "step": 2075
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 5.905325443786982e-05,
-      "loss": 1.5798,
+      "epoch": 9.27,
+      "learning_rate": 8.146428571428572e-05,
+      "loss": 1.1888,
       "step": 2076
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 5.903353057199211e-05,
-      "loss": 1.575,
+      "epoch": 9.27,
+      "learning_rate": 8.145535714285714e-05,
+      "loss": 1.1683,
       "step": 2077
     },
     {
-      "epoch": 6.15,
-      "learning_rate": 5.9013806706114404e-05,
-      "loss": 1.5787,
+      "epoch": 9.28,
+      "learning_rate": 8.144642857142858e-05,
+      "loss": 1.164,
       "step": 2078
     },
     {
-      "epoch": 6.15,
-      "learning_rate": 5.899408284023669e-05,
-      "loss": 1.5432,
+      "epoch": 9.28,
+      "learning_rate": 8.14375e-05,
+      "loss": 1.2701,
       "step": 2079
     },
     {
-      "epoch": 6.15,
-      "learning_rate": 5.897435897435898e-05,
-      "loss": 1.6644,
+      "epoch": 9.29,
+      "learning_rate": 8.142857142857143e-05,
+      "loss": 1.1166,
       "step": 2080
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 5.895463510848126e-05,
-      "loss": 1.7091,
+      "epoch": 9.29,
+      "learning_rate": 8.141964285714286e-05,
+      "loss": 1.1499,
       "step": 2081
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 5.893491124260355e-05,
-      "loss": 1.6619,
+      "epoch": 9.29,
+      "learning_rate": 8.141071428571428e-05,
+      "loss": 1.2695,
       "step": 2082
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 5.8915187376725844e-05,
-      "loss": 1.5301,
+      "epoch": 9.3,
+      "learning_rate": 8.140178571428572e-05,
+      "loss": 1.2955,
       "step": 2083
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 5.889546351084812e-05,
-      "loss": 1.543,
+      "epoch": 9.3,
+      "learning_rate": 8.139285714285715e-05,
+      "loss": 1.27,
       "step": 2084
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 5.8875739644970415e-05,
-      "loss": 1.6549,
+      "epoch": 9.31,
+      "learning_rate": 8.138392857142857e-05,
+      "loss": 1.2575,
       "step": 2085
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 5.885601577909271e-05,
-      "loss": 1.698,
+      "epoch": 9.31,
+      "learning_rate": 8.1375e-05,
+      "loss": 1.2427,
       "step": 2086
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 5.883629191321499e-05,
-      "loss": 1.6191,
+      "epoch": 9.32,
+      "learning_rate": 8.136607142857144e-05,
+      "loss": 1.1399,
       "step": 2087
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 5.8816568047337284e-05,
-      "loss": 1.6626,
+      "epoch": 9.32,
+      "learning_rate": 8.135714285714286e-05,
+      "loss": 1.32,
       "step": 2088
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 5.879684418145956e-05,
-      "loss": 1.6065,
+      "epoch": 9.33,
+      "learning_rate": 8.134821428571429e-05,
+      "loss": 1.2403,
       "step": 2089
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 5.8777120315581855e-05,
-      "loss": 1.6243,
+      "epoch": 9.33,
+      "learning_rate": 8.133928571428571e-05,
+      "loss": 1.2445,
       "step": 2090
     },
     {
-      "epoch": 6.19,
-      "learning_rate": 5.875739644970415e-05,
-      "loss": 1.5242,
+      "epoch": 9.33,
+      "learning_rate": 8.133035714285715e-05,
+      "loss": 1.2623,
       "step": 2091
     },
     {
-      "epoch": 6.19,
-      "learning_rate": 5.873767258382643e-05,
-      "loss": 1.6029,
+      "epoch": 9.34,
+      "learning_rate": 8.132142857142857e-05,
+      "loss": 1.1757,
       "step": 2092
     },
     {
-      "epoch": 6.19,
-      "learning_rate": 5.8717948717948725e-05,
-      "loss": 1.6208,
+      "epoch": 9.34,
+      "learning_rate": 8.13125e-05,
+      "loss": 1.2432,
       "step": 2093
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 5.8698224852071e-05,
-      "loss": 1.6836,
+      "epoch": 9.35,
+      "learning_rate": 8.130357142857143e-05,
+      "loss": 1.3472,
       "step": 2094
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 5.8678500986193295e-05,
-      "loss": 1.5641,
+      "epoch": 9.35,
+      "learning_rate": 8.129464285714287e-05,
+      "loss": 1.2753,
       "step": 2095
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 5.865877712031559e-05,
-      "loss": 1.5688,
+      "epoch": 9.36,
+      "learning_rate": 8.128571428571428e-05,
+      "loss": 1.2157,
       "step": 2096
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 5.863905325443787e-05,
-      "loss": 1.6506,
+      "epoch": 9.36,
+      "learning_rate": 8.127678571428572e-05,
+      "loss": 1.2202,
       "step": 2097
     },
     {
-      "epoch": 6.21,
-      "learning_rate": 5.8619329388560165e-05,
-      "loss": 1.5874,
+      "epoch": 9.37,
+      "learning_rate": 8.126785714285715e-05,
+      "loss": 1.2424,
       "step": 2098
     },
     {
-      "epoch": 6.21,
-      "learning_rate": 5.8599605522682443e-05,
-      "loss": 1.7077,
+      "epoch": 9.37,
+      "learning_rate": 8.125892857142857e-05,
+      "loss": 1.2742,
       "step": 2099
     },
     {
-      "epoch": 6.21,
-      "learning_rate": 5.8579881656804736e-05,
-      "loss": 1.655,
+      "epoch": 9.38,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.2521,
       "step": 2100
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 5.856015779092703e-05,
-      "loss": 1.6922,
+      "epoch": 9.38,
+      "eval_loss": 1.1961336135864258,
+      "eval_runtime": 45.0663,
+      "eval_samples_per_second": 290.55,
+      "eval_steps_per_second": 1.531,
+      "step": 2100
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 8.124107142857144e-05,
+      "loss": 1.2838,
       "step": 2101
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 5.8540433925049306e-05,
-      "loss": 1.6188,
+      "epoch": 9.38,
+      "learning_rate": 8.123214285714286e-05,
+      "loss": 1.3105,
       "step": 2102
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 5.8520710059171605e-05,
-      "loss": 1.5386,
+      "epoch": 9.39,
+      "learning_rate": 8.122321428571429e-05,
+      "loss": 1.1747,
       "step": 2103
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 5.8500986193293884e-05,
-      "loss": 1.6067,
+      "epoch": 9.39,
+      "learning_rate": 8.121428571428573e-05,
+      "loss": 1.2439,
       "step": 2104
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 5.8481262327416176e-05,
-      "loss": 1.5283,
+      "epoch": 9.4,
+      "learning_rate": 8.120535714285714e-05,
+      "loss": 1.1895,
       "step": 2105
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 5.846153846153847e-05,
-      "loss": 1.588,
+      "epoch": 9.4,
+      "learning_rate": 8.119642857142858e-05,
+      "loss": 1.2038,
       "step": 2106
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 5.8441814595660746e-05,
-      "loss": 1.5828,
+      "epoch": 9.41,
+      "learning_rate": 8.11875e-05,
+      "loss": 1.2732,
       "step": 2107
     },
     {
-      "epoch": 6.24,
-      "learning_rate": 5.842209072978304e-05,
-      "loss": 1.609,
+      "epoch": 9.41,
+      "learning_rate": 8.117857142857144e-05,
+      "loss": 1.2149,
       "step": 2108
     },
     {
-      "epoch": 6.24,
-      "learning_rate": 5.840236686390533e-05,
-      "loss": 1.5982,
+      "epoch": 9.42,
+      "learning_rate": 8.116964285714285e-05,
+      "loss": 1.2302,
       "step": 2109
     },
     {
-      "epoch": 6.24,
-      "learning_rate": 5.8382642998027616e-05,
-      "loss": 1.6343,
+      "epoch": 9.42,
+      "learning_rate": 8.116071428571429e-05,
+      "loss": 1.3104,
       "step": 2110
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 5.836291913214991e-05,
-      "loss": 1.498,
+      "epoch": 9.42,
+      "learning_rate": 8.115178571428572e-05,
+      "loss": 1.2809,
       "step": 2111
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 5.834319526627219e-05,
-      "loss": 1.5619,
+      "epoch": 9.43,
+      "learning_rate": 8.114285714285714e-05,
+      "loss": 1.2557,
       "step": 2112
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 5.832347140039448e-05,
-      "loss": 1.6793,
+      "epoch": 9.43,
+      "learning_rate": 8.113392857142858e-05,
+      "loss": 1.373,
       "step": 2113
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 5.830374753451677e-05,
-      "loss": 1.6226,
+      "epoch": 9.44,
+      "learning_rate": 8.112500000000001e-05,
+      "loss": 1.2635,
       "step": 2114
     },
     {
-      "epoch": 6.26,
-      "learning_rate": 5.8284023668639056e-05,
-      "loss": 1.6216,
+      "epoch": 9.44,
+      "learning_rate": 8.111607142857143e-05,
+      "loss": 1.2012,
       "step": 2115
     },
     {
-      "epoch": 6.26,
-      "learning_rate": 5.826429980276135e-05,
-      "loss": 1.5588,
+      "epoch": 9.45,
+      "learning_rate": 8.110714285714286e-05,
+      "loss": 1.2394,
       "step": 2116
     },
     {
-      "epoch": 6.26,
-      "learning_rate": 5.824457593688363e-05,
-      "loss": 1.6249,
+      "epoch": 9.45,
+      "learning_rate": 8.10982142857143e-05,
+      "loss": 1.2729,
       "step": 2117
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 5.822485207100592e-05,
-      "loss": 1.6538,
+      "epoch": 9.46,
+      "learning_rate": 8.108928571428572e-05,
+      "loss": 1.2658,
       "step": 2118
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 5.820512820512821e-05,
-      "loss": 1.592,
+      "epoch": 9.46,
+      "learning_rate": 8.108035714285715e-05,
+      "loss": 1.2774,
       "step": 2119
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 5.8185404339250496e-05,
-      "loss": 1.6058,
+      "epoch": 9.46,
+      "learning_rate": 8.107142857142857e-05,
+      "loss": 1.1453,
       "step": 2120
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 5.816568047337279e-05,
-      "loss": 1.6321,
+      "epoch": 9.47,
+      "learning_rate": 8.106250000000001e-05,
+      "loss": 1.1908,
       "step": 2121
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 5.814595660749507e-05,
-      "loss": 1.5745,
+      "epoch": 9.47,
+      "learning_rate": 8.105357142857142e-05,
+      "loss": 1.1586,
       "step": 2122
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 5.812623274161736e-05,
-      "loss": 1.6497,
+      "epoch": 9.48,
+      "learning_rate": 8.104464285714286e-05,
+      "loss": 1.2089,
       "step": 2123
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 5.810650887573965e-05,
-      "loss": 1.629,
+      "epoch": 9.48,
+      "learning_rate": 8.103571428571429e-05,
+      "loss": 1.2639,
       "step": 2124
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 5.808678500986193e-05,
-      "loss": 1.6455,
+      "epoch": 9.49,
+      "learning_rate": 8.102678571428571e-05,
+      "loss": 1.1989,
       "step": 2125
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 5.806706114398422e-05,
-      "loss": 1.6286,
+      "epoch": 9.49,
+      "learning_rate": 8.101785714285714e-05,
+      "loss": 1.1803,
       "step": 2126
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 5.804733727810651e-05,
-      "loss": 1.563,
+      "epoch": 9.5,
+      "learning_rate": 8.100892857142858e-05,
+      "loss": 1.2778,
       "step": 2127
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 5.80276134122288e-05,
-      "loss": 1.5667,
+      "epoch": 9.5,
+      "learning_rate": 8.1e-05,
+      "loss": 1.2364,
       "step": 2128
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 5.800788954635109e-05,
-      "loss": 1.5795,
+      "epoch": 9.5,
+      "learning_rate": 8.099107142857143e-05,
+      "loss": 1.2418,
       "step": 2129
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 5.798816568047337e-05,
-      "loss": 1.6178,
+      "epoch": 9.51,
+      "learning_rate": 8.098214285714287e-05,
+      "loss": 1.2265,
       "step": 2130
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 5.796844181459566e-05,
-      "loss": 1.6251,
+      "epoch": 9.51,
+      "learning_rate": 8.09732142857143e-05,
+      "loss": 1.1667,
       "step": 2131
     },
     {
-      "epoch": 6.31,
-      "learning_rate": 5.7948717948717954e-05,
-      "loss": 1.6072,
+      "epoch": 9.52,
+      "learning_rate": 8.096428571428572e-05,
+      "loss": 1.2468,
       "step": 2132
     },
     {
-      "epoch": 6.31,
-      "learning_rate": 5.792899408284024e-05,
-      "loss": 1.6705,
+      "epoch": 9.52,
+      "learning_rate": 8.095535714285714e-05,
+      "loss": 1.2633,
       "step": 2133
     },
     {
-      "epoch": 6.31,
-      "learning_rate": 5.790927021696253e-05,
-      "loss": 1.6947,
+      "epoch": 9.53,
+      "learning_rate": 8.094642857142858e-05,
+      "loss": 1.219,
       "step": 2134
     },
     {
-      "epoch": 6.32,
-      "learning_rate": 5.788954635108481e-05,
-      "loss": 1.6439,
+      "epoch": 9.53,
+      "learning_rate": 8.09375e-05,
+      "loss": 1.2024,
       "step": 2135
     },
     {
-      "epoch": 6.32,
-      "learning_rate": 5.78698224852071e-05,
-      "loss": 1.6282,
+      "epoch": 9.54,
+      "learning_rate": 8.092857142857143e-05,
+      "loss": 1.303,
       "step": 2136
     },
     {
-      "epoch": 6.32,
-      "learning_rate": 5.7850098619329395e-05,
-      "loss": 1.5784,
+      "epoch": 9.54,
+      "learning_rate": 8.091964285714286e-05,
+      "loss": 1.2261,
       "step": 2137
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 5.783037475345168e-05,
-      "loss": 1.6451,
+      "epoch": 9.54,
+      "learning_rate": 8.091071428571429e-05,
+      "loss": 1.1815,
       "step": 2138
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 5.781065088757397e-05,
-      "loss": 1.6811,
+      "epoch": 9.55,
+      "learning_rate": 8.090178571428571e-05,
+      "loss": 1.1675,
       "step": 2139
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 5.779092702169625e-05,
-      "loss": 1.7167,
+      "epoch": 9.55,
+      "learning_rate": 8.089285714285715e-05,
+      "loss": 1.2632,
       "step": 2140
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 5.777120315581854e-05,
-      "loss": 1.653,
+      "epoch": 9.56,
+      "learning_rate": 8.088392857142858e-05,
+      "loss": 1.2412,
       "step": 2141
     },
     {
-      "epoch": 6.34,
-      "learning_rate": 5.7751479289940835e-05,
-      "loss": 1.6521,
+      "epoch": 9.56,
+      "learning_rate": 8.0875e-05,
+      "loss": 1.1607,
       "step": 2142
     },
     {
-      "epoch": 6.34,
-      "learning_rate": 5.7731755424063113e-05,
-      "loss": 1.5802,
+      "epoch": 9.57,
+      "learning_rate": 8.086607142857143e-05,
+      "loss": 1.1833,
       "step": 2143
     },
     {
-      "epoch": 6.34,
-      "learning_rate": 5.7712031558185406e-05,
-      "loss": 1.6603,
+      "epoch": 9.57,
+      "learning_rate": 8.085714285714287e-05,
+      "loss": 1.2521,
       "step": 2144
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 5.769230769230769e-05,
-      "loss": 1.6422,
+      "epoch": 9.58,
+      "learning_rate": 8.084821428571429e-05,
+      "loss": 1.2233,
       "step": 2145
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 5.767258382642998e-05,
-      "loss": 1.6256,
+      "epoch": 9.58,
+      "learning_rate": 8.083928571428572e-05,
+      "loss": 1.2183,
       "step": 2146
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 5.7652859960552275e-05,
-      "loss": 1.561,
+      "epoch": 9.58,
+      "learning_rate": 8.083035714285716e-05,
+      "loss": 1.2229,
       "step": 2147
     },
     {
-      "epoch": 6.36,
-      "learning_rate": 5.7633136094674554e-05,
-      "loss": 1.5832,
+      "epoch": 9.59,
+      "learning_rate": 8.082142857142857e-05,
+      "loss": 1.2219,
       "step": 2148
     },
     {
-      "epoch": 6.36,
-      "learning_rate": 5.7613412228796846e-05,
-      "loss": 1.6375,
+      "epoch": 9.59,
+      "learning_rate": 8.08125e-05,
+      "loss": 1.1353,
       "step": 2149
     },
     {
-      "epoch": 6.36,
-      "learning_rate": 5.759368836291914e-05,
-      "loss": 1.6037,
+      "epoch": 9.6,
+      "learning_rate": 8.080357142857143e-05,
+      "loss": 1.2279,
       "step": 2150
     },
     {
-      "epoch": 6.36,
-      "learning_rate": 5.757396449704142e-05,
-      "loss": 1.6044,
+      "epoch": 9.6,
+      "learning_rate": 8.079464285714287e-05,
+      "loss": 1.2585,
       "step": 2151
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 5.7554240631163715e-05,
-      "loss": 1.6619,
+      "epoch": 9.61,
+      "learning_rate": 8.078571428571428e-05,
+      "loss": 1.2801,
       "step": 2152
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 5.7534516765285994e-05,
-      "loss": 1.5219,
+      "epoch": 9.61,
+      "learning_rate": 8.077678571428572e-05,
+      "loss": 1.2127,
       "step": 2153
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 5.7514792899408286e-05,
-      "loss": 1.5851,
+      "epoch": 9.62,
+      "learning_rate": 8.076785714285715e-05,
+      "loss": 1.227,
       "step": 2154
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 5.749506903353058e-05,
-      "loss": 1.6376,
+      "epoch": 9.62,
+      "learning_rate": 8.075892857142857e-05,
+      "loss": 1.2558,
       "step": 2155
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 5.7475345167652863e-05,
-      "loss": 1.5914,
+      "epoch": 9.62,
+      "learning_rate": 8.075e-05,
+      "loss": 1.2842,
       "step": 2156
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 5.7455621301775156e-05,
-      "loss": 1.717,
+      "epoch": 9.63,
+      "learning_rate": 8.074107142857144e-05,
+      "loss": 1.2449,
       "step": 2157
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 5.7435897435897434e-05,
-      "loss": 1.6086,
+      "epoch": 9.63,
+      "learning_rate": 8.073214285714286e-05,
+      "loss": 1.166,
       "step": 2158
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 5.7416173570019726e-05,
-      "loss": 1.5754,
+      "epoch": 9.64,
+      "learning_rate": 8.072321428571429e-05,
+      "loss": 1.2374,
       "step": 2159
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 5.739644970414202e-05,
-      "loss": 1.7116,
+      "epoch": 9.64,
+      "learning_rate": 8.071428571428573e-05,
+      "loss": 1.2439,
       "step": 2160
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 5.73767258382643e-05,
-      "loss": 1.6536,
+      "epoch": 9.65,
+      "learning_rate": 8.070535714285715e-05,
+      "loss": 1.2227,
       "step": 2161
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 5.7357001972386596e-05,
-      "loss": 1.6464,
+      "epoch": 9.65,
+      "learning_rate": 8.069642857142858e-05,
+      "loss": 1.2076,
       "step": 2162
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 5.7337278106508874e-05,
-      "loss": 1.5645,
+      "epoch": 9.66,
+      "learning_rate": 8.06875e-05,
+      "loss": 1.1825,
       "step": 2163
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 5.7317554240631167e-05,
-      "loss": 1.5942,
+      "epoch": 9.66,
+      "learning_rate": 8.067857142857144e-05,
+      "loss": 1.2141,
       "step": 2164
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 5.729783037475346e-05,
-      "loss": 1.6291,
+      "epoch": 9.67,
+      "learning_rate": 8.066964285714285e-05,
+      "loss": 1.2573,
       "step": 2165
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 5.727810650887574e-05,
-      "loss": 1.6629,
+      "epoch": 9.67,
+      "learning_rate": 8.066071428571429e-05,
+      "loss": 1.2007,
       "step": 2166
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 5.725838264299803e-05,
-      "loss": 1.5582,
+      "epoch": 9.67,
+      "learning_rate": 8.065178571428572e-05,
+      "loss": 1.2504,
       "step": 2167
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 5.7238658777120315e-05,
-      "loss": 1.6325,
+      "epoch": 9.68,
+      "learning_rate": 8.064285714285714e-05,
+      "loss": 1.1545,
       "step": 2168
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 5.721893491124261e-05,
-      "loss": 1.6262,
+      "epoch": 9.68,
+      "learning_rate": 8.063392857142857e-05,
+      "loss": 1.2393,
       "step": 2169
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 5.71992110453649e-05,
-      "loss": 1.5959,
+      "epoch": 9.69,
+      "learning_rate": 8.062500000000001e-05,
+      "loss": 1.2242,
       "step": 2170
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 5.717948717948718e-05,
-      "loss": 1.6321,
+      "epoch": 9.69,
+      "learning_rate": 8.061607142857143e-05,
+      "loss": 1.2883,
       "step": 2171
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 5.715976331360947e-05,
-      "loss": 1.6225,
+      "epoch": 9.7,
+      "learning_rate": 8.060714285714286e-05,
+      "loss": 1.2729,
       "step": 2172
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 5.714003944773176e-05,
-      "loss": 1.6231,
+      "epoch": 9.7,
+      "learning_rate": 8.059821428571428e-05,
+      "loss": 1.2313,
       "step": 2173
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 5.712031558185405e-05,
-      "loss": 1.5911,
+      "epoch": 9.71,
+      "learning_rate": 8.058928571428572e-05,
+      "loss": 1.2325,
       "step": 2174
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 5.710059171597634e-05,
-      "loss": 1.5305,
+      "epoch": 9.71,
+      "learning_rate": 8.058035714285715e-05,
+      "loss": 1.1709,
       "step": 2175
     },
     {
-      "epoch": 6.44,
-      "learning_rate": 5.708086785009862e-05,
-      "loss": 1.566,
+      "epoch": 9.71,
+      "learning_rate": 8.057142857142857e-05,
+      "loss": 1.3477,
       "step": 2176
     },
     {
-      "epoch": 6.44,
-      "learning_rate": 5.706114398422091e-05,
-      "loss": 1.5741,
+      "epoch": 9.72,
+      "learning_rate": 8.056250000000001e-05,
+      "loss": 1.1974,
       "step": 2177
     },
     {
-      "epoch": 6.44,
-      "learning_rate": 5.70414201183432e-05,
-      "loss": 1.6367,
+      "epoch": 9.72,
+      "learning_rate": 8.055357142857143e-05,
+      "loss": 1.1552,
       "step": 2178
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 5.702169625246549e-05,
-      "loss": 1.6175,
+      "epoch": 9.73,
+      "learning_rate": 8.054464285714286e-05,
+      "loss": 1.1626,
       "step": 2179
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 5.700197238658778e-05,
-      "loss": 1.5167,
+      "epoch": 9.73,
+      "learning_rate": 8.053571428571429e-05,
+      "loss": 1.2794,
       "step": 2180
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 5.698224852071006e-05,
-      "loss": 1.5728,
+      "epoch": 9.74,
+      "learning_rate": 8.052678571428572e-05,
+      "loss": 1.2231,
       "step": 2181
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 5.696252465483235e-05,
-      "loss": 1.6262,
+      "epoch": 9.74,
+      "learning_rate": 8.051785714285714e-05,
+      "loss": 1.2015,
       "step": 2182
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 5.694280078895464e-05,
-      "loss": 1.6704,
+      "epoch": 9.75,
+      "learning_rate": 8.050892857142858e-05,
+      "loss": 1.1493,
       "step": 2183
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 5.692307692307692e-05,
-      "loss": 1.6757,
+      "epoch": 9.75,
+      "learning_rate": 8.05e-05,
+      "loss": 1.2259,
       "step": 2184
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 5.690335305719921e-05,
-      "loss": 1.6215,
+      "epoch": 9.75,
+      "learning_rate": 8.049107142857143e-05,
+      "loss": 1.2209,
       "step": 2185
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 5.68836291913215e-05,
-      "loss": 1.5856,
+      "epoch": 9.76,
+      "learning_rate": 8.048214285714286e-05,
+      "loss": 1.2101,
       "step": 2186
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 5.686390532544379e-05,
-      "loss": 1.6471,
+      "epoch": 9.76,
+      "learning_rate": 8.04732142857143e-05,
+      "loss": 1.2585,
       "step": 2187
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 5.684418145956608e-05,
-      "loss": 1.613,
+      "epoch": 9.77,
+      "learning_rate": 8.046428571428572e-05,
+      "loss": 1.277,
       "step": 2188
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 5.682445759368836e-05,
-      "loss": 1.6151,
+      "epoch": 9.77,
+      "learning_rate": 8.045535714285715e-05,
+      "loss": 1.2051,
       "step": 2189
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 5.680473372781065e-05,
-      "loss": 1.6174,
+      "epoch": 9.78,
+      "learning_rate": 8.044642857142857e-05,
+      "loss": 1.238,
       "step": 2190
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 5.678500986193294e-05,
-      "loss": 1.7336,
+      "epoch": 9.78,
+      "learning_rate": 8.04375e-05,
+      "loss": 1.22,
       "step": 2191
     },
     {
-      "epoch": 6.49,
-      "learning_rate": 5.676528599605523e-05,
-      "loss": 1.6003,
+      "epoch": 9.79,
+      "learning_rate": 8.042857142857144e-05,
+      "loss": 1.2719,
       "step": 2192
     },
     {
-      "epoch": 6.49,
-      "learning_rate": 5.674556213017752e-05,
-      "loss": 1.4921,
+      "epoch": 9.79,
+      "learning_rate": 8.041964285714286e-05,
+      "loss": 1.247,
       "step": 2193
     },
     {
-      "epoch": 6.49,
-      "learning_rate": 5.67258382642998e-05,
-      "loss": 1.6245,
+      "epoch": 9.79,
+      "learning_rate": 8.04107142857143e-05,
+      "loss": 1.1558,
       "step": 2194
     },
     {
-      "epoch": 6.49,
-      "learning_rate": 5.670611439842209e-05,
-      "loss": 1.6483,
+      "epoch": 9.8,
+      "learning_rate": 8.040178571428571e-05,
+      "loss": 1.2132,
       "step": 2195
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 5.6686390532544385e-05,
-      "loss": 1.6103,
+      "epoch": 9.8,
+      "learning_rate": 8.039285714285715e-05,
+      "loss": 1.2289,
       "step": 2196
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 5.666666666666667e-05,
-      "loss": 1.6053,
+      "epoch": 9.81,
+      "learning_rate": 8.038392857142858e-05,
+      "loss": 1.188,
       "step": 2197
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 5.664694280078896e-05,
-      "loss": 1.7414,
+      "epoch": 9.81,
+      "learning_rate": 8.0375e-05,
+      "loss": 1.2113,
       "step": 2198
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 5.662721893491124e-05,
-      "loss": 1.6037,
+      "epoch": 9.82,
+      "learning_rate": 8.036607142857143e-05,
+      "loss": 1.256,
       "step": 2199
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 5.6607495069033534e-05,
-      "loss": 1.5836,
+      "epoch": 9.82,
+      "learning_rate": 8.035714285714287e-05,
+      "loss": 1.3248,
       "step": 2200
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 5.6587771203155826e-05,
-      "loss": 1.5487,
+      "epoch": 9.83,
+      "learning_rate": 8.034821428571428e-05,
+      "loss": 1.1447,
       "step": 2201
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 5.6568047337278104e-05,
-      "loss": 1.5598,
+      "epoch": 9.83,
+      "learning_rate": 8.033928571428572e-05,
+      "loss": 1.1905,
       "step": 2202
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 5.6548323471400396e-05,
-      "loss": 1.6103,
+      "epoch": 9.83,
+      "learning_rate": 8.033035714285714e-05,
+      "loss": 1.1986,
       "step": 2203
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 5.652859960552268e-05,
-      "loss": 1.6047,
+      "epoch": 9.84,
+      "learning_rate": 8.032142857142858e-05,
+      "loss": 1.1253,
       "step": 2204
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 5.6508875739644974e-05,
-      "loss": 1.5465,
+      "epoch": 9.84,
+      "learning_rate": 8.031250000000001e-05,
+      "loss": 1.2568,
       "step": 2205
     },
     {
-      "epoch": 6.53,
-      "learning_rate": 5.6489151873767266e-05,
-      "loss": 1.5158,
+      "epoch": 9.85,
+      "learning_rate": 8.030357142857143e-05,
+      "loss": 1.22,
       "step": 2206
     },
     {
-      "epoch": 6.53,
-      "learning_rate": 5.6469428007889544e-05,
-      "loss": 1.483,
+      "epoch": 9.85,
+      "learning_rate": 8.029464285714287e-05,
+      "loss": 1.2621,
       "step": 2207
     },
     {
-      "epoch": 6.53,
-      "learning_rate": 5.6449704142011837e-05,
-      "loss": 1.6578,
+      "epoch": 9.86,
+      "learning_rate": 8.028571428571428e-05,
+      "loss": 1.2527,
       "step": 2208
     },
     {
-      "epoch": 6.54,
-      "learning_rate": 5.642998027613412e-05,
-      "loss": 1.6073,
+      "epoch": 9.86,
+      "learning_rate": 8.027678571428572e-05,
+      "loss": 1.1689,
       "step": 2209
     },
     {
-      "epoch": 6.54,
-      "learning_rate": 5.6410256410256414e-05,
-      "loss": 1.642,
+      "epoch": 9.87,
+      "learning_rate": 8.026785714285715e-05,
+      "loss": 1.2664,
       "step": 2210
     },
     {
-      "epoch": 6.54,
-      "learning_rate": 5.6390532544378706e-05,
-      "loss": 1.6735,
+      "epoch": 9.87,
+      "learning_rate": 8.025892857142857e-05,
+      "loss": 1.2277,
       "step": 2211
     },
     {
-      "epoch": 6.54,
-      "learning_rate": 5.6370808678500985e-05,
-      "loss": 1.6256,
+      "epoch": 9.88,
+      "learning_rate": 8.025e-05,
+      "loss": 1.2617,
       "step": 2212
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 5.635108481262328e-05,
-      "loss": 1.5961,
+      "epoch": 9.88,
+      "learning_rate": 8.024107142857144e-05,
+      "loss": 1.216,
       "step": 2213
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 5.633136094674556e-05,
-      "loss": 1.6871,
+      "epoch": 9.88,
+      "learning_rate": 8.023214285714285e-05,
+      "loss": 1.2201,
       "step": 2214
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 5.6311637080867854e-05,
-      "loss": 1.6982,
+      "epoch": 9.89,
+      "learning_rate": 8.022321428571429e-05,
+      "loss": 1.1744,
       "step": 2215
     },
     {
-      "epoch": 6.56,
-      "learning_rate": 5.6291913214990146e-05,
-      "loss": 1.602,
+      "epoch": 9.89,
+      "learning_rate": 8.021428571428571e-05,
+      "loss": 1.2529,
       "step": 2216
     },
     {
-      "epoch": 6.56,
-      "learning_rate": 5.6272189349112425e-05,
-      "loss": 1.5987,
+      "epoch": 9.9,
+      "learning_rate": 8.020535714285715e-05,
+      "loss": 1.2739,
       "step": 2217
     },
     {
-      "epoch": 6.56,
-      "learning_rate": 5.625246548323472e-05,
-      "loss": 1.6333,
+      "epoch": 9.9,
+      "learning_rate": 8.019642857142858e-05,
+      "loss": 1.2472,
       "step": 2218
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 5.623274161735701e-05,
-      "loss": 1.6466,
+      "epoch": 9.91,
+      "learning_rate": 8.01875e-05,
+      "loss": 1.2261,
       "step": 2219
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 5.621301775147929e-05,
-      "loss": 1.5937,
+      "epoch": 9.91,
+      "learning_rate": 8.017857142857143e-05,
+      "loss": 1.195,
       "step": 2220
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 5.6193293885601587e-05,
-      "loss": 1.7089,
+      "epoch": 9.92,
+      "learning_rate": 8.016964285714286e-05,
+      "loss": 1.1783,
       "step": 2221
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 5.6173570019723865e-05,
-      "loss": 1.6593,
+      "epoch": 9.92,
+      "learning_rate": 8.01607142857143e-05,
+      "loss": 1.1422,
       "step": 2222
     },
     {
-      "epoch": 6.58,
-      "learning_rate": 5.615384615384616e-05,
-      "loss": 1.5531,
+      "epoch": 9.92,
+      "learning_rate": 8.015178571428572e-05,
+      "loss": 1.1939,
       "step": 2223
     },
     {
-      "epoch": 6.58,
-      "learning_rate": 5.613412228796845e-05,
-      "loss": 1.5572,
+      "epoch": 9.93,
+      "learning_rate": 8.014285714285715e-05,
+      "loss": 1.185,
       "step": 2224
     },
     {
-      "epoch": 6.58,
-      "learning_rate": 5.611439842209073e-05,
-      "loss": 1.5783,
+      "epoch": 9.93,
+      "learning_rate": 8.013392857142857e-05,
+      "loss": 1.2655,
       "step": 2225
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 5.609467455621302e-05,
-      "loss": 1.611,
+      "epoch": 9.94,
+      "learning_rate": 8.012500000000001e-05,
+      "loss": 1.2542,
       "step": 2226
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 5.6074950690335305e-05,
-      "loss": 1.5844,
+      "epoch": 9.94,
+      "learning_rate": 8.011607142857144e-05,
+      "loss": 1.2229,
       "step": 2227
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 5.60552268244576e-05,
-      "loss": 1.6808,
+      "epoch": 9.95,
+      "learning_rate": 8.010714285714286e-05,
+      "loss": 1.2844,
       "step": 2228
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 5.603550295857989e-05,
-      "loss": 1.6279,
+      "epoch": 9.95,
+      "learning_rate": 8.009821428571429e-05,
+      "loss": 1.2845,
       "step": 2229
     },
     {
-      "epoch": 6.6,
-      "learning_rate": 5.601577909270217e-05,
-      "loss": 1.5226,
+      "epoch": 9.96,
+      "learning_rate": 8.008928571428572e-05,
+      "loss": 1.2251,
       "step": 2230
     },
     {
-      "epoch": 6.6,
-      "learning_rate": 5.599605522682446e-05,
-      "loss": 1.5973,
+      "epoch": 9.96,
+      "learning_rate": 8.008035714285714e-05,
+      "loss": 1.1472,
       "step": 2231
     },
     {
-      "epoch": 6.6,
-      "learning_rate": 5.5976331360946746e-05,
-      "loss": 1.6559,
+      "epoch": 9.96,
+      "learning_rate": 8.007142857142858e-05,
+      "loss": 1.2379,
       "step": 2232
     },
     {
-      "epoch": 6.61,
-      "learning_rate": 5.595660749506904e-05,
-      "loss": 1.5799,
+      "epoch": 9.97,
+      "learning_rate": 8.00625e-05,
+      "loss": 1.2362,
       "step": 2233
     },
     {
-      "epoch": 6.61,
-      "learning_rate": 5.593688362919133e-05,
-      "loss": 1.5554,
+      "epoch": 9.97,
+      "learning_rate": 8.005357142857143e-05,
+      "loss": 1.251,
       "step": 2234
     },
     {
-      "epoch": 6.61,
-      "learning_rate": 5.591715976331361e-05,
-      "loss": 1.5727,
+      "epoch": 9.98,
+      "learning_rate": 8.004464285714287e-05,
+      "loss": 1.178,
       "step": 2235
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 5.58974358974359e-05,
-      "loss": 1.5107,
+      "epoch": 9.98,
+      "learning_rate": 8.003571428571429e-05,
+      "loss": 1.2254,
       "step": 2236
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 5.5877712031558186e-05,
-      "loss": 1.6866,
+      "epoch": 9.99,
+      "learning_rate": 8.002678571428573e-05,
+      "loss": 1.2447,
       "step": 2237
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 5.585798816568048e-05,
-      "loss": 1.6106,
+      "epoch": 9.99,
+      "learning_rate": 8.001785714285714e-05,
+      "loss": 1.2109,
       "step": 2238
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 5.583826429980277e-05,
-      "loss": 1.6143,
+      "epoch": 10.0,
+      "learning_rate": 8.000892857142858e-05,
+      "loss": 1.2391,
       "step": 2239
     },
     {
-      "epoch": 6.63,
-      "learning_rate": 5.581854043392505e-05,
-      "loss": 1.6286,
+      "epoch": 10.0,
+      "learning_rate": 8e-05,
+      "loss": 1.4401,
       "step": 2240
     },
     {
-      "epoch": 6.63,
-      "learning_rate": 5.579881656804734e-05,
-      "loss": 1.6135,
+      "epoch": 10.0,
+      "learning_rate": 7.999107142857143e-05,
+      "loss": 1.2322,
       "step": 2241
     },
     {
-      "epoch": 6.63,
-      "learning_rate": 5.577909270216963e-05,
-      "loss": 1.5536,
+      "epoch": 10.01,
+      "learning_rate": 7.998214285714286e-05,
+      "loss": 1.1165,
       "step": 2242
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 5.575936883629191e-05,
-      "loss": 1.7219,
+      "epoch": 10.01,
+      "learning_rate": 7.99732142857143e-05,
+      "loss": 1.2146,
       "step": 2243
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 5.5739644970414204e-05,
-      "loss": 1.5376,
+      "epoch": 10.02,
+      "learning_rate": 7.996428571428571e-05,
+      "loss": 1.2117,
       "step": 2244
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 5.571992110453649e-05,
-      "loss": 1.6629,
+      "epoch": 10.02,
+      "learning_rate": 7.995535714285715e-05,
+      "loss": 1.2158,
       "step": 2245
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 5.570019723865878e-05,
-      "loss": 1.6586,
+      "epoch": 10.03,
+      "learning_rate": 7.994642857142857e-05,
+      "loss": 1.3208,
       "step": 2246
     },
     {
-      "epoch": 6.65,
-      "learning_rate": 5.568047337278107e-05,
-      "loss": 1.5762,
+      "epoch": 10.03,
+      "learning_rate": 7.99375e-05,
+      "loss": 1.1246,
       "step": 2247
     },
     {
-      "epoch": 6.65,
-      "learning_rate": 5.566074950690335e-05,
-      "loss": 1.638,
+      "epoch": 10.04,
+      "learning_rate": 7.992857142857142e-05,
+      "loss": 1.1547,
       "step": 2248
     },
     {
-      "epoch": 6.65,
-      "learning_rate": 5.5641025641025644e-05,
-      "loss": 1.6351,
+      "epoch": 10.04,
+      "learning_rate": 7.991964285714286e-05,
+      "loss": 1.2098,
       "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,
+      "epoch": 10.04,
+      "learning_rate": 7.991071428571429e-05,
+      "loss": 1.2735,
       "step": 2250
     },
     {
-      "epoch": 6.66,
-      "learning_rate": 5.560157790927022e-05,
-      "loss": 1.6109,
+      "epoch": 10.05,
+      "learning_rate": 7.990178571428571e-05,
+      "loss": 1.1384,
       "step": 2251
     },
     {
-      "epoch": 6.66,
-      "learning_rate": 5.558185404339251e-05,
-      "loss": 1.6852,
+      "epoch": 10.05,
+      "learning_rate": 7.989285714285715e-05,
+      "loss": 1.2324,
       "step": 2252
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 5.556213017751479e-05,
-      "loss": 1.6274,
+      "epoch": 10.06,
+      "learning_rate": 7.988392857142858e-05,
+      "loss": 1.1807,
       "step": 2253
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 5.5542406311637084e-05,
-      "loss": 1.5829,
+      "epoch": 10.06,
+      "learning_rate": 7.9875e-05,
+      "loss": 1.2006,
       "step": 2254
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 5.552268244575937e-05,
-      "loss": 1.5858,
+      "epoch": 10.07,
+      "learning_rate": 7.986607142857143e-05,
+      "loss": 1.2104,
       "step": 2255
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 5.550295857988166e-05,
-      "loss": 1.6228,
+      "epoch": 10.07,
+      "learning_rate": 7.985714285714287e-05,
+      "loss": 1.1771,
       "step": 2256
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 5.5483234714003954e-05,
-      "loss": 1.7294,
+      "epoch": 10.08,
+      "learning_rate": 7.984821428571428e-05,
+      "loss": 1.2338,
       "step": 2257
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 5.546351084812623e-05,
-      "loss": 1.6681,
+      "epoch": 10.08,
+      "learning_rate": 7.983928571428572e-05,
+      "loss": 1.2511,
       "step": 2258
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 5.5443786982248524e-05,
-      "loss": 1.6202,
+      "epoch": 10.08,
+      "learning_rate": 7.983035714285714e-05,
+      "loss": 1.1513,
       "step": 2259
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 5.5424063116370816e-05,
-      "loss": 1.6302,
+      "epoch": 10.09,
+      "learning_rate": 7.982142857142858e-05,
+      "loss": 1.1803,
       "step": 2260
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 5.5404339250493095e-05,
-      "loss": 1.5708,
+      "epoch": 10.09,
+      "learning_rate": 7.98125e-05,
+      "loss": 1.2282,
       "step": 2261
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 5.538461538461539e-05,
-      "loss": 1.5977,
+      "epoch": 10.1,
+      "learning_rate": 7.980357142857143e-05,
+      "loss": 1.3071,
       "step": 2262
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 5.536489151873767e-05,
-      "loss": 1.5321,
+      "epoch": 10.1,
+      "learning_rate": 7.979464285714286e-05,
+      "loss": 1.2134,
       "step": 2263
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 5.5345167652859965e-05,
-      "loss": 1.6043,
+      "epoch": 10.11,
+      "learning_rate": 7.978571428571429e-05,
+      "loss": 1.1695,
       "step": 2264
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 5.532544378698226e-05,
-      "loss": 1.5737,
+      "epoch": 10.11,
+      "learning_rate": 7.977678571428572e-05,
+      "loss": 1.2115,
       "step": 2265
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 5.5305719921104535e-05,
-      "loss": 1.5572,
+      "epoch": 10.12,
+      "learning_rate": 7.976785714285715e-05,
+      "loss": 1.2003,
       "step": 2266
     },
     {
-      "epoch": 6.71,
-      "learning_rate": 5.528599605522683e-05,
-      "loss": 1.5834,
+      "epoch": 10.12,
+      "learning_rate": 7.975892857142858e-05,
+      "loss": 1.2356,
       "step": 2267
     },
     {
-      "epoch": 6.71,
-      "learning_rate": 5.526627218934911e-05,
-      "loss": 1.5669,
+      "epoch": 10.12,
+      "learning_rate": 7.975e-05,
+      "loss": 1.1498,
       "step": 2268
     },
     {
-      "epoch": 6.71,
-      "learning_rate": 5.5246548323471405e-05,
-      "loss": 1.6753,
+      "epoch": 10.13,
+      "learning_rate": 7.974107142857144e-05,
+      "loss": 1.2264,
       "step": 2269
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 5.52268244575937e-05,
-      "loss": 1.5949,
+      "epoch": 10.13,
+      "learning_rate": 7.973214285714286e-05,
+      "loss": 1.1762,
       "step": 2270
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 5.5207100591715975e-05,
-      "loss": 1.592,
+      "epoch": 10.14,
+      "learning_rate": 7.972321428571429e-05,
+      "loss": 1.2156,
       "step": 2271
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 5.518737672583827e-05,
-      "loss": 1.6393,
+      "epoch": 10.14,
+      "learning_rate": 7.971428571428572e-05,
+      "loss": 1.216,
       "step": 2272
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 5.516765285996055e-05,
-      "loss": 1.6183,
+      "epoch": 10.15,
+      "learning_rate": 7.970535714285715e-05,
+      "loss": 1.256,
       "step": 2273
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 5.5147928994082845e-05,
-      "loss": 1.5516,
+      "epoch": 10.15,
+      "learning_rate": 7.969642857142857e-05,
+      "loss": 1.2542,
       "step": 2274
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 5.512820512820514e-05,
-      "loss": 1.6637,
+      "epoch": 10.16,
+      "learning_rate": 7.96875e-05,
+      "loss": 1.1347,
       "step": 2275
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 5.5108481262327416e-05,
-      "loss": 1.5629,
+      "epoch": 10.16,
+      "learning_rate": 7.967857142857143e-05,
+      "loss": 1.1813,
       "step": 2276
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 5.508875739644971e-05,
-      "loss": 1.5463,
+      "epoch": 10.17,
+      "learning_rate": 7.966964285714286e-05,
+      "loss": 1.2984,
       "step": 2277
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 5.5069033530571986e-05,
-      "loss": 1.5953,
+      "epoch": 10.17,
+      "learning_rate": 7.966071428571428e-05,
+      "loss": 1.2009,
       "step": 2278
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 5.5049309664694285e-05,
-      "loss": 1.6574,
+      "epoch": 10.17,
+      "learning_rate": 7.965178571428572e-05,
+      "loss": 1.3209,
       "step": 2279
     },
     {
-      "epoch": 6.75,
-      "learning_rate": 5.502958579881658e-05,
-      "loss": 1.6653,
+      "epoch": 10.18,
+      "learning_rate": 7.964285714285715e-05,
+      "loss": 1.2414,
       "step": 2280
     },
     {
-      "epoch": 6.75,
-      "learning_rate": 5.5009861932938856e-05,
-      "loss": 1.6695,
+      "epoch": 10.18,
+      "learning_rate": 7.963392857142857e-05,
+      "loss": 1.1643,
       "step": 2281
     },
     {
-      "epoch": 6.75,
-      "learning_rate": 5.499013806706115e-05,
-      "loss": 1.5958,
+      "epoch": 10.19,
+      "learning_rate": 7.962500000000001e-05,
+      "loss": 1.2001,
       "step": 2282
     },
     {
-      "epoch": 6.75,
-      "learning_rate": 5.497041420118344e-05,
-      "loss": 1.5628,
+      "epoch": 10.19,
+      "learning_rate": 7.961607142857144e-05,
+      "loss": 1.2842,
       "step": 2283
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 5.495069033530572e-05,
-      "loss": 1.5852,
+      "epoch": 10.2,
+      "learning_rate": 7.960714285714286e-05,
+      "loss": 1.1794,
       "step": 2284
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 5.493096646942801e-05,
-      "loss": 1.6424,
+      "epoch": 10.2,
+      "learning_rate": 7.959821428571429e-05,
+      "loss": 1.2126,
       "step": 2285
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 5.4911242603550296e-05,
-      "loss": 1.6056,
+      "epoch": 10.21,
+      "learning_rate": 7.958928571428573e-05,
+      "loss": 1.1934,
       "step": 2286
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 5.489151873767259e-05,
-      "loss": 1.7359,
+      "epoch": 10.21,
+      "learning_rate": 7.958035714285714e-05,
+      "loss": 1.3287,
       "step": 2287
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 5.487179487179488e-05,
-      "loss": 1.586,
+      "epoch": 10.21,
+      "learning_rate": 7.957142857142858e-05,
+      "loss": 1.1127,
       "step": 2288
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 5.485207100591716e-05,
-      "loss": 1.5643,
+      "epoch": 10.22,
+      "learning_rate": 7.95625e-05,
+      "loss": 1.1876,
       "step": 2289
     },
     {
-      "epoch": 6.78,
-      "learning_rate": 5.483234714003945e-05,
-      "loss": 1.6736,
+      "epoch": 10.22,
+      "learning_rate": 7.955357142857143e-05,
+      "loss": 1.1903,
       "step": 2290
     },
     {
-      "epoch": 6.78,
-      "learning_rate": 5.4812623274161736e-05,
-      "loss": 1.5219,
+      "epoch": 10.23,
+      "learning_rate": 7.954464285714285e-05,
+      "loss": 1.2382,
       "step": 2291
     },
     {
-      "epoch": 6.78,
-      "learning_rate": 5.479289940828403e-05,
-      "loss": 1.8011,
+      "epoch": 10.23,
+      "learning_rate": 7.953571428571429e-05,
+      "loss": 1.2138,
       "step": 2292
     },
     {
-      "epoch": 6.78,
-      "learning_rate": 5.477317554240632e-05,
-      "loss": 1.7237,
+      "epoch": 10.24,
+      "learning_rate": 7.952678571428572e-05,
+      "loss": 1.3138,
       "step": 2293
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 5.47534516765286e-05,
-      "loss": 1.6738,
+      "epoch": 10.24,
+      "learning_rate": 7.951785714285714e-05,
+      "loss": 1.2665,
       "step": 2294
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 5.473372781065089e-05,
-      "loss": 1.5529,
+      "epoch": 10.25,
+      "learning_rate": 7.950892857142858e-05,
+      "loss": 1.2156,
       "step": 2295
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 5.471400394477318e-05,
-      "loss": 1.5506,
+      "epoch": 10.25,
+      "learning_rate": 7.950000000000001e-05,
+      "loss": 1.2406,
       "step": 2296
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 5.469428007889547e-05,
-      "loss": 1.5579,
+      "epoch": 10.25,
+      "learning_rate": 7.949107142857143e-05,
+      "loss": 1.2476,
       "step": 2297
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 5.467455621301776e-05,
-      "loss": 1.5347,
+      "epoch": 10.26,
+      "learning_rate": 7.948214285714286e-05,
+      "loss": 1.1944,
       "step": 2298
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 5.465483234714004e-05,
-      "loss": 1.5628,
+      "epoch": 10.26,
+      "learning_rate": 7.94732142857143e-05,
+      "loss": 1.2241,
       "step": 2299
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 5.463510848126233e-05,
-      "loss": 1.5722,
+      "epoch": 10.27,
+      "learning_rate": 7.946428571428571e-05,
+      "loss": 1.1519,
       "step": 2300
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 5.461538461538461e-05,
-      "loss": 1.6054,
+      "epoch": 10.27,
+      "learning_rate": 7.945535714285715e-05,
+      "loss": 1.226,
       "step": 2301
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 5.45956607495069e-05,
-      "loss": 1.6234,
+      "epoch": 10.28,
+      "learning_rate": 7.944642857142857e-05,
+      "loss": 1.2074,
       "step": 2302
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 5.4575936883629194e-05,
-      "loss": 1.6373,
+      "epoch": 10.28,
+      "learning_rate": 7.943750000000001e-05,
+      "loss": 1.2455,
       "step": 2303
     },
     {
-      "epoch": 6.82,
-      "learning_rate": 5.455621301775148e-05,
-      "loss": 1.5613,
+      "epoch": 10.29,
+      "learning_rate": 7.942857142857143e-05,
+      "loss": 1.1877,
       "step": 2304
     },
     {
-      "epoch": 6.82,
-      "learning_rate": 5.453648915187377e-05,
-      "loss": 1.6901,
+      "epoch": 10.29,
+      "learning_rate": 7.941964285714286e-05,
+      "loss": 1.2019,
       "step": 2305
     },
     {
-      "epoch": 6.82,
-      "learning_rate": 5.4516765285996064e-05,
-      "loss": 1.5735,
+      "epoch": 10.29,
+      "learning_rate": 7.941071428571429e-05,
+      "loss": 1.1613,
       "step": 2306
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 5.449704142011834e-05,
-      "loss": 1.5976,
+      "epoch": 10.3,
+      "learning_rate": 7.940178571428572e-05,
+      "loss": 1.17,
       "step": 2307
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 5.4477317554240635e-05,
-      "loss": 1.6329,
+      "epoch": 10.3,
+      "learning_rate": 7.939285714285714e-05,
+      "loss": 1.2893,
       "step": 2308
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 5.445759368836292e-05,
-      "loss": 1.5565,
+      "epoch": 10.31,
+      "learning_rate": 7.938392857142858e-05,
+      "loss": 1.1696,
       "step": 2309
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 5.443786982248521e-05,
-      "loss": 1.6049,
+      "epoch": 10.31,
+      "learning_rate": 7.9375e-05,
+      "loss": 1.2454,
       "step": 2310
     },
     {
-      "epoch": 6.84,
-      "learning_rate": 5.4418145956607504e-05,
-      "loss": 1.6203,
+      "epoch": 10.32,
+      "learning_rate": 7.936607142857143e-05,
+      "loss": 1.1556,
       "step": 2311
     },
     {
-      "epoch": 6.84,
-      "learning_rate": 5.439842209072978e-05,
-      "loss": 1.6755,
+      "epoch": 10.32,
+      "learning_rate": 7.935714285714287e-05,
+      "loss": 1.1467,
       "step": 2312
     },
     {
-      "epoch": 6.84,
-      "learning_rate": 5.4378698224852075e-05,
-      "loss": 1.6191,
+      "epoch": 10.33,
+      "learning_rate": 7.934821428571428e-05,
+      "loss": 1.1989,
       "step": 2313
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 5.435897435897436e-05,
-      "loss": 1.5839,
+      "epoch": 10.33,
+      "learning_rate": 7.933928571428572e-05,
+      "loss": 1.2452,
       "step": 2314
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 5.433925049309665e-05,
-      "loss": 1.6274,
+      "epoch": 10.33,
+      "learning_rate": 7.933035714285715e-05,
+      "loss": 1.2147,
       "step": 2315
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 5.4319526627218944e-05,
-      "loss": 1.5288,
+      "epoch": 10.34,
+      "learning_rate": 7.932142857142858e-05,
+      "loss": 1.2301,
       "step": 2316
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 5.429980276134122e-05,
-      "loss": 1.6806,
+      "epoch": 10.34,
+      "learning_rate": 7.93125e-05,
+      "loss": 1.2236,
       "step": 2317
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 5.4280078895463515e-05,
-      "loss": 1.5244,
+      "epoch": 10.35,
+      "learning_rate": 7.930357142857144e-05,
+      "loss": 1.2035,
       "step": 2318
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 5.4260355029585794e-05,
-      "loss": 1.618,
+      "epoch": 10.35,
+      "learning_rate": 7.929464285714286e-05,
+      "loss": 1.1722,
       "step": 2319
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 5.4240631163708086e-05,
-      "loss": 1.6673,
+      "epoch": 10.36,
+      "learning_rate": 7.928571428571429e-05,
+      "loss": 1.1315,
       "step": 2320
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 5.422090729783038e-05,
-      "loss": 1.663,
+      "epoch": 10.36,
+      "learning_rate": 7.927678571428571e-05,
+      "loss": 1.2217,
       "step": 2321
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 5.420118343195266e-05,
-      "loss": 1.6004,
+      "epoch": 10.37,
+      "learning_rate": 7.926785714285715e-05,
+      "loss": 1.0848,
       "step": 2322
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 5.4181459566074955e-05,
-      "loss": 1.5649,
+      "epoch": 10.37,
+      "learning_rate": 7.925892857142858e-05,
+      "loss": 1.2205,
       "step": 2323
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 5.4161735700197234e-05,
-      "loss": 1.5782,
+      "epoch": 10.38,
+      "learning_rate": 7.925e-05,
+      "loss": 1.3153,
       "step": 2324
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 5.4142011834319526e-05,
-      "loss": 1.6014,
+      "epoch": 10.38,
+      "learning_rate": 7.924107142857143e-05,
+      "loss": 1.092,
       "step": 2325
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 5.412228796844182e-05,
-      "loss": 1.5304,
+      "epoch": 10.38,
+      "learning_rate": 7.923214285714287e-05,
+      "loss": 1.2386,
       "step": 2326
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 5.41025641025641e-05,
-      "loss": 1.5794,
+      "epoch": 10.39,
+      "learning_rate": 7.922321428571429e-05,
+      "loss": 1.2245,
       "step": 2327
     },
     {
-      "epoch": 6.89,
-      "learning_rate": 5.4082840236686395e-05,
-      "loss": 1.5849,
+      "epoch": 10.39,
+      "learning_rate": 7.921428571428572e-05,
+      "loss": 1.2297,
       "step": 2328
     },
     {
-      "epoch": 6.89,
-      "learning_rate": 5.406311637080869e-05,
-      "loss": 1.5872,
+      "epoch": 10.4,
+      "learning_rate": 7.920535714285716e-05,
+      "loss": 1.2522,
       "step": 2329
     },
     {
-      "epoch": 6.89,
-      "learning_rate": 5.4043392504930966e-05,
-      "loss": 1.5256,
+      "epoch": 10.4,
+      "learning_rate": 7.919642857142857e-05,
+      "loss": 1.2595,
       "step": 2330
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 5.402366863905326e-05,
-      "loss": 1.5696,
+      "epoch": 10.41,
+      "learning_rate": 7.918750000000001e-05,
+      "loss": 1.232,
       "step": 2331
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 5.4003944773175544e-05,
-      "loss": 1.5824,
+      "epoch": 10.41,
+      "learning_rate": 7.917857142857143e-05,
+      "loss": 1.1297,
       "step": 2332
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 5.3984220907297836e-05,
-      "loss": 1.568,
+      "epoch": 10.42,
+      "learning_rate": 7.916964285714286e-05,
+      "loss": 1.2293,
       "step": 2333
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 5.396449704142013e-05,
-      "loss": 1.6071,
+      "epoch": 10.42,
+      "learning_rate": 7.916071428571428e-05,
+      "loss": 1.2777,
       "step": 2334
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 5.3944773175542406e-05,
-      "loss": 1.5164,
+      "epoch": 10.42,
+      "learning_rate": 7.915178571428572e-05,
+      "loss": 1.1782,
       "step": 2335
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 5.39250493096647e-05,
-      "loss": 1.7575,
+      "epoch": 10.43,
+      "learning_rate": 7.914285714285715e-05,
+      "loss": 1.2202,
       "step": 2336
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 5.390532544378698e-05,
-      "loss": 1.6616,
+      "epoch": 10.43,
+      "learning_rate": 7.913392857142857e-05,
+      "loss": 1.2278,
       "step": 2337
     },
     {
-      "epoch": 6.92,
-      "learning_rate": 5.3885601577909276e-05,
-      "loss": 1.6099,
+      "epoch": 10.44,
+      "learning_rate": 7.9125e-05,
+      "loss": 1.151,
       "step": 2338
     },
     {
-      "epoch": 6.92,
-      "learning_rate": 5.386587771203157e-05,
-      "loss": 1.6757,
+      "epoch": 10.44,
+      "learning_rate": 7.911607142857144e-05,
+      "loss": 1.2329,
       "step": 2339
     },
     {
-      "epoch": 6.92,
-      "learning_rate": 5.384615384615385e-05,
-      "loss": 1.5531,
+      "epoch": 10.45,
+      "learning_rate": 7.910714285714286e-05,
+      "loss": 1.2863,
       "step": 2340
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 5.382642998027614e-05,
-      "loss": 1.609,
+      "epoch": 10.45,
+      "learning_rate": 7.909821428571429e-05,
+      "loss": 1.2606,
       "step": 2341
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 5.380670611439842e-05,
-      "loss": 1.5781,
+      "epoch": 10.46,
+      "learning_rate": 7.908928571428573e-05,
+      "loss": 1.1446,
       "step": 2342
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 5.378698224852071e-05,
-      "loss": 1.6105,
+      "epoch": 10.46,
+      "learning_rate": 7.908035714285714e-05,
+      "loss": 1.2063,
       "step": 2343
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 5.3767258382643e-05,
-      "loss": 1.5951,
+      "epoch": 10.46,
+      "learning_rate": 7.907142857142858e-05,
+      "loss": 1.0808,
       "step": 2344
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 5.374753451676529e-05,
-      "loss": 1.5704,
+      "epoch": 10.47,
+      "learning_rate": 7.90625e-05,
+      "loss": 1.1863,
       "step": 2345
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 5.372781065088758e-05,
-      "loss": 1.5163,
+      "epoch": 10.47,
+      "learning_rate": 7.905357142857143e-05,
+      "loss": 1.1079,
       "step": 2346
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 5.370808678500986e-05,
-      "loss": 1.5897,
+      "epoch": 10.48,
+      "learning_rate": 7.904464285714285e-05,
+      "loss": 1.1675,
       "step": 2347
     },
     {
-      "epoch": 6.95,
-      "learning_rate": 5.368836291913215e-05,
-      "loss": 1.6108,
+      "epoch": 10.48,
+      "learning_rate": 7.90357142857143e-05,
+      "loss": 1.1741,
       "step": 2348
     },
     {
-      "epoch": 6.95,
-      "learning_rate": 5.366863905325444e-05,
-      "loss": 1.5421,
+      "epoch": 10.49,
+      "learning_rate": 7.902678571428572e-05,
+      "loss": 1.1891,
       "step": 2349
     },
     {
-      "epoch": 6.95,
-      "learning_rate": 5.364891518737673e-05,
-      "loss": 1.6512,
+      "epoch": 10.49,
+      "learning_rate": 7.901785714285714e-05,
+      "loss": 1.2435,
       "step": 2350
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 5.362919132149902e-05,
-      "loss": 1.6235,
+      "epoch": 10.5,
+      "learning_rate": 7.900892857142857e-05,
+      "loss": 1.2175,
       "step": 2351
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 5.360946745562131e-05,
-      "loss": 1.629,
+      "epoch": 10.5,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.1386,
       "step": 2352
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 5.358974358974359e-05,
-      "loss": 1.5532,
+      "epoch": 10.5,
+      "learning_rate": 7.899107142857143e-05,
+      "loss": 1.1588,
       "step": 2353
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 5.357001972386588e-05,
-      "loss": 1.7182,
+      "epoch": 10.51,
+      "learning_rate": 7.898214285714286e-05,
+      "loss": 1.2098,
       "step": 2354
     },
     {
-      "epoch": 6.97,
-      "learning_rate": 5.355029585798817e-05,
-      "loss": 1.5842,
+      "epoch": 10.51,
+      "learning_rate": 7.897321428571429e-05,
+      "loss": 1.1913,
       "step": 2355
     },
     {
-      "epoch": 6.97,
-      "learning_rate": 5.353057199211046e-05,
-      "loss": 1.6211,
+      "epoch": 10.52,
+      "learning_rate": 7.896428571428571e-05,
+      "loss": 1.2725,
       "step": 2356
     },
     {
-      "epoch": 6.97,
-      "learning_rate": 5.351084812623275e-05,
-      "loss": 1.6419,
+      "epoch": 10.52,
+      "learning_rate": 7.895535714285715e-05,
+      "loss": 1.2792,
       "step": 2357
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 5.349112426035503e-05,
-      "loss": 1.4771,
+      "epoch": 10.53,
+      "learning_rate": 7.894642857142858e-05,
+      "loss": 1.1433,
       "step": 2358
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 5.347140039447732e-05,
-      "loss": 1.6394,
+      "epoch": 10.53,
+      "learning_rate": 7.893750000000001e-05,
+      "loss": 1.1983,
       "step": 2359
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 5.34516765285996e-05,
-      "loss": 1.5242,
+      "epoch": 10.54,
+      "learning_rate": 7.892857142857143e-05,
+      "loss": 1.2217,
       "step": 2360
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 5.343195266272189e-05,
-      "loss": 1.5229,
+      "epoch": 10.54,
+      "learning_rate": 7.891964285714287e-05,
+      "loss": 1.3091,
       "step": 2361
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 5.3412228796844185e-05,
-      "loss": 1.6056,
+      "epoch": 10.54,
+      "learning_rate": 7.891071428571429e-05,
+      "loss": 1.2151,
       "step": 2362
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 5.339250493096647e-05,
-      "loss": 1.6066,
+      "epoch": 10.55,
+      "learning_rate": 7.890178571428572e-05,
+      "loss": 1.1579,
       "step": 2363
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 5.337278106508876e-05,
-      "loss": 1.6204,
+      "epoch": 10.55,
+      "learning_rate": 7.889285714285714e-05,
+      "loss": 1.1751,
       "step": 2364
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 5.335305719921104e-05,
-      "loss": 1.6586,
+      "epoch": 10.56,
+      "learning_rate": 7.888392857142858e-05,
+      "loss": 1.2553,
       "step": 2365
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 5.333333333333333e-05,
-      "loss": 1.6438,
+      "epoch": 10.56,
+      "learning_rate": 7.887499999999999e-05,
+      "loss": 1.2355,
       "step": 2366
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 5.3313609467455625e-05,
-      "loss": 1.557,
+      "epoch": 10.57,
+      "learning_rate": 7.886607142857143e-05,
+      "loss": 1.2443,
       "step": 2367
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 5.329388560157791e-05,
-      "loss": 1.6326,
+      "epoch": 10.57,
+      "learning_rate": 7.885714285714286e-05,
+      "loss": 1.1504,
       "step": 2368
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 5.32741617357002e-05,
-      "loss": 1.548,
+      "epoch": 10.58,
+      "learning_rate": 7.88482142857143e-05,
+      "loss": 1.2455,
       "step": 2369
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 5.3254437869822495e-05,
-      "loss": 1.6138,
+      "epoch": 10.58,
+      "learning_rate": 7.883928571428572e-05,
+      "loss": 1.2226,
       "step": 2370
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 5.3234714003944773e-05,
-      "loss": 1.4844,
+      "epoch": 10.58,
+      "learning_rate": 7.883035714285715e-05,
+      "loss": 1.1571,
       "step": 2371
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 5.3214990138067066e-05,
-      "loss": 1.5519,
+      "epoch": 10.59,
+      "learning_rate": 7.882142857142859e-05,
+      "loss": 1.1889,
       "step": 2372
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 5.319526627218935e-05,
-      "loss": 1.5731,
+      "epoch": 10.59,
+      "learning_rate": 7.88125e-05,
+      "loss": 1.2914,
       "step": 2373
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 5.317554240631164e-05,
-      "loss": 1.7151,
+      "epoch": 10.6,
+      "learning_rate": 7.880357142857144e-05,
+      "loss": 1.2057,
       "step": 2374
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 5.3155818540433935e-05,
-      "loss": 1.6242,
+      "epoch": 10.6,
+      "learning_rate": 7.879464285714286e-05,
+      "loss": 1.1737,
       "step": 2375
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 5.3136094674556214e-05,
-      "loss": 1.554,
+      "epoch": 10.61,
+      "learning_rate": 7.878571428571429e-05,
+      "loss": 1.1883,
       "step": 2376
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 5.3116370808678506e-05,
-      "loss": 1.6007,
+      "epoch": 10.61,
+      "learning_rate": 7.877678571428571e-05,
+      "loss": 1.1666,
       "step": 2377
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 5.3096646942800784e-05,
-      "loss": 1.5668,
+      "epoch": 10.62,
+      "learning_rate": 7.876785714285715e-05,
+      "loss": 1.2097,
       "step": 2378
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 5.3076923076923076e-05,
-      "loss": 1.6583,
+      "epoch": 10.62,
+      "learning_rate": 7.875892857142858e-05,
+      "loss": 1.151,
       "step": 2379
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 5.3057199211045375e-05,
-      "loss": 1.4355,
+      "epoch": 10.62,
+      "learning_rate": 7.875e-05,
+      "loss": 1.1368,
       "step": 2380
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 5.3037475345167654e-05,
-      "loss": 1.4911,
+      "epoch": 10.63,
+      "learning_rate": 7.874107142857143e-05,
+      "loss": 1.2179,
       "step": 2381
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 5.3017751479289946e-05,
-      "loss": 1.5661,
+      "epoch": 10.63,
+      "learning_rate": 7.873214285714287e-05,
+      "loss": 1.1517,
       "step": 2382
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 5.2998027613412225e-05,
-      "loss": 1.6181,
+      "epoch": 10.64,
+      "learning_rate": 7.872321428571428e-05,
+      "loss": 1.2446,
       "step": 2383
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 5.297830374753452e-05,
-      "loss": 1.6415,
+      "epoch": 10.64,
+      "learning_rate": 7.871428571428572e-05,
+      "loss": 1.2371,
       "step": 2384
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 5.295857988165681e-05,
-      "loss": 1.6076,
+      "epoch": 10.65,
+      "learning_rate": 7.870535714285714e-05,
+      "loss": 1.2082,
       "step": 2385
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 5.2938856015779094e-05,
-      "loss": 1.5595,
+      "epoch": 10.65,
+      "learning_rate": 7.869642857142857e-05,
+      "loss": 1.2206,
       "step": 2386
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 5.2919132149901386e-05,
-      "loss": 1.6019,
+      "epoch": 10.66,
+      "learning_rate": 7.868750000000001e-05,
+      "loss": 1.2896,
       "step": 2387
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 5.2899408284023665e-05,
-      "loss": 1.5907,
+      "epoch": 10.66,
+      "learning_rate": 7.867857142857143e-05,
+      "loss": 1.2546,
       "step": 2388
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 5.287968441814596e-05,
-      "loss": 1.555,
+      "epoch": 10.67,
+      "learning_rate": 7.866964285714286e-05,
+      "loss": 1.1534,
       "step": 2389
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 5.285996055226825e-05,
-      "loss": 1.5377,
+      "epoch": 10.67,
+      "learning_rate": 7.866071428571428e-05,
+      "loss": 1.208,
       "step": 2390
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 5.2840236686390534e-05,
-      "loss": 1.5717,
+      "epoch": 10.67,
+      "learning_rate": 7.865178571428572e-05,
+      "loss": 1.1849,
       "step": 2391
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 5.2820512820512826e-05,
-      "loss": 1.6219,
+      "epoch": 10.68,
+      "learning_rate": 7.864285714285715e-05,
+      "loss": 1.2387,
       "step": 2392
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 5.280078895463512e-05,
-      "loss": 1.5771,
+      "epoch": 10.68,
+      "learning_rate": 7.863392857142857e-05,
+      "loss": 1.2702,
       "step": 2393
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 5.27810650887574e-05,
-      "loss": 1.6174,
+      "epoch": 10.69,
+      "learning_rate": 7.8625e-05,
+      "loss": 1.1973,
       "step": 2394
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 5.276134122287969e-05,
-      "loss": 1.6329,
+      "epoch": 10.69,
+      "learning_rate": 7.861607142857144e-05,
+      "loss": 1.1675,
       "step": 2395
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 5.274161735700197e-05,
-      "loss": 1.55,
+      "epoch": 10.7,
+      "learning_rate": 7.860714285714285e-05,
+      "loss": 1.2399,
       "step": 2396
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 5.272189349112427e-05,
-      "loss": 1.6396,
+      "epoch": 10.7,
+      "learning_rate": 7.859821428571429e-05,
+      "loss": 1.1642,
       "step": 2397
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 5.270216962524656e-05,
-      "loss": 1.5326,
+      "epoch": 10.71,
+      "learning_rate": 7.858928571428572e-05,
+      "loss": 1.1586,
       "step": 2398
     },
     {
-      "epoch": 7.1,
-      "learning_rate": 5.268244575936884e-05,
-      "loss": 1.5296,
+      "epoch": 10.71,
+      "learning_rate": 7.858035714285714e-05,
+      "loss": 1.1965,
       "step": 2399
     },
     {
-      "epoch": 7.1,
-      "learning_rate": 5.266272189349113e-05,
-      "loss": 1.5958,
+      "epoch": 10.71,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 1.2048,
       "step": 2400
     },
     {
-      "epoch": 7.1,
-      "learning_rate": 5.264299802761341e-05,
-      "loss": 1.5954,
+      "epoch": 10.71,
+      "eval_loss": 1.1748493909835815,
+      "eval_runtime": 45.6476,
+      "eval_samples_per_second": 286.85,
+      "eval_steps_per_second": 1.512,
+      "step": 2400
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 7.85625e-05,
+      "loss": 1.2259,
       "step": 2401
     },
     {
-      "epoch": 7.11,
-      "learning_rate": 5.26232741617357e-05,
-      "loss": 1.5835,
+      "epoch": 10.72,
+      "learning_rate": 7.855357142857143e-05,
+      "loss": 1.1823,
       "step": 2402
     },
     {
-      "epoch": 7.11,
-      "learning_rate": 5.260355029585799e-05,
-      "loss": 1.5804,
+      "epoch": 10.73,
+      "learning_rate": 7.854464285714286e-05,
+      "loss": 1.1596,
       "step": 2403
     },
     {
-      "epoch": 7.11,
-      "learning_rate": 5.258382642998028e-05,
-      "loss": 1.5872,
+      "epoch": 10.73,
+      "learning_rate": 7.85357142857143e-05,
+      "loss": 1.2162,
       "step": 2404
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 5.256410256410257e-05,
-      "loss": 1.6489,
+      "epoch": 10.74,
+      "learning_rate": 7.852678571428572e-05,
+      "loss": 1.2421,
       "step": 2405
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 5.254437869822485e-05,
-      "loss": 1.5896,
+      "epoch": 10.74,
+      "learning_rate": 7.851785714285715e-05,
+      "loss": 1.1851,
       "step": 2406
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 5.252465483234714e-05,
-      "loss": 1.587,
+      "epoch": 10.75,
+      "learning_rate": 7.850892857142857e-05,
+      "loss": 1.1463,
       "step": 2407
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 5.250493096646943e-05,
-      "loss": 1.6378,
+      "epoch": 10.75,
+      "learning_rate": 7.850000000000001e-05,
+      "loss": 1.258,
       "step": 2408
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 5.248520710059172e-05,
-      "loss": 1.6398,
+      "epoch": 10.75,
+      "learning_rate": 7.849107142857142e-05,
+      "loss": 1.1682,
       "step": 2409
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 5.246548323471401e-05,
-      "loss": 1.5728,
+      "epoch": 10.76,
+      "learning_rate": 7.848214285714286e-05,
+      "loss": 1.19,
       "step": 2410
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 5.244575936883629e-05,
-      "loss": 1.5619,
+      "epoch": 10.76,
+      "learning_rate": 7.847321428571429e-05,
+      "loss": 1.1883,
       "step": 2411
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 5.242603550295858e-05,
-      "loss": 1.5622,
+      "epoch": 10.77,
+      "learning_rate": 7.846428571428573e-05,
+      "loss": 1.1697,
       "step": 2412
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 5.240631163708087e-05,
-      "loss": 1.5956,
+      "epoch": 10.77,
+      "learning_rate": 7.845535714285714e-05,
+      "loss": 1.2484,
       "step": 2413
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 5.238658777120316e-05,
-      "loss": 1.7068,
+      "epoch": 10.78,
+      "learning_rate": 7.844642857142858e-05,
+      "loss": 1.1776,
       "step": 2414
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 5.236686390532545e-05,
-      "loss": 1.5869,
+      "epoch": 10.78,
+      "learning_rate": 7.84375e-05,
+      "loss": 1.2767,
       "step": 2415
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 5.234714003944774e-05,
-      "loss": 1.5638,
+      "epoch": 10.79,
+      "learning_rate": 7.842857142857143e-05,
+      "loss": 1.1423,
       "step": 2416
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 5.232741617357002e-05,
-      "loss": 1.5306,
+      "epoch": 10.79,
+      "learning_rate": 7.841964285714287e-05,
+      "loss": 1.1612,
       "step": 2417
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 5.230769230769231e-05,
-      "loss": 1.5515,
+      "epoch": 10.79,
+      "learning_rate": 7.841071428571429e-05,
+      "loss": 1.1988,
       "step": 2418
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 5.228796844181459e-05,
-      "loss": 1.5729,
+      "epoch": 10.8,
+      "learning_rate": 7.840178571428572e-05,
+      "loss": 1.2087,
       "step": 2419
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 5.2268244575936884e-05,
-      "loss": 1.621,
+      "epoch": 10.8,
+      "learning_rate": 7.839285714285714e-05,
+      "loss": 1.1304,
       "step": 2420
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 5.2248520710059176e-05,
-      "loss": 1.533,
+      "epoch": 10.81,
+      "learning_rate": 7.838392857142858e-05,
+      "loss": 1.2032,
       "step": 2421
     },
     {
-      "epoch": 7.17,
-      "learning_rate": 5.222879684418146e-05,
-      "loss": 1.5682,
+      "epoch": 10.81,
+      "learning_rate": 7.8375e-05,
+      "loss": 1.2108,
       "step": 2422
     },
     {
-      "epoch": 7.17,
-      "learning_rate": 5.220907297830375e-05,
-      "loss": 1.6423,
+      "epoch": 10.82,
+      "learning_rate": 7.836607142857143e-05,
+      "loss": 1.3268,
       "step": 2423
     },
     {
-      "epoch": 7.17,
-      "learning_rate": 5.218934911242603e-05,
-      "loss": 1.5769,
+      "epoch": 10.82,
+      "learning_rate": 7.835714285714286e-05,
+      "loss": 1.2346,
       "step": 2424
     },
     {
-      "epoch": 7.17,
-      "learning_rate": 5.2169625246548324e-05,
-      "loss": 1.6267,
+      "epoch": 10.83,
+      "learning_rate": 7.83482142857143e-05,
+      "loss": 1.1983,
       "step": 2425
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 5.2149901380670616e-05,
-      "loss": 1.5839,
+      "epoch": 10.83,
+      "learning_rate": 7.833928571428571e-05,
+      "loss": 1.131,
       "step": 2426
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 5.21301775147929e-05,
-      "loss": 1.5835,
+      "epoch": 10.83,
+      "learning_rate": 7.833035714285715e-05,
+      "loss": 1.2447,
       "step": 2427
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 5.2110453648915193e-05,
-      "loss": 1.4921,
+      "epoch": 10.84,
+      "learning_rate": 7.832142857142857e-05,
+      "loss": 1.2862,
       "step": 2428
     },
     {
-      "epoch": 7.19,
-      "learning_rate": 5.209072978303747e-05,
-      "loss": 1.6028,
+      "epoch": 10.84,
+      "learning_rate": 7.83125e-05,
+      "loss": 1.2219,
       "step": 2429
     },
     {
-      "epoch": 7.19,
-      "learning_rate": 5.2071005917159764e-05,
-      "loss": 1.6083,
+      "epoch": 10.85,
+      "learning_rate": 7.830357142857144e-05,
+      "loss": 1.1792,
       "step": 2430
     },
     {
-      "epoch": 7.19,
-      "learning_rate": 5.2051282051282056e-05,
-      "loss": 1.6222,
+      "epoch": 10.85,
+      "learning_rate": 7.829464285714286e-05,
+      "loss": 1.1846,
       "step": 2431
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 5.203155818540434e-05,
-      "loss": 1.4607,
+      "epoch": 10.86,
+      "learning_rate": 7.828571428571429e-05,
+      "loss": 1.2285,
       "step": 2432
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 5.2011834319526634e-05,
-      "loss": 1.5086,
+      "epoch": 10.86,
+      "learning_rate": 7.827678571428571e-05,
+      "loss": 1.2274,
       "step": 2433
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 5.199211045364891e-05,
-      "loss": 1.5501,
+      "epoch": 10.87,
+      "learning_rate": 7.826785714285715e-05,
+      "loss": 1.1881,
       "step": 2434
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 5.1972386587771204e-05,
-      "loss": 1.6566,
+      "epoch": 10.87,
+      "learning_rate": 7.825892857142858e-05,
+      "loss": 1.2174,
       "step": 2435
     },
     {
-      "epoch": 7.21,
-      "learning_rate": 5.1952662721893497e-05,
-      "loss": 1.5371,
+      "epoch": 10.88,
+      "learning_rate": 7.825e-05,
+      "loss": 1.1977,
       "step": 2436
     },
     {
-      "epoch": 7.21,
-      "learning_rate": 5.1932938856015775e-05,
-      "loss": 1.5238,
+      "epoch": 10.88,
+      "learning_rate": 7.824107142857143e-05,
+      "loss": 1.1828,
       "step": 2437
     },
     {
-      "epoch": 7.21,
-      "learning_rate": 5.191321499013807e-05,
-      "loss": 1.4676,
+      "epoch": 10.88,
+      "learning_rate": 7.823214285714287e-05,
+      "loss": 1.24,
       "step": 2438
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 5.1893491124260366e-05,
-      "loss": 1.599,
+      "epoch": 10.89,
+      "learning_rate": 7.822321428571428e-05,
+      "loss": 1.3356,
       "step": 2439
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 5.1873767258382645e-05,
-      "loss": 1.5955,
+      "epoch": 10.89,
+      "learning_rate": 7.821428571428572e-05,
+      "loss": 1.2148,
       "step": 2440
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 5.185404339250494e-05,
-      "loss": 1.5569,
+      "epoch": 10.9,
+      "learning_rate": 7.820535714285715e-05,
+      "loss": 1.2198,
       "step": 2441
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 5.1834319526627215e-05,
-      "loss": 1.588,
+      "epoch": 10.9,
+      "learning_rate": 7.819642857142857e-05,
+      "loss": 1.2348,
       "step": 2442
     },
     {
-      "epoch": 7.23,
-      "learning_rate": 5.181459566074951e-05,
-      "loss": 1.6116,
+      "epoch": 10.91,
+      "learning_rate": 7.81875e-05,
+      "loss": 1.1645,
       "step": 2443
     },
     {
-      "epoch": 7.23,
-      "learning_rate": 5.17948717948718e-05,
-      "loss": 1.5511,
+      "epoch": 10.91,
+      "learning_rate": 7.817857142857144e-05,
+      "loss": 1.2573,
       "step": 2444
     },
     {
-      "epoch": 7.23,
-      "learning_rate": 5.1775147928994085e-05,
-      "loss": 1.6564,
+      "epoch": 10.92,
+      "learning_rate": 7.816964285714286e-05,
+      "loss": 1.267,
       "step": 2445
     },
     {
-      "epoch": 7.24,
-      "learning_rate": 5.175542406311638e-05,
-      "loss": 1.5787,
+      "epoch": 10.92,
+      "learning_rate": 7.816071428571429e-05,
+      "loss": 1.2328,
       "step": 2446
     },
     {
-      "epoch": 7.24,
-      "learning_rate": 5.1735700197238656e-05,
-      "loss": 1.7064,
+      "epoch": 10.92,
+      "learning_rate": 7.815178571428573e-05,
+      "loss": 1.1326,
       "step": 2447
     },
     {
-      "epoch": 7.24,
-      "learning_rate": 5.171597633136095e-05,
-      "loss": 1.6236,
+      "epoch": 10.93,
+      "learning_rate": 7.814285714285715e-05,
+      "loss": 1.2079,
       "step": 2448
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 5.169625246548324e-05,
-      "loss": 1.6374,
+      "epoch": 10.93,
+      "learning_rate": 7.813392857142858e-05,
+      "loss": 1.1966,
       "step": 2449
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 5.1676528599605525e-05,
-      "loss": 1.4994,
+      "epoch": 10.94,
+      "learning_rate": 7.8125e-05,
+      "loss": 1.1765,
       "step": 2450
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 5.165680473372782e-05,
-      "loss": 1.5591,
+      "epoch": 10.94,
+      "learning_rate": 7.811607142857144e-05,
+      "loss": 1.1013,
       "step": 2451
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 5.1637080867850096e-05,
-      "loss": 1.511,
+      "epoch": 10.95,
+      "learning_rate": 7.810714285714285e-05,
+      "loss": 1.1885,
       "step": 2452
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 5.161735700197239e-05,
-      "loss": 1.5993,
+      "epoch": 10.95,
+      "learning_rate": 7.809821428571429e-05,
+      "loss": 1.2183,
       "step": 2453
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 5.159763313609468e-05,
-      "loss": 1.5864,
+      "epoch": 10.96,
+      "learning_rate": 7.808928571428572e-05,
+      "loss": 1.2854,
       "step": 2454
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 5.157790927021696e-05,
-      "loss": 1.5499,
+      "epoch": 10.96,
+      "learning_rate": 7.808035714285714e-05,
+      "loss": 1.2194,
       "step": 2455
     },
     {
-      "epoch": 7.27,
-      "learning_rate": 5.155818540433926e-05,
-      "loss": 1.5621,
+      "epoch": 10.96,
+      "learning_rate": 7.807142857142857e-05,
+      "loss": 1.2073,
       "step": 2456
     },
     {
-      "epoch": 7.27,
-      "learning_rate": 5.1538461538461536e-05,
-      "loss": 1.5951,
+      "epoch": 10.97,
+      "learning_rate": 7.806250000000001e-05,
+      "loss": 1.1968,
       "step": 2457
     },
     {
-      "epoch": 7.27,
-      "learning_rate": 5.151873767258383e-05,
-      "loss": 1.5453,
+      "epoch": 10.97,
+      "learning_rate": 7.805357142857143e-05,
+      "loss": 1.1425,
       "step": 2458
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 5.149901380670612e-05,
-      "loss": 1.5588,
+      "epoch": 10.98,
+      "learning_rate": 7.804464285714286e-05,
+      "loss": 1.1791,
       "step": 2459
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 5.14792899408284e-05,
-      "loss": 1.6138,
+      "epoch": 10.98,
+      "learning_rate": 7.803571428571428e-05,
+      "loss": 1.2205,
       "step": 2460
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 5.145956607495069e-05,
-      "loss": 1.5649,
+      "epoch": 10.99,
+      "learning_rate": 7.802678571428572e-05,
+      "loss": 1.1481,
       "step": 2461
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 5.143984220907298e-05,
-      "loss": 1.5677,
+      "epoch": 10.99,
+      "learning_rate": 7.801785714285715e-05,
+      "loss": 1.189,
       "step": 2462
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 5.142011834319527e-05,
-      "loss": 1.52,
+      "epoch": 11.0,
+      "learning_rate": 7.800892857142857e-05,
+      "loss": 1.1496,
       "step": 2463
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 5.140039447731756e-05,
-      "loss": 1.6177,
+      "epoch": 11.0,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.2592,
       "step": 2464
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 5.138067061143984e-05,
-      "loss": 1.6592,
+      "epoch": 11.0,
+      "learning_rate": 7.799107142857142e-05,
+      "loss": 1.2259,
       "step": 2465
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 5.136094674556213e-05,
-      "loss": 1.5731,
+      "epoch": 11.01,
+      "learning_rate": 7.798214285714286e-05,
+      "loss": 1.1261,
       "step": 2466
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 5.134122287968442e-05,
-      "loss": 1.6572,
+      "epoch": 11.01,
+      "learning_rate": 7.797321428571429e-05,
+      "loss": 1.2085,
       "step": 2467
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 5.132149901380671e-05,
-      "loss": 1.6608,
+      "epoch": 11.02,
+      "learning_rate": 7.796428571428573e-05,
+      "loss": 1.1686,
       "step": 2468
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 5.1301775147929e-05,
-      "loss": 1.535,
+      "epoch": 11.02,
+      "learning_rate": 7.795535714285714e-05,
+      "loss": 1.207,
       "step": 2469
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 5.128205128205128e-05,
-      "loss": 1.4749,
+      "epoch": 11.03,
+      "learning_rate": 7.794642857142858e-05,
+      "loss": 1.1603,
       "step": 2470
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 5.126232741617357e-05,
-      "loss": 1.6207,
+      "epoch": 11.03,
+      "learning_rate": 7.79375e-05,
+      "loss": 1.1799,
       "step": 2471
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 5.1242603550295864e-05,
-      "loss": 1.6127,
+      "epoch": 11.04,
+      "learning_rate": 7.792857142857143e-05,
+      "loss": 1.2276,
       "step": 2472
     },
     {
-      "epoch": 7.32,
-      "learning_rate": 5.122287968441815e-05,
-      "loss": 1.5796,
+      "epoch": 11.04,
+      "learning_rate": 7.791964285714285e-05,
+      "loss": 1.1375,
       "step": 2473
     },
     {
-      "epoch": 7.32,
-      "learning_rate": 5.120315581854044e-05,
-      "loss": 1.5719,
+      "epoch": 11.04,
+      "learning_rate": 7.79107142857143e-05,
+      "loss": 1.2396,
       "step": 2474
     },
     {
-      "epoch": 7.32,
-      "learning_rate": 5.118343195266272e-05,
-      "loss": 1.5169,
+      "epoch": 11.05,
+      "learning_rate": 7.790178571428572e-05,
+      "loss": 1.1997,
       "step": 2475
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 5.116370808678501e-05,
-      "loss": 1.4565,
+      "epoch": 11.05,
+      "learning_rate": 7.789285714285714e-05,
+      "loss": 1.1697,
       "step": 2476
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 5.1143984220907304e-05,
-      "loss": 1.738,
+      "epoch": 11.06,
+      "learning_rate": 7.788392857142858e-05,
+      "loss": 1.1782,
       "step": 2477
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 5.112426035502958e-05,
-      "loss": 1.6292,
+      "epoch": 11.06,
+      "learning_rate": 7.787500000000001e-05,
+      "loss": 1.1938,
       "step": 2478
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 5.1104536489151874e-05,
-      "loss": 1.6109,
+      "epoch": 11.07,
+      "learning_rate": 7.786607142857143e-05,
+      "loss": 1.1066,
       "step": 2479
     },
     {
-      "epoch": 7.34,
-      "learning_rate": 5.1084812623274167e-05,
-      "loss": 1.6101,
+      "epoch": 11.07,
+      "learning_rate": 7.785714285714286e-05,
+      "loss": 1.1516,
       "step": 2480
     },
     {
-      "epoch": 7.34,
-      "learning_rate": 5.106508875739645e-05,
-      "loss": 1.6435,
+      "epoch": 11.08,
+      "learning_rate": 7.78482142857143e-05,
+      "loss": 1.1617,
       "step": 2481
     },
     {
-      "epoch": 7.34,
-      "learning_rate": 5.1045364891518744e-05,
-      "loss": 1.5786,
+      "epoch": 11.08,
+      "learning_rate": 7.783928571428571e-05,
+      "loss": 1.1825,
       "step": 2482
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 5.102564102564102e-05,
-      "loss": 1.5508,
+      "epoch": 11.08,
+      "learning_rate": 7.783035714285715e-05,
+      "loss": 1.2126,
       "step": 2483
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 5.1005917159763315e-05,
-      "loss": 1.5764,
+      "epoch": 11.09,
+      "learning_rate": 7.782142857142858e-05,
+      "loss": 1.2141,
       "step": 2484
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 5.098619329388561e-05,
-      "loss": 1.6565,
+      "epoch": 11.09,
+      "learning_rate": 7.78125e-05,
+      "loss": 1.2715,
       "step": 2485
     },
     {
-      "epoch": 7.36,
-      "learning_rate": 5.096646942800789e-05,
-      "loss": 1.6528,
+      "epoch": 11.1,
+      "learning_rate": 7.780357142857143e-05,
+      "loss": 1.2076,
       "step": 2486
     },
     {
-      "epoch": 7.36,
-      "learning_rate": 5.0946745562130184e-05,
-      "loss": 1.5949,
+      "epoch": 11.1,
+      "learning_rate": 7.779464285714287e-05,
+      "loss": 1.1522,
       "step": 2487
     },
     {
-      "epoch": 7.36,
-      "learning_rate": 5.092702169625246e-05,
-      "loss": 1.5998,
+      "epoch": 11.11,
+      "learning_rate": 7.778571428571429e-05,
+      "loss": 1.1656,
       "step": 2488
     },
     {
-      "epoch": 7.36,
-      "learning_rate": 5.0907297830374755e-05,
-      "loss": 1.6078,
+      "epoch": 11.11,
+      "learning_rate": 7.777678571428572e-05,
+      "loss": 1.2469,
       "step": 2489
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 5.088757396449705e-05,
-      "loss": 1.6205,
+      "epoch": 11.12,
+      "learning_rate": 7.776785714285714e-05,
+      "loss": 1.1709,
       "step": 2490
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 5.086785009861933e-05,
-      "loss": 1.5645,
+      "epoch": 11.12,
+      "learning_rate": 7.775892857142858e-05,
+      "loss": 1.1859,
       "step": 2491
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 5.0848126232741624e-05,
-      "loss": 1.6635,
+      "epoch": 11.12,
+      "learning_rate": 7.775e-05,
+      "loss": 1.1845,
       "step": 2492
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 5.08284023668639e-05,
-      "loss": 1.5888,
+      "epoch": 11.13,
+      "learning_rate": 7.774107142857143e-05,
+      "loss": 1.1962,
       "step": 2493
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 5.0808678500986195e-05,
-      "loss": 1.5121,
+      "epoch": 11.13,
+      "learning_rate": 7.773214285714287e-05,
+      "loss": 1.1739,
       "step": 2494
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 5.078895463510849e-05,
-      "loss": 1.6034,
+      "epoch": 11.14,
+      "learning_rate": 7.772321428571428e-05,
+      "loss": 1.1699,
       "step": 2495
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 5.0769230769230766e-05,
-      "loss": 1.652,
+      "epoch": 11.14,
+      "learning_rate": 7.771428571428572e-05,
+      "loss": 1.2323,
       "step": 2496
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 5.074950690335306e-05,
-      "loss": 1.6463,
+      "epoch": 11.15,
+      "learning_rate": 7.770535714285715e-05,
+      "loss": 1.2232,
       "step": 2497
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 5.072978303747534e-05,
-      "loss": 1.5371,
+      "epoch": 11.15,
+      "learning_rate": 7.769642857142857e-05,
+      "loss": 1.2449,
       "step": 2498
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 5.0710059171597635e-05,
-      "loss": 1.5415,
+      "epoch": 11.16,
+      "learning_rate": 7.76875e-05,
+      "loss": 1.211,
       "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,
+      "epoch": 11.16,
+      "learning_rate": 7.767857142857144e-05,
+      "loss": 1.2755,
       "step": 2500
     },
     {
-      "epoch": 7.4,
-      "learning_rate": 5.0670611439842206e-05,
-      "loss": 1.56,
+      "epoch": 11.17,
+      "learning_rate": 7.766964285714286e-05,
+      "loss": 1.1674,
       "step": 2501
     },
     {
-      "epoch": 7.4,
-      "learning_rate": 5.06508875739645e-05,
-      "loss": 1.6422,
+      "epoch": 11.17,
+      "learning_rate": 7.766071428571429e-05,
+      "loss": 1.1781,
       "step": 2502
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 5.063116370808679e-05,
-      "loss": 1.6429,
+      "epoch": 11.17,
+      "learning_rate": 7.765178571428571e-05,
+      "loss": 1.1831,
       "step": 2503
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 5.0611439842209076e-05,
-      "loss": 1.6198,
+      "epoch": 11.18,
+      "learning_rate": 7.764285714285715e-05,
+      "loss": 1.1487,
       "step": 2504
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 5.059171597633137e-05,
-      "loss": 1.6745,
+      "epoch": 11.18,
+      "learning_rate": 7.763392857142858e-05,
+      "loss": 1.2138,
       "step": 2505
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 5.0571992110453646e-05,
-      "loss": 1.6412,
+      "epoch": 11.19,
+      "learning_rate": 7.7625e-05,
+      "loss": 1.1774,
       "step": 2506
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 5.055226824457594e-05,
-      "loss": 1.6325,
+      "epoch": 11.19,
+      "learning_rate": 7.761607142857143e-05,
+      "loss": 1.2772,
       "step": 2507
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 5.053254437869823e-05,
-      "loss": 1.5825,
+      "epoch": 11.2,
+      "learning_rate": 7.760714285714285e-05,
+      "loss": 1.184,
       "step": 2508
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 5.0512820512820516e-05,
-      "loss": 1.6123,
+      "epoch": 11.2,
+      "learning_rate": 7.759821428571429e-05,
+      "loss": 1.1983,
       "step": 2509
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 5.049309664694281e-05,
-      "loss": 1.5409,
+      "epoch": 11.21,
+      "learning_rate": 7.758928571428572e-05,
+      "loss": 1.2523,
       "step": 2510
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 5.0473372781065087e-05,
-      "loss": 1.7114,
+      "epoch": 11.21,
+      "learning_rate": 7.758035714285716e-05,
+      "loss": 1.1624,
       "step": 2511
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 5.045364891518738e-05,
-      "loss": 1.5168,
+      "epoch": 11.21,
+      "learning_rate": 7.757142857142857e-05,
+      "loss": 1.2118,
       "step": 2512
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 5.043392504930967e-05,
-      "loss": 1.6037,
+      "epoch": 11.22,
+      "learning_rate": 7.756250000000001e-05,
+      "loss": 1.1144,
       "step": 2513
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 5.0414201183431956e-05,
-      "loss": 1.5234,
+      "epoch": 11.22,
+      "learning_rate": 7.755357142857143e-05,
+      "loss": 1.1456,
       "step": 2514
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 5.039447731755425e-05,
-      "loss": 1.6489,
+      "epoch": 11.23,
+      "learning_rate": 7.754464285714286e-05,
+      "loss": 1.1,
       "step": 2515
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 5.037475345167653e-05,
-      "loss": 1.5278,
+      "epoch": 11.23,
+      "learning_rate": 7.753571428571428e-05,
+      "loss": 1.2119,
       "step": 2516
     },
     {
-      "epoch": 7.45,
-      "learning_rate": 5.035502958579882e-05,
-      "loss": 1.5757,
+      "epoch": 11.24,
+      "learning_rate": 7.752678571428572e-05,
+      "loss": 1.1815,
       "step": 2517
     },
     {
-      "epoch": 7.45,
-      "learning_rate": 5.033530571992111e-05,
-      "loss": 1.5551,
+      "epoch": 11.24,
+      "learning_rate": 7.751785714285714e-05,
+      "loss": 1.2118,
       "step": 2518
     },
     {
-      "epoch": 7.45,
-      "learning_rate": 5.031558185404339e-05,
-      "loss": 1.5033,
+      "epoch": 11.25,
+      "learning_rate": 7.750892857142857e-05,
+      "loss": 1.1575,
       "step": 2519
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 5.029585798816568e-05,
-      "loss": 1.5316,
+      "epoch": 11.25,
+      "learning_rate": 7.75e-05,
+      "loss": 1.1955,
       "step": 2520
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 5.027613412228797e-05,
-      "loss": 1.5236,
+      "epoch": 11.25,
+      "learning_rate": 7.749107142857143e-05,
+      "loss": 1.1884,
       "step": 2521
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 5.025641025641026e-05,
-      "loss": 1.6251,
+      "epoch": 11.26,
+      "learning_rate": 7.748214285714286e-05,
+      "loss": 1.2626,
       "step": 2522
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 5.023668639053255e-05,
-      "loss": 1.5025,
+      "epoch": 11.26,
+      "learning_rate": 7.747321428571429e-05,
+      "loss": 1.2487,
       "step": 2523
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 5.021696252465483e-05,
-      "loss": 1.526,
+      "epoch": 11.27,
+      "learning_rate": 7.746428571428573e-05,
+      "loss": 1.1313,
       "step": 2524
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 5.019723865877712e-05,
-      "loss": 1.5358,
+      "epoch": 11.27,
+      "learning_rate": 7.745535714285714e-05,
+      "loss": 1.1325,
       "step": 2525
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 5.0177514792899414e-05,
-      "loss": 1.5738,
+      "epoch": 11.28,
+      "learning_rate": 7.744642857142858e-05,
+      "loss": 1.2352,
       "step": 2526
     },
     {
-      "epoch": 7.48,
-      "learning_rate": 5.01577909270217e-05,
-      "loss": 1.5934,
+      "epoch": 11.28,
+      "learning_rate": 7.74375e-05,
+      "loss": 1.1635,
       "step": 2527
     },
     {
-      "epoch": 7.48,
-      "learning_rate": 5.013806706114399e-05,
-      "loss": 1.5143,
+      "epoch": 11.29,
+      "learning_rate": 7.742857142857143e-05,
+      "loss": 1.1778,
       "step": 2528
     },
     {
-      "epoch": 7.48,
-      "learning_rate": 5.011834319526627e-05,
-      "loss": 1.5973,
+      "epoch": 11.29,
+      "learning_rate": 7.741964285714286e-05,
+      "loss": 1.1624,
       "step": 2529
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 5.009861932938856e-05,
-      "loss": 1.5488,
+      "epoch": 11.29,
+      "learning_rate": 7.74107142857143e-05,
+      "loss": 1.3044,
       "step": 2530
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 5.0078895463510854e-05,
-      "loss": 1.5898,
+      "epoch": 11.3,
+      "learning_rate": 7.740178571428571e-05,
+      "loss": 1.1823,
       "step": 2531
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 5.005917159763314e-05,
-      "loss": 1.5416,
+      "epoch": 11.3,
+      "learning_rate": 7.739285714285715e-05,
+      "loss": 1.2721,
       "step": 2532
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 5.003944773175543e-05,
-      "loss": 1.5512,
+      "epoch": 11.31,
+      "learning_rate": 7.738392857142857e-05,
+      "loss": 1.1813,
       "step": 2533
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 5.001972386587771e-05,
-      "loss": 1.6195,
+      "epoch": 11.31,
+      "learning_rate": 7.737500000000001e-05,
+      "loss": 1.214,
       "step": 2534
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 5e-05,
-      "loss": 1.6155,
+      "epoch": 11.32,
+      "learning_rate": 7.736607142857144e-05,
+      "loss": 1.1772,
       "step": 2535
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 4.998027613412229e-05,
-      "loss": 1.5545,
+      "epoch": 11.32,
+      "learning_rate": 7.735714285714286e-05,
+      "loss": 1.147,
       "step": 2536
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 4.996055226824457e-05,
-      "loss": 1.5305,
+      "epoch": 11.33,
+      "learning_rate": 7.734821428571429e-05,
+      "loss": 1.22,
       "step": 2537
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 4.9940828402366865e-05,
-      "loss": 1.6177,
+      "epoch": 11.33,
+      "learning_rate": 7.733928571428571e-05,
+      "loss": 1.1392,
       "step": 2538
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 4.992110453648916e-05,
-      "loss": 1.6389,
+      "epoch": 11.33,
+      "learning_rate": 7.733035714285715e-05,
+      "loss": 1.243,
       "step": 2539
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 4.990138067061144e-05,
-      "loss": 1.5547,
+      "epoch": 11.34,
+      "learning_rate": 7.732142857142858e-05,
+      "loss": 1.1409,
       "step": 2540
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 4.988165680473373e-05,
-      "loss": 1.5735,
+      "epoch": 11.34,
+      "learning_rate": 7.73125e-05,
+      "loss": 1.1088,
       "step": 2541
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 4.986193293885602e-05,
-      "loss": 1.6803,
+      "epoch": 11.35,
+      "learning_rate": 7.730357142857143e-05,
+      "loss": 1.1992,
       "step": 2542
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 4.9842209072978305e-05,
-      "loss": 1.5673,
+      "epoch": 11.35,
+      "learning_rate": 7.729464285714287e-05,
+      "loss": 1.214,
       "step": 2543
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 4.98224852071006e-05,
-      "loss": 1.5917,
+      "epoch": 11.36,
+      "learning_rate": 7.728571428571429e-05,
+      "loss": 1.1695,
       "step": 2544
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 4.980276134122288e-05,
-      "loss": 1.652,
+      "epoch": 11.36,
+      "learning_rate": 7.727678571428572e-05,
+      "loss": 1.1377,
       "step": 2545
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 4.978303747534517e-05,
-      "loss": 1.5253,
+      "epoch": 11.37,
+      "learning_rate": 7.726785714285714e-05,
+      "loss": 1.2193,
       "step": 2546
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 4.976331360946746e-05,
-      "loss": 1.5608,
+      "epoch": 11.37,
+      "learning_rate": 7.725892857142858e-05,
+      "loss": 1.1958,
       "step": 2547
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 4.9743589743589746e-05,
-      "loss": 1.5733,
+      "epoch": 11.38,
+      "learning_rate": 7.725e-05,
+      "loss": 1.1514,
       "step": 2548
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 4.972386587771203e-05,
-      "loss": 1.5754,
+      "epoch": 11.38,
+      "learning_rate": 7.724107142857143e-05,
+      "loss": 1.2014,
       "step": 2549
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 4.970414201183432e-05,
-      "loss": 1.6281,
+      "epoch": 11.38,
+      "learning_rate": 7.723214285714286e-05,
+      "loss": 1.1866,
       "step": 2550
     },
     {
-      "epoch": 7.55,
-      "learning_rate": 4.968441814595661e-05,
-      "loss": 1.4695,
+      "epoch": 11.39,
+      "learning_rate": 7.722321428571428e-05,
+      "loss": 1.2359,
       "step": 2551
     },
     {
-      "epoch": 7.55,
-      "learning_rate": 4.96646942800789e-05,
-      "loss": 1.5733,
+      "epoch": 11.39,
+      "learning_rate": 7.721428571428572e-05,
+      "loss": 1.1368,
       "step": 2552
     },
     {
-      "epoch": 7.55,
-      "learning_rate": 4.9644970414201186e-05,
-      "loss": 1.5198,
+      "epoch": 11.4,
+      "learning_rate": 7.720535714285715e-05,
+      "loss": 1.2324,
       "step": 2553
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 4.962524654832347e-05,
-      "loss": 1.6595,
+      "epoch": 11.4,
+      "learning_rate": 7.719642857142859e-05,
+      "loss": 1.1538,
       "step": 2554
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 4.9605522682445757e-05,
-      "loss": 1.6281,
+      "epoch": 11.41,
+      "learning_rate": 7.71875e-05,
+      "loss": 1.1613,
       "step": 2555
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 4.958579881656805e-05,
-      "loss": 1.513,
+      "epoch": 11.41,
+      "learning_rate": 7.717857142857144e-05,
+      "loss": 1.2375,
       "step": 2556
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 4.956607495069034e-05,
-      "loss": 1.5618,
+      "epoch": 11.42,
+      "learning_rate": 7.716964285714286e-05,
+      "loss": 1.2146,
       "step": 2557
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 4.9546351084812626e-05,
-      "loss": 1.5309,
+      "epoch": 11.42,
+      "learning_rate": 7.716071428571429e-05,
+      "loss": 1.1433,
       "step": 2558
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 4.952662721893491e-05,
-      "loss": 1.632,
+      "epoch": 11.42,
+      "learning_rate": 7.715178571428571e-05,
+      "loss": 1.1626,
       "step": 2559
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 4.95069033530572e-05,
-      "loss": 1.663,
+      "epoch": 11.43,
+      "learning_rate": 7.714285714285715e-05,
+      "loss": 1.3126,
       "step": 2560
     },
     {
-      "epoch": 7.58,
-      "learning_rate": 4.948717948717949e-05,
-      "loss": 1.583,
+      "epoch": 11.43,
+      "learning_rate": 7.713392857142857e-05,
+      "loss": 1.15,
       "step": 2561
     },
     {
-      "epoch": 7.58,
-      "learning_rate": 4.946745562130178e-05,
-      "loss": 1.5299,
+      "epoch": 11.44,
+      "learning_rate": 7.7125e-05,
+      "loss": 1.3004,
       "step": 2562
     },
     {
-      "epoch": 7.58,
-      "learning_rate": 4.9447731755424066e-05,
-      "loss": 1.6134,
+      "epoch": 11.44,
+      "learning_rate": 7.711607142857143e-05,
+      "loss": 1.1639,
       "step": 2563
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 4.942800788954635e-05,
-      "loss": 1.4874,
+      "epoch": 11.45,
+      "learning_rate": 7.710714285714286e-05,
+      "loss": 1.1544,
       "step": 2564
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 4.9408284023668644e-05,
-      "loss": 1.6443,
+      "epoch": 11.45,
+      "learning_rate": 7.709821428571428e-05,
+      "loss": 1.2051,
       "step": 2565
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 4.938856015779093e-05,
-      "loss": 1.5747,
+      "epoch": 11.46,
+      "learning_rate": 7.708928571428572e-05,
+      "loss": 1.1601,
       "step": 2566
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 4.9368836291913215e-05,
-      "loss": 1.5026,
+      "epoch": 11.46,
+      "learning_rate": 7.708035714285715e-05,
+      "loss": 1.2034,
       "step": 2567
     },
     {
-      "epoch": 7.6,
-      "learning_rate": 4.934911242603551e-05,
-      "loss": 1.5959,
+      "epoch": 11.46,
+      "learning_rate": 7.707142857142857e-05,
+      "loss": 1.2682,
       "step": 2568
     },
     {
-      "epoch": 7.6,
-      "learning_rate": 4.932938856015779e-05,
-      "loss": 1.6472,
+      "epoch": 11.47,
+      "learning_rate": 7.706250000000001e-05,
+      "loss": 1.1938,
       "step": 2569
     },
     {
-      "epoch": 7.6,
-      "learning_rate": 4.9309664694280084e-05,
-      "loss": 1.5928,
+      "epoch": 11.47,
+      "learning_rate": 7.705357142857144e-05,
+      "loss": 1.1503,
       "step": 2570
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 4.928994082840237e-05,
-      "loss": 1.6743,
+      "epoch": 11.48,
+      "learning_rate": 7.704464285714286e-05,
+      "loss": 1.2643,
       "step": 2571
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 4.9270216962524655e-05,
-      "loss": 1.5463,
+      "epoch": 11.48,
+      "learning_rate": 7.703571428571429e-05,
+      "loss": 1.263,
       "step": 2572
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 4.925049309664695e-05,
-      "loss": 1.5808,
+      "epoch": 11.49,
+      "learning_rate": 7.702678571428573e-05,
+      "loss": 1.2105,
       "step": 2573
     },
     {
-      "epoch": 7.62,
-      "learning_rate": 4.923076923076924e-05,
-      "loss": 1.5411,
+      "epoch": 11.49,
+      "learning_rate": 7.701785714285714e-05,
+      "loss": 1.2206,
       "step": 2574
     },
     {
-      "epoch": 7.62,
-      "learning_rate": 4.9211045364891524e-05,
-      "loss": 1.5731,
+      "epoch": 11.5,
+      "learning_rate": 7.700892857142858e-05,
+      "loss": 1.2441,
       "step": 2575
     },
     {
-      "epoch": 7.62,
-      "learning_rate": 4.919132149901381e-05,
-      "loss": 1.5319,
+      "epoch": 11.5,
+      "learning_rate": 7.7e-05,
+      "loss": 1.1626,
       "step": 2576
     },
     {
-      "epoch": 7.62,
-      "learning_rate": 4.9171597633136095e-05,
-      "loss": 1.6105,
+      "epoch": 11.5,
+      "learning_rate": 7.699107142857144e-05,
+      "loss": 1.2172,
       "step": 2577
     },
     {
-      "epoch": 7.63,
-      "learning_rate": 4.915187376725838e-05,
-      "loss": 1.5488,
+      "epoch": 11.51,
+      "learning_rate": 7.698214285714285e-05,
+      "loss": 1.2377,
       "step": 2578
     },
     {
-      "epoch": 7.63,
-      "learning_rate": 4.913214990138067e-05,
-      "loss": 1.5737,
+      "epoch": 11.51,
+      "learning_rate": 7.697321428571429e-05,
+      "loss": 1.185,
       "step": 2579
     },
     {
-      "epoch": 7.63,
-      "learning_rate": 4.9112426035502965e-05,
-      "loss": 1.5535,
+      "epoch": 11.52,
+      "learning_rate": 7.696428571428572e-05,
+      "loss": 1.1816,
       "step": 2580
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 4.909270216962525e-05,
-      "loss": 1.6118,
+      "epoch": 11.52,
+      "learning_rate": 7.695535714285714e-05,
+      "loss": 1.2039,
       "step": 2581
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 4.9072978303747535e-05,
-      "loss": 1.6119,
+      "epoch": 11.53,
+      "learning_rate": 7.694642857142858e-05,
+      "loss": 1.18,
       "step": 2582
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 4.905325443786982e-05,
-      "loss": 1.5245,
+      "epoch": 11.53,
+      "learning_rate": 7.69375e-05,
+      "loss": 1.1911,
       "step": 2583
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 4.903353057199211e-05,
-      "loss": 1.5387,
+      "epoch": 11.54,
+      "learning_rate": 7.692857142857143e-05,
+      "loss": 1.0958,
       "step": 2584
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 4.90138067061144e-05,
-      "loss": 1.5857,
+      "epoch": 11.54,
+      "learning_rate": 7.691964285714286e-05,
+      "loss": 1.1537,
       "step": 2585
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 4.899408284023669e-05,
-      "loss": 1.5631,
+      "epoch": 11.54,
+      "learning_rate": 7.69107142857143e-05,
+      "loss": 1.0623,
       "step": 2586
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 4.8974358974358975e-05,
-      "loss": 1.6353,
+      "epoch": 11.55,
+      "learning_rate": 7.690178571428572e-05,
+      "loss": 1.1907,
       "step": 2587
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 4.895463510848127e-05,
-      "loss": 1.5697,
+      "epoch": 11.55,
+      "learning_rate": 7.689285714285715e-05,
+      "loss": 1.1452,
       "step": 2588
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 4.893491124260355e-05,
-      "loss": 1.6323,
+      "epoch": 11.56,
+      "learning_rate": 7.688392857142857e-05,
+      "loss": 1.1584,
       "step": 2589
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 4.891518737672584e-05,
-      "loss": 1.6082,
+      "epoch": 11.56,
+      "learning_rate": 7.687500000000001e-05,
+      "loss": 1.0811,
       "step": 2590
     },
     {
-      "epoch": 7.67,
-      "learning_rate": 4.889546351084813e-05,
-      "loss": 1.6102,
+      "epoch": 11.57,
+      "learning_rate": 7.686607142857142e-05,
+      "loss": 1.0918,
       "step": 2591
     },
     {
-      "epoch": 7.67,
-      "learning_rate": 4.8875739644970416e-05,
-      "loss": 1.543,
+      "epoch": 11.57,
+      "learning_rate": 7.685714285714286e-05,
+      "loss": 1.2082,
       "step": 2592
     },
     {
-      "epoch": 7.67,
-      "learning_rate": 4.885601577909271e-05,
-      "loss": 1.5658,
+      "epoch": 11.58,
+      "learning_rate": 7.684821428571429e-05,
+      "loss": 1.1649,
       "step": 2593
     },
     {
-      "epoch": 7.67,
-      "learning_rate": 4.883629191321499e-05,
-      "loss": 1.611,
+      "epoch": 11.58,
+      "learning_rate": 7.683928571428571e-05,
+      "loss": 1.2073,
       "step": 2594
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 4.881656804733728e-05,
-      "loss": 1.5021,
+      "epoch": 11.58,
+      "learning_rate": 7.683035714285714e-05,
+      "loss": 1.1914,
       "step": 2595
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 4.8796844181459564e-05,
-      "loss": 1.5667,
+      "epoch": 11.59,
+      "learning_rate": 7.682142857142858e-05,
+      "loss": 1.2081,
       "step": 2596
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 4.8777120315581856e-05,
-      "loss": 1.5666,
+      "epoch": 11.59,
+      "learning_rate": 7.68125e-05,
+      "loss": 1.248,
       "step": 2597
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 4.875739644970415e-05,
-      "loss": 1.6753,
+      "epoch": 11.6,
+      "learning_rate": 7.680357142857143e-05,
+      "loss": 1.1394,
       "step": 2598
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 4.873767258382643e-05,
-      "loss": 1.592,
+      "epoch": 11.6,
+      "learning_rate": 7.679464285714287e-05,
+      "loss": 1.1912,
       "step": 2599
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 4.871794871794872e-05,
-      "loss": 1.5678,
+      "epoch": 11.61,
+      "learning_rate": 7.67857142857143e-05,
+      "loss": 1.1257,
       "step": 2600
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 4.8698224852071004e-05,
-      "loss": 1.5199,
+      "epoch": 11.61,
+      "learning_rate": 7.677678571428572e-05,
+      "loss": 1.227,
       "step": 2601
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 4.8678500986193296e-05,
-      "loss": 1.594,
+      "epoch": 11.62,
+      "learning_rate": 7.676785714285714e-05,
+      "loss": 1.2377,
       "step": 2602
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 4.865877712031559e-05,
-      "loss": 1.6114,
+      "epoch": 11.62,
+      "learning_rate": 7.675892857142858e-05,
+      "loss": 1.1632,
       "step": 2603
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 4.8639053254437874e-05,
-      "loss": 1.6616,
+      "epoch": 11.62,
+      "learning_rate": 7.675e-05,
+      "loss": 1.1486,
       "step": 2604
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 4.861932938856016e-05,
-      "loss": 1.6126,
+      "epoch": 11.63,
+      "learning_rate": 7.674107142857143e-05,
+      "loss": 1.2,
       "step": 2605
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 4.8599605522682444e-05,
-      "loss": 1.5815,
+      "epoch": 11.63,
+      "learning_rate": 7.673214285714286e-05,
+      "loss": 1.1933,
       "step": 2606
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 4.8579881656804736e-05,
-      "loss": 1.5875,
+      "epoch": 11.64,
+      "learning_rate": 7.672321428571429e-05,
+      "loss": 1.2397,
       "step": 2607
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 4.856015779092702e-05,
-      "loss": 1.607,
+      "epoch": 11.64,
+      "learning_rate": 7.671428571428571e-05,
+      "loss": 1.21,
       "step": 2608
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 4.8540433925049314e-05,
-      "loss": 1.4878,
+      "epoch": 11.65,
+      "learning_rate": 7.670535714285715e-05,
+      "loss": 1.1798,
       "step": 2609
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 4.85207100591716e-05,
-      "loss": 1.5974,
+      "epoch": 11.65,
+      "learning_rate": 7.669642857142858e-05,
+      "loss": 1.1011,
       "step": 2610
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 4.850098619329389e-05,
-      "loss": 1.58,
+      "epoch": 11.66,
+      "learning_rate": 7.66875e-05,
+      "loss": 1.156,
       "step": 2611
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 4.848126232741618e-05,
-      "loss": 1.5516,
+      "epoch": 11.66,
+      "learning_rate": 7.667857142857144e-05,
+      "loss": 1.2529,
       "step": 2612
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 4.846153846153846e-05,
-      "loss": 1.6167,
+      "epoch": 11.67,
+      "learning_rate": 7.666964285714287e-05,
+      "loss": 1.198,
       "step": 2613
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 4.844181459566075e-05,
-      "loss": 1.5412,
+      "epoch": 11.67,
+      "learning_rate": 7.666071428571429e-05,
+      "loss": 1.1345,
       "step": 2614
     },
     {
-      "epoch": 7.74,
-      "learning_rate": 4.842209072978304e-05,
-      "loss": 1.5168,
+      "epoch": 11.67,
+      "learning_rate": 7.665178571428572e-05,
+      "loss": 1.1161,
       "step": 2615
     },
     {
-      "epoch": 7.74,
-      "learning_rate": 4.840236686390533e-05,
-      "loss": 1.5186,
+      "epoch": 11.68,
+      "learning_rate": 7.664285714285715e-05,
+      "loss": 1.1265,
       "step": 2616
     },
     {
-      "epoch": 7.74,
-      "learning_rate": 4.838264299802762e-05,
-      "loss": 1.6266,
+      "epoch": 11.68,
+      "learning_rate": 7.663392857142857e-05,
+      "loss": 1.1684,
       "step": 2617
     },
     {
-      "epoch": 7.75,
-      "learning_rate": 4.83629191321499e-05,
-      "loss": 1.5566,
+      "epoch": 11.69,
+      "learning_rate": 7.6625e-05,
+      "loss": 1.1848,
       "step": 2618
     },
     {
-      "epoch": 7.75,
-      "learning_rate": 4.834319526627219e-05,
-      "loss": 1.5933,
+      "epoch": 11.69,
+      "learning_rate": 7.661607142857143e-05,
+      "loss": 1.2668,
       "step": 2619
     },
     {
-      "epoch": 7.75,
-      "learning_rate": 4.832347140039448e-05,
-      "loss": 1.6123,
+      "epoch": 11.7,
+      "learning_rate": 7.660714285714287e-05,
+      "loss": 1.22,
       "step": 2620
     },
     {
-      "epoch": 7.75,
-      "learning_rate": 4.830374753451677e-05,
-      "loss": 1.6646,
+      "epoch": 11.7,
+      "learning_rate": 7.659821428571428e-05,
+      "loss": 1.2781,
       "step": 2621
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 4.828402366863906e-05,
-      "loss": 1.6354,
+      "epoch": 11.71,
+      "learning_rate": 7.658928571428572e-05,
+      "loss": 1.2368,
       "step": 2622
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 4.826429980276134e-05,
-      "loss": 1.6297,
+      "epoch": 11.71,
+      "learning_rate": 7.658035714285715e-05,
+      "loss": 1.1844,
       "step": 2623
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 4.824457593688363e-05,
-      "loss": 1.541,
+      "epoch": 11.71,
+      "learning_rate": 7.657142857142857e-05,
+      "loss": 1.2821,
       "step": 2624
     },
     {
-      "epoch": 7.77,
-      "learning_rate": 4.822485207100592e-05,
-      "loss": 1.6938,
+      "epoch": 11.72,
+      "learning_rate": 7.65625e-05,
+      "loss": 1.1341,
       "step": 2625
     },
     {
-      "epoch": 7.77,
-      "learning_rate": 4.8205128205128205e-05,
-      "loss": 1.6025,
+      "epoch": 11.72,
+      "learning_rate": 7.655357142857144e-05,
+      "loss": 1.1809,
       "step": 2626
     },
     {
-      "epoch": 7.77,
-      "learning_rate": 4.81854043392505e-05,
-      "loss": 1.5547,
+      "epoch": 11.73,
+      "learning_rate": 7.654464285714286e-05,
+      "loss": 1.181,
       "step": 2627
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 4.816568047337278e-05,
-      "loss": 1.6419,
+      "epoch": 11.73,
+      "learning_rate": 7.653571428571429e-05,
+      "loss": 1.1351,
       "step": 2628
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 4.8145956607495075e-05,
-      "loss": 1.6041,
+      "epoch": 11.74,
+      "learning_rate": 7.652678571428573e-05,
+      "loss": 1.1705,
       "step": 2629
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 4.812623274161736e-05,
-      "loss": 1.5637,
+      "epoch": 11.74,
+      "learning_rate": 7.651785714285714e-05,
+      "loss": 1.1792,
       "step": 2630
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 4.8106508875739645e-05,
-      "loss": 1.5592,
+      "epoch": 11.75,
+      "learning_rate": 7.650892857142858e-05,
+      "loss": 1.1793,
       "step": 2631
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 4.808678500986194e-05,
-      "loss": 1.668,
+      "epoch": 11.75,
+      "learning_rate": 7.65e-05,
+      "loss": 1.1323,
       "step": 2632
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 4.806706114398422e-05,
-      "loss": 1.6336,
+      "epoch": 11.75,
+      "learning_rate": 7.649107142857144e-05,
+      "loss": 1.1362,
       "step": 2633
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 4.8047337278106515e-05,
-      "loss": 1.5639,
+      "epoch": 11.76,
+      "learning_rate": 7.648214285714285e-05,
+      "loss": 1.1624,
       "step": 2634
     },
     {
-      "epoch": 7.8,
-      "learning_rate": 4.80276134122288e-05,
-      "loss": 1.6436,
+      "epoch": 11.76,
+      "learning_rate": 7.647321428571429e-05,
+      "loss": 1.2285,
       "step": 2635
     },
     {
-      "epoch": 7.8,
-      "learning_rate": 4.8007889546351086e-05,
-      "loss": 1.6108,
+      "epoch": 11.77,
+      "learning_rate": 7.646428571428572e-05,
+      "loss": 1.2504,
       "step": 2636
     },
     {
-      "epoch": 7.8,
-      "learning_rate": 4.798816568047337e-05,
-      "loss": 1.583,
+      "epoch": 11.77,
+      "learning_rate": 7.645535714285714e-05,
+      "loss": 1.1914,
       "step": 2637
     },
     {
-      "epoch": 7.8,
-      "learning_rate": 4.796844181459566e-05,
-      "loss": 1.5617,
+      "epoch": 11.78,
+      "learning_rate": 7.644642857142857e-05,
+      "loss": 1.1603,
       "step": 2638
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 4.7948717948717955e-05,
-      "loss": 1.6691,
+      "epoch": 11.78,
+      "learning_rate": 7.643750000000001e-05,
+      "loss": 1.2082,
       "step": 2639
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 4.792899408284024e-05,
-      "loss": 1.6223,
+      "epoch": 11.79,
+      "learning_rate": 7.642857142857143e-05,
+      "loss": 1.1623,
       "step": 2640
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 4.7909270216962526e-05,
-      "loss": 1.6264,
+      "epoch": 11.79,
+      "learning_rate": 7.641964285714286e-05,
+      "loss": 1.1426,
       "step": 2641
     },
     {
-      "epoch": 7.82,
-      "learning_rate": 4.788954635108481e-05,
-      "loss": 1.5853,
+      "epoch": 11.79,
+      "learning_rate": 7.641071428571428e-05,
+      "loss": 1.1994,
       "step": 2642
     },
     {
-      "epoch": 7.82,
-      "learning_rate": 4.7869822485207103e-05,
-      "loss": 1.5986,
+      "epoch": 11.8,
+      "learning_rate": 7.640178571428572e-05,
+      "loss": 1.1843,
       "step": 2643
     },
     {
-      "epoch": 7.82,
-      "learning_rate": 4.785009861932939e-05,
-      "loss": 1.5375,
+      "epoch": 11.8,
+      "learning_rate": 7.639285714285715e-05,
+      "loss": 1.2203,
       "step": 2644
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 4.783037475345168e-05,
-      "loss": 1.5646,
+      "epoch": 11.81,
+      "learning_rate": 7.638392857142857e-05,
+      "loss": 1.2661,
       "step": 2645
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 4.7810650887573966e-05,
-      "loss": 1.5274,
+      "epoch": 11.81,
+      "learning_rate": 7.637500000000001e-05,
+      "loss": 1.116,
       "step": 2646
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 4.779092702169625e-05,
-      "loss": 1.6055,
+      "epoch": 11.82,
+      "learning_rate": 7.636607142857143e-05,
+      "loss": 1.1567,
       "step": 2647
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 4.7771203155818544e-05,
-      "loss": 1.5748,
+      "epoch": 11.82,
+      "learning_rate": 7.635714285714286e-05,
+      "loss": 1.2394,
       "step": 2648
     },
     {
-      "epoch": 7.84,
-      "learning_rate": 4.775147928994083e-05,
-      "loss": 1.5891,
+      "epoch": 11.83,
+      "learning_rate": 7.634821428571429e-05,
+      "loss": 1.2811,
       "step": 2649
     },
     {
-      "epoch": 7.84,
-      "learning_rate": 4.773175542406312e-05,
-      "loss": 1.5155,
+      "epoch": 11.83,
+      "learning_rate": 7.633928571428572e-05,
+      "loss": 1.0951,
       "step": 2650
     },
     {
-      "epoch": 7.84,
-      "learning_rate": 4.7712031558185406e-05,
-      "loss": 1.6189,
+      "epoch": 11.83,
+      "learning_rate": 7.633035714285714e-05,
+      "loss": 1.2932,
       "step": 2651
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 4.76923076923077e-05,
-      "loss": 1.5857,
+      "epoch": 11.84,
+      "learning_rate": 7.632142857142858e-05,
+      "loss": 1.1885,
       "step": 2652
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 4.7672583826429984e-05,
-      "loss": 1.6701,
+      "epoch": 11.84,
+      "learning_rate": 7.63125e-05,
+      "loss": 1.1485,
       "step": 2653
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 4.765285996055227e-05,
-      "loss": 1.5406,
+      "epoch": 11.85,
+      "learning_rate": 7.630357142857143e-05,
+      "loss": 1.2244,
       "step": 2654
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 4.7633136094674555e-05,
-      "loss": 1.5513,
+      "epoch": 11.85,
+      "learning_rate": 7.629464285714286e-05,
+      "loss": 1.1982,
       "step": 2655
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 4.761341222879685e-05,
-      "loss": 1.6146,
+      "epoch": 11.86,
+      "learning_rate": 7.62857142857143e-05,
+      "loss": 1.1767,
       "step": 2656
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 4.759368836291914e-05,
-      "loss": 1.5121,
+      "epoch": 11.86,
+      "learning_rate": 7.627678571428572e-05,
+      "loss": 1.1496,
       "step": 2657
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 4.7573964497041424e-05,
-      "loss": 1.5404,
+      "epoch": 11.87,
+      "learning_rate": 7.626785714285715e-05,
+      "loss": 1.141,
       "step": 2658
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 4.755424063116371e-05,
-      "loss": 1.5345,
+      "epoch": 11.87,
+      "learning_rate": 7.625892857142858e-05,
+      "loss": 1.181,
       "step": 2659
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 4.7534516765285995e-05,
-      "loss": 1.5665,
+      "epoch": 11.88,
+      "learning_rate": 7.625e-05,
+      "loss": 1.1552,
       "step": 2660
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 4.751479289940829e-05,
-      "loss": 1.6441,
+      "epoch": 11.88,
+      "learning_rate": 7.624107142857144e-05,
+      "loss": 1.1892,
       "step": 2661
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 4.749506903353058e-05,
-      "loss": 1.6337,
+      "epoch": 11.88,
+      "learning_rate": 7.623214285714286e-05,
+      "loss": 1.1825,
       "step": 2662
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 4.7475345167652864e-05,
-      "loss": 1.59,
+      "epoch": 11.89,
+      "learning_rate": 7.622321428571429e-05,
+      "loss": 1.2078,
       "step": 2663
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 4.745562130177515e-05,
-      "loss": 1.636,
+      "epoch": 11.89,
+      "learning_rate": 7.621428571428571e-05,
+      "loss": 1.2388,
       "step": 2664
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 4.7435897435897435e-05,
-      "loss": 1.6231,
+      "epoch": 11.9,
+      "learning_rate": 7.620535714285715e-05,
+      "loss": 1.1962,
       "step": 2665
     },
     {
-      "epoch": 7.89,
-      "learning_rate": 4.741617357001973e-05,
-      "loss": 1.5769,
+      "epoch": 11.9,
+      "learning_rate": 7.619642857142858e-05,
+      "loss": 1.1553,
       "step": 2666
     },
     {
-      "epoch": 7.89,
-      "learning_rate": 4.739644970414201e-05,
-      "loss": 1.5699,
+      "epoch": 11.91,
+      "learning_rate": 7.61875e-05,
+      "loss": 1.1905,
       "step": 2667
     },
     {
-      "epoch": 7.89,
-      "learning_rate": 4.7376725838264305e-05,
-      "loss": 1.5383,
+      "epoch": 11.91,
+      "learning_rate": 7.617857142857143e-05,
+      "loss": 1.1857,
       "step": 2668
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 4.735700197238659e-05,
-      "loss": 1.5106,
+      "epoch": 11.92,
+      "learning_rate": 7.616964285714287e-05,
+      "loss": 1.1432,
       "step": 2669
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 4.7337278106508875e-05,
-      "loss": 1.5785,
+      "epoch": 11.92,
+      "learning_rate": 7.616071428571429e-05,
+      "loss": 1.1425,
       "step": 2670
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 4.731755424063117e-05,
-      "loss": 1.578,
+      "epoch": 11.92,
+      "learning_rate": 7.615178571428572e-05,
+      "loss": 1.0838,
       "step": 2671
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 4.729783037475345e-05,
-      "loss": 1.6099,
+      "epoch": 11.93,
+      "learning_rate": 7.614285714285714e-05,
+      "loss": 1.1657,
       "step": 2672
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 4.727810650887574e-05,
-      "loss": 1.5263,
+      "epoch": 11.93,
+      "learning_rate": 7.613392857142857e-05,
+      "loss": 1.1231,
       "step": 2673
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 4.725838264299803e-05,
-      "loss": 1.6194,
+      "epoch": 11.94,
+      "learning_rate": 7.612500000000001e-05,
+      "loss": 1.1742,
       "step": 2674
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 4.723865877712032e-05,
-      "loss": 1.6472,
+      "epoch": 11.94,
+      "learning_rate": 7.611607142857143e-05,
+      "loss": 1.2652,
       "step": 2675
     },
     {
-      "epoch": 7.92,
-      "learning_rate": 4.721893491124261e-05,
-      "loss": 1.5677,
+      "epoch": 11.95,
+      "learning_rate": 7.610714285714287e-05,
+      "loss": 1.1983,
       "step": 2676
     },
     {
-      "epoch": 7.92,
-      "learning_rate": 4.719921104536489e-05,
-      "loss": 1.5994,
+      "epoch": 11.95,
+      "learning_rate": 7.609821428571428e-05,
+      "loss": 1.1213,
       "step": 2677
     },
     {
-      "epoch": 7.92,
-      "learning_rate": 4.717948717948718e-05,
-      "loss": 1.5242,
+      "epoch": 11.96,
+      "learning_rate": 7.608928571428572e-05,
+      "loss": 1.1691,
       "step": 2678
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 4.715976331360947e-05,
-      "loss": 1.5719,
+      "epoch": 11.96,
+      "learning_rate": 7.608035714285715e-05,
+      "loss": 1.1964,
       "step": 2679
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 4.714003944773176e-05,
-      "loss": 1.5558,
+      "epoch": 11.96,
+      "learning_rate": 7.607142857142857e-05,
+      "loss": 1.1828,
       "step": 2680
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 4.712031558185405e-05,
-      "loss": 1.4538,
+      "epoch": 11.97,
+      "learning_rate": 7.60625e-05,
+      "loss": 1.2257,
       "step": 2681
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 4.710059171597633e-05,
-      "loss": 1.5263,
+      "epoch": 11.97,
+      "learning_rate": 7.605357142857144e-05,
+      "loss": 1.2271,
       "step": 2682
     },
     {
-      "epoch": 7.94,
-      "learning_rate": 4.708086785009862e-05,
-      "loss": 1.5961,
+      "epoch": 11.98,
+      "learning_rate": 7.604464285714285e-05,
+      "loss": 1.1877,
       "step": 2683
     },
     {
-      "epoch": 7.94,
-      "learning_rate": 4.706114398422091e-05,
-      "loss": 1.5931,
+      "epoch": 11.98,
+      "learning_rate": 7.603571428571429e-05,
+      "loss": 1.1446,
       "step": 2684
     },
     {
-      "epoch": 7.94,
-      "learning_rate": 4.7041420118343196e-05,
-      "loss": 1.562,
+      "epoch": 11.99,
+      "learning_rate": 7.602678571428571e-05,
+      "loss": 1.1503,
       "step": 2685
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 4.702169625246549e-05,
-      "loss": 1.6856,
+      "epoch": 11.99,
+      "learning_rate": 7.601785714285715e-05,
+      "loss": 1.1753,
       "step": 2686
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 4.7001972386587773e-05,
-      "loss": 1.589,
+      "epoch": 12.0,
+      "learning_rate": 7.600892857142858e-05,
+      "loss": 1.1766,
       "step": 2687
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 4.698224852071006e-05,
-      "loss": 1.5622,
+      "epoch": 12.0,
+      "learning_rate": 7.6e-05,
+      "loss": 1.2023,
       "step": 2688
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 4.696252465483235e-05,
-      "loss": 1.6076,
+      "epoch": 12.0,
+      "learning_rate": 7.599107142857144e-05,
+      "loss": 1.1493,
       "step": 2689
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 4.6942800788954636e-05,
-      "loss": 1.5278,
+      "epoch": 12.01,
+      "learning_rate": 7.598214285714286e-05,
+      "loss": 1.2321,
       "step": 2690
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 4.692307692307693e-05,
-      "loss": 1.6589,
+      "epoch": 12.01,
+      "learning_rate": 7.59732142857143e-05,
+      "loss": 1.1768,
       "step": 2691
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 4.6903353057199214e-05,
-      "loss": 1.6152,
+      "epoch": 12.02,
+      "learning_rate": 7.596428571428572e-05,
+      "loss": 1.1214,
       "step": 2692
     },
     {
-      "epoch": 7.97,
-      "learning_rate": 4.68836291913215e-05,
-      "loss": 1.5556,
+      "epoch": 12.02,
+      "learning_rate": 7.595535714285714e-05,
+      "loss": 1.2043,
       "step": 2693
     },
     {
-      "epoch": 7.97,
-      "learning_rate": 4.686390532544379e-05,
-      "loss": 1.4502,
+      "epoch": 12.03,
+      "learning_rate": 7.594642857142857e-05,
+      "loss": 1.1901,
       "step": 2694
     },
     {
-      "epoch": 7.97,
-      "learning_rate": 4.6844181459566076e-05,
-      "loss": 1.6657,
+      "epoch": 12.03,
+      "learning_rate": 7.593750000000001e-05,
+      "loss": 1.2054,
       "step": 2695
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 4.682445759368836e-05,
-      "loss": 1.5439,
+      "epoch": 12.04,
+      "learning_rate": 7.592857142857142e-05,
+      "loss": 1.1981,
       "step": 2696
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 4.6804733727810654e-05,
-      "loss": 1.5983,
+      "epoch": 12.04,
+      "learning_rate": 7.591964285714286e-05,
+      "loss": 1.221,
       "step": 2697
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 4.6785009861932946e-05,
-      "loss": 1.64,
+      "epoch": 12.04,
+      "learning_rate": 7.591071428571429e-05,
+      "loss": 1.1513,
       "step": 2698
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 4.676528599605523e-05,
-      "loss": 1.6293,
+      "epoch": 12.05,
+      "learning_rate": 7.590178571428572e-05,
+      "loss": 1.1321,
       "step": 2699
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 4.674556213017752e-05,
-      "loss": 1.6083,
+      "epoch": 12.05,
+      "learning_rate": 7.589285714285714e-05,
+      "loss": 1.2104,
       "step": 2700
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 4.67258382642998e-05,
-      "loss": 1.5188,
+      "epoch": 12.05,
+      "eval_loss": 1.1662105321884155,
+      "eval_runtime": 44.8625,
+      "eval_samples_per_second": 291.87,
+      "eval_steps_per_second": 1.538,
+      "step": 2700
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.588392857142858e-05,
+      "loss": 1.1728,
       "step": 2701
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 4.670611439842209e-05,
-      "loss": 1.645,
+      "epoch": 12.06,
+      "learning_rate": 7.5875e-05,
+      "loss": 1.2338,
       "step": 2702
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 4.668639053254438e-05,
-      "loss": 1.5841,
+      "epoch": 12.07,
+      "learning_rate": 7.586607142857143e-05,
+      "loss": 1.2551,
       "step": 2703
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 4.666666666666667e-05,
-      "loss": 1.66,
+      "epoch": 12.07,
+      "learning_rate": 7.585714285714287e-05,
+      "loss": 1.0771,
       "step": 2704
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 4.664694280078896e-05,
-      "loss": 1.6348,
+      "epoch": 12.08,
+      "learning_rate": 7.584821428571429e-05,
+      "loss": 1.1361,
       "step": 2705
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 4.662721893491124e-05,
-      "loss": 1.6642,
+      "epoch": 12.08,
+      "learning_rate": 7.583928571428572e-05,
+      "loss": 1.2405,
       "step": 2706
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 4.6607495069033534e-05,
-      "loss": 1.6226,
+      "epoch": 12.08,
+      "learning_rate": 7.583035714285714e-05,
+      "loss": 1.087,
       "step": 2707
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 4.658777120315582e-05,
-      "loss": 1.5641,
+      "epoch": 12.09,
+      "learning_rate": 7.582142857142858e-05,
+      "loss": 1.2356,
       "step": 2708
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 4.656804733727811e-05,
-      "loss": 1.5417,
+      "epoch": 12.09,
+      "learning_rate": 7.58125e-05,
+      "loss": 1.1227,
       "step": 2709
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 4.65483234714004e-05,
-      "loss": 1.5438,
+      "epoch": 12.1,
+      "learning_rate": 7.580357142857143e-05,
+      "loss": 1.1364,
       "step": 2710
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 4.652859960552268e-05,
-      "loss": 1.5069,
+      "epoch": 12.1,
+      "learning_rate": 7.579464285714286e-05,
+      "loss": 1.121,
       "step": 2711
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 4.6508875739644975e-05,
-      "loss": 1.5806,
+      "epoch": 12.11,
+      "learning_rate": 7.57857142857143e-05,
+      "loss": 1.2015,
       "step": 2712
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 4.648915187376726e-05,
-      "loss": 1.5709,
+      "epoch": 12.11,
+      "learning_rate": 7.577678571428571e-05,
+      "loss": 1.2115,
       "step": 2713
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 4.6469428007889545e-05,
-      "loss": 1.489,
+      "epoch": 12.12,
+      "learning_rate": 7.576785714285715e-05,
+      "loss": 1.1405,
       "step": 2714
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 4.644970414201184e-05,
-      "loss": 1.487,
+      "epoch": 12.12,
+      "learning_rate": 7.575892857142857e-05,
+      "loss": 1.1654,
       "step": 2715
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 4.642998027613412e-05,
-      "loss": 1.5618,
+      "epoch": 12.12,
+      "learning_rate": 7.575e-05,
+      "loss": 1.1844,
       "step": 2716
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 4.6410256410256415e-05,
-      "loss": 1.6095,
+      "epoch": 12.13,
+      "learning_rate": 7.574107142857144e-05,
+      "loss": 1.2008,
       "step": 2717
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 4.63905325443787e-05,
-      "loss": 1.4637,
+      "epoch": 12.13,
+      "learning_rate": 7.573214285714286e-05,
+      "loss": 1.1604,
       "step": 2718
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 4.6370808678500986e-05,
-      "loss": 1.6213,
+      "epoch": 12.14,
+      "learning_rate": 7.572321428571429e-05,
+      "loss": 1.1964,
       "step": 2719
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 4.635108481262328e-05,
-      "loss": 1.612,
+      "epoch": 12.14,
+      "learning_rate": 7.571428571428571e-05,
+      "loss": 1.1472,
       "step": 2720
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 4.633136094674557e-05,
-      "loss": 1.6858,
+      "epoch": 12.15,
+      "learning_rate": 7.570535714285715e-05,
+      "loss": 1.1717,
       "step": 2721
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 4.6311637080867855e-05,
-      "loss": 1.5221,
+      "epoch": 12.15,
+      "learning_rate": 7.569642857142858e-05,
+      "loss": 1.1698,
       "step": 2722
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 4.629191321499014e-05,
-      "loss": 1.602,
+      "epoch": 12.16,
+      "learning_rate": 7.56875e-05,
+      "loss": 1.0968,
       "step": 2723
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 4.6272189349112426e-05,
-      "loss": 1.5824,
+      "epoch": 12.16,
+      "learning_rate": 7.567857142857143e-05,
+      "loss": 1.1789,
       "step": 2724
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 4.625246548323471e-05,
-      "loss": 1.5974,
+      "epoch": 12.17,
+      "learning_rate": 7.566964285714287e-05,
+      "loss": 1.2152,
       "step": 2725
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 4.6232741617357e-05,
-      "loss": 1.4844,
+      "epoch": 12.17,
+      "learning_rate": 7.566071428571428e-05,
+      "loss": 1.1628,
       "step": 2726
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 4.6213017751479295e-05,
-      "loss": 1.5841,
+      "epoch": 12.17,
+      "learning_rate": 7.565178571428572e-05,
+      "loss": 1.142,
       "step": 2727
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 4.619329388560158e-05,
-      "loss": 1.6699,
+      "epoch": 12.18,
+      "learning_rate": 7.564285714285714e-05,
+      "loss": 1.2631,
       "step": 2728
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 4.6173570019723866e-05,
-      "loss": 1.5632,
+      "epoch": 12.18,
+      "learning_rate": 7.563392857142858e-05,
+      "loss": 1.2028,
       "step": 2729
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 4.615384615384616e-05,
-      "loss": 1.5179,
+      "epoch": 12.19,
+      "learning_rate": 7.5625e-05,
+      "loss": 1.1204,
       "step": 2730
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 4.6134122287968443e-05,
-      "loss": 1.5051,
+      "epoch": 12.19,
+      "learning_rate": 7.561607142857143e-05,
+      "loss": 1.1256,
       "step": 2731
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 4.611439842209073e-05,
-      "loss": 1.5874,
+      "epoch": 12.2,
+      "learning_rate": 7.560714285714286e-05,
+      "loss": 1.1695,
       "step": 2732
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 4.609467455621302e-05,
-      "loss": 1.5707,
+      "epoch": 12.2,
+      "learning_rate": 7.559821428571428e-05,
+      "loss": 1.0878,
       "step": 2733
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 4.6074950690335306e-05,
-      "loss": 1.502,
+      "epoch": 12.21,
+      "learning_rate": 7.558928571428572e-05,
+      "loss": 1.1623,
       "step": 2734
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 4.60552268244576e-05,
-      "loss": 1.5094,
+      "epoch": 12.21,
+      "learning_rate": 7.558035714285715e-05,
+      "loss": 1.1534,
       "step": 2735
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 4.6035502958579884e-05,
-      "loss": 1.5704,
+      "epoch": 12.21,
+      "learning_rate": 7.557142857142857e-05,
+      "loss": 1.1126,
       "step": 2736
     },
     {
-      "epoch": 8.1,
-      "learning_rate": 4.601577909270217e-05,
-      "loss": 1.6142,
+      "epoch": 12.22,
+      "learning_rate": 7.55625e-05,
+      "loss": 1.3015,
       "step": 2737
     },
     {
-      "epoch": 8.1,
-      "learning_rate": 4.599605522682446e-05,
-      "loss": 1.6074,
+      "epoch": 12.22,
+      "learning_rate": 7.555357142857144e-05,
+      "loss": 1.1868,
       "step": 2738
     },
     {
-      "epoch": 8.1,
-      "learning_rate": 4.597633136094675e-05,
-      "loss": 1.563,
+      "epoch": 12.23,
+      "learning_rate": 7.554464285714285e-05,
+      "loss": 1.1509,
       "step": 2739
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 4.595660749506904e-05,
-      "loss": 1.5575,
+      "epoch": 12.23,
+      "learning_rate": 7.553571428571429e-05,
+      "loss": 1.0874,
       "step": 2740
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 4.5936883629191324e-05,
-      "loss": 1.5769,
+      "epoch": 12.24,
+      "learning_rate": 7.552678571428572e-05,
+      "loss": 1.1151,
       "step": 2741
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 4.591715976331361e-05,
-      "loss": 1.503,
+      "epoch": 12.24,
+      "learning_rate": 7.551785714285715e-05,
+      "loss": 1.2232,
       "step": 2742
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 4.5897435897435895e-05,
-      "loss": 1.533,
+      "epoch": 12.25,
+      "learning_rate": 7.550892857142857e-05,
+      "loss": 1.1479,
       "step": 2743
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 4.587771203155819e-05,
-      "loss": 1.5731,
+      "epoch": 12.25,
+      "learning_rate": 7.55e-05,
+      "loss": 1.1075,
       "step": 2744
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 4.585798816568048e-05,
-      "loss": 1.5514,
+      "epoch": 12.25,
+      "learning_rate": 7.549107142857143e-05,
+      "loss": 1.1997,
       "step": 2745
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 4.5838264299802764e-05,
-      "loss": 1.5874,
+      "epoch": 12.26,
+      "learning_rate": 7.548214285714286e-05,
+      "loss": 1.1129,
       "step": 2746
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 4.581854043392505e-05,
-      "loss": 1.6034,
+      "epoch": 12.26,
+      "learning_rate": 7.54732142857143e-05,
+      "loss": 1.1984,
       "step": 2747
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 4.5798816568047335e-05,
-      "loss": 1.5345,
+      "epoch": 12.27,
+      "learning_rate": 7.546428571428572e-05,
+      "loss": 1.2634,
       "step": 2748
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 4.577909270216963e-05,
-      "loss": 1.5036,
+      "epoch": 12.27,
+      "learning_rate": 7.545535714285715e-05,
+      "loss": 1.1436,
       "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,
+      "epoch": 12.28,
+      "learning_rate": 7.544642857142857e-05,
+      "loss": 1.1545,
       "step": 2750
     },
     {
-      "epoch": 8.14,
-      "learning_rate": 4.5739644970414204e-05,
-      "loss": 1.5437,
+      "epoch": 12.28,
+      "learning_rate": 7.543750000000001e-05,
+      "loss": 1.1021,
       "step": 2751
     },
     {
-      "epoch": 8.14,
-      "learning_rate": 4.571992110453649e-05,
-      "loss": 1.5314,
+      "epoch": 12.29,
+      "learning_rate": 7.542857142857144e-05,
+      "loss": 1.1491,
       "step": 2752
     },
     {
-      "epoch": 8.14,
-      "learning_rate": 4.570019723865878e-05,
-      "loss": 1.4844,
+      "epoch": 12.29,
+      "learning_rate": 7.541964285714286e-05,
+      "loss": 1.2734,
       "step": 2753
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 4.568047337278107e-05,
-      "loss": 1.6147,
+      "epoch": 12.29,
+      "learning_rate": 7.541071428571429e-05,
+      "loss": 1.1897,
       "step": 2754
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 4.566074950690335e-05,
-      "loss": 1.6048,
+      "epoch": 12.3,
+      "learning_rate": 7.540178571428573e-05,
+      "loss": 1.1955,
       "step": 2755
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 4.5641025641025645e-05,
-      "loss": 1.6355,
+      "epoch": 12.3,
+      "learning_rate": 7.539285714285714e-05,
+      "loss": 1.2002,
       "step": 2756
     },
     {
-      "epoch": 8.16,
-      "learning_rate": 4.562130177514793e-05,
-      "loss": 1.5588,
+      "epoch": 12.31,
+      "learning_rate": 7.538392857142858e-05,
+      "loss": 1.1563,
       "step": 2757
     },
     {
-      "epoch": 8.16,
-      "learning_rate": 4.560157790927022e-05,
-      "loss": 1.638,
+      "epoch": 12.31,
+      "learning_rate": 7.5375e-05,
+      "loss": 1.1674,
       "step": 2758
     },
     {
-      "epoch": 8.16,
-      "learning_rate": 4.558185404339251e-05,
-      "loss": 1.6198,
+      "epoch": 12.32,
+      "learning_rate": 7.536607142857143e-05,
+      "loss": 1.2177,
       "step": 2759
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 4.556213017751479e-05,
-      "loss": 1.5291,
+      "epoch": 12.32,
+      "learning_rate": 7.535714285714285e-05,
+      "loss": 1.1644,
       "step": 2760
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 4.554240631163708e-05,
-      "loss": 1.568,
+      "epoch": 12.33,
+      "learning_rate": 7.534821428571429e-05,
+      "loss": 1.176,
       "step": 2761
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 4.552268244575938e-05,
-      "loss": 1.5123,
+      "epoch": 12.33,
+      "learning_rate": 7.533928571428572e-05,
+      "loss": 1.2048,
       "step": 2762
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 4.550295857988166e-05,
-      "loss": 1.5304,
+      "epoch": 12.33,
+      "learning_rate": 7.533035714285714e-05,
+      "loss": 1.2108,
       "step": 2763
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 4.548323471400395e-05,
-      "loss": 1.5373,
+      "epoch": 12.34,
+      "learning_rate": 7.532142857142858e-05,
+      "loss": 1.2189,
       "step": 2764
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 4.546351084812623e-05,
-      "loss": 1.5204,
+      "epoch": 12.34,
+      "learning_rate": 7.531250000000001e-05,
+      "loss": 1.1996,
       "step": 2765
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 4.544378698224852e-05,
-      "loss": 1.5549,
+      "epoch": 12.35,
+      "learning_rate": 7.530357142857143e-05,
+      "loss": 1.187,
       "step": 2766
     },
     {
-      "epoch": 8.19,
-      "learning_rate": 4.542406311637081e-05,
-      "loss": 1.5293,
+      "epoch": 12.35,
+      "learning_rate": 7.529464285714286e-05,
+      "loss": 1.1879,
       "step": 2767
     },
     {
-      "epoch": 8.19,
-      "learning_rate": 4.54043392504931e-05,
-      "loss": 1.6036,
+      "epoch": 12.36,
+      "learning_rate": 7.52857142857143e-05,
+      "loss": 1.2058,
       "step": 2768
     },
     {
-      "epoch": 8.19,
-      "learning_rate": 4.538461538461539e-05,
-      "loss": 1.5415,
+      "epoch": 12.36,
+      "learning_rate": 7.527678571428571e-05,
+      "loss": 1.1848,
       "step": 2769
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 4.536489151873767e-05,
-      "loss": 1.4765,
+      "epoch": 12.37,
+      "learning_rate": 7.526785714285715e-05,
+      "loss": 1.2279,
       "step": 2770
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 4.534516765285996e-05,
-      "loss": 1.6607,
+      "epoch": 12.37,
+      "learning_rate": 7.525892857142857e-05,
+      "loss": 1.2448,
       "step": 2771
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 4.532544378698225e-05,
-      "loss": 1.5772,
+      "epoch": 12.38,
+      "learning_rate": 7.525e-05,
+      "loss": 1.0802,
       "step": 2772
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 4.5305719921104536e-05,
-      "loss": 1.6257,
+      "epoch": 12.38,
+      "learning_rate": 7.524107142857142e-05,
+      "loss": 1.1511,
       "step": 2773
     },
     {
-      "epoch": 8.21,
-      "learning_rate": 4.528599605522683e-05,
-      "loss": 1.4935,
+      "epoch": 12.38,
+      "learning_rate": 7.523214285714286e-05,
+      "loss": 1.1327,
       "step": 2774
     },
     {
-      "epoch": 8.21,
-      "learning_rate": 4.5266272189349114e-05,
-      "loss": 1.5563,
+      "epoch": 12.39,
+      "learning_rate": 7.522321428571429e-05,
+      "loss": 1.1268,
       "step": 2775
     },
     {
-      "epoch": 8.21,
-      "learning_rate": 4.5246548323471406e-05,
-      "loss": 1.6774,
+      "epoch": 12.39,
+      "learning_rate": 7.521428571428571e-05,
+      "loss": 1.1765,
       "step": 2776
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 4.522682445759369e-05,
-      "loss": 1.5793,
+      "epoch": 12.4,
+      "learning_rate": 7.520535714285714e-05,
+      "loss": 1.1774,
       "step": 2777
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 4.5207100591715976e-05,
-      "loss": 1.6289,
+      "epoch": 12.4,
+      "learning_rate": 7.519642857142858e-05,
+      "loss": 1.171,
       "step": 2778
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 4.518737672583827e-05,
-      "loss": 1.631,
+      "epoch": 12.41,
+      "learning_rate": 7.51875e-05,
+      "loss": 1.1514,
       "step": 2779
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 4.5167652859960554e-05,
-      "loss": 1.5763,
+      "epoch": 12.41,
+      "learning_rate": 7.517857142857143e-05,
+      "loss": 1.2264,
       "step": 2780
     },
     {
-      "epoch": 8.23,
-      "learning_rate": 4.5147928994082846e-05,
-      "loss": 1.5766,
+      "epoch": 12.42,
+      "learning_rate": 7.516964285714287e-05,
+      "loss": 1.0898,
       "step": 2781
     },
     {
-      "epoch": 8.23,
-      "learning_rate": 4.512820512820513e-05,
-      "loss": 1.6216,
+      "epoch": 12.42,
+      "learning_rate": 7.516071428571428e-05,
+      "loss": 1.0978,
       "step": 2782
     },
     {
-      "epoch": 8.23,
-      "learning_rate": 4.5108481262327417e-05,
-      "loss": 1.5911,
+      "epoch": 12.42,
+      "learning_rate": 7.515178571428572e-05,
+      "loss": 1.1758,
       "step": 2783
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 4.50887573964497e-05,
-      "loss": 1.6,
+      "epoch": 12.43,
+      "learning_rate": 7.514285714285715e-05,
+      "loss": 1.1867,
       "step": 2784
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 4.5069033530571994e-05,
-      "loss": 1.5748,
+      "epoch": 12.43,
+      "learning_rate": 7.513392857142858e-05,
+      "loss": 1.1559,
       "step": 2785
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 4.5049309664694286e-05,
-      "loss": 1.6229,
+      "epoch": 12.44,
+      "learning_rate": 7.5125e-05,
+      "loss": 1.1397,
       "step": 2786
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 4.502958579881657e-05,
-      "loss": 1.6346,
+      "epoch": 12.44,
+      "learning_rate": 7.511607142857144e-05,
+      "loss": 1.1789,
       "step": 2787
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 4.500986193293886e-05,
-      "loss": 1.5731,
+      "epoch": 12.45,
+      "learning_rate": 7.510714285714286e-05,
+      "loss": 1.1505,
       "step": 2788
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 4.499013806706114e-05,
-      "loss": 1.5048,
+      "epoch": 12.45,
+      "learning_rate": 7.509821428571429e-05,
+      "loss": 1.1757,
       "step": 2789
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 4.4970414201183434e-05,
-      "loss": 1.5813,
+      "epoch": 12.46,
+      "learning_rate": 7.508928571428571e-05,
+      "loss": 1.1336,
       "step": 2790
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 4.495069033530572e-05,
-      "loss": 1.6381,
+      "epoch": 12.46,
+      "learning_rate": 7.508035714285715e-05,
+      "loss": 1.1479,
       "step": 2791
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 4.493096646942801e-05,
-      "loss": 1.5002,
+      "epoch": 12.46,
+      "learning_rate": 7.507142857142858e-05,
+      "loss": 1.1375,
       "step": 2792
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 4.49112426035503e-05,
-      "loss": 1.5372,
+      "epoch": 12.47,
+      "learning_rate": 7.50625e-05,
+      "loss": 1.1328,
       "step": 2793
     },
     {
-      "epoch": 8.27,
-      "learning_rate": 4.489151873767258e-05,
-      "loss": 1.5644,
+      "epoch": 12.47,
+      "learning_rate": 7.505357142857144e-05,
+      "loss": 1.1603,
       "step": 2794
     },
     {
-      "epoch": 8.27,
-      "learning_rate": 4.4871794871794874e-05,
-      "loss": 1.6434,
+      "epoch": 12.48,
+      "learning_rate": 7.504464285714287e-05,
+      "loss": 1.2071,
       "step": 2795
     },
     {
-      "epoch": 8.27,
-      "learning_rate": 4.485207100591716e-05,
-      "loss": 1.5682,
+      "epoch": 12.48,
+      "learning_rate": 7.503571428571429e-05,
+      "loss": 1.1149,
       "step": 2796
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 4.483234714003945e-05,
-      "loss": 1.6144,
+      "epoch": 12.49,
+      "learning_rate": 7.502678571428572e-05,
+      "loss": 1.1601,
       "step": 2797
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 4.481262327416174e-05,
-      "loss": 1.4969,
+      "epoch": 12.49,
+      "learning_rate": 7.501785714285716e-05,
+      "loss": 1.2013,
       "step": 2798
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 4.479289940828403e-05,
-      "loss": 1.5724,
+      "epoch": 12.5,
+      "learning_rate": 7.500892857142857e-05,
+      "loss": 1.1377,
       "step": 2799
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 4.4773175542406315e-05,
-      "loss": 1.558,
+      "epoch": 12.5,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.162,
       "step": 2800
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 4.47534516765286e-05,
-      "loss": 1.6054,
+      "epoch": 12.5,
+      "learning_rate": 7.499107142857143e-05,
+      "loss": 1.1393,
       "step": 2801
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 4.4733727810650885e-05,
-      "loss": 1.5862,
+      "epoch": 12.51,
+      "learning_rate": 7.498214285714286e-05,
+      "loss": 1.1002,
       "step": 2802
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 4.471400394477318e-05,
-      "loss": 1.5659,
+      "epoch": 12.51,
+      "learning_rate": 7.497321428571428e-05,
+      "loss": 1.0828,
       "step": 2803
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 4.469428007889547e-05,
-      "loss": 1.5996,
+      "epoch": 12.52,
+      "learning_rate": 7.496428571428572e-05,
+      "loss": 1.1769,
       "step": 2804
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 4.4674556213017755e-05,
-      "loss": 1.5771,
+      "epoch": 12.52,
+      "learning_rate": 7.495535714285715e-05,
+      "loss": 1.1755,
       "step": 2805
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 4.465483234714004e-05,
-      "loss": 1.6214,
+      "epoch": 12.53,
+      "learning_rate": 7.494642857142857e-05,
+      "loss": 1.2134,
       "step": 2806
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 4.4635108481262326e-05,
-      "loss": 1.6571,
+      "epoch": 12.53,
+      "learning_rate": 7.49375e-05,
+      "loss": 1.1587,
       "step": 2807
     },
     {
-      "epoch": 8.31,
-      "learning_rate": 4.461538461538462e-05,
-      "loss": 1.6018,
+      "epoch": 12.54,
+      "learning_rate": 7.492857142857144e-05,
+      "loss": 1.1464,
       "step": 2808
     },
     {
-      "epoch": 8.31,
-      "learning_rate": 4.459566074950691e-05,
-      "loss": 1.548,
+      "epoch": 12.54,
+      "learning_rate": 7.491964285714286e-05,
+      "loss": 1.1346,
       "step": 2809
     },
     {
-      "epoch": 8.31,
-      "learning_rate": 4.4575936883629195e-05,
-      "loss": 1.4964,
+      "epoch": 12.54,
+      "learning_rate": 7.491071428571429e-05,
+      "loss": 1.1301,
       "step": 2810
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 4.455621301775148e-05,
-      "loss": 1.489,
+      "epoch": 12.55,
+      "learning_rate": 7.490178571428573e-05,
+      "loss": 1.0409,
       "step": 2811
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 4.4536489151873766e-05,
-      "loss": 1.5784,
+      "epoch": 12.55,
+      "learning_rate": 7.489285714285714e-05,
+      "loss": 1.1357,
       "step": 2812
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 4.451676528599606e-05,
-      "loss": 1.4835,
+      "epoch": 12.56,
+      "learning_rate": 7.488392857142858e-05,
+      "loss": 1.1366,
       "step": 2813
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 4.449704142011834e-05,
-      "loss": 1.5982,
+      "epoch": 12.56,
+      "learning_rate": 7.4875e-05,
+      "loss": 1.2126,
       "step": 2814
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 4.4477317554240635e-05,
-      "loss": 1.6174,
+      "epoch": 12.57,
+      "learning_rate": 7.486607142857143e-05,
+      "loss": 1.1317,
       "step": 2815
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 4.445759368836292e-05,
-      "loss": 1.6509,
+      "epoch": 12.57,
+      "learning_rate": 7.485714285714285e-05,
+      "loss": 1.1695,
       "step": 2816
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 4.443786982248521e-05,
-      "loss": 1.5494,
+      "epoch": 12.58,
+      "learning_rate": 7.48482142857143e-05,
+      "loss": 1.1166,
       "step": 2817
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 4.44181459566075e-05,
-      "loss": 1.6078,
+      "epoch": 12.58,
+      "learning_rate": 7.483928571428572e-05,
+      "loss": 1.1845,
       "step": 2818
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 4.4398422090729784e-05,
-      "loss": 1.685,
+      "epoch": 12.58,
+      "learning_rate": 7.483035714285714e-05,
+      "loss": 1.1282,
       "step": 2819
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 4.437869822485207e-05,
-      "loss": 1.6272,
+      "epoch": 12.59,
+      "learning_rate": 7.482142857142857e-05,
+      "loss": 1.2279,
       "step": 2820
     },
     {
-      "epoch": 8.35,
-      "learning_rate": 4.435897435897436e-05,
-      "loss": 1.5914,
+      "epoch": 12.59,
+      "learning_rate": 7.481250000000001e-05,
+      "loss": 1.1577,
       "step": 2821
     },
     {
-      "epoch": 8.35,
-      "learning_rate": 4.433925049309665e-05,
-      "loss": 1.645,
+      "epoch": 12.6,
+      "learning_rate": 7.480357142857143e-05,
+      "loss": 1.1877,
       "step": 2822
     },
     {
-      "epoch": 8.35,
-      "learning_rate": 4.431952662721894e-05,
-      "loss": 1.582,
+      "epoch": 12.6,
+      "learning_rate": 7.479464285714286e-05,
+      "loss": 1.1617,
       "step": 2823
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 4.4299802761341224e-05,
-      "loss": 1.5965,
+      "epoch": 12.61,
+      "learning_rate": 7.478571428571429e-05,
+      "loss": 1.1201,
       "step": 2824
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 4.428007889546351e-05,
-      "loss": 1.4944,
+      "epoch": 12.61,
+      "learning_rate": 7.477678571428571e-05,
+      "loss": 1.1885,
       "step": 2825
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 4.42603550295858e-05,
-      "loss": 1.6258,
+      "epoch": 12.62,
+      "learning_rate": 7.476785714285715e-05,
+      "loss": 1.2262,
       "step": 2826
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 4.424063116370809e-05,
-      "loss": 1.5329,
+      "epoch": 12.62,
+      "learning_rate": 7.475892857142858e-05,
+      "loss": 1.1576,
       "step": 2827
     },
     {
-      "epoch": 8.37,
-      "learning_rate": 4.422090729783038e-05,
-      "loss": 1.5792,
+      "epoch": 12.62,
+      "learning_rate": 7.475000000000001e-05,
+      "loss": 1.2067,
       "step": 2828
     },
     {
-      "epoch": 8.37,
-      "learning_rate": 4.4201183431952664e-05,
-      "loss": 1.6054,
+      "epoch": 12.63,
+      "learning_rate": 7.474107142857143e-05,
+      "loss": 1.1633,
       "step": 2829
     },
     {
-      "epoch": 8.37,
-      "learning_rate": 4.418145956607495e-05,
-      "loss": 1.5927,
+      "epoch": 12.63,
+      "learning_rate": 7.473214285714287e-05,
+      "loss": 1.1342,
       "step": 2830
     },
     {
-      "epoch": 8.38,
-      "learning_rate": 4.416173570019724e-05,
-      "loss": 1.6204,
+      "epoch": 12.64,
+      "learning_rate": 7.472321428571429e-05,
+      "loss": 1.2392,
       "step": 2831
     },
     {
-      "epoch": 8.38,
-      "learning_rate": 4.414201183431953e-05,
-      "loss": 1.6657,
+      "epoch": 12.64,
+      "learning_rate": 7.471428571428572e-05,
+      "loss": 1.1619,
       "step": 2832
     },
     {
-      "epoch": 8.38,
-      "learning_rate": 4.412228796844182e-05,
-      "loss": 1.6263,
+      "epoch": 12.65,
+      "learning_rate": 7.470535714285714e-05,
+      "loss": 1.1741,
       "step": 2833
     },
     {
-      "epoch": 8.38,
-      "learning_rate": 4.4102564102564104e-05,
-      "loss": 1.6835,
+      "epoch": 12.65,
+      "learning_rate": 7.469642857142858e-05,
+      "loss": 1.1971,
       "step": 2834
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 4.408284023668639e-05,
-      "loss": 1.6444,
+      "epoch": 12.66,
+      "learning_rate": 7.468749999999999e-05,
+      "loss": 1.1821,
       "step": 2835
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 4.406311637080868e-05,
-      "loss": 1.4505,
+      "epoch": 12.66,
+      "learning_rate": 7.467857142857143e-05,
+      "loss": 1.1163,
       "step": 2836
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 4.404339250493097e-05,
-      "loss": 1.5611,
+      "epoch": 12.67,
+      "learning_rate": 7.466964285714286e-05,
+      "loss": 1.1083,
       "step": 2837
     },
     {
-      "epoch": 8.4,
-      "learning_rate": 4.402366863905326e-05,
-      "loss": 1.5271,
+      "epoch": 12.67,
+      "learning_rate": 7.466071428571428e-05,
+      "loss": 1.2008,
       "step": 2838
     },
     {
-      "epoch": 8.4,
-      "learning_rate": 4.4003944773175544e-05,
-      "loss": 1.6081,
+      "epoch": 12.67,
+      "learning_rate": 7.465178571428572e-05,
+      "loss": 1.1655,
       "step": 2839
     },
     {
-      "epoch": 8.4,
-      "learning_rate": 4.398422090729784e-05,
-      "loss": 1.5489,
+      "epoch": 12.68,
+      "learning_rate": 7.464285714285715e-05,
+      "loss": 1.1084,
       "step": 2840
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 4.396449704142012e-05,
-      "loss": 1.5124,
+      "epoch": 12.68,
+      "learning_rate": 7.463392857142859e-05,
+      "loss": 1.2082,
       "step": 2841
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 4.394477317554241e-05,
-      "loss": 1.4926,
+      "epoch": 12.69,
+      "learning_rate": 7.4625e-05,
+      "loss": 1.2061,
       "step": 2842
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 4.392504930966469e-05,
-      "loss": 1.5619,
+      "epoch": 12.69,
+      "learning_rate": 7.461607142857144e-05,
+      "loss": 1.2288,
       "step": 2843
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 4.3905325443786985e-05,
-      "loss": 1.5936,
+      "epoch": 12.7,
+      "learning_rate": 7.460714285714286e-05,
+      "loss": 1.144,
       "step": 2844
     },
     {
-      "epoch": 8.42,
-      "learning_rate": 4.388560157790928e-05,
-      "loss": 1.5662,
+      "epoch": 12.7,
+      "learning_rate": 7.459821428571429e-05,
+      "loss": 1.1764,
       "step": 2845
     },
     {
-      "epoch": 8.42,
-      "learning_rate": 4.386587771203156e-05,
-      "loss": 1.523,
+      "epoch": 12.71,
+      "learning_rate": 7.458928571428571e-05,
+      "loss": 1.2472,
       "step": 2846
     },
     {
-      "epoch": 8.42,
-      "learning_rate": 4.384615384615385e-05,
-      "loss": 1.559,
+      "epoch": 12.71,
+      "learning_rate": 7.458035714285715e-05,
+      "loss": 1.1707,
       "step": 2847
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 4.382642998027613e-05,
-      "loss": 1.5933,
+      "epoch": 12.71,
+      "learning_rate": 7.457142857142856e-05,
+      "loss": 1.1369,
       "step": 2848
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 4.380670611439842e-05,
-      "loss": 1.5559,
+      "epoch": 12.72,
+      "learning_rate": 7.45625e-05,
+      "loss": 1.1921,
       "step": 2849
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 4.378698224852072e-05,
-      "loss": 1.5346,
+      "epoch": 12.72,
+      "learning_rate": 7.455357142857143e-05,
+      "loss": 1.1691,
       "step": 2850
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 4.3767258382643e-05,
-      "loss": 1.5999,
+      "epoch": 12.73,
+      "learning_rate": 7.454464285714287e-05,
+      "loss": 1.2365,
       "step": 2851
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 4.374753451676529e-05,
-      "loss": 1.5506,
+      "epoch": 12.73,
+      "learning_rate": 7.453571428571429e-05,
+      "loss": 1.1323,
       "step": 2852
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 4.372781065088757e-05,
-      "loss": 1.5293,
+      "epoch": 12.74,
+      "learning_rate": 7.452678571428572e-05,
+      "loss": 1.1056,
       "step": 2853
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 4.3708086785009865e-05,
-      "loss": 1.5117,
+      "epoch": 12.74,
+      "learning_rate": 7.451785714285714e-05,
+      "loss": 1.1782,
       "step": 2854
     },
     {
-      "epoch": 8.45,
-      "learning_rate": 4.368836291913215e-05,
-      "loss": 1.63,
+      "epoch": 12.75,
+      "learning_rate": 7.450892857142857e-05,
+      "loss": 1.0932,
       "step": 2855
     },
     {
-      "epoch": 8.45,
-      "learning_rate": 4.366863905325444e-05,
-      "loss": 1.6238,
+      "epoch": 12.75,
+      "learning_rate": 7.450000000000001e-05,
+      "loss": 1.1848,
       "step": 2856
     },
     {
-      "epoch": 8.45,
-      "learning_rate": 4.364891518737673e-05,
-      "loss": 1.5829,
+      "epoch": 12.75,
+      "learning_rate": 7.449107142857143e-05,
+      "loss": 1.1633,
       "step": 2857
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 4.362919132149901e-05,
-      "loss": 1.5539,
+      "epoch": 12.76,
+      "learning_rate": 7.448214285714286e-05,
+      "loss": 1.1251,
       "step": 2858
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 4.3609467455621305e-05,
-      "loss": 1.5503,
+      "epoch": 12.76,
+      "learning_rate": 7.447321428571428e-05,
+      "loss": 1.1143,
       "step": 2859
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 4.358974358974359e-05,
-      "loss": 1.5541,
+      "epoch": 12.77,
+      "learning_rate": 7.446428571428572e-05,
+      "loss": 1.2219,
       "step": 2860
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 4.3570019723865876e-05,
-      "loss": 1.584,
+      "epoch": 12.77,
+      "learning_rate": 7.445535714285715e-05,
+      "loss": 1.1022,
       "step": 2861
     },
     {
-      "epoch": 8.47,
-      "learning_rate": 4.355029585798817e-05,
-      "loss": 1.4504,
+      "epoch": 12.78,
+      "learning_rate": 7.444642857142857e-05,
+      "loss": 1.1839,
       "step": 2862
     },
     {
-      "epoch": 8.47,
-      "learning_rate": 4.353057199211046e-05,
-      "loss": 1.4238,
+      "epoch": 12.78,
+      "learning_rate": 7.44375e-05,
+      "loss": 1.1764,
       "step": 2863
     },
     {
-      "epoch": 8.47,
-      "learning_rate": 4.3510848126232746e-05,
-      "loss": 1.5022,
+      "epoch": 12.79,
+      "learning_rate": 7.442857142857144e-05,
+      "loss": 1.2442,
       "step": 2864
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 4.349112426035503e-05,
-      "loss": 1.5196,
+      "epoch": 12.79,
+      "learning_rate": 7.441964285714285e-05,
+      "loss": 1.1748,
       "step": 2865
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 4.3471400394477316e-05,
-      "loss": 1.5495,
+      "epoch": 12.79,
+      "learning_rate": 7.441071428571429e-05,
+      "loss": 1.1353,
       "step": 2866
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 4.345167652859961e-05,
-      "loss": 1.5647,
+      "epoch": 12.8,
+      "learning_rate": 7.440178571428572e-05,
+      "loss": 1.1612,
       "step": 2867
     },
     {
-      "epoch": 8.49,
-      "learning_rate": 4.34319526627219e-05,
-      "loss": 1.4863,
+      "epoch": 12.8,
+      "learning_rate": 7.439285714285714e-05,
+      "loss": 1.2795,
       "step": 2868
     },
     {
-      "epoch": 8.49,
-      "learning_rate": 4.3412228796844186e-05,
-      "loss": 1.6388,
+      "epoch": 12.81,
+      "learning_rate": 7.438392857142858e-05,
+      "loss": 1.2581,
       "step": 2869
     },
     {
-      "epoch": 8.49,
-      "learning_rate": 4.339250493096647e-05,
-      "loss": 1.6072,
+      "epoch": 12.81,
+      "learning_rate": 7.4375e-05,
+      "loss": 1.121,
       "step": 2870
     },
     {
-      "epoch": 8.49,
-      "learning_rate": 4.3372781065088757e-05,
-      "loss": 1.4989,
+      "epoch": 12.82,
+      "learning_rate": 7.436607142857143e-05,
+      "loss": 1.1765,
       "step": 2871
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 4.335305719921105e-05,
-      "loss": 1.4485,
+      "epoch": 12.82,
+      "learning_rate": 7.435714285714286e-05,
+      "loss": 1.2339,
       "step": 2872
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 4.3333333333333334e-05,
-      "loss": 1.5938,
+      "epoch": 12.83,
+      "learning_rate": 7.43482142857143e-05,
+      "loss": 1.1616,
       "step": 2873
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 4.3313609467455626e-05,
-      "loss": 1.4972,
+      "epoch": 12.83,
+      "learning_rate": 7.433928571428572e-05,
+      "loss": 1.1494,
       "step": 2874
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 4.329388560157791e-05,
-      "loss": 1.68,
+      "epoch": 12.83,
+      "learning_rate": 7.433035714285715e-05,
+      "loss": 1.1552,
       "step": 2875
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 4.32741617357002e-05,
-      "loss": 1.5847,
+      "epoch": 12.84,
+      "learning_rate": 7.432142857142857e-05,
+      "loss": 1.1541,
       "step": 2876
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 4.325443786982249e-05,
-      "loss": 1.4753,
+      "epoch": 12.84,
+      "learning_rate": 7.431250000000001e-05,
+      "loss": 1.1706,
       "step": 2877
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 4.3234714003944774e-05,
-      "loss": 1.6212,
+      "epoch": 12.85,
+      "learning_rate": 7.430357142857142e-05,
+      "loss": 1.1436,
       "step": 2878
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 4.321499013806706e-05,
-      "loss": 1.4954,
+      "epoch": 12.85,
+      "learning_rate": 7.429464285714286e-05,
+      "loss": 1.0791,
       "step": 2879
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 4.319526627218935e-05,
-      "loss": 1.4881,
+      "epoch": 12.86,
+      "learning_rate": 7.428571428571429e-05,
+      "loss": 1.2133,
       "step": 2880
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 4.317554240631164e-05,
-      "loss": 1.61,
+      "epoch": 12.86,
+      "learning_rate": 7.427678571428571e-05,
+      "loss": 1.2029,
       "step": 2881
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 4.315581854043393e-05,
-      "loss": 1.5301,
+      "epoch": 12.87,
+      "learning_rate": 7.426785714285714e-05,
+      "loss": 1.1029,
       "step": 2882
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 4.3136094674556215e-05,
-      "loss": 1.5835,
+      "epoch": 12.87,
+      "learning_rate": 7.425892857142858e-05,
+      "loss": 1.2127,
       "step": 2883
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 4.31163708086785e-05,
-      "loss": 1.4849,
+      "epoch": 12.88,
+      "learning_rate": 7.425e-05,
+      "loss": 1.1759,
       "step": 2884
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.309664694280079e-05,
-      "loss": 1.5922,
+      "epoch": 12.88,
+      "learning_rate": 7.424107142857143e-05,
+      "loss": 1.2207,
       "step": 2885
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.3076923076923084e-05,
-      "loss": 1.6262,
+      "epoch": 12.88,
+      "learning_rate": 7.423214285714287e-05,
+      "loss": 1.1615,
       "step": 2886
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.305719921104537e-05,
-      "loss": 1.5763,
+      "epoch": 12.89,
+      "learning_rate": 7.422321428571429e-05,
+      "loss": 1.2184,
       "step": 2887
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.3037475345167655e-05,
-      "loss": 1.5266,
+      "epoch": 12.89,
+      "learning_rate": 7.421428571428572e-05,
+      "loss": 1.1107,
       "step": 2888
     },
     {
-      "epoch": 8.55,
-      "learning_rate": 4.301775147928994e-05,
-      "loss": 1.6227,
+      "epoch": 12.9,
+      "learning_rate": 7.420535714285714e-05,
+      "loss": 1.1171,
       "step": 2889
     },
     {
-      "epoch": 8.55,
-      "learning_rate": 4.2998027613412225e-05,
-      "loss": 1.4404,
+      "epoch": 12.9,
+      "learning_rate": 7.419642857142858e-05,
+      "loss": 1.1484,
       "step": 2890
     },
     {
-      "epoch": 8.55,
-      "learning_rate": 4.297830374753452e-05,
-      "loss": 1.5905,
+      "epoch": 12.91,
+      "learning_rate": 7.41875e-05,
+      "loss": 1.1731,
       "step": 2891
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 4.295857988165681e-05,
-      "loss": 1.5076,
+      "epoch": 12.91,
+      "learning_rate": 7.417857142857143e-05,
+      "loss": 1.1681,
       "step": 2892
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 4.2938856015779095e-05,
-      "loss": 1.5658,
+      "epoch": 12.92,
+      "learning_rate": 7.416964285714286e-05,
+      "loss": 1.1816,
       "step": 2893
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 4.291913214990138e-05,
-      "loss": 1.6037,
+      "epoch": 12.92,
+      "learning_rate": 7.41607142857143e-05,
+      "loss": 1.1843,
       "step": 2894
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.289940828402367e-05,
-      "loss": 1.4758,
+      "epoch": 12.92,
+      "learning_rate": 7.415178571428571e-05,
+      "loss": 1.1542,
       "step": 2895
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.287968441814596e-05,
-      "loss": 1.6158,
+      "epoch": 12.93,
+      "learning_rate": 7.414285714285715e-05,
+      "loss": 1.117,
       "step": 2896
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.285996055226825e-05,
-      "loss": 1.5917,
+      "epoch": 12.93,
+      "learning_rate": 7.413392857142857e-05,
+      "loss": 1.1722,
       "step": 2897
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.2840236686390535e-05,
-      "loss": 1.5032,
+      "epoch": 12.94,
+      "learning_rate": 7.4125e-05,
+      "loss": 1.1505,
       "step": 2898
     },
     {
-      "epoch": 8.58,
-      "learning_rate": 4.282051282051282e-05,
-      "loss": 1.4997,
+      "epoch": 12.94,
+      "learning_rate": 7.411607142857144e-05,
+      "loss": 1.0996,
       "step": 2899
     },
     {
-      "epoch": 8.58,
-      "learning_rate": 4.280078895463511e-05,
-      "loss": 1.5596,
+      "epoch": 12.95,
+      "learning_rate": 7.410714285714286e-05,
+      "loss": 1.2297,
       "step": 2900
     },
     {
-      "epoch": 8.58,
-      "learning_rate": 4.27810650887574e-05,
-      "loss": 1.4513,
+      "epoch": 12.95,
+      "learning_rate": 7.409821428571429e-05,
+      "loss": 1.2226,
       "step": 2901
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.276134122287968e-05,
-      "loss": 1.607,
+      "epoch": 12.96,
+      "learning_rate": 7.408928571428571e-05,
+      "loss": 1.1218,
       "step": 2902
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.2741617357001975e-05,
-      "loss": 1.5855,
+      "epoch": 12.96,
+      "learning_rate": 7.408035714285715e-05,
+      "loss": 1.1341,
       "step": 2903
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.272189349112426e-05,
-      "loss": 1.6101,
+      "epoch": 12.96,
+      "learning_rate": 7.407142857142858e-05,
+      "loss": 1.1874,
       "step": 2904
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.270216962524655e-05,
-      "loss": 1.556,
+      "epoch": 12.97,
+      "learning_rate": 7.40625e-05,
+      "loss": 1.2136,
       "step": 2905
     },
     {
-      "epoch": 8.6,
-      "learning_rate": 4.268244575936884e-05,
-      "loss": 1.481,
+      "epoch": 12.97,
+      "learning_rate": 7.405357142857143e-05,
+      "loss": 1.2162,
       "step": 2906
     },
     {
-      "epoch": 8.6,
-      "learning_rate": 4.2662721893491124e-05,
-      "loss": 1.4872,
+      "epoch": 12.98,
+      "learning_rate": 7.404464285714287e-05,
+      "loss": 1.1374,
       "step": 2907
     },
     {
-      "epoch": 8.6,
-      "learning_rate": 4.264299802761341e-05,
-      "loss": 1.5988,
+      "epoch": 12.98,
+      "learning_rate": 7.403571428571428e-05,
+      "loss": 1.2473,
       "step": 2908
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 4.262327416173571e-05,
-      "loss": 1.5171,
+      "epoch": 12.99,
+      "learning_rate": 7.402678571428572e-05,
+      "loss": 1.2366,
       "step": 2909
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 4.260355029585799e-05,
-      "loss": 1.5211,
+      "epoch": 12.99,
+      "learning_rate": 7.401785714285715e-05,
+      "loss": 1.1882,
       "step": 2910
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 4.258382642998028e-05,
-      "loss": 1.5506,
+      "epoch": 13.0,
+      "learning_rate": 7.400892857142857e-05,
+      "loss": 1.2214,
       "step": 2911
     },
     {
-      "epoch": 8.62,
-      "learning_rate": 4.2564102564102564e-05,
-      "loss": 1.5946,
+      "epoch": 13.0,
+      "learning_rate": 7.4e-05,
+      "loss": 1.1169,
       "step": 2912
     },
     {
-      "epoch": 8.62,
-      "learning_rate": 4.254437869822485e-05,
-      "loss": 1.5533,
+      "epoch": 13.0,
+      "learning_rate": 7.399107142857144e-05,
+      "loss": 1.236,
       "step": 2913
     },
     {
-      "epoch": 8.62,
-      "learning_rate": 4.252465483234714e-05,
-      "loss": 1.5597,
+      "epoch": 13.01,
+      "learning_rate": 7.398214285714286e-05,
+      "loss": 1.146,
       "step": 2914
     },
     {
-      "epoch": 8.62,
-      "learning_rate": 4.2504930966469433e-05,
-      "loss": 1.5213,
+      "epoch": 13.01,
+      "learning_rate": 7.397321428571429e-05,
+      "loss": 1.1288,
       "step": 2915
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 4.248520710059172e-05,
-      "loss": 1.5775,
+      "epoch": 13.02,
+      "learning_rate": 7.396428571428573e-05,
+      "loss": 1.2012,
       "step": 2916
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 4.2465483234714004e-05,
-      "loss": 1.6257,
+      "epoch": 13.02,
+      "learning_rate": 7.395535714285715e-05,
+      "loss": 1.0538,
       "step": 2917
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 4.2445759368836296e-05,
-      "loss": 1.4139,
+      "epoch": 13.03,
+      "learning_rate": 7.394642857142858e-05,
+      "loss": 1.1907,
       "step": 2918
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 4.242603550295858e-05,
-      "loss": 1.5957,
+      "epoch": 13.03,
+      "learning_rate": 7.39375e-05,
+      "loss": 1.1665,
       "step": 2919
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 4.240631163708087e-05,
-      "loss": 1.5521,
+      "epoch": 13.04,
+      "learning_rate": 7.392857142857144e-05,
+      "loss": 1.059,
       "step": 2920
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 4.238658777120316e-05,
-      "loss": 1.6679,
+      "epoch": 13.04,
+      "learning_rate": 7.391964285714285e-05,
+      "loss": 1.112,
       "step": 2921
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 4.2366863905325444e-05,
-      "loss": 1.573,
+      "epoch": 13.04,
+      "learning_rate": 7.391071428571429e-05,
+      "loss": 1.1657,
       "step": 2922
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 4.2347140039447736e-05,
-      "loss": 1.5343,
+      "epoch": 13.05,
+      "learning_rate": 7.390178571428572e-05,
+      "loss": 1.1674,
       "step": 2923
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 4.232741617357002e-05,
-      "loss": 1.5355,
+      "epoch": 13.05,
+      "learning_rate": 7.389285714285714e-05,
+      "loss": 1.2049,
       "step": 2924
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 4.230769230769231e-05,
-      "loss": 1.6277,
+      "epoch": 13.06,
+      "learning_rate": 7.388392857142857e-05,
+      "loss": 1.0693,
       "step": 2925
     },
     {
-      "epoch": 8.66,
-      "learning_rate": 4.22879684418146e-05,
-      "loss": 1.6213,
+      "epoch": 13.06,
+      "learning_rate": 7.3875e-05,
+      "loss": 1.1532,
       "step": 2926
     },
     {
-      "epoch": 8.66,
-      "learning_rate": 4.226824457593689e-05,
-      "loss": 1.5211,
+      "epoch": 13.07,
+      "learning_rate": 7.386607142857143e-05,
+      "loss": 1.1619,
       "step": 2927
     },
     {
-      "epoch": 8.66,
-      "learning_rate": 4.224852071005918e-05,
-      "loss": 1.6396,
+      "epoch": 13.07,
+      "learning_rate": 7.385714285714286e-05,
+      "loss": 1.2022,
       "step": 2928
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.222879684418146e-05,
-      "loss": 1.5625,
+      "epoch": 13.08,
+      "learning_rate": 7.38482142857143e-05,
+      "loss": 1.1417,
       "step": 2929
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.220907297830375e-05,
-      "loss": 1.4434,
+      "epoch": 13.08,
+      "learning_rate": 7.383928571428572e-05,
+      "loss": 1.1435,
       "step": 2930
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.218934911242603e-05,
-      "loss": 1.6081,
+      "epoch": 13.08,
+      "learning_rate": 7.383035714285715e-05,
+      "loss": 1.086,
       "step": 2931
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.2169625246548325e-05,
-      "loss": 1.6177,
+      "epoch": 13.09,
+      "learning_rate": 7.382142857142857e-05,
+      "loss": 1.0973,
       "step": 2932
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 4.214990138067062e-05,
-      "loss": 1.5705,
+      "epoch": 13.09,
+      "learning_rate": 7.381250000000001e-05,
+      "loss": 1.1697,
       "step": 2933
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 4.21301775147929e-05,
-      "loss": 1.6803,
+      "epoch": 13.1,
+      "learning_rate": 7.380357142857142e-05,
+      "loss": 1.1313,
       "step": 2934
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 4.211045364891519e-05,
-      "loss": 1.5303,
+      "epoch": 13.1,
+      "learning_rate": 7.379464285714286e-05,
+      "loss": 1.125,
       "step": 2935
     },
     {
-      "epoch": 8.69,
-      "learning_rate": 4.209072978303747e-05,
-      "loss": 1.5839,
+      "epoch": 13.11,
+      "learning_rate": 7.378571428571429e-05,
+      "loss": 1.1842,
       "step": 2936
     },
     {
-      "epoch": 8.69,
-      "learning_rate": 4.2071005917159765e-05,
-      "loss": 1.5507,
+      "epoch": 13.11,
+      "learning_rate": 7.377678571428573e-05,
+      "loss": 1.0958,
       "step": 2937
     },
     {
-      "epoch": 8.69,
-      "learning_rate": 4.205128205128206e-05,
-      "loss": 1.5969,
+      "epoch": 13.12,
+      "learning_rate": 7.376785714285714e-05,
+      "loss": 1.1382,
       "step": 2938
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.203155818540434e-05,
-      "loss": 1.5103,
+      "epoch": 13.12,
+      "learning_rate": 7.375892857142858e-05,
+      "loss": 1.1793,
       "step": 2939
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.201183431952663e-05,
-      "loss": 1.562,
+      "epoch": 13.12,
+      "learning_rate": 7.375e-05,
+      "loss": 1.1319,
       "step": 2940
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.199211045364892e-05,
-      "loss": 1.4152,
+      "epoch": 13.13,
+      "learning_rate": 7.374107142857143e-05,
+      "loss": 1.2109,
       "step": 2941
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.1972386587771205e-05,
-      "loss": 1.5368,
+      "epoch": 13.13,
+      "learning_rate": 7.373214285714285e-05,
+      "loss": 1.134,
       "step": 2942
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 4.195266272189349e-05,
-      "loss": 1.509,
+      "epoch": 13.14,
+      "learning_rate": 7.37232142857143e-05,
+      "loss": 1.2208,
       "step": 2943
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 4.193293885601578e-05,
-      "loss": 1.5795,
+      "epoch": 13.14,
+      "learning_rate": 7.371428571428572e-05,
+      "loss": 1.2049,
       "step": 2944
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 4.191321499013807e-05,
-      "loss": 1.6065,
+      "epoch": 13.15,
+      "learning_rate": 7.370535714285714e-05,
+      "loss": 1.2018,
       "step": 2945
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.189349112426036e-05,
-      "loss": 1.527,
+      "epoch": 13.15,
+      "learning_rate": 7.369642857142858e-05,
+      "loss": 1.1339,
       "step": 2946
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.1873767258382646e-05,
-      "loss": 1.4939,
+      "epoch": 13.16,
+      "learning_rate": 7.36875e-05,
+      "loss": 1.1411,
       "step": 2947
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.185404339250493e-05,
-      "loss": 1.5622,
+      "epoch": 13.16,
+      "learning_rate": 7.367857142857143e-05,
+      "loss": 1.1536,
       "step": 2948
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.1834319526627216e-05,
-      "loss": 1.5788,
+      "epoch": 13.17,
+      "learning_rate": 7.366964285714286e-05,
+      "loss": 1.1101,
       "step": 2949
     },
     {
-      "epoch": 8.73,
-      "learning_rate": 4.181459566074951e-05,
-      "loss": 1.5973,
+      "epoch": 13.17,
+      "learning_rate": 7.36607142857143e-05,
+      "loss": 1.2319,
       "step": 2950
     },
     {
-      "epoch": 8.73,
-      "learning_rate": 4.17948717948718e-05,
-      "loss": 1.5017,
+      "epoch": 13.17,
+      "learning_rate": 7.365178571428571e-05,
+      "loss": 1.1696,
       "step": 2951
     },
     {
-      "epoch": 8.73,
-      "learning_rate": 4.1775147928994086e-05,
-      "loss": 1.6045,
+      "epoch": 13.18,
+      "learning_rate": 7.364285714285715e-05,
+      "loss": 1.1987,
       "step": 2952
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 4.175542406311637e-05,
-      "loss": 1.6384,
+      "epoch": 13.18,
+      "learning_rate": 7.363392857142858e-05,
+      "loss": 1.2006,
       "step": 2953
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 4.1735700197238656e-05,
-      "loss": 1.5073,
+      "epoch": 13.19,
+      "learning_rate": 7.3625e-05,
+      "loss": 1.1794,
       "step": 2954
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 4.171597633136095e-05,
-      "loss": 1.5103,
+      "epoch": 13.19,
+      "learning_rate": 7.361607142857143e-05,
+      "loss": 1.138,
       "step": 2955
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 4.169625246548324e-05,
-      "loss": 1.5249,
+      "epoch": 13.2,
+      "learning_rate": 7.360714285714287e-05,
+      "loss": 1.1585,
       "step": 2956
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 4.1676528599605526e-05,
-      "loss": 1.5459,
+      "epoch": 13.2,
+      "learning_rate": 7.359821428571429e-05,
+      "loss": 1.1135,
       "step": 2957
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 4.165680473372781e-05,
-      "loss": 1.5855,
+      "epoch": 13.21,
+      "learning_rate": 7.358928571428572e-05,
+      "loss": 1.1858,
       "step": 2958
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 4.16370808678501e-05,
-      "loss": 1.6178,
+      "epoch": 13.21,
+      "learning_rate": 7.358035714285714e-05,
+      "loss": 1.1402,
       "step": 2959
     },
     {
-      "epoch": 8.76,
-      "learning_rate": 4.161735700197239e-05,
-      "loss": 1.5156,
+      "epoch": 13.21,
+      "learning_rate": 7.357142857142858e-05,
+      "loss": 1.17,
       "step": 2960
     },
     {
-      "epoch": 8.76,
-      "learning_rate": 4.1597633136094674e-05,
-      "loss": 1.4776,
+      "epoch": 13.22,
+      "learning_rate": 7.35625e-05,
+      "loss": 1.1593,
       "step": 2961
     },
     {
-      "epoch": 8.76,
-      "learning_rate": 4.1577909270216966e-05,
-      "loss": 1.557,
+      "epoch": 13.22,
+      "learning_rate": 7.355357142857143e-05,
+      "loss": 1.1373,
       "step": 2962
     },
     {
-      "epoch": 8.77,
-      "learning_rate": 4.155818540433925e-05,
-      "loss": 1.6104,
+      "epoch": 13.23,
+      "learning_rate": 7.354464285714287e-05,
+      "loss": 1.2286,
       "step": 2963
     },
     {
-      "epoch": 8.77,
-      "learning_rate": 4.1538461538461544e-05,
-      "loss": 1.5765,
+      "epoch": 13.23,
+      "learning_rate": 7.353571428571428e-05,
+      "loss": 1.1883,
       "step": 2964
     },
     {
-      "epoch": 8.77,
-      "learning_rate": 4.151873767258383e-05,
-      "loss": 1.5608,
+      "epoch": 13.24,
+      "learning_rate": 7.352678571428572e-05,
+      "loss": 1.1267,
       "step": 2965
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.1499013806706114e-05,
-      "loss": 1.5848,
+      "epoch": 13.24,
+      "learning_rate": 7.351785714285715e-05,
+      "loss": 1.0902,
       "step": 2966
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.14792899408284e-05,
-      "loss": 1.5948,
+      "epoch": 13.25,
+      "learning_rate": 7.350892857142857e-05,
+      "loss": 1.0872,
       "step": 2967
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.145956607495069e-05,
-      "loss": 1.5428,
+      "epoch": 13.25,
+      "learning_rate": 7.35e-05,
+      "loss": 1.1567,
       "step": 2968
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.1439842209072984e-05,
-      "loss": 1.5041,
+      "epoch": 13.25,
+      "learning_rate": 7.349107142857144e-05,
+      "loss": 1.199,
       "step": 2969
     },
     {
-      "epoch": 8.79,
-      "learning_rate": 4.142011834319527e-05,
-      "loss": 1.5674,
+      "epoch": 13.26,
+      "learning_rate": 7.348214285714286e-05,
+      "loss": 1.2191,
       "step": 2970
     },
     {
-      "epoch": 8.79,
-      "learning_rate": 4.1400394477317555e-05,
-      "loss": 1.5568,
+      "epoch": 13.26,
+      "learning_rate": 7.347321428571429e-05,
+      "loss": 1.1279,
       "step": 2971
     },
     {
-      "epoch": 8.79,
-      "learning_rate": 4.138067061143984e-05,
-      "loss": 1.5431,
+      "epoch": 13.27,
+      "learning_rate": 7.346428571428571e-05,
+      "loss": 1.1598,
       "step": 2972
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.136094674556213e-05,
-      "loss": 1.5963,
+      "epoch": 13.27,
+      "learning_rate": 7.345535714285715e-05,
+      "loss": 1.1679,
       "step": 2973
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.1341222879684424e-05,
-      "loss": 1.5339,
+      "epoch": 13.28,
+      "learning_rate": 7.344642857142858e-05,
+      "loss": 1.1574,
       "step": 2974
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.132149901380671e-05,
-      "loss": 1.5813,
+      "epoch": 13.28,
+      "learning_rate": 7.34375e-05,
+      "loss": 1.0599,
       "step": 2975
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.1301775147928995e-05,
-      "loss": 1.619,
+      "epoch": 13.29,
+      "learning_rate": 7.342857142857144e-05,
+      "loss": 1.0561,
       "step": 2976
     },
     {
-      "epoch": 8.81,
-      "learning_rate": 4.128205128205128e-05,
-      "loss": 1.4316,
+      "epoch": 13.29,
+      "learning_rate": 7.341964285714285e-05,
+      "loss": 1.1506,
       "step": 2977
     },
     {
-      "epoch": 8.81,
-      "learning_rate": 4.126232741617357e-05,
-      "loss": 1.5671,
+      "epoch": 13.29,
+      "learning_rate": 7.341071428571429e-05,
+      "loss": 1.051,
       "step": 2978
     },
     {
-      "epoch": 8.81,
-      "learning_rate": 4.124260355029586e-05,
-      "loss": 1.4708,
+      "epoch": 13.3,
+      "learning_rate": 7.340178571428572e-05,
+      "loss": 1.1965,
       "step": 2979
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 4.122287968441815e-05,
-      "loss": 1.5883,
+      "epoch": 13.3,
+      "learning_rate": 7.339285714285714e-05,
+      "loss": 1.0782,
       "step": 2980
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 4.1203155818540435e-05,
-      "loss": 1.5604,
+      "epoch": 13.31,
+      "learning_rate": 7.338392857142857e-05,
+      "loss": 1.2183,
       "step": 2981
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 4.118343195266273e-05,
-      "loss": 1.5283,
+      "epoch": 13.31,
+      "learning_rate": 7.337500000000001e-05,
+      "loss": 1.1304,
       "step": 2982
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 4.116370808678501e-05,
-      "loss": 1.6185,
+      "epoch": 13.32,
+      "learning_rate": 7.336607142857143e-05,
+      "loss": 1.0526,
       "step": 2983
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 4.11439842209073e-05,
-      "loss": 1.4773,
+      "epoch": 13.32,
+      "learning_rate": 7.335714285714286e-05,
+      "loss": 1.1997,
       "step": 2984
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 4.112426035502959e-05,
-      "loss": 1.5274,
+      "epoch": 13.33,
+      "learning_rate": 7.334821428571428e-05,
+      "loss": 1.169,
       "step": 2985
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 4.1104536489151875e-05,
-      "loss": 1.6186,
+      "epoch": 13.33,
+      "learning_rate": 7.333928571428572e-05,
+      "loss": 1.1839,
       "step": 2986
     },
     {
-      "epoch": 8.84,
-      "learning_rate": 4.108481262327417e-05,
-      "loss": 1.6322,
+      "epoch": 13.33,
+      "learning_rate": 7.333035714285715e-05,
+      "loss": 1.0903,
       "step": 2987
     },
     {
-      "epoch": 8.84,
-      "learning_rate": 4.106508875739645e-05,
-      "loss": 1.5911,
+      "epoch": 13.34,
+      "learning_rate": 7.332142857142857e-05,
+      "loss": 1.1763,
       "step": 2988
     },
     {
-      "epoch": 8.84,
-      "learning_rate": 4.104536489151874e-05,
-      "loss": 1.5531,
+      "epoch": 13.34,
+      "learning_rate": 7.33125e-05,
+      "loss": 1.2121,
       "step": 2989
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 4.1025641025641023e-05,
-      "loss": 1.4761,
+      "epoch": 13.35,
+      "learning_rate": 7.330357142857143e-05,
+      "loss": 1.2034,
       "step": 2990
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 4.1005917159763316e-05,
-      "loss": 1.4432,
+      "epoch": 13.35,
+      "learning_rate": 7.329464285714286e-05,
+      "loss": 1.2336,
       "step": 2991
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 4.098619329388561e-05,
-      "loss": 1.5329,
+      "epoch": 13.36,
+      "learning_rate": 7.328571428571429e-05,
+      "loss": 1.1672,
       "step": 2992
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 4.096646942800789e-05,
-      "loss": 1.5849,
+      "epoch": 13.36,
+      "learning_rate": 7.327678571428573e-05,
+      "loss": 1.1781,
       "step": 2993
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 4.094674556213018e-05,
-      "loss": 1.626,
+      "epoch": 13.37,
+      "learning_rate": 7.326785714285714e-05,
+      "loss": 1.1004,
       "step": 2994
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 4.0927021696252464e-05,
-      "loss": 1.519,
+      "epoch": 13.37,
+      "learning_rate": 7.325892857142858e-05,
+      "loss": 1.1318,
       "step": 2995
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 4.0907297830374756e-05,
-      "loss": 1.5259,
+      "epoch": 13.38,
+      "learning_rate": 7.325e-05,
+      "loss": 1.1583,
       "step": 2996
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 4.088757396449705e-05,
-      "loss": 1.5248,
+      "epoch": 13.38,
+      "learning_rate": 7.324107142857143e-05,
+      "loss": 1.1604,
       "step": 2997
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 4.086785009861933e-05,
-      "loss": 1.6588,
+      "epoch": 13.38,
+      "learning_rate": 7.323214285714286e-05,
+      "loss": 1.1707,
       "step": 2998
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 4.084812623274162e-05,
-      "loss": 1.5237,
+      "epoch": 13.39,
+      "learning_rate": 7.32232142857143e-05,
+      "loss": 1.1987,
       "step": 2999
     },
     {
-      "epoch": 8.88,
-      "learning_rate": 4.0828402366863904e-05,
-      "loss": 1.5347,
+      "epoch": 13.39,
+      "learning_rate": 7.321428571428571e-05,
+      "loss": 1.1535,
       "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,
+      "epoch": 13.39,
+      "eval_loss": 1.1609708070755005,
+      "eval_runtime": 45.2125,
+      "eval_samples_per_second": 289.61,
+      "eval_steps_per_second": 1.526,
       "step": 3000
     },
     {
-      "epoch": 8.88,
-      "learning_rate": 4.0808678500986196e-05,
-      "loss": 1.5793,
+      "epoch": 13.4,
+      "learning_rate": 7.320535714285715e-05,
+      "loss": 1.1348,
       "step": 3001
     },
     {
-      "epoch": 8.88,
-      "learning_rate": 4.078895463510848e-05,
-      "loss": 1.6102,
+      "epoch": 13.4,
+      "learning_rate": 7.319642857142857e-05,
+      "loss": 1.2229,
       "step": 3002
     },
     {
-      "epoch": 8.88,
-      "learning_rate": 4.0769230769230773e-05,
-      "loss": 1.446,
+      "epoch": 13.41,
+      "learning_rate": 7.318750000000001e-05,
+      "loss": 1.1164,
       "step": 3003
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 4.074950690335306e-05,
-      "loss": 1.5826,
+      "epoch": 13.41,
+      "learning_rate": 7.317857142857144e-05,
+      "loss": 1.1323,
       "step": 3004
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 4.072978303747535e-05,
-      "loss": 1.5602,
+      "epoch": 13.42,
+      "learning_rate": 7.316964285714286e-05,
+      "loss": 1.1276,
       "step": 3005
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 4.0710059171597636e-05,
-      "loss": 1.5756,
+      "epoch": 13.42,
+      "learning_rate": 7.31607142857143e-05,
+      "loss": 1.133,
       "step": 3006
     },
     {
-      "epoch": 8.9,
-      "learning_rate": 4.069033530571992e-05,
-      "loss": 1.5139,
+      "epoch": 13.42,
+      "learning_rate": 7.315178571428571e-05,
+      "loss": 1.156,
       "step": 3007
     },
     {
-      "epoch": 8.9,
-      "learning_rate": 4.067061143984221e-05,
-      "loss": 1.6895,
+      "epoch": 13.43,
+      "learning_rate": 7.314285714285715e-05,
+      "loss": 1.198,
       "step": 3008
     },
     {
-      "epoch": 8.9,
-      "learning_rate": 4.06508875739645e-05,
-      "loss": 1.5488,
+      "epoch": 13.43,
+      "learning_rate": 7.313392857142858e-05,
+      "loss": 1.0676,
       "step": 3009
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 4.063116370808679e-05,
-      "loss": 1.5559,
+      "epoch": 13.44,
+      "learning_rate": 7.3125e-05,
+      "loss": 1.1759,
       "step": 3010
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 4.0611439842209077e-05,
-      "loss": 1.4673,
+      "epoch": 13.44,
+      "learning_rate": 7.311607142857143e-05,
+      "loss": 1.0723,
       "step": 3011
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 4.059171597633136e-05,
-      "loss": 1.5717,
+      "epoch": 13.45,
+      "learning_rate": 7.310714285714287e-05,
+      "loss": 1.1092,
       "step": 3012
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 4.057199211045365e-05,
-      "loss": 1.4654,
+      "epoch": 13.45,
+      "learning_rate": 7.309821428571428e-05,
+      "loss": 1.1434,
       "step": 3013
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 4.055226824457594e-05,
-      "loss": 1.5716,
+      "epoch": 13.46,
+      "learning_rate": 7.308928571428572e-05,
+      "loss": 1.1762,
       "step": 3014
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 4.053254437869823e-05,
-      "loss": 1.5812,
+      "epoch": 13.46,
+      "learning_rate": 7.308035714285714e-05,
+      "loss": 1.1605,
       "step": 3015
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 4.051282051282052e-05,
-      "loss": 1.6005,
+      "epoch": 13.46,
+      "learning_rate": 7.307142857142858e-05,
+      "loss": 1.1589,
       "step": 3016
     },
     {
-      "epoch": 8.93,
-      "learning_rate": 4.04930966469428e-05,
-      "loss": 1.5633,
+      "epoch": 13.47,
+      "learning_rate": 7.30625e-05,
+      "loss": 1.1107,
       "step": 3017
     },
     {
-      "epoch": 8.93,
-      "learning_rate": 4.047337278106509e-05,
-      "loss": 1.6101,
+      "epoch": 13.47,
+      "learning_rate": 7.305357142857143e-05,
+      "loss": 1.1937,
       "step": 3018
     },
     {
-      "epoch": 8.93,
-      "learning_rate": 4.045364891518738e-05,
-      "loss": 1.5105,
+      "epoch": 13.48,
+      "learning_rate": 7.304464285714286e-05,
+      "loss": 1.1928,
       "step": 3019
     },
     {
-      "epoch": 8.93,
-      "learning_rate": 4.0433925049309665e-05,
-      "loss": 1.6047,
+      "epoch": 13.48,
+      "learning_rate": 7.303571428571428e-05,
+      "loss": 1.2345,
       "step": 3020
     },
     {
-      "epoch": 8.94,
-      "learning_rate": 4.041420118343196e-05,
-      "loss": 1.4727,
+      "epoch": 13.49,
+      "learning_rate": 7.302678571428572e-05,
+      "loss": 1.2109,
       "step": 3021
     },
     {
-      "epoch": 8.94,
-      "learning_rate": 4.039447731755424e-05,
-      "loss": 1.6329,
+      "epoch": 13.49,
+      "learning_rate": 7.301785714285715e-05,
+      "loss": 1.1287,
       "step": 3022
     },
     {
-      "epoch": 8.94,
-      "learning_rate": 4.037475345167653e-05,
-      "loss": 1.5104,
+      "epoch": 13.5,
+      "learning_rate": 7.300892857142857e-05,
+      "loss": 1.2032,
       "step": 3023
     },
     {
-      "epoch": 8.95,
-      "learning_rate": 4.035502958579882e-05,
-      "loss": 1.5747,
+      "epoch": 13.5,
+      "learning_rate": 7.3e-05,
+      "loss": 1.1549,
       "step": 3024
     },
     {
-      "epoch": 8.95,
-      "learning_rate": 4.0335305719921105e-05,
-      "loss": 1.555,
+      "epoch": 13.5,
+      "learning_rate": 7.299107142857144e-05,
+      "loss": 1.2687,
       "step": 3025
     },
     {
-      "epoch": 8.95,
-      "learning_rate": 4.031558185404339e-05,
-      "loss": 1.5769,
+      "epoch": 13.51,
+      "learning_rate": 7.298214285714286e-05,
+      "loss": 1.1376,
       "step": 3026
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 4.029585798816568e-05,
-      "loss": 1.6514,
+      "epoch": 13.51,
+      "learning_rate": 7.297321428571429e-05,
+      "loss": 1.1821,
       "step": 3027
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 4.0276134122287975e-05,
-      "loss": 1.5405,
+      "epoch": 13.52,
+      "learning_rate": 7.296428571428571e-05,
+      "loss": 1.1951,
       "step": 3028
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 4.025641025641026e-05,
-      "loss": 1.5859,
+      "epoch": 13.52,
+      "learning_rate": 7.295535714285715e-05,
+      "loss": 1.1297,
       "step": 3029
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 4.0236686390532545e-05,
-      "loss": 1.5197,
+      "epoch": 13.53,
+      "learning_rate": 7.294642857142857e-05,
+      "loss": 1.1174,
       "step": 3030
     },
     {
-      "epoch": 8.97,
-      "learning_rate": 4.021696252465483e-05,
-      "loss": 1.5322,
+      "epoch": 13.53,
+      "learning_rate": 7.29375e-05,
+      "loss": 1.1638,
       "step": 3031
     },
     {
-      "epoch": 8.97,
-      "learning_rate": 4.019723865877712e-05,
-      "loss": 1.5638,
+      "epoch": 13.54,
+      "learning_rate": 7.292857142857143e-05,
+      "loss": 1.2135,
       "step": 3032
     },
     {
-      "epoch": 8.97,
-      "learning_rate": 4.0177514792899415e-05,
-      "loss": 1.5157,
+      "epoch": 13.54,
+      "learning_rate": 7.291964285714286e-05,
+      "loss": 1.1308,
       "step": 3033
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 4.01577909270217e-05,
-      "loss": 1.5597,
+      "epoch": 13.54,
+      "learning_rate": 7.29107142857143e-05,
+      "loss": 1.14,
       "step": 3034
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 4.0138067061143986e-05,
-      "loss": 1.5804,
+      "epoch": 13.55,
+      "learning_rate": 7.290178571428572e-05,
+      "loss": 1.1405,
       "step": 3035
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 4.011834319526627e-05,
-      "loss": 1.5735,
+      "epoch": 13.55,
+      "learning_rate": 7.289285714285715e-05,
+      "loss": 1.1156,
       "step": 3036
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 4.009861932938856e-05,
-      "loss": 1.6214,
+      "epoch": 13.56,
+      "learning_rate": 7.288392857142857e-05,
+      "loss": 1.1408,
       "step": 3037
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 4.007889546351085e-05,
-      "loss": 1.5735,
+      "epoch": 13.56,
+      "learning_rate": 7.287500000000001e-05,
+      "loss": 1.2613,
       "step": 3038
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 4.005917159763314e-05,
-      "loss": 1.5336,
+      "epoch": 13.57,
+      "learning_rate": 7.286607142857143e-05,
+      "loss": 1.1006,
       "step": 3039
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 4.0039447731755426e-05,
-      "loss": 1.4175,
+      "epoch": 13.57,
+      "learning_rate": 7.285714285714286e-05,
+      "loss": 1.1612,
       "step": 3040
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 4.001972386587771e-05,
-      "loss": 1.6547,
+      "epoch": 13.58,
+      "learning_rate": 7.284821428571429e-05,
+      "loss": 1.1356,
       "step": 3041
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 4e-05,
-      "loss": 1.5879,
+      "epoch": 13.58,
+      "learning_rate": 7.283928571428572e-05,
+      "loss": 1.0825,
       "step": 3042
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 3.998027613412229e-05,
-      "loss": 1.6097,
+      "epoch": 13.58,
+      "learning_rate": 7.283035714285714e-05,
+      "loss": 1.1682,
       "step": 3043
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 3.996055226824458e-05,
-      "loss": 1.7091,
+      "epoch": 13.59,
+      "learning_rate": 7.282142857142858e-05,
+      "loss": 1.1949,
       "step": 3044
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 3.9940828402366866e-05,
-      "loss": 1.5003,
+      "epoch": 13.59,
+      "learning_rate": 7.28125e-05,
+      "loss": 1.1096,
       "step": 3045
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 3.992110453648915e-05,
-      "loss": 1.4911,
+      "epoch": 13.6,
+      "learning_rate": 7.280357142857143e-05,
+      "loss": 1.1048,
       "step": 3046
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 3.9901380670611444e-05,
-      "loss": 1.5997,
+      "epoch": 13.6,
+      "learning_rate": 7.279464285714285e-05,
+      "loss": 1.1999,
       "step": 3047
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 3.988165680473373e-05,
-      "loss": 1.5976,
+      "epoch": 13.61,
+      "learning_rate": 7.278571428571429e-05,
+      "loss": 1.1371,
       "step": 3048
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 3.9861932938856014e-05,
-      "loss": 1.5035,
+      "epoch": 13.61,
+      "learning_rate": 7.277678571428572e-05,
+      "loss": 1.2282,
       "step": 3049
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 3.9842209072978306e-05,
-      "loss": 1.5185,
+      "epoch": 13.62,
+      "learning_rate": 7.276785714285714e-05,
+      "loss": 1.1711,
       "step": 3050
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 3.98224852071006e-05,
-      "loss": 1.6087,
+      "epoch": 13.62,
+      "learning_rate": 7.275892857142858e-05,
+      "loss": 1.0476,
       "step": 3051
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 3.9802761341222884e-05,
-      "loss": 1.4409,
+      "epoch": 13.62,
+      "learning_rate": 7.275e-05,
+      "loss": 1.1566,
       "step": 3052
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 3.978303747534517e-05,
-      "loss": 1.4535,
+      "epoch": 13.63,
+      "learning_rate": 7.274107142857143e-05,
+      "loss": 1.1756,
       "step": 3053
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.9763313609467454e-05,
-      "loss": 1.5369,
+      "epoch": 13.63,
+      "learning_rate": 7.273214285714286e-05,
+      "loss": 1.1242,
       "step": 3054
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.974358974358974e-05,
-      "loss": 1.5544,
+      "epoch": 13.64,
+      "learning_rate": 7.27232142857143e-05,
+      "loss": 1.1387,
       "step": 3055
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.972386587771204e-05,
-      "loss": 1.5959,
+      "epoch": 13.64,
+      "learning_rate": 7.271428571428571e-05,
+      "loss": 1.1579,
       "step": 3056
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.9704142011834324e-05,
-      "loss": 1.5999,
+      "epoch": 13.65,
+      "learning_rate": 7.270535714285715e-05,
+      "loss": 1.2386,
       "step": 3057
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 3.968441814595661e-05,
-      "loss": 1.5381,
+      "epoch": 13.65,
+      "learning_rate": 7.269642857142857e-05,
+      "loss": 1.0795,
       "step": 3058
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 3.9664694280078895e-05,
-      "loss": 1.5741,
+      "epoch": 13.66,
+      "learning_rate": 7.268750000000001e-05,
+      "loss": 1.1694,
       "step": 3059
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 3.964497041420119e-05,
-      "loss": 1.5688,
+      "epoch": 13.66,
+      "learning_rate": 7.267857142857142e-05,
+      "loss": 1.1025,
       "step": 3060
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 3.962524654832347e-05,
-      "loss": 1.4473,
+      "epoch": 13.67,
+      "learning_rate": 7.266964285714286e-05,
+      "loss": 1.058,
       "step": 3061
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 3.9605522682445764e-05,
-      "loss": 1.5307,
+      "epoch": 13.67,
+      "learning_rate": 7.266071428571429e-05,
+      "loss": 1.1718,
       "step": 3062
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 3.958579881656805e-05,
-      "loss": 1.6018,
+      "epoch": 13.67,
+      "learning_rate": 7.265178571428571e-05,
+      "loss": 1.1151,
       "step": 3063
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.9566074950690335e-05,
-      "loss": 1.4977,
+      "epoch": 13.68,
+      "learning_rate": 7.264285714285715e-05,
+      "loss": 1.1788,
       "step": 3064
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.954635108481263e-05,
-      "loss": 1.4538,
+      "epoch": 13.68,
+      "learning_rate": 7.263392857142858e-05,
+      "loss": 1.1643,
       "step": 3065
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.952662721893491e-05,
-      "loss": 1.5896,
+      "epoch": 13.69,
+      "learning_rate": 7.2625e-05,
+      "loss": 1.1477,
       "step": 3066
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.95069033530572e-05,
-      "loss": 1.5537,
+      "epoch": 13.69,
+      "learning_rate": 7.261607142857143e-05,
+      "loss": 1.1586,
       "step": 3067
     },
     {
-      "epoch": 9.08,
-      "learning_rate": 3.948717948717949e-05,
-      "loss": 1.5163,
+      "epoch": 13.7,
+      "learning_rate": 7.260714285714287e-05,
+      "loss": 1.084,
       "step": 3068
     },
     {
-      "epoch": 9.08,
-      "learning_rate": 3.9467455621301775e-05,
-      "loss": 1.5602,
+      "epoch": 13.7,
+      "learning_rate": 7.25982142857143e-05,
+      "loss": 1.1631,
       "step": 3069
     },
     {
-      "epoch": 9.08,
-      "learning_rate": 3.944773175542407e-05,
-      "loss": 1.5689,
+      "epoch": 13.71,
+      "learning_rate": 7.258928571428572e-05,
+      "loss": 1.171,
       "step": 3070
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.942800788954635e-05,
-      "loss": 1.6069,
+      "epoch": 13.71,
+      "learning_rate": 7.258035714285714e-05,
+      "loss": 1.1874,
       "step": 3071
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.940828402366864e-05,
-      "loss": 1.5643,
+      "epoch": 13.71,
+      "learning_rate": 7.257142857142858e-05,
+      "loss": 1.055,
       "step": 3072
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.938856015779093e-05,
-      "loss": 1.4874,
+      "epoch": 13.72,
+      "learning_rate": 7.25625e-05,
+      "loss": 1.2185,
       "step": 3073
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.936883629191322e-05,
-      "loss": 1.5542,
+      "epoch": 13.72,
+      "learning_rate": 7.255357142857143e-05,
+      "loss": 1.2246,
       "step": 3074
     },
     {
-      "epoch": 9.1,
-      "learning_rate": 3.934911242603551e-05,
-      "loss": 1.5281,
+      "epoch": 13.73,
+      "learning_rate": 7.254464285714286e-05,
+      "loss": 1.1856,
       "step": 3075
     },
     {
-      "epoch": 9.1,
-      "learning_rate": 3.932938856015779e-05,
-      "loss": 1.4737,
+      "epoch": 13.73,
+      "learning_rate": 7.253571428571429e-05,
+      "loss": 1.2168,
       "step": 3076
     },
     {
-      "epoch": 9.1,
-      "learning_rate": 3.930966469428008e-05,
-      "loss": 1.62,
+      "epoch": 13.74,
+      "learning_rate": 7.252678571428571e-05,
+      "loss": 1.1904,
       "step": 3077
     },
     {
-      "epoch": 9.11,
-      "learning_rate": 3.9289940828402364e-05,
-      "loss": 1.5474,
+      "epoch": 13.74,
+      "learning_rate": 7.251785714285715e-05,
+      "loss": 1.1516,
       "step": 3078
     },
     {
-      "epoch": 9.11,
-      "learning_rate": 3.9270216962524656e-05,
-      "loss": 1.5544,
+      "epoch": 13.75,
+      "learning_rate": 7.250892857142857e-05,
+      "loss": 1.1592,
       "step": 3079
     },
     {
-      "epoch": 9.11,
-      "learning_rate": 3.925049309664695e-05,
-      "loss": 1.5887,
+      "epoch": 13.75,
+      "learning_rate": 7.25e-05,
+      "loss": 1.0985,
       "step": 3080
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 3.923076923076923e-05,
-      "loss": 1.548,
+      "epoch": 13.75,
+      "learning_rate": 7.249107142857144e-05,
+      "loss": 1.1826,
       "step": 3081
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 3.921104536489152e-05,
-      "loss": 1.5453,
+      "epoch": 13.76,
+      "learning_rate": 7.248214285714286e-05,
+      "loss": 1.0849,
       "step": 3082
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 3.919132149901381e-05,
-      "loss": 1.6304,
+      "epoch": 13.76,
+      "learning_rate": 7.247321428571429e-05,
+      "loss": 1.1749,
       "step": 3083
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 3.9171597633136096e-05,
-      "loss": 1.5454,
+      "epoch": 13.77,
+      "learning_rate": 7.246428571428572e-05,
+      "loss": 1.2166,
       "step": 3084
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 3.915187376725839e-05,
-      "loss": 1.545,
+      "epoch": 13.77,
+      "learning_rate": 7.245535714285715e-05,
+      "loss": 1.145,
       "step": 3085
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 3.913214990138067e-05,
-      "loss": 1.5868,
+      "epoch": 13.78,
+      "learning_rate": 7.244642857142857e-05,
+      "loss": 1.245,
       "step": 3086
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 3.911242603550296e-05,
-      "loss": 1.6242,
+      "epoch": 13.78,
+      "learning_rate": 7.24375e-05,
+      "loss": 1.1685,
       "step": 3087
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 3.909270216962525e-05,
-      "loss": 1.5425,
+      "epoch": 13.79,
+      "learning_rate": 7.242857142857143e-05,
+      "loss": 1.0747,
       "step": 3088
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 3.9072978303747536e-05,
-      "loss": 1.5394,
+      "epoch": 13.79,
+      "learning_rate": 7.241964285714286e-05,
+      "loss": 1.1387,
       "step": 3089
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 3.905325443786982e-05,
-      "loss": 1.5882,
+      "epoch": 13.79,
+      "learning_rate": 7.241071428571428e-05,
+      "loss": 1.1336,
       "step": 3090
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 3.9033530571992114e-05,
-      "loss": 1.6046,
+      "epoch": 13.8,
+      "learning_rate": 7.240178571428572e-05,
+      "loss": 1.1536,
       "step": 3091
     },
     {
-      "epoch": 9.15,
-      "learning_rate": 3.9013806706114406e-05,
-      "loss": 1.58,
+      "epoch": 13.8,
+      "learning_rate": 7.239285714285715e-05,
+      "loss": 1.1561,
       "step": 3092
     },
     {
-      "epoch": 9.15,
-      "learning_rate": 3.899408284023669e-05,
-      "loss": 1.5476,
+      "epoch": 13.81,
+      "learning_rate": 7.238392857142857e-05,
+      "loss": 1.1552,
       "step": 3093
     },
     {
-      "epoch": 9.15,
-      "learning_rate": 3.8974358974358976e-05,
-      "loss": 1.5051,
+      "epoch": 13.81,
+      "learning_rate": 7.2375e-05,
+      "loss": 1.1396,
       "step": 3094
     },
     {
-      "epoch": 9.16,
-      "learning_rate": 3.895463510848126e-05,
-      "loss": 1.5057,
+      "epoch": 13.82,
+      "learning_rate": 7.236607142857144e-05,
+      "loss": 1.2177,
       "step": 3095
     },
     {
-      "epoch": 9.16,
-      "learning_rate": 3.893491124260355e-05,
-      "loss": 1.575,
+      "epoch": 13.82,
+      "learning_rate": 7.235714285714286e-05,
+      "loss": 1.0957,
       "step": 3096
     },
     {
-      "epoch": 9.16,
-      "learning_rate": 3.891518737672584e-05,
-      "loss": 1.5495,
+      "epoch": 13.83,
+      "learning_rate": 7.234821428571429e-05,
+      "loss": 1.1436,
       "step": 3097
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 3.889546351084813e-05,
-      "loss": 1.4767,
+      "epoch": 13.83,
+      "learning_rate": 7.233928571428573e-05,
+      "loss": 1.1768,
       "step": 3098
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 3.8875739644970417e-05,
-      "loss": 1.5655,
+      "epoch": 13.83,
+      "learning_rate": 7.233035714285714e-05,
+      "loss": 1.1812,
       "step": 3099
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 3.88560157790927e-05,
-      "loss": 1.5465,
+      "epoch": 13.84,
+      "learning_rate": 7.232142857142858e-05,
+      "loss": 1.0703,
       "step": 3100
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 3.883629191321499e-05,
-      "loss": 1.6056,
+      "epoch": 13.84,
+      "learning_rate": 7.23125e-05,
+      "loss": 1.1676,
       "step": 3101
     },
     {
-      "epoch": 9.18,
-      "learning_rate": 3.881656804733728e-05,
-      "loss": 1.4579,
+      "epoch": 13.85,
+      "learning_rate": 7.230357142857144e-05,
+      "loss": 1.1056,
       "step": 3102
     },
     {
-      "epoch": 9.18,
-      "learning_rate": 3.879684418145957e-05,
-      "loss": 1.6231,
+      "epoch": 13.85,
+      "learning_rate": 7.229464285714285e-05,
+      "loss": 1.1182,
       "step": 3103
     },
     {
-      "epoch": 9.18,
-      "learning_rate": 3.877712031558186e-05,
-      "loss": 1.5228,
+      "epoch": 13.86,
+      "learning_rate": 7.228571428571429e-05,
+      "loss": 1.1876,
       "step": 3104
     },
     {
-      "epoch": 9.19,
-      "learning_rate": 3.875739644970414e-05,
-      "loss": 1.5912,
+      "epoch": 13.86,
+      "learning_rate": 7.227678571428572e-05,
+      "loss": 1.1658,
       "step": 3105
     },
     {
-      "epoch": 9.19,
-      "learning_rate": 3.8737672583826434e-05,
-      "loss": 1.572,
+      "epoch": 13.87,
+      "learning_rate": 7.226785714285714e-05,
+      "loss": 1.213,
       "step": 3106
     },
     {
-      "epoch": 9.19,
-      "learning_rate": 3.871794871794872e-05,
-      "loss": 1.5337,
+      "epoch": 13.87,
+      "learning_rate": 7.225892857142857e-05,
+      "loss": 1.2001,
       "step": 3107
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 3.8698224852071005e-05,
-      "loss": 1.4812,
+      "epoch": 13.88,
+      "learning_rate": 7.225000000000001e-05,
+      "loss": 1.151,
       "step": 3108
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 3.86785009861933e-05,
-      "loss": 1.5358,
+      "epoch": 13.88,
+      "learning_rate": 7.224107142857143e-05,
+      "loss": 1.165,
       "step": 3109
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 3.865877712031558e-05,
-      "loss": 1.5502,
+      "epoch": 13.88,
+      "learning_rate": 7.223214285714286e-05,
+      "loss": 1.2017,
       "step": 3110
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 3.8639053254437874e-05,
-      "loss": 1.6087,
+      "epoch": 13.89,
+      "learning_rate": 7.22232142857143e-05,
+      "loss": 1.2006,
       "step": 3111
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 3.861932938856016e-05,
-      "loss": 1.5066,
+      "epoch": 13.89,
+      "learning_rate": 7.221428571428572e-05,
+      "loss": 1.2099,
       "step": 3112
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 3.8599605522682445e-05,
-      "loss": 1.5532,
+      "epoch": 13.9,
+      "learning_rate": 7.220535714285715e-05,
+      "loss": 1.1721,
       "step": 3113
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 3.857988165680473e-05,
-      "loss": 1.6146,
+      "epoch": 13.9,
+      "learning_rate": 7.219642857142857e-05,
+      "loss": 1.1056,
       "step": 3114
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 3.856015779092703e-05,
-      "loss": 1.4853,
+      "epoch": 13.91,
+      "learning_rate": 7.218750000000001e-05,
+      "loss": 1.1577,
       "step": 3115
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 3.8540433925049315e-05,
-      "loss": 1.4852,
+      "epoch": 13.91,
+      "learning_rate": 7.217857142857143e-05,
+      "loss": 1.1334,
       "step": 3116
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 3.85207100591716e-05,
-      "loss": 1.6478,
+      "epoch": 13.92,
+      "learning_rate": 7.216964285714286e-05,
+      "loss": 1.1121,
       "step": 3117
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 3.8500986193293885e-05,
-      "loss": 1.5359,
+      "epoch": 13.92,
+      "learning_rate": 7.216071428571429e-05,
+      "loss": 1.1562,
       "step": 3118
     },
     {
-      "epoch": 9.23,
-      "learning_rate": 3.848126232741617e-05,
-      "loss": 1.5902,
+      "epoch": 13.92,
+      "learning_rate": 7.215178571428571e-05,
+      "loss": 1.1121,
       "step": 3119
     },
     {
-      "epoch": 9.23,
-      "learning_rate": 3.846153846153846e-05,
-      "loss": 1.5082,
+      "epoch": 13.93,
+      "learning_rate": 7.214285714285714e-05,
+      "loss": 1.1052,
       "step": 3120
     },
     {
-      "epoch": 9.23,
-      "learning_rate": 3.8441814595660755e-05,
-      "loss": 1.4772,
+      "epoch": 13.93,
+      "learning_rate": 7.213392857142858e-05,
+      "loss": 1.2355,
       "step": 3121
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 3.842209072978304e-05,
-      "loss": 1.5694,
+      "epoch": 13.94,
+      "learning_rate": 7.2125e-05,
+      "loss": 1.1486,
       "step": 3122
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 3.8402366863905326e-05,
-      "loss": 1.6885,
+      "epoch": 13.94,
+      "learning_rate": 7.211607142857143e-05,
+      "loss": 1.1096,
       "step": 3123
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 3.838264299802761e-05,
-      "loss": 1.6287,
+      "epoch": 13.95,
+      "learning_rate": 7.210714285714286e-05,
+      "loss": 1.1834,
       "step": 3124
     },
     {
-      "epoch": 9.25,
-      "learning_rate": 3.83629191321499e-05,
-      "loss": 1.4981,
+      "epoch": 13.95,
+      "learning_rate": 7.20982142857143e-05,
+      "loss": 1.1016,
       "step": 3125
     },
     {
-      "epoch": 9.25,
-      "learning_rate": 3.834319526627219e-05,
-      "loss": 1.5669,
+      "epoch": 13.96,
+      "learning_rate": 7.208928571428572e-05,
+      "loss": 1.1274,
       "step": 3126
     },
     {
-      "epoch": 9.25,
-      "learning_rate": 3.832347140039448e-05,
-      "loss": 1.603,
+      "epoch": 13.96,
+      "learning_rate": 7.208035714285715e-05,
+      "loss": 1.162,
       "step": 3127
     },
     {
-      "epoch": 9.25,
-      "learning_rate": 3.8303747534516766e-05,
-      "loss": 1.6242,
+      "epoch": 13.96,
+      "learning_rate": 7.207142857142858e-05,
+      "loss": 1.1272,
       "step": 3128
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 3.828402366863906e-05,
-      "loss": 1.5587,
+      "epoch": 13.97,
+      "learning_rate": 7.20625e-05,
+      "loss": 1.191,
       "step": 3129
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 3.826429980276134e-05,
-      "loss": 1.6115,
+      "epoch": 13.97,
+      "learning_rate": 7.205357142857144e-05,
+      "loss": 1.1608,
       "step": 3130
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 3.824457593688363e-05,
-      "loss": 1.4985,
+      "epoch": 13.98,
+      "learning_rate": 7.204464285714286e-05,
+      "loss": 1.0922,
       "step": 3131
     },
     {
-      "epoch": 9.27,
-      "learning_rate": 3.822485207100592e-05,
-      "loss": 1.6299,
+      "epoch": 13.98,
+      "learning_rate": 7.203571428571429e-05,
+      "loss": 1.1694,
       "step": 3132
     },
     {
-      "epoch": 9.27,
-      "learning_rate": 3.8205128205128206e-05,
-      "loss": 1.5685,
+      "epoch": 13.99,
+      "learning_rate": 7.202678571428571e-05,
+      "loss": 1.1137,
       "step": 3133
     },
     {
-      "epoch": 9.27,
-      "learning_rate": 3.81854043392505e-05,
-      "loss": 1.5173,
+      "epoch": 13.99,
+      "learning_rate": 7.201785714285715e-05,
+      "loss": 1.106,
       "step": 3134
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 3.8165680473372784e-05,
-      "loss": 1.532,
+      "epoch": 14.0,
+      "learning_rate": 7.200892857142858e-05,
+      "loss": 1.203,
       "step": 3135
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 3.814595660749507e-05,
-      "loss": 1.4838,
+      "epoch": 14.0,
+      "learning_rate": 7.2e-05,
+      "loss": 1.0873,
       "step": 3136
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 3.8126232741617354e-05,
-      "loss": 1.5372,
+      "epoch": 14.0,
+      "learning_rate": 7.199107142857143e-05,
+      "loss": 1.1553,
       "step": 3137
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 3.8106508875739646e-05,
-      "loss": 1.6158,
+      "epoch": 14.01,
+      "learning_rate": 7.198214285714287e-05,
+      "loss": 1.2125,
       "step": 3138
     },
     {
-      "epoch": 9.29,
-      "learning_rate": 3.808678500986194e-05,
-      "loss": 1.5015,
+      "epoch": 14.01,
+      "learning_rate": 7.197321428571429e-05,
+      "loss": 1.1474,
       "step": 3139
     },
     {
-      "epoch": 9.29,
-      "learning_rate": 3.8067061143984224e-05,
-      "loss": 1.5996,
+      "epoch": 14.02,
+      "learning_rate": 7.196428571428572e-05,
+      "loss": 1.2048,
       "step": 3140
     },
     {
-      "epoch": 9.29,
-      "learning_rate": 3.804733727810651e-05,
-      "loss": 1.5879,
+      "epoch": 14.02,
+      "learning_rate": 7.195535714285714e-05,
+      "loss": 1.161,
       "step": 3141
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 3.8027613412228794e-05,
-      "loss": 1.5172,
+      "epoch": 14.03,
+      "learning_rate": 7.194642857142857e-05,
+      "loss": 1.0776,
       "step": 3142
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 3.8007889546351087e-05,
-      "loss": 1.4851,
+      "epoch": 14.03,
+      "learning_rate": 7.193750000000001e-05,
+      "loss": 1.1194,
       "step": 3143
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 3.798816568047338e-05,
-      "loss": 1.4757,
+      "epoch": 14.04,
+      "learning_rate": 7.192857142857143e-05,
+      "loss": 1.177,
       "step": 3144
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 3.7968441814595664e-05,
-      "loss": 1.4725,
+      "epoch": 14.04,
+      "learning_rate": 7.191964285714287e-05,
+      "loss": 1.1168,
       "step": 3145
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 3.794871794871795e-05,
-      "loss": 1.4971,
+      "epoch": 14.04,
+      "learning_rate": 7.191071428571428e-05,
+      "loss": 1.1562,
       "step": 3146
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 3.792899408284024e-05,
-      "loss": 1.5986,
+      "epoch": 14.05,
+      "learning_rate": 7.190178571428572e-05,
+      "loss": 1.0996,
       "step": 3147
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 3.790927021696253e-05,
-      "loss": 1.5853,
+      "epoch": 14.05,
+      "learning_rate": 7.189285714285715e-05,
+      "loss": 1.1778,
       "step": 3148
     },
     {
-      "epoch": 9.32,
-      "learning_rate": 3.788954635108481e-05,
-      "loss": 1.5863,
+      "epoch": 14.06,
+      "learning_rate": 7.188392857142857e-05,
+      "loss": 1.0177,
       "step": 3149
     },
     {
-      "epoch": 9.32,
-      "learning_rate": 3.7869822485207104e-05,
-      "loss": 1.3966,
+      "epoch": 14.06,
+      "learning_rate": 7.1875e-05,
+      "loss": 1.0527,
       "step": 3150
     },
     {
-      "epoch": 9.32,
-      "learning_rate": 3.785009861932939e-05,
-      "loss": 1.5716,
+      "epoch": 14.07,
+      "learning_rate": 7.186607142857144e-05,
+      "loss": 1.1036,
       "step": 3151
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 3.783037475345168e-05,
-      "loss": 1.6478,
+      "epoch": 14.07,
+      "learning_rate": 7.185714285714285e-05,
+      "loss": 1.0797,
       "step": 3152
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 3.781065088757397e-05,
-      "loss": 1.633,
+      "epoch": 14.08,
+      "learning_rate": 7.184821428571429e-05,
+      "loss": 1.0822,
       "step": 3153
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 3.779092702169625e-05,
-      "loss": 1.5774,
+      "epoch": 14.08,
+      "learning_rate": 7.183928571428571e-05,
+      "loss": 1.1122,
       "step": 3154
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 3.777120315581854e-05,
-      "loss": 1.5661,
+      "epoch": 14.08,
+      "learning_rate": 7.183035714285714e-05,
+      "loss": 1.1959,
       "step": 3155
     },
     {
-      "epoch": 9.34,
-      "learning_rate": 3.775147928994083e-05,
-      "loss": 1.4983,
+      "epoch": 14.09,
+      "learning_rate": 7.182142857142858e-05,
+      "loss": 1.124,
       "step": 3156
     },
     {
-      "epoch": 9.34,
-      "learning_rate": 3.773175542406312e-05,
-      "loss": 1.6791,
+      "epoch": 14.09,
+      "learning_rate": 7.18125e-05,
+      "loss": 1.1061,
       "step": 3157
     },
     {
-      "epoch": 9.34,
-      "learning_rate": 3.771203155818541e-05,
-      "loss": 1.5948,
+      "epoch": 14.1,
+      "learning_rate": 7.180357142857144e-05,
+      "loss": 1.092,
       "step": 3158
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 3.769230769230769e-05,
-      "loss": 1.5055,
+      "epoch": 14.1,
+      "learning_rate": 7.179464285714285e-05,
+      "loss": 1.1068,
       "step": 3159
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 3.767258382642998e-05,
-      "loss": 1.4671,
+      "epoch": 14.11,
+      "learning_rate": 7.17857142857143e-05,
+      "loss": 1.1058,
       "step": 3160
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 3.765285996055227e-05,
-      "loss": 1.5678,
+      "epoch": 14.11,
+      "learning_rate": 7.177678571428572e-05,
+      "loss": 1.1944,
       "step": 3161
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 3.763313609467456e-05,
-      "loss": 1.4923,
+      "epoch": 14.12,
+      "learning_rate": 7.176785714285714e-05,
+      "loss": 1.131,
       "step": 3162
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 3.761341222879685e-05,
-      "loss": 1.6175,
+      "epoch": 14.12,
+      "learning_rate": 7.175892857142857e-05,
+      "loss": 1.1801,
       "step": 3163
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 3.759368836291913e-05,
-      "loss": 1.4966,
+      "epoch": 14.12,
+      "learning_rate": 7.175000000000001e-05,
+      "loss": 1.1439,
       "step": 3164
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 3.757396449704142e-05,
-      "loss": 1.4948,
+      "epoch": 14.13,
+      "learning_rate": 7.174107142857142e-05,
+      "loss": 1.1316,
       "step": 3165
     },
     {
-      "epoch": 9.37,
-      "learning_rate": 3.755424063116371e-05,
-      "loss": 1.6119,
+      "epoch": 14.13,
+      "learning_rate": 7.173214285714286e-05,
+      "loss": 1.1458,
       "step": 3166
     },
     {
-      "epoch": 9.37,
-      "learning_rate": 3.7534516765285996e-05,
-      "loss": 1.5598,
+      "epoch": 14.14,
+      "learning_rate": 7.172321428571429e-05,
+      "loss": 1.1851,
       "step": 3167
     },
     {
-      "epoch": 9.37,
-      "learning_rate": 3.751479289940829e-05,
-      "loss": 1.577,
+      "epoch": 14.14,
+      "learning_rate": 7.171428571428572e-05,
+      "loss": 1.1827,
       "step": 3168
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 3.749506903353057e-05,
-      "loss": 1.5998,
+      "epoch": 14.15,
+      "learning_rate": 7.170535714285715e-05,
+      "loss": 1.1787,
       "step": 3169
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 3.7475345167652865e-05,
-      "loss": 1.5736,
+      "epoch": 14.15,
+      "learning_rate": 7.169642857142858e-05,
+      "loss": 1.2166,
       "step": 3170
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 3.745562130177515e-05,
-      "loss": 1.5183,
+      "epoch": 14.16,
+      "learning_rate": 7.16875e-05,
+      "loss": 1.0987,
       "step": 3171
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 3.7435897435897436e-05,
-      "loss": 1.5407,
+      "epoch": 14.16,
+      "learning_rate": 7.167857142857143e-05,
+      "loss": 1.1339,
       "step": 3172
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 3.741617357001973e-05,
-      "loss": 1.5436,
+      "epoch": 14.17,
+      "learning_rate": 7.166964285714287e-05,
+      "loss": 1.0916,
       "step": 3173
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 3.739644970414201e-05,
-      "loss": 1.5078,
+      "epoch": 14.17,
+      "learning_rate": 7.166071428571429e-05,
+      "loss": 1.1952,
       "step": 3174
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 3.7376725838264305e-05,
-      "loss": 1.5411,
+      "epoch": 14.17,
+      "learning_rate": 7.165178571428572e-05,
+      "loss": 1.1849,
       "step": 3175
     },
     {
-      "epoch": 9.4,
-      "learning_rate": 3.735700197238659e-05,
-      "loss": 1.5082,
+      "epoch": 14.18,
+      "learning_rate": 7.164285714285714e-05,
+      "loss": 1.1683,
       "step": 3176
     },
     {
-      "epoch": 9.4,
-      "learning_rate": 3.7337278106508876e-05,
-      "loss": 1.5585,
+      "epoch": 14.18,
+      "learning_rate": 7.163392857142858e-05,
+      "loss": 1.0978,
       "step": 3177
     },
     {
-      "epoch": 9.4,
-      "learning_rate": 3.731755424063116e-05,
-      "loss": 1.5902,
+      "epoch": 14.19,
+      "learning_rate": 7.1625e-05,
+      "loss": 1.1284,
       "step": 3178
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 3.7297830374753454e-05,
-      "loss": 1.596,
+      "epoch": 14.19,
+      "learning_rate": 7.161607142857143e-05,
+      "loss": 1.0351,
       "step": 3179
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 3.7278106508875746e-05,
-      "loss": 1.422,
+      "epoch": 14.2,
+      "learning_rate": 7.160714285714286e-05,
+      "loss": 1.1238,
       "step": 3180
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 3.725838264299803e-05,
-      "loss": 1.5789,
+      "epoch": 14.2,
+      "learning_rate": 7.15982142857143e-05,
+      "loss": 1.2038,
       "step": 3181
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 3.7238658777120316e-05,
-      "loss": 1.6676,
+      "epoch": 14.21,
+      "learning_rate": 7.158928571428571e-05,
+      "loss": 1.1063,
       "step": 3182
     },
     {
-      "epoch": 9.42,
-      "learning_rate": 3.72189349112426e-05,
-      "loss": 1.5298,
+      "epoch": 14.21,
+      "learning_rate": 7.158035714285715e-05,
+      "loss": 1.2198,
       "step": 3183
     },
     {
-      "epoch": 9.42,
-      "learning_rate": 3.7199211045364894e-05,
-      "loss": 1.4417,
+      "epoch": 14.21,
+      "learning_rate": 7.157142857142857e-05,
+      "loss": 1.1392,
       "step": 3184
     },
     {
-      "epoch": 9.42,
-      "learning_rate": 3.717948717948718e-05,
-      "loss": 1.5404,
+      "epoch": 14.22,
+      "learning_rate": 7.15625e-05,
+      "loss": 1.1222,
       "step": 3185
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 3.715976331360947e-05,
-      "loss": 1.5469,
+      "epoch": 14.22,
+      "learning_rate": 7.155357142857144e-05,
+      "loss": 1.1693,
       "step": 3186
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 3.714003944773176e-05,
-      "loss": 1.552,
+      "epoch": 14.23,
+      "learning_rate": 7.154464285714286e-05,
+      "loss": 1.192,
       "step": 3187
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 3.712031558185404e-05,
-      "loss": 1.5144,
+      "epoch": 14.23,
+      "learning_rate": 7.153571428571429e-05,
+      "loss": 1.1931,
       "step": 3188
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 3.7100591715976334e-05,
-      "loss": 1.5324,
+      "epoch": 14.24,
+      "learning_rate": 7.152678571428571e-05,
+      "loss": 1.1133,
       "step": 3189
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 3.708086785009862e-05,
-      "loss": 1.5837,
+      "epoch": 14.24,
+      "learning_rate": 7.151785714285715e-05,
+      "loss": 1.1048,
       "step": 3190
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 3.706114398422091e-05,
-      "loss": 1.5437,
+      "epoch": 14.25,
+      "learning_rate": 7.150892857142858e-05,
+      "loss": 1.1107,
       "step": 3191
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 3.70414201183432e-05,
-      "loss": 1.5105,
+      "epoch": 14.25,
+      "learning_rate": 7.15e-05,
+      "loss": 1.2682,
       "step": 3192
     },
     {
-      "epoch": 9.45,
-      "learning_rate": 3.702169625246549e-05,
-      "loss": 1.559,
+      "epoch": 14.25,
+      "learning_rate": 7.149107142857143e-05,
+      "loss": 1.1762,
       "step": 3193
     },
     {
-      "epoch": 9.45,
-      "learning_rate": 3.7001972386587774e-05,
-      "loss": 1.5477,
+      "epoch": 14.26,
+      "learning_rate": 7.148214285714287e-05,
+      "loss": 1.0977,
       "step": 3194
     },
     {
-      "epoch": 9.45,
-      "learning_rate": 3.698224852071006e-05,
-      "loss": 1.4668,
+      "epoch": 14.26,
+      "learning_rate": 7.147321428571428e-05,
+      "loss": 1.0981,
       "step": 3195
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 3.6962524654832345e-05,
-      "loss": 1.4858,
+      "epoch": 14.27,
+      "learning_rate": 7.146428571428572e-05,
+      "loss": 1.1697,
       "step": 3196
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 3.694280078895464e-05,
-      "loss": 1.5414,
+      "epoch": 14.27,
+      "learning_rate": 7.145535714285714e-05,
+      "loss": 1.1109,
       "step": 3197
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 3.692307692307693e-05,
-      "loss": 1.504,
+      "epoch": 14.28,
+      "learning_rate": 7.144642857142857e-05,
+      "loss": 1.1472,
       "step": 3198
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 3.6903353057199215e-05,
-      "loss": 1.5778,
+      "epoch": 14.28,
+      "learning_rate": 7.14375e-05,
+      "loss": 1.1442,
       "step": 3199
     },
     {
-      "epoch": 9.47,
-      "learning_rate": 3.68836291913215e-05,
-      "loss": 1.4725,
+      "epoch": 14.29,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.1277,
       "step": 3200
     },
     {
-      "epoch": 9.47,
-      "learning_rate": 3.6863905325443785e-05,
-      "loss": 1.5154,
+      "epoch": 14.29,
+      "learning_rate": 7.141964285714286e-05,
+      "loss": 1.0548,
       "step": 3201
     },
     {
-      "epoch": 9.47,
-      "learning_rate": 3.684418145956608e-05,
-      "loss": 1.5659,
+      "epoch": 14.29,
+      "learning_rate": 7.141071428571428e-05,
+      "loss": 1.1515,
       "step": 3202
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 3.682445759368837e-05,
-      "loss": 1.6065,
+      "epoch": 14.3,
+      "learning_rate": 7.140178571428572e-05,
+      "loss": 1.1378,
       "step": 3203
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 3.6804733727810655e-05,
-      "loss": 1.5441,
+      "epoch": 14.3,
+      "learning_rate": 7.139285714285715e-05,
+      "loss": 1.0796,
       "step": 3204
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 3.678500986193294e-05,
-      "loss": 1.4903,
+      "epoch": 14.31,
+      "learning_rate": 7.138392857142857e-05,
+      "loss": 1.1106,
       "step": 3205
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 3.6765285996055225e-05,
-      "loss": 1.6062,
+      "epoch": 14.31,
+      "learning_rate": 7.1375e-05,
+      "loss": 1.0887,
       "step": 3206
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 3.674556213017752e-05,
-      "loss": 1.4839,
+      "epoch": 14.32,
+      "learning_rate": 7.136607142857144e-05,
+      "loss": 1.1385,
       "step": 3207
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 3.67258382642998e-05,
-      "loss": 1.5547,
+      "epoch": 14.32,
+      "learning_rate": 7.135714285714285e-05,
+      "loss": 1.1934,
       "step": 3208
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 3.6706114398422095e-05,
-      "loss": 1.4894,
+      "epoch": 14.33,
+      "learning_rate": 7.134821428571429e-05,
+      "loss": 1.2234,
       "step": 3209
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 3.668639053254438e-05,
-      "loss": 1.5241,
+      "epoch": 14.33,
+      "learning_rate": 7.133928571428572e-05,
+      "loss": 1.1322,
       "step": 3210
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 3.6666666666666666e-05,
-      "loss": 1.5718,
+      "epoch": 14.33,
+      "learning_rate": 7.133035714285715e-05,
+      "loss": 1.0992,
       "step": 3211
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 3.664694280078896e-05,
-      "loss": 1.4922,
+      "epoch": 14.34,
+      "learning_rate": 7.132142857142857e-05,
+      "loss": 1.0573,
       "step": 3212
     },
     {
-      "epoch": 9.51,
-      "learning_rate": 3.662721893491124e-05,
-      "loss": 1.4938,
+      "epoch": 14.34,
+      "learning_rate": 7.13125e-05,
+      "loss": 1.1739,
       "step": 3213
     },
     {
-      "epoch": 9.51,
-      "learning_rate": 3.660749506903353e-05,
-      "loss": 1.6023,
+      "epoch": 14.35,
+      "learning_rate": 7.130357142857143e-05,
+      "loss": 1.1258,
       "step": 3214
     },
     {
-      "epoch": 9.51,
-      "learning_rate": 3.658777120315582e-05,
-      "loss": 1.6036,
+      "epoch": 14.35,
+      "learning_rate": 7.129464285714286e-05,
+      "loss": 1.0731,
       "step": 3215
     },
     {
-      "epoch": 9.51,
-      "learning_rate": 3.656804733727811e-05,
-      "loss": 1.5665,
+      "epoch": 14.36,
+      "learning_rate": 7.12857142857143e-05,
+      "loss": 1.1873,
       "step": 3216
     },
     {
-      "epoch": 9.52,
-      "learning_rate": 3.65483234714004e-05,
-      "loss": 1.4976,
+      "epoch": 14.36,
+      "learning_rate": 7.127678571428572e-05,
+      "loss": 1.1481,
       "step": 3217
     },
     {
-      "epoch": 9.52,
-      "learning_rate": 3.6528599605522683e-05,
-      "loss": 1.5764,
+      "epoch": 14.37,
+      "learning_rate": 7.126785714285715e-05,
+      "loss": 1.1094,
       "step": 3218
     },
     {
-      "epoch": 9.52,
-      "learning_rate": 3.650887573964497e-05,
-      "loss": 1.572,
+      "epoch": 14.37,
+      "learning_rate": 7.125892857142857e-05,
+      "loss": 1.1297,
       "step": 3219
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 3.648915187376726e-05,
-      "loss": 1.5082,
+      "epoch": 14.38,
+      "learning_rate": 7.125000000000001e-05,
+      "loss": 1.1267,
       "step": 3220
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 3.646942800788955e-05,
-      "loss": 1.5624,
+      "epoch": 14.38,
+      "learning_rate": 7.124107142857142e-05,
+      "loss": 1.1127,
       "step": 3221
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 3.644970414201184e-05,
-      "loss": 1.4584,
+      "epoch": 14.38,
+      "learning_rate": 7.123214285714286e-05,
+      "loss": 1.1003,
       "step": 3222
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 3.6429980276134124e-05,
-      "loss": 1.5318,
+      "epoch": 14.39,
+      "learning_rate": 7.122321428571429e-05,
+      "loss": 1.1619,
       "step": 3223
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 3.641025641025641e-05,
-      "loss": 1.4668,
+      "epoch": 14.39,
+      "learning_rate": 7.121428571428573e-05,
+      "loss": 1.1418,
       "step": 3224
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 3.63905325443787e-05,
-      "loss": 1.558,
+      "epoch": 14.4,
+      "learning_rate": 7.120535714285714e-05,
+      "loss": 1.1172,
       "step": 3225
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 3.6370808678500986e-05,
-      "loss": 1.6236,
+      "epoch": 14.4,
+      "learning_rate": 7.119642857142858e-05,
+      "loss": 1.1224,
       "step": 3226
     },
     {
-      "epoch": 9.55,
-      "learning_rate": 3.635108481262328e-05,
-      "loss": 1.5503,
+      "epoch": 14.41,
+      "learning_rate": 7.11875e-05,
+      "loss": 1.1649,
       "step": 3227
     },
     {
-      "epoch": 9.55,
-      "learning_rate": 3.6331360946745564e-05,
-      "loss": 1.4751,
+      "epoch": 14.41,
+      "learning_rate": 7.117857142857143e-05,
+      "loss": 1.1051,
       "step": 3228
     },
     {
-      "epoch": 9.55,
-      "learning_rate": 3.631163708086785e-05,
-      "loss": 1.4487,
+      "epoch": 14.42,
+      "learning_rate": 7.116964285714285e-05,
+      "loss": 1.118,
       "step": 3229
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 3.629191321499014e-05,
-      "loss": 1.6519,
+      "epoch": 14.42,
+      "learning_rate": 7.116071428571429e-05,
+      "loss": 1.2402,
       "step": 3230
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 3.627218934911243e-05,
-      "loss": 1.5876,
+      "epoch": 14.42,
+      "learning_rate": 7.115178571428572e-05,
+      "loss": 1.164,
       "step": 3231
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 3.625246548323472e-05,
-      "loss": 1.5318,
+      "epoch": 14.43,
+      "learning_rate": 7.114285714285714e-05,
+      "loss": 1.1053,
       "step": 3232
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 3.6232741617357004e-05,
-      "loss": 1.5113,
+      "epoch": 14.43,
+      "learning_rate": 7.113392857142858e-05,
+      "loss": 1.1001,
       "step": 3233
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 3.621301775147929e-05,
-      "loss": 1.4911,
+      "epoch": 14.44,
+      "learning_rate": 7.112500000000001e-05,
+      "loss": 1.1253,
       "step": 3234
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 3.619329388560158e-05,
-      "loss": 1.51,
+      "epoch": 14.44,
+      "learning_rate": 7.111607142857143e-05,
+      "loss": 1.0652,
       "step": 3235
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 3.617357001972387e-05,
-      "loss": 1.6253,
+      "epoch": 14.45,
+      "learning_rate": 7.110714285714286e-05,
+      "loss": 1.1405,
       "step": 3236
     },
     {
-      "epoch": 9.58,
-      "learning_rate": 3.615384615384615e-05,
-      "loss": 1.4886,
+      "epoch": 14.45,
+      "learning_rate": 7.10982142857143e-05,
+      "loss": 1.1166,
       "step": 3237
     },
     {
-      "epoch": 9.58,
-      "learning_rate": 3.6134122287968444e-05,
-      "loss": 1.4994,
+      "epoch": 14.46,
+      "learning_rate": 7.108928571428571e-05,
+      "loss": 1.128,
       "step": 3238
     },
     {
-      "epoch": 9.58,
-      "learning_rate": 3.6114398422090736e-05,
-      "loss": 1.5102,
+      "epoch": 14.46,
+      "learning_rate": 7.108035714285715e-05,
+      "loss": 1.1257,
       "step": 3239
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 3.609467455621302e-05,
-      "loss": 1.4806,
+      "epoch": 14.46,
+      "learning_rate": 7.107142857142857e-05,
+      "loss": 1.1025,
       "step": 3240
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 3.607495069033531e-05,
-      "loss": 1.5059,
+      "epoch": 14.47,
+      "learning_rate": 7.10625e-05,
+      "loss": 1.1465,
       "step": 3241
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 3.605522682445759e-05,
-      "loss": 1.4678,
+      "epoch": 14.47,
+      "learning_rate": 7.105357142857142e-05,
+      "loss": 1.1188,
       "step": 3242
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 3.603550295857988e-05,
-      "loss": 1.5654,
+      "epoch": 14.48,
+      "learning_rate": 7.104464285714286e-05,
+      "loss": 1.1302,
       "step": 3243
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 3.601577909270217e-05,
-      "loss": 1.4643,
+      "epoch": 14.48,
+      "learning_rate": 7.103571428571429e-05,
+      "loss": 1.1864,
       "step": 3244
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 3.599605522682446e-05,
-      "loss": 1.4392,
+      "epoch": 14.49,
+      "learning_rate": 7.102678571428571e-05,
+      "loss": 1.1213,
       "step": 3245
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 3.597633136094675e-05,
-      "loss": 1.5416,
+      "epoch": 14.49,
+      "learning_rate": 7.101785714285715e-05,
+      "loss": 1.2375,
       "step": 3246
     },
     {
-      "epoch": 9.61,
-      "learning_rate": 3.595660749506903e-05,
-      "loss": 1.62,
+      "epoch": 14.5,
+      "learning_rate": 7.100892857142858e-05,
+      "loss": 1.1931,
       "step": 3247
     },
     {
-      "epoch": 9.61,
-      "learning_rate": 3.5936883629191325e-05,
-      "loss": 1.6068,
+      "epoch": 14.5,
+      "learning_rate": 7.1e-05,
+      "loss": 1.1497,
       "step": 3248
     },
     {
-      "epoch": 9.61,
-      "learning_rate": 3.591715976331361e-05,
-      "loss": 1.6062,
+      "epoch": 14.5,
+      "learning_rate": 7.099107142857143e-05,
+      "loss": 1.0891,
       "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,
+      "epoch": 14.51,
+      "learning_rate": 7.098214285714287e-05,
+      "loss": 1.1953,
       "step": 3250
     },
     {
-      "epoch": 9.62,
-      "learning_rate": 3.587771203155819e-05,
-      "loss": 1.5883,
+      "epoch": 14.51,
+      "learning_rate": 7.097321428571428e-05,
+      "loss": 1.0595,
       "step": 3251
     },
     {
-      "epoch": 9.62,
-      "learning_rate": 3.585798816568047e-05,
-      "loss": 1.4992,
+      "epoch": 14.52,
+      "learning_rate": 7.096428571428572e-05,
+      "loss": 1.2118,
       "step": 3252
     },
     {
-      "epoch": 9.62,
-      "learning_rate": 3.5838264299802765e-05,
-      "loss": 1.6249,
+      "epoch": 14.52,
+      "learning_rate": 7.095535714285715e-05,
+      "loss": 1.141,
       "step": 3253
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 3.581854043392505e-05,
-      "loss": 1.4722,
+      "epoch": 14.53,
+      "learning_rate": 7.094642857142858e-05,
+      "loss": 1.1858,
       "step": 3254
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 3.5798816568047336e-05,
-      "loss": 1.5256,
+      "epoch": 14.53,
+      "learning_rate": 7.09375e-05,
+      "loss": 1.1137,
       "step": 3255
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 3.577909270216963e-05,
-      "loss": 1.4851,
+      "epoch": 14.54,
+      "learning_rate": 7.092857142857144e-05,
+      "loss": 1.1038,
       "step": 3256
     },
     {
-      "epoch": 9.64,
-      "learning_rate": 3.575936883629192e-05,
-      "loss": 1.4547,
+      "epoch": 14.54,
+      "learning_rate": 7.091964285714286e-05,
+      "loss": 1.119,
       "step": 3257
     },
     {
-      "epoch": 9.64,
-      "learning_rate": 3.5739644970414205e-05,
-      "loss": 1.5067,
+      "epoch": 14.54,
+      "learning_rate": 7.091071428571429e-05,
+      "loss": 1.152,
       "step": 3258
     },
     {
-      "epoch": 9.64,
-      "learning_rate": 3.571992110453649e-05,
-      "loss": 1.5155,
+      "epoch": 14.55,
+      "learning_rate": 7.090178571428571e-05,
+      "loss": 1.0828,
       "step": 3259
     },
     {
-      "epoch": 9.64,
-      "learning_rate": 3.5700197238658776e-05,
-      "loss": 1.5306,
+      "epoch": 14.55,
+      "learning_rate": 7.089285714285715e-05,
+      "loss": 1.0852,
       "step": 3260
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 3.568047337278107e-05,
-      "loss": 1.5105,
+      "epoch": 14.56,
+      "learning_rate": 7.088392857142858e-05,
+      "loss": 1.1608,
       "step": 3261
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 3.566074950690336e-05,
-      "loss": 1.5331,
+      "epoch": 14.56,
+      "learning_rate": 7.0875e-05,
+      "loss": 1.1318,
       "step": 3262
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 3.5641025641025646e-05,
-      "loss": 1.5691,
+      "epoch": 14.57,
+      "learning_rate": 7.086607142857144e-05,
+      "loss": 1.1742,
       "step": 3263
     },
     {
-      "epoch": 9.66,
-      "learning_rate": 3.562130177514793e-05,
-      "loss": 1.5833,
+      "epoch": 14.57,
+      "learning_rate": 7.085714285714285e-05,
+      "loss": 1.1014,
       "step": 3264
     },
     {
-      "epoch": 9.66,
-      "learning_rate": 3.5601577909270216e-05,
-      "loss": 1.4245,
+      "epoch": 14.58,
+      "learning_rate": 7.084821428571429e-05,
+      "loss": 1.1083,
       "step": 3265
     },
     {
-      "epoch": 9.66,
-      "learning_rate": 3.55818540433925e-05,
-      "loss": 1.5589,
+      "epoch": 14.58,
+      "learning_rate": 7.083928571428572e-05,
+      "loss": 1.1685,
       "step": 3266
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 3.5562130177514794e-05,
-      "loss": 1.4587,
+      "epoch": 14.58,
+      "learning_rate": 7.083035714285716e-05,
+      "loss": 1.1779,
       "step": 3267
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 3.5542406311637086e-05,
-      "loss": 1.5965,
+      "epoch": 14.59,
+      "learning_rate": 7.082142857142857e-05,
+      "loss": 1.1057,
       "step": 3268
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 3.552268244575937e-05,
-      "loss": 1.562,
+      "epoch": 14.59,
+      "learning_rate": 7.081250000000001e-05,
+      "loss": 1.1576,
       "step": 3269
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 3.5502958579881656e-05,
-      "loss": 1.5733,
+      "epoch": 14.6,
+      "learning_rate": 7.080357142857143e-05,
+      "loss": 1.1881,
       "step": 3270
     },
     {
-      "epoch": 9.68,
-      "learning_rate": 3.548323471400395e-05,
-      "loss": 1.4886,
+      "epoch": 14.6,
+      "learning_rate": 7.079464285714286e-05,
+      "loss": 1.1551,
       "step": 3271
     },
     {
-      "epoch": 9.68,
-      "learning_rate": 3.5463510848126234e-05,
-      "loss": 1.5501,
+      "epoch": 14.61,
+      "learning_rate": 7.078571428571428e-05,
+      "loss": 1.1293,
       "step": 3272
     },
     {
-      "epoch": 9.68,
-      "learning_rate": 3.544378698224852e-05,
-      "loss": 1.6245,
+      "epoch": 14.61,
+      "learning_rate": 7.077678571428572e-05,
+      "loss": 1.1373,
       "step": 3273
     },
     {
-      "epoch": 9.69,
-      "learning_rate": 3.542406311637081e-05,
-      "loss": 1.5399,
+      "epoch": 14.62,
+      "learning_rate": 7.076785714285715e-05,
+      "loss": 1.1589,
       "step": 3274
     },
     {
-      "epoch": 9.69,
-      "learning_rate": 3.54043392504931e-05,
-      "loss": 1.633,
+      "epoch": 14.62,
+      "learning_rate": 7.075892857142857e-05,
+      "loss": 1.1405,
       "step": 3275
     },
     {
-      "epoch": 9.69,
-      "learning_rate": 3.538461538461539e-05,
-      "loss": 1.5672,
+      "epoch": 14.62,
+      "learning_rate": 7.075e-05,
+      "loss": 1.193,
       "step": 3276
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 3.5364891518737674e-05,
-      "loss": 1.5234,
+      "epoch": 14.63,
+      "learning_rate": 7.074107142857144e-05,
+      "loss": 1.1004,
       "step": 3277
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 3.534516765285996e-05,
-      "loss": 1.5191,
+      "epoch": 14.63,
+      "learning_rate": 7.073214285714286e-05,
+      "loss": 1.1478,
       "step": 3278
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 3.532544378698225e-05,
-      "loss": 1.5722,
+      "epoch": 14.64,
+      "learning_rate": 7.072321428571429e-05,
+      "loss": 1.1301,
       "step": 3279
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 3.5305719921104544e-05,
-      "loss": 1.5787,
+      "epoch": 14.64,
+      "learning_rate": 7.071428571428573e-05,
+      "loss": 1.1557,
       "step": 3280
     },
     {
-      "epoch": 9.71,
-      "learning_rate": 3.528599605522683e-05,
-      "loss": 1.5575,
+      "epoch": 14.65,
+      "learning_rate": 7.070535714285714e-05,
+      "loss": 1.1527,
       "step": 3281
     },
     {
-      "epoch": 9.71,
-      "learning_rate": 3.5266272189349114e-05,
-      "loss": 1.5621,
+      "epoch": 14.65,
+      "learning_rate": 7.069642857142858e-05,
+      "loss": 1.2071,
       "step": 3282
     },
     {
-      "epoch": 9.71,
-      "learning_rate": 3.52465483234714e-05,
-      "loss": 1.4546,
+      "epoch": 14.66,
+      "learning_rate": 7.06875e-05,
+      "loss": 1.1637,
       "step": 3283
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 3.5226824457593685e-05,
-      "loss": 1.5618,
+      "epoch": 14.66,
+      "learning_rate": 7.067857142857143e-05,
+      "loss": 1.2151,
       "step": 3284
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 3.520710059171598e-05,
-      "loss": 1.5162,
+      "epoch": 14.67,
+      "learning_rate": 7.066964285714285e-05,
+      "loss": 1.139,
       "step": 3285
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 3.518737672583827e-05,
-      "loss": 1.5031,
+      "epoch": 14.67,
+      "learning_rate": 7.06607142857143e-05,
+      "loss": 1.2151,
       "step": 3286
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 3.5167652859960555e-05,
-      "loss": 1.5204,
+      "epoch": 14.67,
+      "learning_rate": 7.065178571428572e-05,
+      "loss": 1.2332,
       "step": 3287
     },
     {
-      "epoch": 9.73,
-      "learning_rate": 3.514792899408284e-05,
-      "loss": 1.5367,
+      "epoch": 14.68,
+      "learning_rate": 7.064285714285714e-05,
+      "loss": 1.168,
       "step": 3288
     },
     {
-      "epoch": 9.73,
-      "learning_rate": 3.5128205128205125e-05,
-      "loss": 1.5163,
+      "epoch": 14.68,
+      "learning_rate": 7.063392857142857e-05,
+      "loss": 1.1191,
       "step": 3289
     },
     {
-      "epoch": 9.73,
-      "learning_rate": 3.510848126232742e-05,
-      "loss": 1.5895,
+      "epoch": 14.69,
+      "learning_rate": 7.062500000000001e-05,
+      "loss": 1.1872,
       "step": 3290
     },
     {
-      "epoch": 9.74,
-      "learning_rate": 3.508875739644971e-05,
-      "loss": 1.4683,
+      "epoch": 14.69,
+      "learning_rate": 7.061607142857143e-05,
+      "loss": 1.1272,
       "step": 3291
     },
     {
-      "epoch": 9.74,
-      "learning_rate": 3.5069033530571995e-05,
-      "loss": 1.6268,
+      "epoch": 14.7,
+      "learning_rate": 7.060714285714286e-05,
+      "loss": 1.118,
       "step": 3292
     },
     {
-      "epoch": 9.74,
-      "learning_rate": 3.504930966469428e-05,
-      "loss": 1.5348,
+      "epoch": 14.7,
+      "learning_rate": 7.05982142857143e-05,
+      "loss": 1.1553,
       "step": 3293
     },
     {
-      "epoch": 9.75,
-      "learning_rate": 3.502958579881657e-05,
-      "loss": 1.5806,
+      "epoch": 14.71,
+      "learning_rate": 7.058928571428571e-05,
+      "loss": 1.1326,
       "step": 3294
     },
     {
-      "epoch": 9.75,
-      "learning_rate": 3.500986193293886e-05,
-      "loss": 1.5552,
+      "epoch": 14.71,
+      "learning_rate": 7.058035714285715e-05,
+      "loss": 1.2061,
       "step": 3295
     },
     {
-      "epoch": 9.75,
-      "learning_rate": 3.499013806706114e-05,
-      "loss": 1.4595,
+      "epoch": 14.71,
+      "learning_rate": 7.057142857142858e-05,
+      "loss": 1.1794,
       "step": 3296
     },
     {
-      "epoch": 9.75,
-      "learning_rate": 3.4970414201183435e-05,
-      "loss": 1.542,
+      "epoch": 14.72,
+      "learning_rate": 7.05625e-05,
+      "loss": 1.1955,
       "step": 3297
     },
     {
-      "epoch": 9.76,
-      "learning_rate": 3.495069033530572e-05,
-      "loss": 1.5846,
+      "epoch": 14.72,
+      "learning_rate": 7.055357142857143e-05,
+      "loss": 1.0699,
       "step": 3298
     },
     {
-      "epoch": 9.76,
-      "learning_rate": 3.493096646942801e-05,
-      "loss": 1.5057,
+      "epoch": 14.73,
+      "learning_rate": 7.054464285714287e-05,
+      "loss": 1.141,
       "step": 3299
     },
     {
-      "epoch": 9.76,
-      "learning_rate": 3.49112426035503e-05,
-      "loss": 1.5888,
+      "epoch": 14.73,
+      "learning_rate": 7.053571428571429e-05,
+      "loss": 1.097,
       "step": 3300
     },
     {
-      "epoch": 9.77,
-      "learning_rate": 3.489151873767258e-05,
-      "loss": 1.4485,
+      "epoch": 14.73,
+      "eval_loss": 1.1503467559814453,
+      "eval_runtime": 44.8456,
+      "eval_samples_per_second": 291.98,
+      "eval_steps_per_second": 1.539,
+      "step": 3300
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 7.052678571428572e-05,
+      "loss": 1.1908,
       "step": 3301
     },
     {
-      "epoch": 9.77,
-      "learning_rate": 3.487179487179487e-05,
-      "loss": 1.5359,
+      "epoch": 14.74,
+      "learning_rate": 7.051785714285714e-05,
+      "loss": 1.2246,
       "step": 3302
     },
     {
-      "epoch": 9.77,
-      "learning_rate": 3.485207100591716e-05,
-      "loss": 1.6716,
+      "epoch": 14.75,
+      "learning_rate": 7.050892857142858e-05,
+      "loss": 1.0738,
       "step": 3303
     },
     {
-      "epoch": 9.78,
-      "learning_rate": 3.483234714003945e-05,
-      "loss": 1.5512,
+      "epoch": 14.75,
+      "learning_rate": 7.05e-05,
+      "loss": 1.2032,
       "step": 3304
     },
     {
-      "epoch": 9.78,
-      "learning_rate": 3.481262327416174e-05,
-      "loss": 1.5448,
+      "epoch": 14.75,
+      "learning_rate": 7.049107142857143e-05,
+      "loss": 1.0707,
       "step": 3305
     },
     {
-      "epoch": 9.78,
-      "learning_rate": 3.4792899408284023e-05,
-      "loss": 1.5472,
+      "epoch": 14.76,
+      "learning_rate": 7.048214285714286e-05,
+      "loss": 1.1698,
       "step": 3306
     },
     {
-      "epoch": 9.78,
-      "learning_rate": 3.477317554240631e-05,
-      "loss": 1.5648,
+      "epoch": 14.76,
+      "learning_rate": 7.047321428571428e-05,
+      "loss": 1.1634,
       "step": 3307
     },
     {
-      "epoch": 9.79,
-      "learning_rate": 3.47534516765286e-05,
-      "loss": 1.4858,
+      "epoch": 14.77,
+      "learning_rate": 7.046428571428572e-05,
+      "loss": 1.1343,
       "step": 3308
     },
     {
-      "epoch": 9.79,
-      "learning_rate": 3.473372781065089e-05,
-      "loss": 1.5814,
+      "epoch": 14.77,
+      "learning_rate": 7.045535714285715e-05,
+      "loss": 1.0674,
       "step": 3309
     },
     {
-      "epoch": 9.79,
-      "learning_rate": 3.471400394477318e-05,
-      "loss": 1.522,
+      "epoch": 14.78,
+      "learning_rate": 7.044642857142859e-05,
+      "loss": 1.1672,
       "step": 3310
     },
     {
-      "epoch": 9.8,
-      "learning_rate": 3.4694280078895464e-05,
-      "loss": 1.5704,
+      "epoch": 14.78,
+      "learning_rate": 7.04375e-05,
+      "loss": 1.0368,
       "step": 3311
     },
     {
-      "epoch": 9.8,
-      "learning_rate": 3.467455621301775e-05,
-      "loss": 1.5132,
+      "epoch": 14.79,
+      "learning_rate": 7.042857142857144e-05,
+      "loss": 1.0922,
       "step": 3312
     },
     {
-      "epoch": 9.8,
-      "learning_rate": 3.465483234714004e-05,
-      "loss": 1.5651,
+      "epoch": 14.79,
+      "learning_rate": 7.041964285714286e-05,
+      "loss": 1.1684,
       "step": 3313
     },
     {
-      "epoch": 9.8,
-      "learning_rate": 3.4635108481262326e-05,
-      "loss": 1.4726,
+      "epoch": 14.79,
+      "learning_rate": 7.041071428571429e-05,
+      "loss": 1.1186,
       "step": 3314
     },
     {
-      "epoch": 9.81,
-      "learning_rate": 3.461538461538462e-05,
-      "loss": 1.5136,
+      "epoch": 14.8,
+      "learning_rate": 7.040178571428571e-05,
+      "loss": 1.1603,
       "step": 3315
     },
     {
-      "epoch": 9.81,
-      "learning_rate": 3.4595660749506904e-05,
-      "loss": 1.4024,
+      "epoch": 14.8,
+      "learning_rate": 7.039285714285715e-05,
+      "loss": 1.0843,
       "step": 3316
     },
     {
-      "epoch": 9.81,
-      "learning_rate": 3.4575936883629196e-05,
-      "loss": 1.5449,
+      "epoch": 14.81,
+      "learning_rate": 7.038392857142856e-05,
+      "loss": 1.1224,
       "step": 3317
     },
     {
-      "epoch": 9.82,
-      "learning_rate": 3.455621301775148e-05,
-      "loss": 1.6117,
+      "epoch": 14.81,
+      "learning_rate": 7.0375e-05,
+      "loss": 1.1978,
       "step": 3318
     },
     {
-      "epoch": 9.82,
-      "learning_rate": 3.453648915187377e-05,
-      "loss": 1.5558,
+      "epoch": 14.82,
+      "learning_rate": 7.036607142857143e-05,
+      "loss": 1.1345,
       "step": 3319
     },
     {
-      "epoch": 9.82,
-      "learning_rate": 3.451676528599606e-05,
-      "loss": 1.5656,
+      "epoch": 14.82,
+      "learning_rate": 7.035714285714287e-05,
+      "loss": 1.2177,
       "step": 3320
     },
     {
-      "epoch": 9.83,
-      "learning_rate": 3.4497041420118344e-05,
-      "loss": 1.4923,
+      "epoch": 14.83,
+      "learning_rate": 7.034821428571429e-05,
+      "loss": 1.1608,
       "step": 3321
     },
     {
-      "epoch": 9.83,
-      "learning_rate": 3.4477317554240636e-05,
-      "loss": 1.573,
+      "epoch": 14.83,
+      "learning_rate": 7.033928571428572e-05,
+      "loss": 1.0712,
       "step": 3322
     },
     {
-      "epoch": 9.83,
-      "learning_rate": 3.445759368836292e-05,
-      "loss": 1.526,
+      "epoch": 14.83,
+      "learning_rate": 7.033035714285716e-05,
+      "loss": 1.1477,
       "step": 3323
     },
     {
-      "epoch": 9.83,
-      "learning_rate": 3.443786982248521e-05,
-      "loss": 1.5943,
+      "epoch": 14.84,
+      "learning_rate": 7.032142857142857e-05,
+      "loss": 1.1265,
       "step": 3324
     },
     {
-      "epoch": 9.84,
-      "learning_rate": 3.441814595660749e-05,
-      "loss": 1.6335,
+      "epoch": 14.84,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 1.1064,
       "step": 3325
     },
     {
-      "epoch": 9.84,
-      "learning_rate": 3.4398422090729784e-05,
-      "loss": 1.5846,
+      "epoch": 14.85,
+      "learning_rate": 7.030357142857143e-05,
+      "loss": 1.1844,
       "step": 3326
     },
     {
-      "epoch": 9.84,
-      "learning_rate": 3.4378698224852077e-05,
-      "loss": 1.642,
+      "epoch": 14.85,
+      "learning_rate": 7.029464285714286e-05,
+      "loss": 1.1356,
       "step": 3327
     },
     {
-      "epoch": 9.85,
-      "learning_rate": 3.435897435897436e-05,
-      "loss": 1.5129,
+      "epoch": 14.86,
+      "learning_rate": 7.028571428571428e-05,
+      "loss": 1.1406,
       "step": 3328
     },
     {
-      "epoch": 9.85,
-      "learning_rate": 3.433925049309665e-05,
-      "loss": 1.6131,
+      "epoch": 14.86,
+      "learning_rate": 7.027678571428572e-05,
+      "loss": 1.1083,
       "step": 3329
     },
     {
-      "epoch": 9.85,
-      "learning_rate": 3.431952662721893e-05,
-      "loss": 1.5212,
+      "epoch": 14.87,
+      "learning_rate": 7.026785714285714e-05,
+      "loss": 1.1321,
       "step": 3330
     },
     {
-      "epoch": 9.86,
-      "learning_rate": 3.4299802761341225e-05,
-      "loss": 1.586,
+      "epoch": 14.87,
+      "learning_rate": 7.025892857142857e-05,
+      "loss": 1.1907,
       "step": 3331
     },
     {
-      "epoch": 9.86,
-      "learning_rate": 3.428007889546351e-05,
-      "loss": 1.5487,
+      "epoch": 14.88,
+      "learning_rate": 7.025e-05,
+      "loss": 1.087,
       "step": 3332
     },
     {
-      "epoch": 9.86,
-      "learning_rate": 3.42603550295858e-05,
-      "loss": 1.5574,
+      "epoch": 14.88,
+      "learning_rate": 7.024107142857144e-05,
+      "loss": 1.1654,
       "step": 3333
     },
     {
-      "epoch": 9.86,
-      "learning_rate": 3.424063116370809e-05,
-      "loss": 1.5476,
+      "epoch": 14.88,
+      "learning_rate": 7.023214285714285e-05,
+      "loss": 1.1775,
       "step": 3334
     },
     {
-      "epoch": 9.87,
-      "learning_rate": 3.422090729783038e-05,
-      "loss": 1.4726,
+      "epoch": 14.89,
+      "learning_rate": 7.022321428571429e-05,
+      "loss": 1.0861,
       "step": 3335
     },
     {
-      "epoch": 9.87,
-      "learning_rate": 3.4201183431952665e-05,
-      "loss": 1.556,
+      "epoch": 14.89,
+      "learning_rate": 7.021428571428572e-05,
+      "loss": 1.1451,
       "step": 3336
     },
     {
-      "epoch": 9.87,
-      "learning_rate": 3.418145956607495e-05,
-      "loss": 1.5672,
+      "epoch": 14.9,
+      "learning_rate": 7.020535714285714e-05,
+      "loss": 1.0754,
       "step": 3337
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 3.416173570019724e-05,
-      "loss": 1.5332,
+      "epoch": 14.9,
+      "learning_rate": 7.019642857142858e-05,
+      "loss": 1.1798,
       "step": 3338
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 3.414201183431953e-05,
-      "loss": 1.5252,
+      "epoch": 14.91,
+      "learning_rate": 7.01875e-05,
+      "loss": 1.27,
       "step": 3339
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 3.412228796844182e-05,
-      "loss": 1.4874,
+      "epoch": 14.91,
+      "learning_rate": 7.017857142857143e-05,
+      "loss": 1.1397,
       "step": 3340
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 3.4102564102564105e-05,
-      "loss": 1.484,
+      "epoch": 14.92,
+      "learning_rate": 7.016964285714286e-05,
+      "loss": 1.0888,
       "step": 3341
     },
     {
-      "epoch": 9.89,
-      "learning_rate": 3.408284023668639e-05,
-      "loss": 1.5331,
+      "epoch": 14.92,
+      "learning_rate": 7.01607142857143e-05,
+      "loss": 1.1637,
       "step": 3342
     },
     {
-      "epoch": 9.89,
-      "learning_rate": 3.4063116370808676e-05,
-      "loss": 1.4808,
+      "epoch": 14.92,
+      "learning_rate": 7.015178571428572e-05,
+      "loss": 1.1564,
       "step": 3343
     },
     {
-      "epoch": 9.89,
-      "learning_rate": 3.404339250493097e-05,
-      "loss": 1.4746,
+      "epoch": 14.93,
+      "learning_rate": 7.014285714285715e-05,
+      "loss": 1.1055,
       "step": 3344
     },
     {
-      "epoch": 9.9,
-      "learning_rate": 3.402366863905326e-05,
-      "loss": 1.5255,
+      "epoch": 14.93,
+      "learning_rate": 7.013392857142857e-05,
+      "loss": 1.1498,
       "step": 3345
     },
     {
-      "epoch": 9.9,
-      "learning_rate": 3.4003944773175545e-05,
-      "loss": 1.5398,
+      "epoch": 14.94,
+      "learning_rate": 7.012500000000001e-05,
+      "loss": 1.1662,
       "step": 3346
     },
     {
-      "epoch": 9.9,
-      "learning_rate": 3.398422090729783e-05,
-      "loss": 1.5862,
+      "epoch": 14.94,
+      "learning_rate": 7.011607142857142e-05,
+      "loss": 1.1663,
       "step": 3347
     },
     {
-      "epoch": 9.91,
-      "learning_rate": 3.3964497041420116e-05,
-      "loss": 1.5257,
+      "epoch": 14.95,
+      "learning_rate": 7.010714285714286e-05,
+      "loss": 1.1925,
       "step": 3348
     },
     {
-      "epoch": 9.91,
-      "learning_rate": 3.394477317554241e-05,
-      "loss": 1.5618,
+      "epoch": 14.95,
+      "learning_rate": 7.009821428571429e-05,
+      "loss": 1.1089,
       "step": 3349
     },
     {
-      "epoch": 9.91,
-      "learning_rate": 3.39250493096647e-05,
-      "loss": 1.5621,
+      "epoch": 14.96,
+      "learning_rate": 7.008928571428571e-05,
+      "loss": 1.1395,
       "step": 3350
     },
     {
-      "epoch": 9.91,
-      "learning_rate": 3.3905325443786986e-05,
-      "loss": 1.6478,
+      "epoch": 14.96,
+      "learning_rate": 7.008035714285715e-05,
+      "loss": 1.1764,
       "step": 3351
     },
     {
-      "epoch": 9.92,
-      "learning_rate": 3.388560157790927e-05,
-      "loss": 1.5761,
+      "epoch": 14.96,
+      "learning_rate": 7.007142857142858e-05,
+      "loss": 1.058,
       "step": 3352
     },
     {
-      "epoch": 9.92,
-      "learning_rate": 3.3865877712031556e-05,
-      "loss": 1.4994,
+      "epoch": 14.97,
+      "learning_rate": 7.00625e-05,
+      "loss": 1.0982,
       "step": 3353
     },
     {
-      "epoch": 9.92,
-      "learning_rate": 3.384615384615385e-05,
-      "loss": 1.5439,
+      "epoch": 14.97,
+      "learning_rate": 7.005357142857143e-05,
+      "loss": 1.1055,
       "step": 3354
     },
     {
-      "epoch": 9.93,
-      "learning_rate": 3.3826429980276134e-05,
-      "loss": 1.4953,
+      "epoch": 14.98,
+      "learning_rate": 7.004464285714287e-05,
+      "loss": 1.1111,
       "step": 3355
     },
     {
-      "epoch": 9.93,
-      "learning_rate": 3.3806706114398426e-05,
-      "loss": 1.5531,
+      "epoch": 14.98,
+      "learning_rate": 7.003571428571429e-05,
+      "loss": 1.2018,
       "step": 3356
     },
     {
-      "epoch": 9.93,
-      "learning_rate": 3.378698224852071e-05,
-      "loss": 1.6189,
+      "epoch": 14.99,
+      "learning_rate": 7.002678571428572e-05,
+      "loss": 1.19,
       "step": 3357
     },
     {
-      "epoch": 9.93,
-      "learning_rate": 3.3767258382643e-05,
-      "loss": 1.5825,
+      "epoch": 14.99,
+      "learning_rate": 7.001785714285714e-05,
+      "loss": 1.1208,
       "step": 3358
     },
     {
-      "epoch": 9.94,
-      "learning_rate": 3.374753451676529e-05,
-      "loss": 1.651,
+      "epoch": 15.0,
+      "learning_rate": 7.000892857142858e-05,
+      "loss": 1.127,
       "step": 3359
     },
     {
-      "epoch": 9.94,
-      "learning_rate": 3.3727810650887574e-05,
-      "loss": 1.4302,
+      "epoch": 15.0,
+      "learning_rate": 7e-05,
+      "loss": 1.1322,
       "step": 3360
     },
     {
-      "epoch": 9.94,
-      "learning_rate": 3.370808678500986e-05,
-      "loss": 1.5408,
+      "epoch": 15.0,
+      "learning_rate": 6.999107142857143e-05,
+      "loss": 1.1402,
       "step": 3361
     },
     {
-      "epoch": 9.95,
-      "learning_rate": 3.368836291913215e-05,
-      "loss": 1.5861,
+      "epoch": 15.01,
+      "learning_rate": 6.998214285714286e-05,
+      "loss": 1.088,
       "step": 3362
     },
     {
-      "epoch": 9.95,
-      "learning_rate": 3.3668639053254444e-05,
-      "loss": 1.5981,
+      "epoch": 15.01,
+      "learning_rate": 6.997321428571428e-05,
+      "loss": 1.0464,
       "step": 3363
     },
     {
-      "epoch": 9.95,
-      "learning_rate": 3.364891518737673e-05,
-      "loss": 1.5869,
+      "epoch": 15.02,
+      "learning_rate": 6.996428571428571e-05,
+      "loss": 1.1362,
       "step": 3364
     },
     {
-      "epoch": 9.96,
-      "learning_rate": 3.3629191321499014e-05,
-      "loss": 1.4853,
+      "epoch": 15.02,
+      "learning_rate": 6.995535714285715e-05,
+      "loss": 1.1287,
       "step": 3365
     },
     {
-      "epoch": 9.96,
-      "learning_rate": 3.36094674556213e-05,
-      "loss": 1.5196,
+      "epoch": 15.03,
+      "learning_rate": 6.994642857142857e-05,
+      "loss": 1.1028,
       "step": 3366
     },
     {
-      "epoch": 9.96,
-      "learning_rate": 3.358974358974359e-05,
-      "loss": 1.5741,
+      "epoch": 15.03,
+      "learning_rate": 6.99375e-05,
+      "loss": 1.12,
       "step": 3367
     },
     {
-      "epoch": 9.96,
-      "learning_rate": 3.3570019723865884e-05,
-      "loss": 1.4911,
+      "epoch": 15.04,
+      "learning_rate": 6.992857142857144e-05,
+      "loss": 0.9899,
       "step": 3368
     },
     {
-      "epoch": 9.97,
-      "learning_rate": 3.355029585798817e-05,
-      "loss": 1.5875,
+      "epoch": 15.04,
+      "learning_rate": 6.991964285714286e-05,
+      "loss": 1.1666,
       "step": 3369
     },
     {
-      "epoch": 9.97,
-      "learning_rate": 3.3530571992110454e-05,
-      "loss": 1.5059,
+      "epoch": 15.04,
+      "learning_rate": 6.991071428571429e-05,
+      "loss": 1.0743,
       "step": 3370
     },
     {
-      "epoch": 9.97,
-      "learning_rate": 3.351084812623274e-05,
-      "loss": 1.4908,
+      "epoch": 15.05,
+      "learning_rate": 6.990178571428571e-05,
+      "loss": 1.0074,
       "step": 3371
     },
     {
-      "epoch": 9.98,
-      "learning_rate": 3.349112426035503e-05,
-      "loss": 1.5752,
+      "epoch": 15.05,
+      "learning_rate": 6.989285714285715e-05,
+      "loss": 1.1154,
       "step": 3372
     },
     {
-      "epoch": 9.98,
-      "learning_rate": 3.347140039447732e-05,
-      "loss": 1.4817,
+      "epoch": 15.06,
+      "learning_rate": 6.988392857142857e-05,
+      "loss": 1.175,
       "step": 3373
     },
     {
-      "epoch": 9.98,
-      "learning_rate": 3.345167652859961e-05,
-      "loss": 1.5595,
+      "epoch": 15.06,
+      "learning_rate": 6.9875e-05,
+      "loss": 1.1357,
       "step": 3374
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 3.3431952662721895e-05,
-      "loss": 1.5416,
+      "epoch": 15.07,
+      "learning_rate": 6.986607142857143e-05,
+      "loss": 1.0882,
       "step": 3375
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 3.341222879684418e-05,
-      "loss": 1.6161,
+      "epoch": 15.07,
+      "learning_rate": 6.985714285714287e-05,
+      "loss": 1.125,
       "step": 3376
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 3.339250493096647e-05,
-      "loss": 1.5472,
+      "epoch": 15.08,
+      "learning_rate": 6.984821428571428e-05,
+      "loss": 1.0795,
       "step": 3377
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 3.337278106508876e-05,
-      "loss": 1.5452,
+      "epoch": 15.08,
+      "learning_rate": 6.983928571428572e-05,
+      "loss": 1.1089,
       "step": 3378
     },
     {
-      "epoch": 10.0,
-      "learning_rate": 3.335305719921105e-05,
-      "loss": 1.53,
+      "epoch": 15.08,
+      "learning_rate": 6.983035714285715e-05,
+      "loss": 1.185,
       "step": 3379
     },
     {
-      "epoch": 10.0,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 1.2721,
+      "epoch": 15.09,
+      "learning_rate": 6.982142857142857e-05,
+      "loss": 1.1452,
       "step": 3380
     },
     {
-      "epoch": 10.0,
-      "learning_rate": 3.331360946745563e-05,
-      "loss": 1.5989,
+      "epoch": 15.09,
+      "learning_rate": 6.981250000000001e-05,
+      "loss": 1.2246,
       "step": 3381
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 3.329388560157791e-05,
-      "loss": 1.6194,
+      "epoch": 15.1,
+      "learning_rate": 6.980357142857144e-05,
+      "loss": 1.1574,
       "step": 3382
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 3.32741617357002e-05,
-      "loss": 1.5551,
+      "epoch": 15.1,
+      "learning_rate": 6.979464285714286e-05,
+      "loss": 1.1566,
       "step": 3383
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 3.325443786982248e-05,
-      "loss": 1.5339,
+      "epoch": 15.11,
+      "learning_rate": 6.978571428571429e-05,
+      "loss": 1.0775,
       "step": 3384
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 3.3234714003944775e-05,
-      "loss": 1.6192,
+      "epoch": 15.11,
+      "learning_rate": 6.977678571428572e-05,
+      "loss": 1.0788,
       "step": 3385
     },
     {
-      "epoch": 10.02,
-      "learning_rate": 3.321499013806707e-05,
-      "loss": 1.5154,
+      "epoch": 15.12,
+      "learning_rate": 6.976785714285715e-05,
+      "loss": 1.18,
       "step": 3386
     },
     {
-      "epoch": 10.02,
-      "learning_rate": 3.319526627218935e-05,
-      "loss": 1.4981,
+      "epoch": 15.12,
+      "learning_rate": 6.975892857142858e-05,
+      "loss": 1.1228,
       "step": 3387
     },
     {
-      "epoch": 10.02,
-      "learning_rate": 3.317554240631164e-05,
-      "loss": 1.5273,
+      "epoch": 15.12,
+      "learning_rate": 6.975e-05,
+      "loss": 1.1774,
       "step": 3388
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 3.315581854043392e-05,
-      "loss": 1.5956,
+      "epoch": 15.13,
+      "learning_rate": 6.974107142857144e-05,
+      "loss": 1.1276,
       "step": 3389
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 3.3136094674556215e-05,
-      "loss": 1.4963,
+      "epoch": 15.13,
+      "learning_rate": 6.973214285714285e-05,
+      "loss": 1.0898,
       "step": 3390
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 3.31163708086785e-05,
-      "loss": 1.51,
+      "epoch": 15.14,
+      "learning_rate": 6.972321428571429e-05,
+      "loss": 1.1027,
       "step": 3391
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 3.309664694280079e-05,
-      "loss": 1.5607,
+      "epoch": 15.14,
+      "learning_rate": 6.971428571428572e-05,
+      "loss": 1.1265,
       "step": 3392
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 3.307692307692308e-05,
-      "loss": 1.628,
+      "epoch": 15.15,
+      "learning_rate": 6.970535714285714e-05,
+      "loss": 1.1511,
       "step": 3393
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 3.3057199211045364e-05,
-      "loss": 1.5161,
+      "epoch": 15.15,
+      "learning_rate": 6.969642857142857e-05,
+      "loss": 1.157,
       "step": 3394
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 3.3037475345167656e-05,
-      "loss": 1.5041,
+      "epoch": 15.16,
+      "learning_rate": 6.96875e-05,
+      "loss": 1.1249,
       "step": 3395
     },
     {
-      "epoch": 10.05,
-      "learning_rate": 3.301775147928994e-05,
-      "loss": 1.5866,
+      "epoch": 15.16,
+      "learning_rate": 6.967857142857143e-05,
+      "loss": 1.0615,
       "step": 3396
     },
     {
-      "epoch": 10.05,
-      "learning_rate": 3.299802761341223e-05,
-      "loss": 1.4476,
+      "epoch": 15.17,
+      "learning_rate": 6.966964285714286e-05,
+      "loss": 1.0795,
       "step": 3397
     },
     {
-      "epoch": 10.05,
-      "learning_rate": 3.297830374753452e-05,
-      "loss": 1.4254,
+      "epoch": 15.17,
+      "learning_rate": 6.96607142857143e-05,
+      "loss": 1.1692,
       "step": 3398
     },
     {
-      "epoch": 10.06,
-      "learning_rate": 3.2958579881656804e-05,
-      "loss": 1.5533,
+      "epoch": 15.17,
+      "learning_rate": 6.965178571428572e-05,
+      "loss": 1.1538,
       "step": 3399
     },
     {
-      "epoch": 10.06,
-      "learning_rate": 3.2938856015779096e-05,
-      "loss": 1.4887,
+      "epoch": 15.18,
+      "learning_rate": 6.964285714285715e-05,
+      "loss": 1.1308,
       "step": 3400
     },
     {
-      "epoch": 10.06,
-      "learning_rate": 3.291913214990138e-05,
-      "loss": 1.5489,
+      "epoch": 15.18,
+      "learning_rate": 6.963392857142857e-05,
+      "loss": 1.1062,
       "step": 3401
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 3.2899408284023667e-05,
-      "loss": 1.4956,
+      "epoch": 15.19,
+      "learning_rate": 6.962500000000001e-05,
+      "loss": 1.0773,
       "step": 3402
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 3.287968441814596e-05,
-      "loss": 1.4985,
+      "epoch": 15.19,
+      "learning_rate": 6.961607142857142e-05,
+      "loss": 1.1806,
       "step": 3403
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 3.285996055226825e-05,
-      "loss": 1.4674,
+      "epoch": 15.2,
+      "learning_rate": 6.960714285714286e-05,
+      "loss": 1.1074,
       "step": 3404
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 3.2840236686390536e-05,
-      "loss": 1.5,
+      "epoch": 15.2,
+      "learning_rate": 6.959821428571429e-05,
+      "loss": 1.0769,
       "step": 3405
     },
     {
-      "epoch": 10.08,
-      "learning_rate": 3.282051282051282e-05,
-      "loss": 1.5355,
+      "epoch": 15.21,
+      "learning_rate": 6.958928571428571e-05,
+      "loss": 1.1358,
       "step": 3406
     },
     {
-      "epoch": 10.08,
-      "learning_rate": 3.280078895463511e-05,
-      "loss": 1.4869,
+      "epoch": 15.21,
+      "learning_rate": 6.958035714285714e-05,
+      "loss": 1.1369,
       "step": 3407
     },
     {
-      "epoch": 10.08,
-      "learning_rate": 3.27810650887574e-05,
-      "loss": 1.5488,
+      "epoch": 15.21,
+      "learning_rate": 6.957142857142858e-05,
+      "loss": 1.0637,
       "step": 3408
     },
     {
-      "epoch": 10.09,
-      "learning_rate": 3.276134122287969e-05,
-      "loss": 1.5946,
+      "epoch": 15.22,
+      "learning_rate": 6.95625e-05,
+      "loss": 1.1238,
       "step": 3409
     },
     {
-      "epoch": 10.09,
-      "learning_rate": 3.2741617357001976e-05,
-      "loss": 1.5261,
+      "epoch": 15.22,
+      "learning_rate": 6.955357142857143e-05,
+      "loss": 1.0509,
       "step": 3410
     },
     {
-      "epoch": 10.09,
-      "learning_rate": 3.272189349112426e-05,
-      "loss": 1.545,
+      "epoch": 15.23,
+      "learning_rate": 6.954464285714285e-05,
+      "loss": 1.1309,
       "step": 3411
     },
     {
-      "epoch": 10.09,
-      "learning_rate": 3.270216962524655e-05,
-      "loss": 1.5686,
+      "epoch": 15.23,
+      "learning_rate": 6.95357142857143e-05,
+      "loss": 1.0886,
       "step": 3412
     },
     {
-      "epoch": 10.1,
-      "learning_rate": 3.268244575936884e-05,
-      "loss": 1.5634,
+      "epoch": 15.24,
+      "learning_rate": 6.952678571428572e-05,
+      "loss": 1.102,
       "step": 3413
     },
     {
-      "epoch": 10.1,
-      "learning_rate": 3.2662721893491124e-05,
-      "loss": 1.4827,
+      "epoch": 15.24,
+      "learning_rate": 6.951785714285714e-05,
+      "loss": 1.0873,
       "step": 3414
     },
     {
-      "epoch": 10.1,
-      "learning_rate": 3.2642998027613417e-05,
-      "loss": 1.5042,
+      "epoch": 15.25,
+      "learning_rate": 6.950892857142858e-05,
+      "loss": 1.0588,
       "step": 3415
     },
     {
-      "epoch": 10.11,
-      "learning_rate": 3.26232741617357e-05,
-      "loss": 1.5372,
+      "epoch": 15.25,
+      "learning_rate": 6.95e-05,
+      "loss": 1.14,
       "step": 3416
     },
     {
-      "epoch": 10.11,
-      "learning_rate": 3.260355029585799e-05,
-      "loss": 1.5907,
+      "epoch": 15.25,
+      "learning_rate": 6.949107142857143e-05,
+      "loss": 1.0928,
       "step": 3417
     },
     {
-      "epoch": 10.11,
-      "learning_rate": 3.258382642998028e-05,
-      "loss": 1.4247,
+      "epoch": 15.26,
+      "learning_rate": 6.948214285714286e-05,
+      "loss": 1.131,
       "step": 3418
     },
     {
-      "epoch": 10.12,
-      "learning_rate": 3.2564102564102565e-05,
-      "loss": 1.5379,
+      "epoch": 15.26,
+      "learning_rate": 6.94732142857143e-05,
+      "loss": 1.2258,
       "step": 3419
     },
     {
-      "epoch": 10.12,
-      "learning_rate": 3.254437869822485e-05,
-      "loss": 1.4803,
+      "epoch": 15.27,
+      "learning_rate": 6.946428571428571e-05,
+      "loss": 1.1115,
       "step": 3420
     },
     {
-      "epoch": 10.12,
-      "learning_rate": 3.252465483234714e-05,
-      "loss": 1.5734,
+      "epoch": 15.27,
+      "learning_rate": 6.945535714285715e-05,
+      "loss": 1.1356,
       "step": 3421
     },
     {
-      "epoch": 10.12,
-      "learning_rate": 3.250493096646943e-05,
-      "loss": 1.4542,
+      "epoch": 15.28,
+      "learning_rate": 6.944642857142858e-05,
+      "loss": 1.1451,
       "step": 3422
     },
     {
-      "epoch": 10.13,
-      "learning_rate": 3.248520710059172e-05,
-      "loss": 1.3683,
+      "epoch": 15.28,
+      "learning_rate": 6.94375e-05,
+      "loss": 1.1789,
       "step": 3423
     },
     {
-      "epoch": 10.13,
-      "learning_rate": 3.2465483234714005e-05,
-      "loss": 1.5039,
+      "epoch": 15.29,
+      "learning_rate": 6.942857142857143e-05,
+      "loss": 1.1271,
       "step": 3424
     },
     {
-      "epoch": 10.13,
-      "learning_rate": 3.244575936883629e-05,
-      "loss": 1.5679,
+      "epoch": 15.29,
+      "learning_rate": 6.941964285714286e-05,
+      "loss": 1.1637,
       "step": 3425
     },
     {
-      "epoch": 10.14,
-      "learning_rate": 3.242603550295858e-05,
-      "loss": 1.6378,
+      "epoch": 15.29,
+      "learning_rate": 6.941071428571429e-05,
+      "loss": 1.2082,
       "step": 3426
     },
     {
-      "epoch": 10.14,
-      "learning_rate": 3.2406311637080875e-05,
-      "loss": 1.4933,
+      "epoch": 15.3,
+      "learning_rate": 6.940178571428572e-05,
+      "loss": 1.0707,
       "step": 3427
     },
     {
-      "epoch": 10.14,
-      "learning_rate": 3.238658777120316e-05,
-      "loss": 1.5176,
+      "epoch": 15.3,
+      "learning_rate": 6.939285714285715e-05,
+      "loss": 1.1729,
       "step": 3428
     },
     {
-      "epoch": 10.14,
-      "learning_rate": 3.2366863905325445e-05,
-      "loss": 1.4542,
+      "epoch": 15.31,
+      "learning_rate": 6.938392857142858e-05,
+      "loss": 1.1803,
       "step": 3429
     },
     {
-      "epoch": 10.15,
-      "learning_rate": 3.234714003944773e-05,
-      "loss": 1.5378,
+      "epoch": 15.31,
+      "learning_rate": 6.9375e-05,
+      "loss": 1.1173,
       "step": 3430
     },
     {
-      "epoch": 10.15,
-      "learning_rate": 3.2327416173570016e-05,
-      "loss": 1.5982,
+      "epoch": 15.32,
+      "learning_rate": 6.936607142857143e-05,
+      "loss": 1.1587,
       "step": 3431
     },
     {
-      "epoch": 10.15,
-      "learning_rate": 3.230769230769231e-05,
-      "loss": 1.5111,
+      "epoch": 15.32,
+      "learning_rate": 6.935714285714287e-05,
+      "loss": 1.127,
       "step": 3432
     },
     {
-      "epoch": 10.16,
-      "learning_rate": 3.22879684418146e-05,
-      "loss": 1.5593,
+      "epoch": 15.33,
+      "learning_rate": 6.934821428571428e-05,
+      "loss": 1.1403,
       "step": 3433
     },
     {
-      "epoch": 10.16,
-      "learning_rate": 3.2268244575936885e-05,
-      "loss": 1.5253,
+      "epoch": 15.33,
+      "learning_rate": 6.933928571428572e-05,
+      "loss": 1.0965,
       "step": 3434
     },
     {
-      "epoch": 10.16,
-      "learning_rate": 3.224852071005917e-05,
-      "loss": 1.5219,
+      "epoch": 15.33,
+      "learning_rate": 6.933035714285715e-05,
+      "loss": 1.0894,
       "step": 3435
     },
     {
-      "epoch": 10.17,
-      "learning_rate": 3.222879684418146e-05,
-      "loss": 1.5267,
+      "epoch": 15.34,
+      "learning_rate": 6.932142857142857e-05,
+      "loss": 1.1729,
       "step": 3436
     },
     {
-      "epoch": 10.17,
-      "learning_rate": 3.220907297830375e-05,
-      "loss": 1.5461,
+      "epoch": 15.34,
+      "learning_rate": 6.93125e-05,
+      "loss": 1.0664,
       "step": 3437
     },
     {
-      "epoch": 10.17,
-      "learning_rate": 3.218934911242604e-05,
-      "loss": 1.5837,
+      "epoch": 15.35,
+      "learning_rate": 6.930357142857144e-05,
+      "loss": 1.1323,
       "step": 3438
     },
     {
-      "epoch": 10.17,
-      "learning_rate": 3.2169625246548326e-05,
-      "loss": 1.5521,
+      "epoch": 15.35,
+      "learning_rate": 6.929464285714286e-05,
+      "loss": 1.1268,
       "step": 3439
     },
     {
-      "epoch": 10.18,
-      "learning_rate": 3.214990138067061e-05,
-      "loss": 1.516,
+      "epoch": 15.36,
+      "learning_rate": 6.928571428571429e-05,
+      "loss": 1.0804,
       "step": 3440
     },
     {
-      "epoch": 10.18,
-      "learning_rate": 3.21301775147929e-05,
-      "loss": 1.5494,
+      "epoch": 15.36,
+      "learning_rate": 6.927678571428571e-05,
+      "loss": 1.0955,
       "step": 3441
     },
     {
-      "epoch": 10.18,
-      "learning_rate": 3.211045364891519e-05,
-      "loss": 1.5013,
+      "epoch": 15.37,
+      "learning_rate": 6.926785714285715e-05,
+      "loss": 1.2022,
       "step": 3442
     },
     {
-      "epoch": 10.19,
-      "learning_rate": 3.2090729783037474e-05,
-      "loss": 1.5469,
+      "epoch": 15.37,
+      "learning_rate": 6.925892857142858e-05,
+      "loss": 1.1205,
       "step": 3443
     },
     {
-      "epoch": 10.19,
-      "learning_rate": 3.2071005917159766e-05,
-      "loss": 1.5494,
+      "epoch": 15.38,
+      "learning_rate": 6.925e-05,
+      "loss": 1.1758,
       "step": 3444
     },
     {
-      "epoch": 10.19,
-      "learning_rate": 3.205128205128206e-05,
-      "loss": 1.5031,
+      "epoch": 15.38,
+      "learning_rate": 6.924107142857144e-05,
+      "loss": 1.1894,
       "step": 3445
     },
     {
-      "epoch": 10.2,
-      "learning_rate": 3.203155818540434e-05,
-      "loss": 1.4928,
+      "epoch": 15.38,
+      "learning_rate": 6.923214285714285e-05,
+      "loss": 1.096,
       "step": 3446
     },
     {
-      "epoch": 10.2,
-      "learning_rate": 3.201183431952663e-05,
-      "loss": 1.4752,
+      "epoch": 15.39,
+      "learning_rate": 6.922321428571429e-05,
+      "loss": 1.184,
       "step": 3447
     },
     {
-      "epoch": 10.2,
-      "learning_rate": 3.1992110453648914e-05,
-      "loss": 1.5546,
+      "epoch": 15.39,
+      "learning_rate": 6.921428571428572e-05,
+      "loss": 1.0997,
       "step": 3448
     },
     {
-      "epoch": 10.2,
-      "learning_rate": 3.19723865877712e-05,
-      "loss": 1.491,
+      "epoch": 15.4,
+      "learning_rate": 6.920535714285714e-05,
+      "loss": 1.1173,
       "step": 3449
     },
     {
-      "epoch": 10.21,
-      "learning_rate": 3.195266272189349e-05,
-      "loss": 1.4241,
+      "epoch": 15.4,
+      "learning_rate": 6.919642857142857e-05,
+      "loss": 1.1406,
       "step": 3450
     },
     {
-      "epoch": 10.21,
-      "learning_rate": 3.1932938856015784e-05,
-      "loss": 1.5031,
+      "epoch": 15.41,
+      "learning_rate": 6.918750000000001e-05,
+      "loss": 1.087,
       "step": 3451
     },
     {
-      "epoch": 10.21,
-      "learning_rate": 3.191321499013807e-05,
-      "loss": 1.4482,
+      "epoch": 15.41,
+      "learning_rate": 6.917857142857143e-05,
+      "loss": 1.1035,
       "step": 3452
     },
     {
-      "epoch": 10.22,
-      "learning_rate": 3.1893491124260354e-05,
-      "loss": 1.6384,
+      "epoch": 15.42,
+      "learning_rate": 6.916964285714286e-05,
+      "loss": 1.1214,
       "step": 3453
     },
     {
-      "epoch": 10.22,
-      "learning_rate": 3.187376725838264e-05,
-      "loss": 1.5158,
+      "epoch": 15.42,
+      "learning_rate": 6.916071428571428e-05,
+      "loss": 1.1624,
       "step": 3454
     },
     {
-      "epoch": 10.22,
-      "learning_rate": 3.185404339250493e-05,
-      "loss": 1.5904,
+      "epoch": 15.42,
+      "learning_rate": 6.915178571428572e-05,
+      "loss": 1.1018,
       "step": 3455
     },
     {
-      "epoch": 10.22,
-      "learning_rate": 3.1834319526627224e-05,
-      "loss": 1.5985,
+      "epoch": 15.43,
+      "learning_rate": 6.914285714285715e-05,
+      "loss": 1.0788,
       "step": 3456
     },
     {
-      "epoch": 10.23,
-      "learning_rate": 3.181459566074951e-05,
-      "loss": 1.5117,
+      "epoch": 15.43,
+      "learning_rate": 6.913392857142857e-05,
+      "loss": 1.1424,
       "step": 3457
     },
     {
-      "epoch": 10.23,
-      "learning_rate": 3.1794871794871795e-05,
-      "loss": 1.5607,
+      "epoch": 15.44,
+      "learning_rate": 6.9125e-05,
+      "loss": 1.1918,
       "step": 3458
     },
     {
-      "epoch": 10.23,
-      "learning_rate": 3.177514792899409e-05,
-      "loss": 1.5747,
+      "epoch": 15.44,
+      "learning_rate": 6.911607142857143e-05,
+      "loss": 1.1572,
       "step": 3459
     },
     {
-      "epoch": 10.24,
-      "learning_rate": 3.175542406311637e-05,
-      "loss": 1.4874,
+      "epoch": 15.45,
+      "learning_rate": 6.910714285714286e-05,
+      "loss": 1.0934,
       "step": 3460
     },
     {
-      "epoch": 10.24,
-      "learning_rate": 3.173570019723866e-05,
-      "loss": 1.5568,
+      "epoch": 15.45,
+      "learning_rate": 6.909821428571429e-05,
+      "loss": 1.128,
       "step": 3461
     },
     {
-      "epoch": 10.24,
-      "learning_rate": 3.171597633136095e-05,
-      "loss": 1.5093,
+      "epoch": 15.46,
+      "learning_rate": 6.908928571428573e-05,
+      "loss": 1.0651,
       "step": 3462
     },
     {
-      "epoch": 10.25,
-      "learning_rate": 3.1696252465483235e-05,
-      "loss": 1.5376,
+      "epoch": 15.46,
+      "learning_rate": 6.908035714285714e-05,
+      "loss": 1.0813,
       "step": 3463
     },
     {
-      "epoch": 10.25,
-      "learning_rate": 3.167652859960553e-05,
-      "loss": 1.5561,
+      "epoch": 15.46,
+      "learning_rate": 6.907142857142858e-05,
+      "loss": 1.0627,
       "step": 3464
     },
     {
-      "epoch": 10.25,
-      "learning_rate": 3.165680473372781e-05,
-      "loss": 1.5051,
+      "epoch": 15.47,
+      "learning_rate": 6.90625e-05,
+      "loss": 1.1548,
       "step": 3465
     },
     {
-      "epoch": 10.25,
-      "learning_rate": 3.16370808678501e-05,
-      "loss": 1.516,
+      "epoch": 15.47,
+      "learning_rate": 6.905357142857143e-05,
+      "loss": 1.075,
       "step": 3466
     },
     {
-      "epoch": 10.26,
-      "learning_rate": 3.161735700197239e-05,
-      "loss": 1.4518,
+      "epoch": 15.48,
+      "learning_rate": 6.904464285714286e-05,
+      "loss": 1.1624,
       "step": 3467
     },
     {
-      "epoch": 10.26,
-      "learning_rate": 3.159763313609468e-05,
-      "loss": 1.5867,
+      "epoch": 15.48,
+      "learning_rate": 6.90357142857143e-05,
+      "loss": 1.1436,
       "step": 3468
     },
     {
-      "epoch": 10.26,
-      "learning_rate": 3.157790927021697e-05,
-      "loss": 1.5351,
+      "epoch": 15.49,
+      "learning_rate": 6.90267857142857e-05,
+      "loss": 1.1538,
       "step": 3469
     },
     {
-      "epoch": 10.27,
-      "learning_rate": 3.155818540433925e-05,
-      "loss": 1.5367,
+      "epoch": 15.49,
+      "learning_rate": 6.901785714285715e-05,
+      "loss": 1.1676,
       "step": 3470
     },
     {
-      "epoch": 10.27,
-      "learning_rate": 3.153846153846154e-05,
-      "loss": 1.5745,
+      "epoch": 15.5,
+      "learning_rate": 6.900892857142857e-05,
+      "loss": 1.0779,
       "step": 3471
     },
     {
-      "epoch": 10.27,
-      "learning_rate": 3.151873767258382e-05,
-      "loss": 1.5279,
+      "epoch": 15.5,
+      "learning_rate": 6.9e-05,
+      "loss": 1.1156,
       "step": 3472
     },
     {
-      "epoch": 10.28,
-      "learning_rate": 3.1499013806706115e-05,
-      "loss": 1.571,
+      "epoch": 15.5,
+      "learning_rate": 6.899107142857144e-05,
+      "loss": 1.0927,
       "step": 3473
     },
     {
-      "epoch": 10.28,
-      "learning_rate": 3.147928994082841e-05,
-      "loss": 1.5501,
+      "epoch": 15.51,
+      "learning_rate": 6.898214285714286e-05,
+      "loss": 1.1603,
       "step": 3474
     },
     {
-      "epoch": 10.28,
-      "learning_rate": 3.145956607495069e-05,
-      "loss": 1.5977,
+      "epoch": 15.51,
+      "learning_rate": 6.89732142857143e-05,
+      "loss": 1.076,
       "step": 3475
     },
     {
-      "epoch": 10.28,
-      "learning_rate": 3.143984220907298e-05,
-      "loss": 1.4951,
+      "epoch": 15.52,
+      "learning_rate": 6.896428571428571e-05,
+      "loss": 1.1402,
       "step": 3476
     },
     {
-      "epoch": 10.29,
-      "learning_rate": 3.142011834319526e-05,
-      "loss": 1.525,
+      "epoch": 15.52,
+      "learning_rate": 6.895535714285715e-05,
+      "loss": 1.2214,
       "step": 3477
     },
     {
-      "epoch": 10.29,
-      "learning_rate": 3.1400394477317555e-05,
-      "loss": 1.5636,
+      "epoch": 15.53,
+      "learning_rate": 6.894642857142858e-05,
+      "loss": 1.2231,
       "step": 3478
     },
     {
-      "epoch": 10.29,
-      "learning_rate": 3.138067061143984e-05,
-      "loss": 1.519,
+      "epoch": 15.53,
+      "learning_rate": 6.89375e-05,
+      "loss": 1.108,
       "step": 3479
     },
     {
-      "epoch": 10.3,
-      "learning_rate": 3.136094674556213e-05,
-      "loss": 1.4588,
+      "epoch": 15.54,
+      "learning_rate": 6.892857142857143e-05,
+      "loss": 1.1273,
       "step": 3480
     },
     {
-      "epoch": 10.3,
-      "learning_rate": 3.134122287968442e-05,
-      "loss": 1.5145,
+      "epoch": 15.54,
+      "learning_rate": 6.891964285714287e-05,
+      "loss": 1.0949,
       "step": 3481
     },
     {
-      "epoch": 10.3,
-      "learning_rate": 3.132149901380671e-05,
-      "loss": 1.5179,
+      "epoch": 15.54,
+      "learning_rate": 6.891071428571428e-05,
+      "loss": 1.153,
       "step": 3482
     },
     {
-      "epoch": 10.3,
-      "learning_rate": 3.1301775147928996e-05,
-      "loss": 1.4455,
+      "epoch": 15.55,
+      "learning_rate": 6.890178571428572e-05,
+      "loss": 1.0661,
       "step": 3483
     },
     {
-      "epoch": 10.31,
-      "learning_rate": 3.128205128205128e-05,
-      "loss": 1.4993,
+      "epoch": 15.55,
+      "learning_rate": 6.889285714285714e-05,
+      "loss": 1.1254,
       "step": 3484
     },
     {
-      "epoch": 10.31,
-      "learning_rate": 3.126232741617357e-05,
-      "loss": 1.5427,
+      "epoch": 15.56,
+      "learning_rate": 6.888392857142858e-05,
+      "loss": 1.1693,
       "step": 3485
     },
     {
-      "epoch": 10.31,
-      "learning_rate": 3.124260355029586e-05,
-      "loss": 1.475,
+      "epoch": 15.56,
+      "learning_rate": 6.887500000000001e-05,
+      "loss": 1.1471,
       "step": 3486
     },
     {
-      "epoch": 10.32,
-      "learning_rate": 3.122287968441815e-05,
-      "loss": 1.4151,
+      "epoch": 15.57,
+      "learning_rate": 6.886607142857143e-05,
+      "loss": 1.0956,
       "step": 3487
     },
     {
-      "epoch": 10.32,
-      "learning_rate": 3.1203155818540436e-05,
-      "loss": 1.4675,
+      "epoch": 15.57,
+      "learning_rate": 6.885714285714286e-05,
+      "loss": 1.1213,
       "step": 3488
     },
     {
-      "epoch": 10.32,
-      "learning_rate": 3.118343195266272e-05,
-      "loss": 1.5941,
+      "epoch": 15.58,
+      "learning_rate": 6.884821428571428e-05,
+      "loss": 1.1089,
       "step": 3489
     },
     {
-      "epoch": 10.33,
-      "learning_rate": 3.116370808678501e-05,
-      "loss": 1.4751,
+      "epoch": 15.58,
+      "learning_rate": 6.883928571428572e-05,
+      "loss": 1.1172,
       "step": 3490
     },
     {
-      "epoch": 10.33,
-      "learning_rate": 3.11439842209073e-05,
-      "loss": 1.4469,
+      "epoch": 15.58,
+      "learning_rate": 6.883035714285715e-05,
+      "loss": 1.1101,
       "step": 3491
     },
     {
-      "epoch": 10.33,
-      "learning_rate": 3.112426035502959e-05,
-      "loss": 1.5247,
+      "epoch": 15.59,
+      "learning_rate": 6.882142857142857e-05,
+      "loss": 1.0956,
       "step": 3492
     },
     {
-      "epoch": 10.33,
-      "learning_rate": 3.1104536489151876e-05,
-      "loss": 1.6142,
+      "epoch": 15.59,
+      "learning_rate": 6.88125e-05,
+      "loss": 1.1318,
       "step": 3493
     },
     {
-      "epoch": 10.34,
-      "learning_rate": 3.108481262327416e-05,
-      "loss": 1.4219,
+      "epoch": 15.6,
+      "learning_rate": 6.880357142857144e-05,
+      "loss": 1.1576,
       "step": 3494
     },
     {
-      "epoch": 10.34,
-      "learning_rate": 3.106508875739645e-05,
-      "loss": 1.4932,
+      "epoch": 15.6,
+      "learning_rate": 6.879464285714286e-05,
+      "loss": 1.1308,
       "step": 3495
     },
     {
-      "epoch": 10.34,
-      "learning_rate": 3.104536489151874e-05,
-      "loss": 1.5612,
+      "epoch": 15.61,
+      "learning_rate": 6.878571428571429e-05,
+      "loss": 1.1255,
       "step": 3496
     },
     {
-      "epoch": 10.35,
-      "learning_rate": 3.102564102564103e-05,
-      "loss": 1.5159,
+      "epoch": 15.61,
+      "learning_rate": 6.877678571428571e-05,
+      "loss": 1.078,
       "step": 3497
     },
     {
-      "epoch": 10.35,
-      "learning_rate": 3.1005917159763316e-05,
-      "loss": 1.613,
+      "epoch": 15.62,
+      "learning_rate": 6.876785714285715e-05,
+      "loss": 1.1886,
       "step": 3498
     },
     {
-      "epoch": 10.35,
-      "learning_rate": 3.09861932938856e-05,
-      "loss": 1.5355,
+      "epoch": 15.62,
+      "learning_rate": 6.875892857142857e-05,
+      "loss": 1.1626,
       "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,
+      "epoch": 15.62,
+      "learning_rate": 6.875e-05,
+      "loss": 1.0797,
       "step": 3500
     },
     {
-      "epoch": 10.36,
-      "learning_rate": 3.094674556213018e-05,
-      "loss": 1.6542,
+      "epoch": 15.63,
+      "learning_rate": 6.874107142857143e-05,
+      "loss": 1.1157,
       "step": 3501
     },
     {
-      "epoch": 10.36,
-      "learning_rate": 3.0927021696252465e-05,
-      "loss": 1.5627,
+      "epoch": 15.63,
+      "learning_rate": 6.873214285714286e-05,
+      "loss": 1.1743,
       "step": 3502
     },
     {
-      "epoch": 10.36,
-      "learning_rate": 3.090729783037476e-05,
-      "loss": 1.5776,
+      "epoch": 15.64,
+      "learning_rate": 6.87232142857143e-05,
+      "loss": 1.1185,
       "step": 3503
     },
     {
-      "epoch": 10.37,
-      "learning_rate": 3.088757396449704e-05,
-      "loss": 1.5314,
+      "epoch": 15.64,
+      "learning_rate": 6.871428571428572e-05,
+      "loss": 1.1279,
       "step": 3504
     },
     {
-      "epoch": 10.37,
-      "learning_rate": 3.0867850098619334e-05,
-      "loss": 1.44,
+      "epoch": 15.65,
+      "learning_rate": 6.870535714285714e-05,
+      "loss": 1.1091,
       "step": 3505
     },
     {
-      "epoch": 10.37,
-      "learning_rate": 3.084812623274162e-05,
-      "loss": 1.6514,
+      "epoch": 15.65,
+      "learning_rate": 6.869642857142857e-05,
+      "loss": 1.1687,
       "step": 3506
     },
     {
-      "epoch": 10.38,
-      "learning_rate": 3.0828402366863905e-05,
-      "loss": 1.5413,
+      "epoch": 15.66,
+      "learning_rate": 6.868750000000001e-05,
+      "loss": 1.1346,
       "step": 3507
     },
     {
-      "epoch": 10.38,
-      "learning_rate": 3.080867850098619e-05,
-      "loss": 1.6208,
+      "epoch": 15.66,
+      "learning_rate": 6.867857142857143e-05,
+      "loss": 1.1759,
       "step": 3508
     },
     {
-      "epoch": 10.38,
-      "learning_rate": 3.078895463510848e-05,
-      "loss": 1.5506,
+      "epoch": 15.67,
+      "learning_rate": 6.866964285714286e-05,
+      "loss": 1.2036,
       "step": 3509
     },
     {
-      "epoch": 10.38,
-      "learning_rate": 3.0769230769230774e-05,
-      "loss": 1.5627,
+      "epoch": 15.67,
+      "learning_rate": 6.866071428571429e-05,
+      "loss": 1.1567,
       "step": 3510
     },
     {
-      "epoch": 10.39,
-      "learning_rate": 3.074950690335306e-05,
-      "loss": 1.4799,
+      "epoch": 15.67,
+      "learning_rate": 6.865178571428572e-05,
+      "loss": 0.9894,
       "step": 3511
     },
     {
-      "epoch": 10.39,
-      "learning_rate": 3.0729783037475345e-05,
-      "loss": 1.4755,
+      "epoch": 15.68,
+      "learning_rate": 6.864285714285714e-05,
+      "loss": 1.0704,
       "step": 3512
     },
     {
-      "epoch": 10.39,
-      "learning_rate": 3.071005917159763e-05,
-      "loss": 1.5916,
+      "epoch": 15.68,
+      "learning_rate": 6.863392857142858e-05,
+      "loss": 1.1683,
       "step": 3513
     },
     {
-      "epoch": 10.4,
-      "learning_rate": 3.069033530571992e-05,
-      "loss": 1.4316,
+      "epoch": 15.69,
+      "learning_rate": 6.8625e-05,
+      "loss": 1.166,
       "step": 3514
     },
     {
-      "epoch": 10.4,
-      "learning_rate": 3.0670611439842215e-05,
-      "loss": 1.5374,
+      "epoch": 15.69,
+      "learning_rate": 6.861607142857143e-05,
+      "loss": 1.2364,
       "step": 3515
     },
     {
-      "epoch": 10.4,
-      "learning_rate": 3.06508875739645e-05,
-      "loss": 1.5939,
+      "epoch": 15.7,
+      "learning_rate": 6.860714285714285e-05,
+      "loss": 1.1398,
       "step": 3516
     },
     {
-      "epoch": 10.41,
-      "learning_rate": 3.0631163708086785e-05,
-      "loss": 1.5836,
+      "epoch": 15.7,
+      "learning_rate": 6.859821428571429e-05,
+      "loss": 1.117,
       "step": 3517
     },
     {
-      "epoch": 10.41,
-      "learning_rate": 3.061143984220907e-05,
-      "loss": 1.6008,
+      "epoch": 15.71,
+      "learning_rate": 6.858928571428572e-05,
+      "loss": 1.1776,
       "step": 3518
     },
     {
-      "epoch": 10.41,
-      "learning_rate": 3.059171597633136e-05,
-      "loss": 1.4781,
+      "epoch": 15.71,
+      "learning_rate": 6.858035714285714e-05,
+      "loss": 1.0951,
       "step": 3519
     },
     {
-      "epoch": 10.41,
-      "learning_rate": 3.057199211045365e-05,
-      "loss": 1.4561,
+      "epoch": 15.71,
+      "learning_rate": 6.857142857142858e-05,
+      "loss": 1.1235,
       "step": 3520
     },
     {
-      "epoch": 10.42,
-      "learning_rate": 3.055226824457594e-05,
-      "loss": 1.4335,
+      "epoch": 15.72,
+      "learning_rate": 6.85625e-05,
+      "loss": 1.0536,
       "step": 3521
     },
     {
-      "epoch": 10.42,
-      "learning_rate": 3.0532544378698226e-05,
-      "loss": 1.5363,
+      "epoch": 15.72,
+      "learning_rate": 6.855357142857143e-05,
+      "loss": 1.0867,
       "step": 3522
     },
     {
-      "epoch": 10.42,
-      "learning_rate": 3.0512820512820518e-05,
-      "loss": 1.5746,
+      "epoch": 15.73,
+      "learning_rate": 6.854464285714286e-05,
+      "loss": 1.1127,
       "step": 3523
     },
     {
-      "epoch": 10.43,
-      "learning_rate": 3.0493096646942803e-05,
-      "loss": 1.5291,
+      "epoch": 15.73,
+      "learning_rate": 6.85357142857143e-05,
+      "loss": 1.124,
       "step": 3524
     },
     {
-      "epoch": 10.43,
-      "learning_rate": 3.047337278106509e-05,
-      "loss": 1.5459,
+      "epoch": 15.74,
+      "learning_rate": 6.852678571428571e-05,
+      "loss": 1.1198,
       "step": 3525
     },
     {
-      "epoch": 10.43,
-      "learning_rate": 3.0453648915187377e-05,
-      "loss": 1.5609,
+      "epoch": 15.74,
+      "learning_rate": 6.851785714285715e-05,
+      "loss": 1.0854,
       "step": 3526
     },
     {
-      "epoch": 10.43,
-      "learning_rate": 3.0433925049309662e-05,
-      "loss": 1.5585,
+      "epoch": 15.75,
+      "learning_rate": 6.850892857142857e-05,
+      "loss": 1.1086,
       "step": 3527
     },
     {
-      "epoch": 10.44,
-      "learning_rate": 3.0414201183431958e-05,
-      "loss": 1.4702,
+      "epoch": 15.75,
+      "learning_rate": 6.850000000000001e-05,
+      "loss": 1.0882,
       "step": 3528
     },
     {
-      "epoch": 10.44,
-      "learning_rate": 3.0394477317554243e-05,
-      "loss": 1.561,
+      "epoch": 15.75,
+      "learning_rate": 6.849107142857142e-05,
+      "loss": 1.0901,
       "step": 3529
     },
     {
-      "epoch": 10.44,
-      "learning_rate": 3.037475345167653e-05,
-      "loss": 1.5363,
+      "epoch": 15.76,
+      "learning_rate": 6.848214285714286e-05,
+      "loss": 1.2235,
       "step": 3530
     },
     {
-      "epoch": 10.45,
-      "learning_rate": 3.0355029585798817e-05,
-      "loss": 1.5827,
+      "epoch": 15.76,
+      "learning_rate": 6.847321428571429e-05,
+      "loss": 1.1279,
       "step": 3531
     },
     {
-      "epoch": 10.45,
-      "learning_rate": 3.0335305719921103e-05,
-      "loss": 1.468,
+      "epoch": 15.77,
+      "learning_rate": 6.846428571428571e-05,
+      "loss": 1.1904,
       "step": 3532
     },
     {
-      "epoch": 10.45,
-      "learning_rate": 3.0315581854043395e-05,
-      "loss": 1.6421,
+      "epoch": 15.77,
+      "learning_rate": 6.845535714285715e-05,
+      "loss": 1.0639,
       "step": 3533
     },
     {
-      "epoch": 10.46,
-      "learning_rate": 3.0295857988165683e-05,
-      "loss": 1.5592,
+      "epoch": 15.78,
+      "learning_rate": 6.844642857142858e-05,
+      "loss": 1.1109,
       "step": 3534
     },
     {
-      "epoch": 10.46,
-      "learning_rate": 3.027613412228797e-05,
-      "loss": 1.5378,
+      "epoch": 15.78,
+      "learning_rate": 6.84375e-05,
+      "loss": 1.1509,
       "step": 3535
     },
     {
-      "epoch": 10.46,
-      "learning_rate": 3.0256410256410257e-05,
-      "loss": 1.5508,
+      "epoch": 15.79,
+      "learning_rate": 6.842857142857143e-05,
+      "loss": 1.1089,
       "step": 3536
     },
     {
-      "epoch": 10.46,
-      "learning_rate": 3.023668639053255e-05,
-      "loss": 1.4125,
+      "epoch": 15.79,
+      "learning_rate": 6.841964285714287e-05,
+      "loss": 1.1939,
       "step": 3537
     },
     {
-      "epoch": 10.47,
-      "learning_rate": 3.0216962524654835e-05,
-      "loss": 1.5793,
+      "epoch": 15.79,
+      "learning_rate": 6.841071428571428e-05,
+      "loss": 1.1213,
       "step": 3538
     },
     {
-      "epoch": 10.47,
-      "learning_rate": 3.019723865877712e-05,
-      "loss": 1.5848,
+      "epoch": 15.8,
+      "learning_rate": 6.840178571428572e-05,
+      "loss": 1.0806,
       "step": 3539
     },
     {
-      "epoch": 10.47,
-      "learning_rate": 3.017751479289941e-05,
-      "loss": 1.4227,
+      "epoch": 15.8,
+      "learning_rate": 6.839285714285714e-05,
+      "loss": 1.1405,
       "step": 3540
     },
     {
-      "epoch": 10.48,
-      "learning_rate": 3.0157790927021694e-05,
-      "loss": 1.5197,
+      "epoch": 15.81,
+      "learning_rate": 6.838392857142858e-05,
+      "loss": 1.1779,
       "step": 3541
     },
     {
-      "epoch": 10.48,
-      "learning_rate": 3.0138067061143986e-05,
-      "loss": 1.5711,
+      "epoch": 15.81,
+      "learning_rate": 6.8375e-05,
+      "loss": 1.2021,
       "step": 3542
     },
     {
-      "epoch": 10.48,
-      "learning_rate": 3.0118343195266275e-05,
-      "loss": 1.3745,
+      "epoch": 15.82,
+      "learning_rate": 6.836607142857143e-05,
+      "loss": 1.1463,
       "step": 3543
     },
     {
-      "epoch": 10.49,
-      "learning_rate": 3.009861932938856e-05,
-      "loss": 1.5391,
+      "epoch": 15.82,
+      "learning_rate": 6.835714285714286e-05,
+      "loss": 1.1121,
       "step": 3544
     },
     {
-      "epoch": 10.49,
-      "learning_rate": 3.007889546351085e-05,
-      "loss": 1.5347,
+      "epoch": 15.83,
+      "learning_rate": 6.834821428571428e-05,
+      "loss": 1.1372,
       "step": 3545
     },
     {
-      "epoch": 10.49,
-      "learning_rate": 3.005917159763314e-05,
-      "loss": 1.5624,
+      "epoch": 15.83,
+      "learning_rate": 6.833928571428571e-05,
+      "loss": 1.1222,
       "step": 3546
     },
     {
-      "epoch": 10.49,
-      "learning_rate": 3.0039447731755427e-05,
-      "loss": 1.4337,
+      "epoch": 15.83,
+      "learning_rate": 6.833035714285715e-05,
+      "loss": 1.1453,
       "step": 3547
     },
     {
-      "epoch": 10.5,
-      "learning_rate": 3.0019723865877712e-05,
-      "loss": 1.5516,
+      "epoch": 15.84,
+      "learning_rate": 6.832142857142857e-05,
+      "loss": 1.1394,
       "step": 3548
     },
     {
-      "epoch": 10.5,
-      "learning_rate": 3e-05,
-      "loss": 1.4535,
+      "epoch": 15.84,
+      "learning_rate": 6.83125e-05,
+      "loss": 1.0813,
       "step": 3549
     },
     {
-      "epoch": 10.5,
-      "learning_rate": 2.9980276134122286e-05,
-      "loss": 1.5161,
+      "epoch": 15.85,
+      "learning_rate": 6.830357142857144e-05,
+      "loss": 1.1064,
       "step": 3550
     },
     {
-      "epoch": 10.51,
-      "learning_rate": 2.9960552268244578e-05,
-      "loss": 1.5192,
+      "epoch": 15.85,
+      "learning_rate": 6.829464285714286e-05,
+      "loss": 1.1788,
       "step": 3551
     },
     {
-      "epoch": 10.51,
-      "learning_rate": 2.9940828402366867e-05,
-      "loss": 1.6124,
+      "epoch": 15.86,
+      "learning_rate": 6.828571428571429e-05,
+      "loss": 1.1937,
       "step": 3552
     },
     {
-      "epoch": 10.51,
-      "learning_rate": 2.9921104536489152e-05,
-      "loss": 1.5605,
+      "epoch": 15.86,
+      "learning_rate": 6.827678571428572e-05,
+      "loss": 1.1369,
       "step": 3553
     },
     {
-      "epoch": 10.51,
-      "learning_rate": 2.990138067061144e-05,
-      "loss": 1.5195,
+      "epoch": 15.87,
+      "learning_rate": 6.826785714285715e-05,
+      "loss": 1.1785,
       "step": 3554
     },
     {
-      "epoch": 10.52,
-      "learning_rate": 2.9881656804733733e-05,
-      "loss": 1.5202,
+      "epoch": 15.87,
+      "learning_rate": 6.825892857142857e-05,
+      "loss": 1.101,
       "step": 3555
     },
     {
-      "epoch": 10.52,
-      "learning_rate": 2.986193293885602e-05,
-      "loss": 1.5125,
+      "epoch": 15.88,
+      "learning_rate": 6.825e-05,
+      "loss": 1.0787,
       "step": 3556
     },
     {
-      "epoch": 10.52,
-      "learning_rate": 2.9842209072978307e-05,
-      "loss": 1.5164,
+      "epoch": 15.88,
+      "learning_rate": 6.824107142857143e-05,
+      "loss": 1.0837,
       "step": 3557
     },
     {
-      "epoch": 10.53,
-      "learning_rate": 2.9822485207100593e-05,
-      "loss": 1.4881,
+      "epoch": 15.88,
+      "learning_rate": 6.823214285714286e-05,
+      "loss": 1.098,
       "step": 3558
     },
     {
-      "epoch": 10.53,
-      "learning_rate": 2.9802761341222878e-05,
-      "loss": 1.5173,
+      "epoch": 15.89,
+      "learning_rate": 6.822321428571428e-05,
+      "loss": 1.0934,
       "step": 3559
     },
     {
-      "epoch": 10.53,
-      "learning_rate": 2.978303747534517e-05,
-      "loss": 1.4304,
+      "epoch": 15.89,
+      "learning_rate": 6.821428571428572e-05,
+      "loss": 1.1119,
       "step": 3560
     },
     {
-      "epoch": 10.54,
-      "learning_rate": 2.976331360946746e-05,
-      "loss": 1.4912,
+      "epoch": 15.9,
+      "learning_rate": 6.820535714285715e-05,
+      "loss": 1.0661,
       "step": 3561
     },
     {
-      "epoch": 10.54,
-      "learning_rate": 2.9743589743589744e-05,
-      "loss": 1.5221,
+      "epoch": 15.9,
+      "learning_rate": 6.819642857142857e-05,
+      "loss": 1.0725,
       "step": 3562
     },
     {
-      "epoch": 10.54,
-      "learning_rate": 2.9723865877712033e-05,
-      "loss": 1.5276,
+      "epoch": 15.91,
+      "learning_rate": 6.818750000000001e-05,
+      "loss": 1.0622,
       "step": 3563
     },
     {
-      "epoch": 10.54,
-      "learning_rate": 2.9704142011834318e-05,
-      "loss": 1.5275,
+      "epoch": 15.91,
+      "learning_rate": 6.817857142857144e-05,
+      "loss": 1.122,
       "step": 3564
     },
     {
-      "epoch": 10.55,
-      "learning_rate": 2.968441814595661e-05,
-      "loss": 1.5936,
+      "epoch": 15.92,
+      "learning_rate": 6.816964285714286e-05,
+      "loss": 1.193,
       "step": 3565
     },
     {
-      "epoch": 10.55,
-      "learning_rate": 2.96646942800789e-05,
-      "loss": 1.5841,
+      "epoch": 15.92,
+      "learning_rate": 6.816071428571429e-05,
+      "loss": 1.1035,
       "step": 3566
     },
     {
-      "epoch": 10.55,
-      "learning_rate": 2.9644970414201184e-05,
-      "loss": 1.3831,
+      "epoch": 15.92,
+      "learning_rate": 6.815178571428573e-05,
+      "loss": 1.0267,
       "step": 3567
     },
     {
-      "epoch": 10.56,
-      "learning_rate": 2.962524654832347e-05,
-      "loss": 1.3958,
+      "epoch": 15.93,
+      "learning_rate": 6.814285714285714e-05,
+      "loss": 1.1782,
       "step": 3568
     },
     {
-      "epoch": 10.56,
-      "learning_rate": 2.9605522682445762e-05,
-      "loss": 1.4824,
+      "epoch": 15.93,
+      "learning_rate": 6.813392857142858e-05,
+      "loss": 1.1144,
       "step": 3569
     },
     {
-      "epoch": 10.56,
-      "learning_rate": 2.958579881656805e-05,
-      "loss": 1.5896,
+      "epoch": 15.94,
+      "learning_rate": 6.8125e-05,
+      "loss": 1.1438,
       "step": 3570
     },
     {
-      "epoch": 10.57,
-      "learning_rate": 2.9566074950690336e-05,
-      "loss": 1.4243,
+      "epoch": 15.94,
+      "learning_rate": 6.811607142857143e-05,
+      "loss": 1.2122,
       "step": 3571
     },
     {
-      "epoch": 10.57,
-      "learning_rate": 2.9546351084812625e-05,
-      "loss": 1.4982,
+      "epoch": 15.95,
+      "learning_rate": 6.810714285714285e-05,
+      "loss": 1.1416,
       "step": 3572
     },
     {
-      "epoch": 10.57,
-      "learning_rate": 2.952662721893491e-05,
-      "loss": 1.5514,
+      "epoch": 15.95,
+      "learning_rate": 6.809821428571429e-05,
+      "loss": 1.0775,
       "step": 3573
     },
     {
-      "epoch": 10.57,
-      "learning_rate": 2.9506903353057202e-05,
-      "loss": 1.4735,
+      "epoch": 15.96,
+      "learning_rate": 6.808928571428572e-05,
+      "loss": 1.1502,
       "step": 3574
     },
     {
-      "epoch": 10.58,
-      "learning_rate": 2.948717948717949e-05,
-      "loss": 1.4486,
+      "epoch": 15.96,
+      "learning_rate": 6.808035714285714e-05,
+      "loss": 1.0652,
       "step": 3575
     },
     {
-      "epoch": 10.58,
-      "learning_rate": 2.9467455621301776e-05,
-      "loss": 1.5311,
+      "epoch": 15.96,
+      "learning_rate": 6.807142857142857e-05,
+      "loss": 1.15,
       "step": 3576
     },
     {
-      "epoch": 10.58,
-      "learning_rate": 2.944773175542406e-05,
-      "loss": 1.4919,
+      "epoch": 15.97,
+      "learning_rate": 6.806250000000001e-05,
+      "loss": 1.1416,
       "step": 3577
     },
     {
-      "epoch": 10.59,
-      "learning_rate": 2.9428007889546353e-05,
-      "loss": 1.5512,
+      "epoch": 15.97,
+      "learning_rate": 6.805357142857143e-05,
+      "loss": 1.128,
       "step": 3578
     },
     {
-      "epoch": 10.59,
-      "learning_rate": 2.9408284023668642e-05,
-      "loss": 1.5282,
+      "epoch": 15.98,
+      "learning_rate": 6.804464285714286e-05,
+      "loss": 1.0954,
       "step": 3579
     },
     {
-      "epoch": 10.59,
-      "learning_rate": 2.9388560157790928e-05,
-      "loss": 1.6092,
+      "epoch": 15.98,
+      "learning_rate": 6.80357142857143e-05,
+      "loss": 1.2083,
       "step": 3580
     },
     {
-      "epoch": 10.59,
-      "learning_rate": 2.9368836291913216e-05,
-      "loss": 1.5604,
+      "epoch": 15.99,
+      "learning_rate": 6.802678571428571e-05,
+      "loss": 1.1139,
       "step": 3581
     },
     {
-      "epoch": 10.6,
-      "learning_rate": 2.93491124260355e-05,
-      "loss": 1.5878,
+      "epoch": 15.99,
+      "learning_rate": 6.801785714285715e-05,
+      "loss": 1.1001,
       "step": 3582
     },
     {
-      "epoch": 10.6,
-      "learning_rate": 2.9329388560157794e-05,
-      "loss": 1.5473,
+      "epoch": 16.0,
+      "learning_rate": 6.800892857142857e-05,
+      "loss": 1.2221,
       "step": 3583
     },
     {
-      "epoch": 10.6,
-      "learning_rate": 2.9309664694280082e-05,
-      "loss": 1.6231,
+      "epoch": 16.0,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.0189,
       "step": 3584
     },
     {
-      "epoch": 10.61,
-      "learning_rate": 2.9289940828402368e-05,
-      "loss": 1.6238,
+      "epoch": 16.0,
+      "learning_rate": 6.799107142857142e-05,
+      "loss": 1.1144,
       "step": 3585
     },
     {
-      "epoch": 10.61,
-      "learning_rate": 2.9270216962524653e-05,
-      "loss": 1.5918,
+      "epoch": 16.01,
+      "learning_rate": 6.798214285714286e-05,
+      "loss": 1.1943,
       "step": 3586
     },
     {
-      "epoch": 10.61,
-      "learning_rate": 2.9250493096646942e-05,
-      "loss": 1.5922,
+      "epoch": 16.01,
+      "learning_rate": 6.797321428571429e-05,
+      "loss": 1.1398,
       "step": 3587
     },
     {
-      "epoch": 10.62,
-      "learning_rate": 2.9230769230769234e-05,
-      "loss": 1.5254,
+      "epoch": 16.02,
+      "learning_rate": 6.796428571428571e-05,
+      "loss": 1.1294,
       "step": 3588
     },
     {
-      "epoch": 10.62,
-      "learning_rate": 2.921104536489152e-05,
-      "loss": 1.5017,
+      "epoch": 16.02,
+      "learning_rate": 6.795535714285714e-05,
+      "loss": 1.0848,
       "step": 3589
     },
     {
-      "epoch": 10.62,
-      "learning_rate": 2.9191321499013808e-05,
-      "loss": 1.6447,
+      "epoch": 16.03,
+      "learning_rate": 6.794642857142858e-05,
+      "loss": 1.1943,
       "step": 3590
     },
     {
-      "epoch": 10.62,
-      "learning_rate": 2.9171597633136093e-05,
-      "loss": 1.5452,
+      "epoch": 16.03,
+      "learning_rate": 6.79375e-05,
+      "loss": 1.1094,
       "step": 3591
     },
     {
-      "epoch": 10.63,
-      "learning_rate": 2.9151873767258385e-05,
-      "loss": 1.6336,
+      "epoch": 16.04,
+      "learning_rate": 6.792857142857143e-05,
+      "loss": 1.1634,
       "step": 3592
     },
     {
-      "epoch": 10.63,
-      "learning_rate": 2.9132149901380674e-05,
-      "loss": 1.5082,
+      "epoch": 16.04,
+      "learning_rate": 6.791964285714286e-05,
+      "loss": 1.19,
       "step": 3593
     },
     {
-      "epoch": 10.63,
-      "learning_rate": 2.911242603550296e-05,
-      "loss": 1.4503,
+      "epoch": 16.04,
+      "learning_rate": 6.79107142857143e-05,
+      "loss": 1.058,
       "step": 3594
     },
     {
-      "epoch": 10.64,
-      "learning_rate": 2.9092702169625248e-05,
-      "loss": 1.5017,
+      "epoch": 16.05,
+      "learning_rate": 6.790178571428572e-05,
+      "loss": 1.0918,
       "step": 3595
     },
     {
-      "epoch": 10.64,
-      "learning_rate": 2.9072978303747534e-05,
-      "loss": 1.5083,
+      "epoch": 16.05,
+      "learning_rate": 6.789285714285715e-05,
+      "loss": 1.1446,
       "step": 3596
     },
     {
-      "epoch": 10.64,
-      "learning_rate": 2.9053254437869826e-05,
-      "loss": 1.5745,
+      "epoch": 16.06,
+      "learning_rate": 6.788392857142858e-05,
+      "loss": 1.098,
       "step": 3597
     },
     {
-      "epoch": 10.64,
-      "learning_rate": 2.903353057199211e-05,
-      "loss": 1.5859,
+      "epoch": 16.06,
+      "learning_rate": 6.7875e-05,
+      "loss": 1.1299,
       "step": 3598
     },
     {
-      "epoch": 10.65,
-      "learning_rate": 2.90138067061144e-05,
-      "loss": 1.4949,
+      "epoch": 16.07,
+      "learning_rate": 6.786607142857144e-05,
+      "loss": 1.1435,
       "step": 3599
     },
     {
-      "epoch": 10.65,
-      "learning_rate": 2.8994082840236685e-05,
-      "loss": 1.657,
+      "epoch": 16.07,
+      "learning_rate": 6.785714285714286e-05,
+      "loss": 1.1051,
       "step": 3600
     },
     {
-      "epoch": 10.65,
-      "learning_rate": 2.8974358974358977e-05,
-      "loss": 1.569,
+      "epoch": 16.07,
+      "eval_loss": 1.1463580131530762,
+      "eval_runtime": 45.1212,
+      "eval_samples_per_second": 290.196,
+      "eval_steps_per_second": 1.529,
+      "step": 3600
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 6.784821428571429e-05,
+      "loss": 1.1537,
       "step": 3601
     },
     {
-      "epoch": 10.66,
-      "learning_rate": 2.8954635108481266e-05,
-      "loss": 1.6132,
+      "epoch": 16.08,
+      "learning_rate": 6.783928571428571e-05,
+      "loss": 1.1462,
       "step": 3602
     },
     {
-      "epoch": 10.66,
-      "learning_rate": 2.893491124260355e-05,
-      "loss": 1.5582,
+      "epoch": 16.08,
+      "learning_rate": 6.783035714285715e-05,
+      "loss": 1.0623,
       "step": 3603
     },
     {
-      "epoch": 10.66,
-      "learning_rate": 2.891518737672584e-05,
-      "loss": 1.4496,
+      "epoch": 16.09,
+      "learning_rate": 6.782142857142858e-05,
+      "loss": 1.0968,
       "step": 3604
     },
     {
-      "epoch": 10.67,
-      "learning_rate": 2.8895463510848125e-05,
-      "loss": 1.6329,
+      "epoch": 16.09,
+      "learning_rate": 6.78125e-05,
+      "loss": 1.0921,
       "step": 3605
     },
     {
-      "epoch": 10.67,
-      "learning_rate": 2.8875739644970417e-05,
-      "loss": 1.5437,
+      "epoch": 16.1,
+      "learning_rate": 6.780357142857143e-05,
+      "loss": 1.1099,
       "step": 3606
     },
     {
-      "epoch": 10.67,
-      "learning_rate": 2.8856015779092703e-05,
-      "loss": 1.4964,
+      "epoch": 16.1,
+      "learning_rate": 6.779464285714287e-05,
+      "loss": 1.0735,
       "step": 3607
     },
     {
-      "epoch": 10.67,
-      "learning_rate": 2.883629191321499e-05,
-      "loss": 1.4217,
+      "epoch": 16.11,
+      "learning_rate": 6.778571428571429e-05,
+      "loss": 1.0507,
       "step": 3608
     },
     {
-      "epoch": 10.68,
-      "learning_rate": 2.8816568047337277e-05,
-      "loss": 1.5547,
+      "epoch": 16.11,
+      "learning_rate": 6.777678571428572e-05,
+      "loss": 1.0958,
       "step": 3609
     },
     {
-      "epoch": 10.68,
-      "learning_rate": 2.879684418145957e-05,
-      "loss": 1.5228,
+      "epoch": 16.12,
+      "learning_rate": 6.776785714285716e-05,
+      "loss": 1.0754,
       "step": 3610
     },
     {
-      "epoch": 10.68,
-      "learning_rate": 2.8777120315581858e-05,
-      "loss": 1.5261,
+      "epoch": 16.12,
+      "learning_rate": 6.775892857142857e-05,
+      "loss": 1.0642,
       "step": 3611
     },
     {
-      "epoch": 10.69,
-      "learning_rate": 2.8757396449704143e-05,
-      "loss": 1.544,
+      "epoch": 16.12,
+      "learning_rate": 6.775000000000001e-05,
+      "loss": 1.1483,
       "step": 3612
     },
     {
-      "epoch": 10.69,
-      "learning_rate": 2.8737672583826432e-05,
-      "loss": 1.5515,
+      "epoch": 16.13,
+      "learning_rate": 6.774107142857143e-05,
+      "loss": 1.0727,
       "step": 3613
     },
     {
-      "epoch": 10.69,
-      "learning_rate": 2.8717948717948717e-05,
-      "loss": 1.5206,
+      "epoch": 16.13,
+      "learning_rate": 6.773214285714286e-05,
+      "loss": 1.1376,
       "step": 3614
     },
     {
-      "epoch": 10.7,
-      "learning_rate": 2.869822485207101e-05,
-      "loss": 1.5459,
+      "epoch": 16.14,
+      "learning_rate": 6.772321428571428e-05,
+      "loss": 1.0133,
       "step": 3615
     },
     {
-      "epoch": 10.7,
-      "learning_rate": 2.8678500986193298e-05,
-      "loss": 1.5795,
+      "epoch": 16.14,
+      "learning_rate": 6.771428571428572e-05,
+      "loss": 1.1173,
       "step": 3616
     },
     {
-      "epoch": 10.7,
-      "learning_rate": 2.8658777120315583e-05,
-      "loss": 1.4926,
+      "epoch": 16.15,
+      "learning_rate": 6.770535714285715e-05,
+      "loss": 1.0845,
       "step": 3617
     },
     {
-      "epoch": 10.7,
-      "learning_rate": 2.863905325443787e-05,
-      "loss": 1.4456,
+      "epoch": 16.15,
+      "learning_rate": 6.769642857142857e-05,
+      "loss": 1.1067,
       "step": 3618
     },
     {
-      "epoch": 10.71,
-      "learning_rate": 2.8619329388560157e-05,
-      "loss": 1.5498,
+      "epoch": 16.16,
+      "learning_rate": 6.76875e-05,
+      "loss": 1.1272,
       "step": 3619
     },
     {
-      "epoch": 10.71,
-      "learning_rate": 2.859960552268245e-05,
-      "loss": 1.54,
+      "epoch": 16.16,
+      "learning_rate": 6.767857142857144e-05,
+      "loss": 1.0764,
       "step": 3620
     },
     {
-      "epoch": 10.71,
-      "learning_rate": 2.8579881656804735e-05,
-      "loss": 1.4699,
+      "epoch": 16.17,
+      "learning_rate": 6.766964285714286e-05,
+      "loss": 1.1142,
       "step": 3621
     },
     {
-      "epoch": 10.72,
-      "learning_rate": 2.8560157790927023e-05,
-      "loss": 1.5545,
+      "epoch": 16.17,
+      "learning_rate": 6.766071428571429e-05,
+      "loss": 1.0877,
       "step": 3622
     },
     {
-      "epoch": 10.72,
-      "learning_rate": 2.854043392504931e-05,
-      "loss": 1.5318,
+      "epoch": 16.17,
+      "learning_rate": 6.765178571428571e-05,
+      "loss": 1.1308,
       "step": 3623
     },
     {
-      "epoch": 10.72,
-      "learning_rate": 2.85207100591716e-05,
-      "loss": 1.3886,
+      "epoch": 16.18,
+      "learning_rate": 6.764285714285714e-05,
+      "loss": 1.1044,
       "step": 3624
     },
     {
-      "epoch": 10.72,
-      "learning_rate": 2.850098619329389e-05,
-      "loss": 1.4744,
+      "epoch": 16.18,
+      "learning_rate": 6.763392857142858e-05,
+      "loss": 1.0995,
       "step": 3625
     },
     {
-      "epoch": 10.73,
-      "learning_rate": 2.8481262327416175e-05,
-      "loss": 1.5006,
+      "epoch": 16.19,
+      "learning_rate": 6.7625e-05,
+      "loss": 1.1747,
       "step": 3626
     },
     {
-      "epoch": 10.73,
-      "learning_rate": 2.846153846153846e-05,
-      "loss": 1.5539,
+      "epoch": 16.19,
+      "learning_rate": 6.761607142857144e-05,
+      "loss": 1.0874,
       "step": 3627
     },
     {
-      "epoch": 10.73,
-      "learning_rate": 2.844181459566075e-05,
-      "loss": 1.5892,
+      "epoch": 16.2,
+      "learning_rate": 6.760714285714285e-05,
+      "loss": 1.0495,
       "step": 3628
     },
     {
-      "epoch": 10.74,
-      "learning_rate": 2.842209072978304e-05,
-      "loss": 1.524,
+      "epoch": 16.2,
+      "learning_rate": 6.75982142857143e-05,
+      "loss": 1.1284,
       "step": 3629
     },
     {
-      "epoch": 10.74,
-      "learning_rate": 2.8402366863905327e-05,
-      "loss": 1.538,
+      "epoch": 16.21,
+      "learning_rate": 6.758928571428572e-05,
+      "loss": 1.0323,
       "step": 3630
     },
     {
-      "epoch": 10.74,
-      "learning_rate": 2.8382642998027615e-05,
-      "loss": 1.4624,
+      "epoch": 16.21,
+      "learning_rate": 6.758035714285714e-05,
+      "loss": 1.0693,
       "step": 3631
     },
     {
-      "epoch": 10.75,
-      "learning_rate": 2.83629191321499e-05,
-      "loss": 1.5244,
+      "epoch": 16.21,
+      "learning_rate": 6.757142857142857e-05,
+      "loss": 1.155,
       "step": 3632
     },
     {
-      "epoch": 10.75,
-      "learning_rate": 2.8343195266272193e-05,
-      "loss": 1.5242,
+      "epoch": 16.22,
+      "learning_rate": 6.756250000000001e-05,
+      "loss": 1.1396,
       "step": 3633
     },
     {
-      "epoch": 10.75,
-      "learning_rate": 2.832347140039448e-05,
-      "loss": 1.5052,
+      "epoch": 16.22,
+      "learning_rate": 6.755357142857142e-05,
+      "loss": 1.1117,
       "step": 3634
     },
     {
-      "epoch": 10.75,
-      "learning_rate": 2.8303747534516767e-05,
-      "loss": 1.5849,
+      "epoch": 16.23,
+      "learning_rate": 6.754464285714286e-05,
+      "loss": 1.1922,
       "step": 3635
     },
     {
-      "epoch": 10.76,
-      "learning_rate": 2.8284023668639052e-05,
-      "loss": 1.5021,
+      "epoch": 16.23,
+      "learning_rate": 6.753571428571429e-05,
+      "loss": 1.1712,
       "step": 3636
     },
     {
-      "epoch": 10.76,
-      "learning_rate": 2.826429980276134e-05,
-      "loss": 1.5554,
+      "epoch": 16.24,
+      "learning_rate": 6.752678571428572e-05,
+      "loss": 1.1242,
       "step": 3637
     },
     {
-      "epoch": 10.76,
-      "learning_rate": 2.8244575936883633e-05,
-      "loss": 1.5515,
+      "epoch": 16.24,
+      "learning_rate": 6.751785714285715e-05,
+      "loss": 1.1036,
       "step": 3638
     },
     {
-      "epoch": 10.77,
-      "learning_rate": 2.8224852071005918e-05,
-      "loss": 1.5607,
+      "epoch": 16.25,
+      "learning_rate": 6.750892857142858e-05,
+      "loss": 1.121,
       "step": 3639
     },
     {
-      "epoch": 10.77,
-      "learning_rate": 2.8205128205128207e-05,
-      "loss": 1.6001,
+      "epoch": 16.25,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 1.0565,
       "step": 3640
     },
     {
-      "epoch": 10.77,
-      "learning_rate": 2.8185404339250492e-05,
-      "loss": 1.5292,
+      "epoch": 16.25,
+      "learning_rate": 6.749107142857143e-05,
+      "loss": 1.0869,
       "step": 3641
     },
     {
-      "epoch": 10.78,
-      "learning_rate": 2.816568047337278e-05,
-      "loss": 1.574,
+      "epoch": 16.26,
+      "learning_rate": 6.748214285714287e-05,
+      "loss": 1.0931,
       "step": 3642
     },
     {
-      "epoch": 10.78,
-      "learning_rate": 2.8145956607495073e-05,
-      "loss": 1.5073,
+      "epoch": 16.26,
+      "learning_rate": 6.747321428571429e-05,
+      "loss": 1.1349,
       "step": 3643
     },
     {
-      "epoch": 10.78,
-      "learning_rate": 2.812623274161736e-05,
-      "loss": 1.5273,
+      "epoch": 16.27,
+      "learning_rate": 6.746428571428572e-05,
+      "loss": 1.1604,
       "step": 3644
     },
     {
-      "epoch": 10.78,
-      "learning_rate": 2.8106508875739644e-05,
-      "loss": 1.5525,
+      "epoch": 16.27,
+      "learning_rate": 6.745535714285714e-05,
+      "loss": 1.1175,
       "step": 3645
     },
     {
-      "epoch": 10.79,
-      "learning_rate": 2.8086785009861933e-05,
-      "loss": 1.545,
+      "epoch": 16.28,
+      "learning_rate": 6.744642857142858e-05,
+      "loss": 1.11,
       "step": 3646
     },
     {
-      "epoch": 10.79,
-      "learning_rate": 2.8067061143984225e-05,
-      "loss": 1.4363,
+      "epoch": 16.28,
+      "learning_rate": 6.743749999999999e-05,
+      "loss": 1.1626,
       "step": 3647
     },
     {
-      "epoch": 10.79,
-      "learning_rate": 2.804733727810651e-05,
-      "loss": 1.5159,
+      "epoch": 16.29,
+      "learning_rate": 6.742857142857143e-05,
+      "loss": 1.0996,
       "step": 3648
     },
     {
-      "epoch": 10.8,
-      "learning_rate": 2.80276134122288e-05,
-      "loss": 1.5527,
+      "epoch": 16.29,
+      "learning_rate": 6.741964285714286e-05,
+      "loss": 1.1777,
       "step": 3649
     },
     {
-      "epoch": 10.8,
-      "learning_rate": 2.8007889546351084e-05,
-      "loss": 1.5477,
+      "epoch": 16.29,
+      "learning_rate": 6.74107142857143e-05,
+      "loss": 1.1184,
       "step": 3650
     },
     {
-      "epoch": 10.8,
-      "learning_rate": 2.7988165680473373e-05,
-      "loss": 1.5835,
+      "epoch": 16.3,
+      "learning_rate": 6.740178571428571e-05,
+      "loss": 1.1142,
       "step": 3651
     },
     {
-      "epoch": 10.8,
-      "learning_rate": 2.7968441814595665e-05,
-      "loss": 1.5365,
+      "epoch": 16.3,
+      "learning_rate": 6.739285714285715e-05,
+      "loss": 1.1141,
       "step": 3652
     },
     {
-      "epoch": 10.81,
-      "learning_rate": 2.794871794871795e-05,
-      "loss": 1.5521,
+      "epoch": 16.31,
+      "learning_rate": 6.738392857142857e-05,
+      "loss": 1.0485,
       "step": 3653
     },
     {
-      "epoch": 10.81,
-      "learning_rate": 2.792899408284024e-05,
-      "loss": 1.6245,
+      "epoch": 16.31,
+      "learning_rate": 6.7375e-05,
+      "loss": 1.1246,
       "step": 3654
     },
     {
-      "epoch": 10.81,
-      "learning_rate": 2.7909270216962524e-05,
-      "loss": 1.5672,
+      "epoch": 16.32,
+      "learning_rate": 6.736607142857144e-05,
+      "loss": 1.1927,
       "step": 3655
     },
     {
-      "epoch": 10.82,
-      "learning_rate": 2.7889546351084816e-05,
-      "loss": 1.4541,
+      "epoch": 16.32,
+      "learning_rate": 6.735714285714286e-05,
+      "loss": 1.1249,
       "step": 3656
     },
     {
-      "epoch": 10.82,
-      "learning_rate": 2.7869822485207102e-05,
-      "loss": 1.5547,
+      "epoch": 16.33,
+      "learning_rate": 6.734821428571429e-05,
+      "loss": 1.1093,
       "step": 3657
     },
     {
-      "epoch": 10.82,
-      "learning_rate": 2.785009861932939e-05,
-      "loss": 1.5718,
+      "epoch": 16.33,
+      "learning_rate": 6.733928571428571e-05,
+      "loss": 1.0884,
       "step": 3658
     },
     {
-      "epoch": 10.83,
-      "learning_rate": 2.7830374753451676e-05,
-      "loss": 1.5141,
+      "epoch": 16.33,
+      "learning_rate": 6.733035714285715e-05,
+      "loss": 1.0658,
       "step": 3659
     },
     {
-      "epoch": 10.83,
-      "learning_rate": 2.7810650887573965e-05,
-      "loss": 1.5231,
+      "epoch": 16.34,
+      "learning_rate": 6.732142857142858e-05,
+      "loss": 1.049,
       "step": 3660
     },
     {
-      "epoch": 10.83,
-      "learning_rate": 2.7790927021696257e-05,
-      "loss": 1.5294,
+      "epoch": 16.34,
+      "learning_rate": 6.73125e-05,
+      "loss": 1.1366,
       "step": 3661
     },
     {
-      "epoch": 10.83,
-      "learning_rate": 2.7771203155818542e-05,
-      "loss": 1.5544,
+      "epoch": 16.35,
+      "learning_rate": 6.730357142857143e-05,
+      "loss": 1.0152,
       "step": 3662
     },
     {
-      "epoch": 10.84,
-      "learning_rate": 2.775147928994083e-05,
-      "loss": 1.5608,
+      "epoch": 16.35,
+      "learning_rate": 6.729464285714287e-05,
+      "loss": 1.14,
       "step": 3663
     },
     {
-      "epoch": 10.84,
-      "learning_rate": 2.7731755424063116e-05,
-      "loss": 1.6343,
+      "epoch": 16.36,
+      "learning_rate": 6.728571428571428e-05,
+      "loss": 1.1055,
       "step": 3664
     },
     {
-      "epoch": 10.84,
-      "learning_rate": 2.7712031558185408e-05,
-      "loss": 1.5258,
+      "epoch": 16.36,
+      "learning_rate": 6.727678571428572e-05,
+      "loss": 1.0963,
       "step": 3665
     },
     {
-      "epoch": 10.85,
-      "learning_rate": 2.7692307692307694e-05,
-      "loss": 1.528,
+      "epoch": 16.37,
+      "learning_rate": 6.726785714285714e-05,
+      "loss": 1.0693,
       "step": 3666
     },
     {
-      "epoch": 10.85,
-      "learning_rate": 2.7672583826429982e-05,
-      "loss": 1.5529,
+      "epoch": 16.37,
+      "learning_rate": 6.725892857142857e-05,
+      "loss": 1.055,
       "step": 3667
     },
     {
-      "epoch": 10.85,
-      "learning_rate": 2.7652859960552268e-05,
-      "loss": 1.4489,
+      "epoch": 16.38,
+      "learning_rate": 6.725000000000001e-05,
+      "loss": 1.1801,
       "step": 3668
     },
     {
-      "epoch": 10.86,
-      "learning_rate": 2.7633136094674556e-05,
-      "loss": 1.5355,
+      "epoch": 16.38,
+      "learning_rate": 6.724107142857143e-05,
+      "loss": 1.0524,
       "step": 3669
     },
     {
-      "epoch": 10.86,
-      "learning_rate": 2.761341222879685e-05,
-      "loss": 1.5133,
+      "epoch": 16.38,
+      "learning_rate": 6.723214285714286e-05,
+      "loss": 1.0767,
       "step": 3670
     },
     {
-      "epoch": 10.86,
-      "learning_rate": 2.7593688362919134e-05,
-      "loss": 1.5471,
+      "epoch": 16.39,
+      "learning_rate": 6.722321428571428e-05,
+      "loss": 1.1355,
       "step": 3671
     },
     {
-      "epoch": 10.86,
-      "learning_rate": 2.7573964497041422e-05,
-      "loss": 1.6117,
+      "epoch": 16.39,
+      "learning_rate": 6.721428571428572e-05,
+      "loss": 1.0943,
       "step": 3672
     },
     {
-      "epoch": 10.87,
-      "learning_rate": 2.7554240631163708e-05,
-      "loss": 1.5523,
+      "epoch": 16.4,
+      "learning_rate": 6.720535714285715e-05,
+      "loss": 1.138,
       "step": 3673
     },
     {
-      "epoch": 10.87,
-      "learning_rate": 2.7534516765285993e-05,
-      "loss": 1.5119,
+      "epoch": 16.4,
+      "learning_rate": 6.719642857142857e-05,
+      "loss": 1.14,
       "step": 3674
     },
     {
-      "epoch": 10.87,
-      "learning_rate": 2.751479289940829e-05,
-      "loss": 1.5067,
+      "epoch": 16.41,
+      "learning_rate": 6.71875e-05,
+      "loss": 1.0734,
       "step": 3675
     },
     {
-      "epoch": 10.88,
-      "learning_rate": 2.7495069033530574e-05,
-      "loss": 1.4485,
+      "epoch": 16.41,
+      "learning_rate": 6.717857142857144e-05,
+      "loss": 1.1161,
       "step": 3676
     },
     {
-      "epoch": 10.88,
-      "learning_rate": 2.747534516765286e-05,
-      "loss": 1.5223,
+      "epoch": 16.42,
+      "learning_rate": 6.716964285714285e-05,
+      "loss": 1.0688,
       "step": 3677
     },
     {
-      "epoch": 10.88,
-      "learning_rate": 2.7455621301775148e-05,
-      "loss": 1.4588,
+      "epoch": 16.42,
+      "learning_rate": 6.716071428571429e-05,
+      "loss": 1.1643,
       "step": 3678
     },
     {
-      "epoch": 10.88,
-      "learning_rate": 2.743589743589744e-05,
-      "loss": 1.439,
+      "epoch": 16.42,
+      "learning_rate": 6.715178571428572e-05,
+      "loss": 1.1739,
       "step": 3679
     },
     {
-      "epoch": 10.89,
-      "learning_rate": 2.7416173570019726e-05,
-      "loss": 1.5502,
+      "epoch": 16.43,
+      "learning_rate": 6.714285714285714e-05,
+      "loss": 1.1198,
       "step": 3680
     },
     {
-      "epoch": 10.89,
-      "learning_rate": 2.7396449704142014e-05,
-      "loss": 1.4638,
+      "epoch": 16.43,
+      "learning_rate": 6.713392857142857e-05,
+      "loss": 1.1276,
       "step": 3681
     },
     {
-      "epoch": 10.89,
-      "learning_rate": 2.73767258382643e-05,
-      "loss": 1.5442,
+      "epoch": 16.44,
+      "learning_rate": 6.7125e-05,
+      "loss": 1.1055,
       "step": 3682
     },
     {
-      "epoch": 10.9,
-      "learning_rate": 2.735700197238659e-05,
-      "loss": 1.566,
+      "epoch": 16.44,
+      "learning_rate": 6.711607142857143e-05,
+      "loss": 1.1787,
       "step": 3683
     },
     {
-      "epoch": 10.9,
-      "learning_rate": 2.733727810650888e-05,
-      "loss": 1.4338,
+      "epoch": 16.45,
+      "learning_rate": 6.710714285714286e-05,
+      "loss": 1.0975,
       "step": 3684
     },
     {
-      "epoch": 10.9,
-      "learning_rate": 2.7317554240631166e-05,
-      "loss": 1.4113,
+      "epoch": 16.45,
+      "learning_rate": 6.70982142857143e-05,
+      "loss": 1.1506,
       "step": 3685
     },
     {
-      "epoch": 10.91,
-      "learning_rate": 2.729783037475345e-05,
-      "loss": 1.5207,
+      "epoch": 16.46,
+      "learning_rate": 6.708928571428572e-05,
+      "loss": 1.1039,
       "step": 3686
     },
     {
-      "epoch": 10.91,
-      "learning_rate": 2.727810650887574e-05,
-      "loss": 1.5036,
+      "epoch": 16.46,
+      "learning_rate": 6.708035714285715e-05,
+      "loss": 1.0571,
       "step": 3687
     },
     {
-      "epoch": 10.91,
-      "learning_rate": 2.7258382642998032e-05,
-      "loss": 1.5414,
+      "epoch": 16.46,
+      "learning_rate": 6.707142857142857e-05,
+      "loss": 1.0813,
       "step": 3688
     },
     {
-      "epoch": 10.91,
-      "learning_rate": 2.7238658777120317e-05,
-      "loss": 1.4985,
+      "epoch": 16.47,
+      "learning_rate": 6.706250000000001e-05,
+      "loss": 1.0911,
       "step": 3689
     },
     {
-      "epoch": 10.92,
-      "learning_rate": 2.7218934911242606e-05,
-      "loss": 1.5309,
+      "epoch": 16.47,
+      "learning_rate": 6.705357142857142e-05,
+      "loss": 1.1612,
       "step": 3690
     },
     {
-      "epoch": 10.92,
-      "learning_rate": 2.719921104536489e-05,
-      "loss": 1.5537,
+      "epoch": 16.48,
+      "learning_rate": 6.704464285714286e-05,
+      "loss": 1.1731,
       "step": 3691
     },
     {
-      "epoch": 10.92,
-      "learning_rate": 2.717948717948718e-05,
-      "loss": 1.6118,
+      "epoch": 16.48,
+      "learning_rate": 6.703571428571429e-05,
+      "loss": 1.1519,
       "step": 3692
     },
     {
-      "epoch": 10.93,
-      "learning_rate": 2.7159763313609472e-05,
-      "loss": 1.597,
+      "epoch": 16.49,
+      "learning_rate": 6.702678571428573e-05,
+      "loss": 1.095,
       "step": 3693
     },
     {
-      "epoch": 10.93,
-      "learning_rate": 2.7140039447731758e-05,
-      "loss": 1.5513,
+      "epoch": 16.49,
+      "learning_rate": 6.701785714285714e-05,
+      "loss": 1.0662,
       "step": 3694
     },
     {
-      "epoch": 10.93,
-      "learning_rate": 2.7120315581854043e-05,
-      "loss": 1.5372,
+      "epoch": 16.5,
+      "learning_rate": 6.700892857142858e-05,
+      "loss": 1.0552,
       "step": 3695
     },
     {
-      "epoch": 10.93,
-      "learning_rate": 2.710059171597633e-05,
-      "loss": 1.5694,
+      "epoch": 16.5,
+      "learning_rate": 6.7e-05,
+      "loss": 1.1128,
       "step": 3696
     },
     {
-      "epoch": 10.94,
-      "learning_rate": 2.7080867850098617e-05,
-      "loss": 1.5667,
+      "epoch": 16.5,
+      "learning_rate": 6.699107142857143e-05,
+      "loss": 1.0869,
       "step": 3697
     },
     {
-      "epoch": 10.94,
-      "learning_rate": 2.706114398422091e-05,
-      "loss": 1.5232,
+      "epoch": 16.51,
+      "learning_rate": 6.698214285714287e-05,
+      "loss": 1.1642,
       "step": 3698
     },
     {
-      "epoch": 10.94,
-      "learning_rate": 2.7041420118343198e-05,
-      "loss": 1.5013,
+      "epoch": 16.51,
+      "learning_rate": 6.697321428571429e-05,
+      "loss": 1.1656,
       "step": 3699
     },
     {
-      "epoch": 10.95,
-      "learning_rate": 2.7021696252465483e-05,
-      "loss": 1.5752,
+      "epoch": 16.52,
+      "learning_rate": 6.696428571428572e-05,
+      "loss": 1.1522,
       "step": 3700
     },
     {
-      "epoch": 10.95,
-      "learning_rate": 2.7001972386587772e-05,
-      "loss": 1.5563,
+      "epoch": 16.52,
+      "learning_rate": 6.695535714285714e-05,
+      "loss": 1.2248,
       "step": 3701
     },
     {
-      "epoch": 10.95,
-      "learning_rate": 2.6982248520710064e-05,
-      "loss": 1.5107,
+      "epoch": 16.53,
+      "learning_rate": 6.694642857142858e-05,
+      "loss": 1.1014,
       "step": 3702
     },
     {
-      "epoch": 10.96,
-      "learning_rate": 2.696252465483235e-05,
-      "loss": 1.5217,
+      "epoch": 16.53,
+      "learning_rate": 6.693750000000001e-05,
+      "loss": 1.0941,
       "step": 3703
     },
     {
-      "epoch": 10.96,
-      "learning_rate": 2.6942800788954638e-05,
-      "loss": 1.5045,
+      "epoch": 16.54,
+      "learning_rate": 6.692857142857143e-05,
+      "loss": 1.0787,
       "step": 3704
     },
     {
-      "epoch": 10.96,
-      "learning_rate": 2.6923076923076923e-05,
-      "loss": 1.5058,
+      "epoch": 16.54,
+      "learning_rate": 6.691964285714286e-05,
+      "loss": 1.1634,
       "step": 3705
     },
     {
-      "epoch": 10.96,
-      "learning_rate": 2.690335305719921e-05,
-      "loss": 1.5308,
+      "epoch": 16.54,
+      "learning_rate": 6.69107142857143e-05,
+      "loss": 1.1355,
       "step": 3706
     },
     {
-      "epoch": 10.97,
-      "learning_rate": 2.68836291913215e-05,
-      "loss": 1.5213,
+      "epoch": 16.55,
+      "learning_rate": 6.690178571428571e-05,
+      "loss": 1.1458,
       "step": 3707
     },
     {
-      "epoch": 10.97,
-      "learning_rate": 2.686390532544379e-05,
-      "loss": 1.5683,
+      "epoch": 16.55,
+      "learning_rate": 6.689285714285715e-05,
+      "loss": 1.1739,
       "step": 3708
     },
     {
-      "epoch": 10.97,
-      "learning_rate": 2.6844181459566075e-05,
-      "loss": 1.5111,
+      "epoch": 16.56,
+      "learning_rate": 6.688392857142857e-05,
+      "loss": 1.0835,
       "step": 3709
     },
     {
-      "epoch": 10.98,
-      "learning_rate": 2.6824457593688364e-05,
-      "loss": 1.4495,
+      "epoch": 16.56,
+      "learning_rate": 6.6875e-05,
+      "loss": 1.0977,
       "step": 3710
     },
     {
-      "epoch": 10.98,
-      "learning_rate": 2.6804733727810656e-05,
-      "loss": 1.6226,
+      "epoch": 16.57,
+      "learning_rate": 6.686607142857142e-05,
+      "loss": 1.139,
       "step": 3711
     },
     {
-      "epoch": 10.98,
-      "learning_rate": 2.678500986193294e-05,
-      "loss": 1.5459,
+      "epoch": 16.57,
+      "learning_rate": 6.685714285714286e-05,
+      "loss": 1.1302,
       "step": 3712
     },
     {
-      "epoch": 10.99,
-      "learning_rate": 2.676528599605523e-05,
-      "loss": 1.6076,
+      "epoch": 16.58,
+      "learning_rate": 6.684821428571429e-05,
+      "loss": 1.0411,
       "step": 3713
     },
     {
-      "epoch": 10.99,
-      "learning_rate": 2.6745562130177515e-05,
-      "loss": 1.5896,
+      "epoch": 16.58,
+      "learning_rate": 6.683928571428571e-05,
+      "loss": 1.1666,
       "step": 3714
     },
     {
-      "epoch": 10.99,
-      "learning_rate": 2.67258382642998e-05,
-      "loss": 1.5557,
+      "epoch": 16.58,
+      "learning_rate": 6.683035714285715e-05,
+      "loss": 1.0975,
       "step": 3715
     },
     {
-      "epoch": 10.99,
-      "learning_rate": 2.6706114398422093e-05,
-      "loss": 1.538,
+      "epoch": 16.59,
+      "learning_rate": 6.682142857142858e-05,
+      "loss": 1.0301,
       "step": 3716
     },
     {
-      "epoch": 11.0,
-      "learning_rate": 2.668639053254438e-05,
-      "loss": 1.5623,
+      "epoch": 16.59,
+      "learning_rate": 6.68125e-05,
+      "loss": 1.1008,
       "step": 3717
     },
     {
-      "epoch": 11.0,
-      "learning_rate": 2.6666666666666667e-05,
-      "loss": 1.4831,
+      "epoch": 16.6,
+      "learning_rate": 6.680357142857143e-05,
+      "loss": 1.1102,
       "step": 3718
     },
     {
-      "epoch": 11.0,
-      "learning_rate": 2.6646942800788955e-05,
-      "loss": 1.524,
+      "epoch": 16.6,
+      "learning_rate": 6.679464285714287e-05,
+      "loss": 1.0598,
       "step": 3719
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 2.6627218934911247e-05,
-      "loss": 1.5606,
+      "epoch": 16.61,
+      "learning_rate": 6.678571428571428e-05,
+      "loss": 1.1311,
       "step": 3720
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 2.6607495069033533e-05,
-      "loss": 1.5041,
+      "epoch": 16.61,
+      "learning_rate": 6.677678571428572e-05,
+      "loss": 1.1256,
       "step": 3721
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 2.658777120315582e-05,
-      "loss": 1.5058,
+      "epoch": 16.62,
+      "learning_rate": 6.676785714285715e-05,
+      "loss": 1.0696,
       "step": 3722
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 2.6568047337278107e-05,
-      "loss": 1.5743,
+      "epoch": 16.62,
+      "learning_rate": 6.675892857142857e-05,
+      "loss": 1.1712,
       "step": 3723
     },
     {
-      "epoch": 11.02,
-      "learning_rate": 2.6548323471400392e-05,
-      "loss": 1.526,
+      "epoch": 16.62,
+      "learning_rate": 6.675e-05,
+      "loss": 1.0834,
       "step": 3724
     },
     {
-      "epoch": 11.02,
-      "learning_rate": 2.6528599605522688e-05,
-      "loss": 1.5911,
+      "epoch": 16.63,
+      "learning_rate": 6.674107142857144e-05,
+      "loss": 1.0929,
       "step": 3725
     },
     {
-      "epoch": 11.02,
-      "learning_rate": 2.6508875739644973e-05,
-      "loss": 1.5384,
+      "epoch": 16.63,
+      "learning_rate": 6.673214285714286e-05,
+      "loss": 1.1224,
       "step": 3726
     },
     {
-      "epoch": 11.03,
-      "learning_rate": 2.648915187376726e-05,
-      "loss": 1.4172,
+      "epoch": 16.64,
+      "learning_rate": 6.672321428571429e-05,
+      "loss": 1.1172,
       "step": 3727
     },
     {
-      "epoch": 11.03,
-      "learning_rate": 2.6469428007889547e-05,
-      "loss": 1.5522,
+      "epoch": 16.64,
+      "learning_rate": 6.671428571428571e-05,
+      "loss": 1.1482,
       "step": 3728
     },
     {
-      "epoch": 11.03,
-      "learning_rate": 2.6449704142011832e-05,
-      "loss": 1.5337,
+      "epoch": 16.65,
+      "learning_rate": 6.670535714285715e-05,
+      "loss": 1.1908,
       "step": 3729
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 2.6429980276134125e-05,
-      "loss": 1.5068,
+      "epoch": 16.65,
+      "learning_rate": 6.669642857142858e-05,
+      "loss": 1.1228,
       "step": 3730
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 2.6410256410256413e-05,
-      "loss": 1.3999,
+      "epoch": 16.66,
+      "learning_rate": 6.66875e-05,
+      "loss": 1.1457,
       "step": 3731
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 2.63905325443787e-05,
-      "loss": 1.5943,
+      "epoch": 16.66,
+      "learning_rate": 6.667857142857144e-05,
+      "loss": 1.127,
       "step": 3732
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 2.6370808678500984e-05,
-      "loss": 1.5324,
+      "epoch": 16.67,
+      "learning_rate": 6.666964285714285e-05,
+      "loss": 1.0713,
       "step": 3733
     },
     {
-      "epoch": 11.05,
-      "learning_rate": 2.635108481262328e-05,
-      "loss": 1.4795,
+      "epoch": 16.67,
+      "learning_rate": 6.666071428571429e-05,
+      "loss": 1.0847,
       "step": 3734
     },
     {
-      "epoch": 11.05,
-      "learning_rate": 2.6331360946745565e-05,
-      "loss": 1.513,
+      "epoch": 16.67,
+      "learning_rate": 6.665178571428572e-05,
+      "loss": 1.128,
       "step": 3735
     },
     {
-      "epoch": 11.05,
-      "learning_rate": 2.631163708086785e-05,
-      "loss": 1.5976,
+      "epoch": 16.68,
+      "learning_rate": 6.664285714285716e-05,
+      "loss": 1.1249,
       "step": 3736
     },
     {
-      "epoch": 11.06,
-      "learning_rate": 2.629191321499014e-05,
-      "loss": 1.4744,
+      "epoch": 16.68,
+      "learning_rate": 6.663392857142857e-05,
+      "loss": 1.0582,
       "step": 3737
     },
     {
-      "epoch": 11.06,
-      "learning_rate": 2.6272189349112424e-05,
-      "loss": 1.4328,
+      "epoch": 16.69,
+      "learning_rate": 6.6625e-05,
+      "loss": 1.1511,
       "step": 3738
     },
     {
-      "epoch": 11.06,
-      "learning_rate": 2.6252465483234716e-05,
-      "loss": 1.4651,
+      "epoch": 16.69,
+      "learning_rate": 6.661607142857143e-05,
+      "loss": 1.1298,
       "step": 3739
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.6232741617357005e-05,
-      "loss": 1.4932,
+      "epoch": 16.7,
+      "learning_rate": 6.660714285714286e-05,
+      "loss": 1.1547,
       "step": 3740
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.621301775147929e-05,
-      "loss": 1.5073,
+      "epoch": 16.7,
+      "learning_rate": 6.659821428571428e-05,
+      "loss": 1.1114,
       "step": 3741
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.619329388560158e-05,
-      "loss": 1.5482,
+      "epoch": 16.71,
+      "learning_rate": 6.658928571428572e-05,
+      "loss": 1.104,
       "step": 3742
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.617357001972387e-05,
-      "loss": 1.5206,
+      "epoch": 16.71,
+      "learning_rate": 6.658035714285715e-05,
+      "loss": 1.1162,
       "step": 3743
     },
     {
-      "epoch": 11.08,
-      "learning_rate": 2.6153846153846157e-05,
-      "loss": 1.6045,
+      "epoch": 16.71,
+      "learning_rate": 6.657142857142857e-05,
+      "loss": 1.0792,
       "step": 3744
     },
     {
-      "epoch": 11.08,
-      "learning_rate": 2.6134122287968442e-05,
-      "loss": 1.5185,
+      "epoch": 16.72,
+      "learning_rate": 6.656250000000001e-05,
+      "loss": 1.1088,
       "step": 3745
     },
     {
-      "epoch": 11.08,
-      "learning_rate": 2.611439842209073e-05,
-      "loss": 1.504,
+      "epoch": 16.72,
+      "learning_rate": 6.655357142857142e-05,
+      "loss": 1.1094,
       "step": 3746
     },
     {
-      "epoch": 11.09,
-      "learning_rate": 2.6094674556213016e-05,
-      "loss": 1.4362,
+      "epoch": 16.73,
+      "learning_rate": 6.654464285714286e-05,
+      "loss": 1.1506,
       "step": 3747
     },
     {
-      "epoch": 11.09,
-      "learning_rate": 2.6074950690335308e-05,
-      "loss": 1.4653,
+      "epoch": 16.73,
+      "learning_rate": 6.653571428571429e-05,
+      "loss": 1.1253,
       "step": 3748
     },
     {
-      "epoch": 11.09,
-      "learning_rate": 2.6055226824457597e-05,
-      "loss": 1.5387,
+      "epoch": 16.74,
+      "learning_rate": 6.652678571428573e-05,
+      "loss": 1.151,
       "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,
+      "epoch": 16.74,
+      "learning_rate": 6.651785714285714e-05,
+      "loss": 1.0965,
       "step": 3750
     },
     {
-      "epoch": 11.1,
-      "learning_rate": 2.601577909270217e-05,
-      "loss": 1.4919,
+      "epoch": 16.75,
+      "learning_rate": 6.650892857142858e-05,
+      "loss": 1.1471,
       "step": 3751
     },
     {
-      "epoch": 11.1,
-      "learning_rate": 2.5996055226824456e-05,
-      "loss": 1.5342,
+      "epoch": 16.75,
+      "learning_rate": 6.65e-05,
+      "loss": 1.0451,
       "step": 3752
     },
     {
-      "epoch": 11.1,
-      "learning_rate": 2.5976331360946748e-05,
-      "loss": 1.6541,
+      "epoch": 16.75,
+      "learning_rate": 6.649107142857143e-05,
+      "loss": 1.1128,
       "step": 3753
     },
     {
-      "epoch": 11.11,
-      "learning_rate": 2.5956607495069034e-05,
-      "loss": 1.5163,
+      "epoch": 16.76,
+      "learning_rate": 6.648214285714285e-05,
+      "loss": 1.1167,
       "step": 3754
     },
     {
-      "epoch": 11.11,
-      "learning_rate": 2.5936883629191322e-05,
-      "loss": 1.4143,
+      "epoch": 16.76,
+      "learning_rate": 6.64732142857143e-05,
+      "loss": 1.1195,
       "step": 3755
     },
     {
-      "epoch": 11.11,
-      "learning_rate": 2.5917159763313608e-05,
-      "loss": 1.5188,
+      "epoch": 16.77,
+      "learning_rate": 6.646428571428572e-05,
+      "loss": 1.1099,
       "step": 3756
     },
     {
-      "epoch": 11.12,
-      "learning_rate": 2.58974358974359e-05,
-      "loss": 1.5655,
+      "epoch": 16.77,
+      "learning_rate": 6.645535714285714e-05,
+      "loss": 1.1138,
       "step": 3757
     },
     {
-      "epoch": 11.12,
-      "learning_rate": 2.587771203155819e-05,
-      "loss": 1.429,
+      "epoch": 16.78,
+      "learning_rate": 6.644642857142857e-05,
+      "loss": 1.0966,
       "step": 3758
     },
     {
-      "epoch": 11.12,
-      "learning_rate": 2.5857988165680474e-05,
-      "loss": 1.478,
+      "epoch": 16.78,
+      "learning_rate": 6.643750000000001e-05,
+      "loss": 1.1734,
       "step": 3759
     },
     {
-      "epoch": 11.12,
-      "learning_rate": 2.5838264299802763e-05,
-      "loss": 1.5319,
+      "epoch": 16.79,
+      "learning_rate": 6.642857142857143e-05,
+      "loss": 1.1597,
       "step": 3760
     },
     {
-      "epoch": 11.13,
-      "learning_rate": 2.5818540433925048e-05,
-      "loss": 1.5191,
+      "epoch": 16.79,
+      "learning_rate": 6.641964285714286e-05,
+      "loss": 1.1512,
       "step": 3761
     },
     {
-      "epoch": 11.13,
-      "learning_rate": 2.579881656804734e-05,
-      "loss": 1.567,
+      "epoch": 16.79,
+      "learning_rate": 6.64107142857143e-05,
+      "loss": 1.1233,
       "step": 3762
     },
     {
-      "epoch": 11.13,
-      "learning_rate": 2.577909270216963e-05,
-      "loss": 1.5798,
+      "epoch": 16.8,
+      "learning_rate": 6.640178571428571e-05,
+      "loss": 1.0821,
       "step": 3763
     },
     {
-      "epoch": 11.14,
-      "learning_rate": 2.5759368836291914e-05,
-      "loss": 1.5914,
+      "epoch": 16.8,
+      "learning_rate": 6.639285714285715e-05,
+      "loss": 1.1081,
       "step": 3764
     },
     {
-      "epoch": 11.14,
-      "learning_rate": 2.57396449704142e-05,
-      "loss": 1.4832,
+      "epoch": 16.81,
+      "learning_rate": 6.638392857142858e-05,
+      "loss": 1.0829,
       "step": 3765
     },
     {
-      "epoch": 11.14,
-      "learning_rate": 2.571992110453649e-05,
-      "loss": 1.5404,
+      "epoch": 16.81,
+      "learning_rate": 6.6375e-05,
+      "loss": 1.1512,
       "step": 3766
     },
     {
-      "epoch": 11.14,
-      "learning_rate": 2.570019723865878e-05,
-      "loss": 1.5002,
+      "epoch": 16.82,
+      "learning_rate": 6.636607142857143e-05,
+      "loss": 1.1532,
       "step": 3767
     },
     {
-      "epoch": 11.15,
-      "learning_rate": 2.5680473372781066e-05,
-      "loss": 1.621,
+      "epoch": 16.82,
+      "learning_rate": 6.635714285714287e-05,
+      "loss": 1.1814,
       "step": 3768
     },
     {
-      "epoch": 11.15,
-      "learning_rate": 2.5660749506903354e-05,
-      "loss": 1.5687,
+      "epoch": 16.83,
+      "learning_rate": 6.634821428571429e-05,
+      "loss": 1.0231,
       "step": 3769
     },
     {
-      "epoch": 11.15,
-      "learning_rate": 2.564102564102564e-05,
-      "loss": 1.4335,
+      "epoch": 16.83,
+      "learning_rate": 6.633928571428572e-05,
+      "loss": 1.1758,
       "step": 3770
     },
     {
-      "epoch": 11.16,
-      "learning_rate": 2.5621301775147932e-05,
-      "loss": 1.4665,
+      "epoch": 16.83,
+      "learning_rate": 6.633035714285714e-05,
+      "loss": 1.1366,
       "step": 3771
     },
     {
-      "epoch": 11.16,
-      "learning_rate": 2.560157790927022e-05,
-      "loss": 1.5645,
+      "epoch": 16.84,
+      "learning_rate": 6.632142857142858e-05,
+      "loss": 1.1369,
       "step": 3772
     },
     {
-      "epoch": 11.16,
-      "learning_rate": 2.5581854043392506e-05,
-      "loss": 1.6061,
+      "epoch": 16.84,
+      "learning_rate": 6.63125e-05,
+      "loss": 1.1105,
       "step": 3773
     },
     {
-      "epoch": 11.17,
-      "learning_rate": 2.556213017751479e-05,
-      "loss": 1.5408,
+      "epoch": 16.85,
+      "learning_rate": 6.630357142857143e-05,
+      "loss": 1.1179,
       "step": 3774
     },
     {
-      "epoch": 11.17,
-      "learning_rate": 2.5542406311637083e-05,
-      "loss": 1.5171,
+      "epoch": 16.85,
+      "learning_rate": 6.629464285714286e-05,
+      "loss": 1.1128,
       "step": 3775
     },
     {
-      "epoch": 11.17,
-      "learning_rate": 2.5522682445759372e-05,
-      "loss": 1.4927,
+      "epoch": 16.86,
+      "learning_rate": 6.628571428571428e-05,
+      "loss": 1.0869,
       "step": 3776
     },
     {
-      "epoch": 11.17,
-      "learning_rate": 2.5502958579881657e-05,
-      "loss": 1.5321,
+      "epoch": 16.86,
+      "learning_rate": 6.627678571428572e-05,
+      "loss": 1.1268,
       "step": 3777
     },
     {
-      "epoch": 11.18,
-      "learning_rate": 2.5483234714003946e-05,
-      "loss": 1.5455,
+      "epoch": 16.87,
+      "learning_rate": 6.626785714285715e-05,
+      "loss": 1.0984,
       "step": 3778
     },
     {
-      "epoch": 11.18,
-      "learning_rate": 2.546351084812623e-05,
-      "loss": 1.5219,
+      "epoch": 16.87,
+      "learning_rate": 6.625892857142859e-05,
+      "loss": 1.1175,
       "step": 3779
     },
     {
-      "epoch": 11.18,
-      "learning_rate": 2.5443786982248524e-05,
-      "loss": 1.5673,
+      "epoch": 16.88,
+      "learning_rate": 6.625e-05,
+      "loss": 1.1254,
       "step": 3780
     },
     {
-      "epoch": 11.19,
-      "learning_rate": 2.5424063116370812e-05,
-      "loss": 1.4837,
+      "epoch": 16.88,
+      "learning_rate": 6.624107142857144e-05,
+      "loss": 1.1184,
       "step": 3781
     },
     {
-      "epoch": 11.19,
-      "learning_rate": 2.5404339250493098e-05,
-      "loss": 1.5371,
+      "epoch": 16.88,
+      "learning_rate": 6.623214285714286e-05,
+      "loss": 1.1335,
       "step": 3782
     },
     {
-      "epoch": 11.19,
-      "learning_rate": 2.5384615384615383e-05,
-      "loss": 1.4384,
+      "epoch": 16.89,
+      "learning_rate": 6.622321428571429e-05,
+      "loss": 1.0477,
       "step": 3783
     },
     {
-      "epoch": 11.2,
-      "learning_rate": 2.536489151873767e-05,
-      "loss": 1.5029,
+      "epoch": 16.89,
+      "learning_rate": 6.621428571428571e-05,
+      "loss": 1.1385,
       "step": 3784
     },
     {
-      "epoch": 11.2,
-      "learning_rate": 2.5345167652859964e-05,
-      "loss": 1.5853,
+      "epoch": 16.9,
+      "learning_rate": 6.620535714285715e-05,
+      "loss": 1.0836,
       "step": 3785
     },
     {
-      "epoch": 11.2,
-      "learning_rate": 2.532544378698225e-05,
-      "loss": 1.4937,
+      "epoch": 16.9,
+      "learning_rate": 6.619642857142856e-05,
+      "loss": 1.1458,
       "step": 3786
     },
     {
-      "epoch": 11.2,
-      "learning_rate": 2.5305719921104538e-05,
-      "loss": 1.5099,
+      "epoch": 16.91,
+      "learning_rate": 6.61875e-05,
+      "loss": 1.1035,
       "step": 3787
     },
     {
-      "epoch": 11.21,
-      "learning_rate": 2.5285996055226823e-05,
-      "loss": 1.5342,
+      "epoch": 16.91,
+      "learning_rate": 6.617857142857143e-05,
+      "loss": 1.0113,
       "step": 3788
     },
     {
-      "epoch": 11.21,
-      "learning_rate": 2.5266272189349115e-05,
-      "loss": 1.5933,
+      "epoch": 16.92,
+      "learning_rate": 6.616964285714285e-05,
+      "loss": 1.1009,
       "step": 3789
     },
     {
-      "epoch": 11.21,
-      "learning_rate": 2.5246548323471404e-05,
-      "loss": 1.6151,
+      "epoch": 16.92,
+      "learning_rate": 6.616071428571429e-05,
+      "loss": 1.1047,
       "step": 3790
     },
     {
-      "epoch": 11.22,
-      "learning_rate": 2.522682445759369e-05,
-      "loss": 1.5265,
+      "epoch": 16.92,
+      "learning_rate": 6.615178571428572e-05,
+      "loss": 1.0568,
       "step": 3791
     },
     {
-      "epoch": 11.22,
-      "learning_rate": 2.5207100591715978e-05,
-      "loss": 1.459,
+      "epoch": 16.93,
+      "learning_rate": 6.614285714285716e-05,
+      "loss": 1.1329,
       "step": 3792
     },
     {
-      "epoch": 11.22,
-      "learning_rate": 2.5187376725838263e-05,
-      "loss": 1.5081,
+      "epoch": 16.93,
+      "learning_rate": 6.613392857142857e-05,
+      "loss": 1.1003,
       "step": 3793
     },
     {
-      "epoch": 11.22,
-      "learning_rate": 2.5167652859960556e-05,
-      "loss": 1.4831,
+      "epoch": 16.94,
+      "learning_rate": 6.612500000000001e-05,
+      "loss": 1.038,
       "step": 3794
     },
     {
-      "epoch": 11.23,
-      "learning_rate": 2.514792899408284e-05,
-      "loss": 1.5611,
+      "epoch": 16.94,
+      "learning_rate": 6.611607142857143e-05,
+      "loss": 1.1796,
       "step": 3795
     },
     {
-      "epoch": 11.23,
-      "learning_rate": 2.512820512820513e-05,
-      "loss": 1.5441,
+      "epoch": 16.95,
+      "learning_rate": 6.610714285714286e-05,
+      "loss": 1.1155,
       "step": 3796
     },
     {
-      "epoch": 11.23,
-      "learning_rate": 2.5108481262327415e-05,
-      "loss": 1.4822,
+      "epoch": 16.95,
+      "learning_rate": 6.609821428571428e-05,
+      "loss": 1.1045,
       "step": 3797
     },
     {
-      "epoch": 11.24,
-      "learning_rate": 2.5088757396449707e-05,
-      "loss": 1.4833,
+      "epoch": 16.96,
+      "learning_rate": 6.608928571428572e-05,
+      "loss": 1.1699,
       "step": 3798
     },
     {
-      "epoch": 11.24,
-      "learning_rate": 2.5069033530571996e-05,
-      "loss": 1.5501,
+      "epoch": 16.96,
+      "learning_rate": 6.608035714285714e-05,
+      "loss": 1.1298,
       "step": 3799
     },
     {
-      "epoch": 11.24,
-      "learning_rate": 2.504930966469428e-05,
-      "loss": 1.4574,
+      "epoch": 16.96,
+      "learning_rate": 6.607142857142857e-05,
+      "loss": 1.1181,
       "step": 3800
     },
     {
-      "epoch": 11.25,
-      "learning_rate": 2.502958579881657e-05,
-      "loss": 1.4842,
+      "epoch": 16.97,
+      "learning_rate": 6.60625e-05,
+      "loss": 1.0485,
       "step": 3801
     },
     {
-      "epoch": 11.25,
-      "learning_rate": 2.5009861932938855e-05,
-      "loss": 1.5941,
+      "epoch": 16.97,
+      "learning_rate": 6.605357142857144e-05,
+      "loss": 1.0645,
       "step": 3802
     },
     {
-      "epoch": 11.25,
-      "learning_rate": 2.4990138067061144e-05,
-      "loss": 1.5049,
+      "epoch": 16.98,
+      "learning_rate": 6.604464285714286e-05,
+      "loss": 1.0915,
       "step": 3803
     },
     {
-      "epoch": 11.25,
-      "learning_rate": 2.4970414201183433e-05,
-      "loss": 1.5371,
+      "epoch": 16.98,
+      "learning_rate": 6.603571428571429e-05,
+      "loss": 1.0918,
       "step": 3804
     },
     {
-      "epoch": 11.26,
-      "learning_rate": 2.495069033530572e-05,
-      "loss": 1.5914,
+      "epoch": 16.99,
+      "learning_rate": 6.602678571428572e-05,
+      "loss": 1.0316,
       "step": 3805
     },
     {
-      "epoch": 11.26,
-      "learning_rate": 2.493096646942801e-05,
-      "loss": 1.4083,
+      "epoch": 16.99,
+      "learning_rate": 6.601785714285714e-05,
+      "loss": 1.0704,
       "step": 3806
     },
     {
-      "epoch": 11.26,
-      "learning_rate": 2.49112426035503e-05,
-      "loss": 1.4795,
+      "epoch": 17.0,
+      "learning_rate": 6.600892857142858e-05,
+      "loss": 1.1105,
       "step": 3807
     },
     {
-      "epoch": 11.27,
-      "learning_rate": 2.4891518737672584e-05,
-      "loss": 1.4984,
+      "epoch": 17.0,
+      "learning_rate": 6.6e-05,
+      "loss": 1.1919,
       "step": 3808
     },
     {
-      "epoch": 11.27,
-      "learning_rate": 2.4871794871794873e-05,
-      "loss": 1.4559,
+      "epoch": 17.0,
+      "learning_rate": 6.599107142857143e-05,
+      "loss": 1.0821,
       "step": 3809
     },
     {
-      "epoch": 11.27,
-      "learning_rate": 2.485207100591716e-05,
-      "loss": 1.5304,
+      "epoch": 17.01,
+      "learning_rate": 6.598214285714286e-05,
+      "loss": 1.1833,
       "step": 3810
     },
     {
-      "epoch": 11.28,
-      "learning_rate": 2.483234714003945e-05,
-      "loss": 1.4901,
+      "epoch": 17.01,
+      "learning_rate": 6.59732142857143e-05,
+      "loss": 1.002,
       "step": 3811
     },
     {
-      "epoch": 11.28,
-      "learning_rate": 2.4812623274161736e-05,
-      "loss": 1.6006,
+      "epoch": 17.02,
+      "learning_rate": 6.596428571428572e-05,
+      "loss": 1.0811,
       "step": 3812
     },
     {
-      "epoch": 11.28,
-      "learning_rate": 2.4792899408284024e-05,
-      "loss": 1.4493,
+      "epoch": 17.02,
+      "learning_rate": 6.595535714285715e-05,
+      "loss": 1.0933,
       "step": 3813
     },
     {
-      "epoch": 11.28,
-      "learning_rate": 2.4773175542406313e-05,
-      "loss": 1.442,
+      "epoch": 17.03,
+      "learning_rate": 6.594642857142857e-05,
+      "loss": 1.0983,
       "step": 3814
     },
     {
-      "epoch": 11.29,
-      "learning_rate": 2.47534516765286e-05,
-      "loss": 1.5658,
+      "epoch": 17.03,
+      "learning_rate": 6.593750000000001e-05,
+      "loss": 1.1323,
       "step": 3815
     },
     {
-      "epoch": 11.29,
-      "learning_rate": 2.473372781065089e-05,
-      "loss": 1.5164,
+      "epoch": 17.04,
+      "learning_rate": 6.592857142857142e-05,
+      "loss": 1.0772,
       "step": 3816
     },
     {
-      "epoch": 11.29,
-      "learning_rate": 2.4714003944773176e-05,
-      "loss": 1.4131,
+      "epoch": 17.04,
+      "learning_rate": 6.591964285714286e-05,
+      "loss": 1.0428,
       "step": 3817
     },
     {
-      "epoch": 11.3,
-      "learning_rate": 2.4694280078895465e-05,
-      "loss": 1.518,
+      "epoch": 17.04,
+      "learning_rate": 6.591071428571429e-05,
+      "loss": 1.103,
       "step": 3818
     },
     {
-      "epoch": 11.3,
-      "learning_rate": 2.4674556213017753e-05,
-      "loss": 1.5181,
+      "epoch": 17.05,
+      "learning_rate": 6.590178571428571e-05,
+      "loss": 1.0836,
       "step": 3819
     },
     {
-      "epoch": 11.3,
-      "learning_rate": 2.4654832347140042e-05,
-      "loss": 1.5011,
+      "epoch": 17.05,
+      "learning_rate": 6.589285714285715e-05,
+      "loss": 1.047,
       "step": 3820
     },
     {
-      "epoch": 11.3,
-      "learning_rate": 2.4635108481262327e-05,
-      "loss": 1.5612,
+      "epoch": 17.06,
+      "learning_rate": 6.588392857142858e-05,
+      "loss": 1.0804,
       "step": 3821
     },
     {
-      "epoch": 11.31,
-      "learning_rate": 2.461538461538462e-05,
-      "loss": 1.6012,
+      "epoch": 17.06,
+      "learning_rate": 6.5875e-05,
+      "loss": 1.0821,
       "step": 3822
     },
     {
-      "epoch": 11.31,
-      "learning_rate": 2.4595660749506905e-05,
-      "loss": 1.5509,
+      "epoch": 17.07,
+      "learning_rate": 6.586607142857143e-05,
+      "loss": 1.0926,
       "step": 3823
     },
     {
-      "epoch": 11.31,
-      "learning_rate": 2.457593688362919e-05,
-      "loss": 1.4795,
+      "epoch": 17.07,
+      "learning_rate": 6.585714285714287e-05,
+      "loss": 1.18,
       "step": 3824
     },
     {
-      "epoch": 11.32,
-      "learning_rate": 2.4556213017751482e-05,
-      "loss": 1.5048,
+      "epoch": 17.08,
+      "learning_rate": 6.584821428571429e-05,
+      "loss": 1.1266,
       "step": 3825
     },
     {
-      "epoch": 11.32,
-      "learning_rate": 2.4536489151873768e-05,
-      "loss": 1.5672,
+      "epoch": 17.08,
+      "learning_rate": 6.583928571428572e-05,
+      "loss": 1.1081,
       "step": 3826
     },
     {
-      "epoch": 11.32,
-      "learning_rate": 2.4516765285996056e-05,
-      "loss": 1.4869,
+      "epoch": 17.08,
+      "learning_rate": 6.583035714285714e-05,
+      "loss": 1.1354,
       "step": 3827
     },
     {
-      "epoch": 11.33,
-      "learning_rate": 2.4497041420118345e-05,
-      "loss": 1.5516,
+      "epoch": 17.09,
+      "learning_rate": 6.582142857142858e-05,
+      "loss": 1.0057,
       "step": 3828
     },
     {
-      "epoch": 11.33,
-      "learning_rate": 2.4477317554240634e-05,
-      "loss": 1.5001,
+      "epoch": 17.09,
+      "learning_rate": 6.58125e-05,
+      "loss": 1.0653,
       "step": 3829
     },
     {
-      "epoch": 11.33,
-      "learning_rate": 2.445759368836292e-05,
-      "loss": 1.486,
+      "epoch": 17.1,
+      "learning_rate": 6.580357142857143e-05,
+      "loss": 1.1033,
       "step": 3830
     },
     {
-      "epoch": 11.33,
-      "learning_rate": 2.4437869822485208e-05,
-      "loss": 1.5286,
+      "epoch": 17.1,
+      "learning_rate": 6.579464285714286e-05,
+      "loss": 1.1096,
       "step": 3831
     },
     {
-      "epoch": 11.34,
-      "learning_rate": 2.4418145956607497e-05,
-      "loss": 1.4732,
+      "epoch": 17.11,
+      "learning_rate": 6.578571428571428e-05,
+      "loss": 1.1422,
       "step": 3832
     },
     {
-      "epoch": 11.34,
-      "learning_rate": 2.4398422090729782e-05,
-      "loss": 1.5875,
+      "epoch": 17.11,
+      "learning_rate": 6.577678571428571e-05,
+      "loss": 1.0316,
       "step": 3833
     },
     {
-      "epoch": 11.34,
-      "learning_rate": 2.4378698224852074e-05,
-      "loss": 1.4337,
+      "epoch": 17.12,
+      "learning_rate": 6.576785714285715e-05,
+      "loss": 1.0298,
       "step": 3834
     },
     {
-      "epoch": 11.35,
-      "learning_rate": 2.435897435897436e-05,
-      "loss": 1.6015,
+      "epoch": 17.12,
+      "learning_rate": 6.575892857142857e-05,
+      "loss": 1.1474,
       "step": 3835
     },
     {
-      "epoch": 11.35,
-      "learning_rate": 2.4339250493096648e-05,
-      "loss": 1.68,
+      "epoch": 17.12,
+      "learning_rate": 6.575e-05,
+      "loss": 1.0318,
       "step": 3836
     },
     {
-      "epoch": 11.35,
-      "learning_rate": 2.4319526627218937e-05,
-      "loss": 1.6564,
+      "epoch": 17.13,
+      "learning_rate": 6.574107142857144e-05,
+      "loss": 1.0726,
       "step": 3837
     },
     {
-      "epoch": 11.36,
-      "learning_rate": 2.4299802761341222e-05,
-      "loss": 1.5301,
+      "epoch": 17.13,
+      "learning_rate": 6.573214285714286e-05,
+      "loss": 1.0556,
       "step": 3838
     },
     {
-      "epoch": 11.36,
-      "learning_rate": 2.428007889546351e-05,
-      "loss": 1.5559,
+      "epoch": 17.14,
+      "learning_rate": 6.572321428571429e-05,
+      "loss": 1.0755,
       "step": 3839
     },
     {
-      "epoch": 11.36,
-      "learning_rate": 2.42603550295858e-05,
-      "loss": 1.467,
+      "epoch": 17.14,
+      "learning_rate": 6.571428571428571e-05,
+      "loss": 1.1337,
       "step": 3840
     },
     {
-      "epoch": 11.36,
-      "learning_rate": 2.424063116370809e-05,
-      "loss": 1.5005,
+      "epoch": 17.15,
+      "learning_rate": 6.570535714285715e-05,
+      "loss": 1.1818,
       "step": 3841
     },
     {
-      "epoch": 11.37,
-      "learning_rate": 2.4220907297830374e-05,
-      "loss": 1.6289,
+      "epoch": 17.15,
+      "learning_rate": 6.569642857142857e-05,
+      "loss": 1.1508,
       "step": 3842
     },
     {
-      "epoch": 11.37,
-      "learning_rate": 2.4201183431952666e-05,
-      "loss": 1.5174,
+      "epoch": 17.16,
+      "learning_rate": 6.56875e-05,
+      "loss": 1.1424,
       "step": 3843
     },
     {
-      "epoch": 11.37,
-      "learning_rate": 2.418145956607495e-05,
-      "loss": 1.532,
+      "epoch": 17.16,
+      "learning_rate": 6.567857142857143e-05,
+      "loss": 1.1405,
       "step": 3844
     },
     {
-      "epoch": 11.38,
-      "learning_rate": 2.416173570019724e-05,
-      "loss": 1.6316,
+      "epoch": 17.17,
+      "learning_rate": 6.566964285714287e-05,
+      "loss": 1.068,
       "step": 3845
     },
     {
-      "epoch": 11.38,
-      "learning_rate": 2.414201183431953e-05,
-      "loss": 1.4573,
+      "epoch": 17.17,
+      "learning_rate": 6.566071428571428e-05,
+      "loss": 1.0514,
       "step": 3846
     },
     {
-      "epoch": 11.38,
-      "learning_rate": 2.4122287968441814e-05,
-      "loss": 1.6001,
+      "epoch": 17.17,
+      "learning_rate": 6.565178571428572e-05,
+      "loss": 1.1023,
       "step": 3847
     },
     {
-      "epoch": 11.38,
-      "learning_rate": 2.4102564102564103e-05,
-      "loss": 1.5116,
+      "epoch": 17.18,
+      "learning_rate": 6.564285714285715e-05,
+      "loss": 1.1076,
       "step": 3848
     },
     {
-      "epoch": 11.39,
-      "learning_rate": 2.408284023668639e-05,
-      "loss": 1.5313,
+      "epoch": 17.18,
+      "learning_rate": 6.563392857142857e-05,
+      "loss": 1.1221,
       "step": 3849
     },
     {
-      "epoch": 11.39,
-      "learning_rate": 2.406311637080868e-05,
-      "loss": 1.5197,
+      "epoch": 17.19,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 1.1335,
       "step": 3850
     },
     {
-      "epoch": 11.39,
-      "learning_rate": 2.404339250493097e-05,
-      "loss": 1.5454,
+      "epoch": 17.19,
+      "learning_rate": 6.561607142857143e-05,
+      "loss": 1.099,
       "step": 3851
     },
     {
-      "epoch": 11.4,
-      "learning_rate": 2.4023668639053258e-05,
-      "loss": 1.4549,
+      "epoch": 17.2,
+      "learning_rate": 6.560714285714286e-05,
+      "loss": 1.1729,
       "step": 3852
     },
     {
-      "epoch": 11.4,
-      "learning_rate": 2.4003944773175543e-05,
-      "loss": 1.4995,
+      "epoch": 17.2,
+      "learning_rate": 6.559821428571429e-05,
+      "loss": 1.0663,
       "step": 3853
     },
     {
-      "epoch": 11.4,
-      "learning_rate": 2.398422090729783e-05,
-      "loss": 1.5604,
+      "epoch": 17.21,
+      "learning_rate": 6.558928571428572e-05,
+      "loss": 1.0541,
       "step": 3854
     },
     {
-      "epoch": 11.41,
-      "learning_rate": 2.396449704142012e-05,
-      "loss": 1.4891,
+      "epoch": 17.21,
+      "learning_rate": 6.558035714285714e-05,
+      "loss": 1.0853,
       "step": 3855
     },
     {
-      "epoch": 11.41,
-      "learning_rate": 2.3944773175542406e-05,
-      "loss": 1.518,
+      "epoch": 17.21,
+      "learning_rate": 6.557142857142858e-05,
+      "loss": 1.1314,
       "step": 3856
     },
     {
-      "epoch": 11.41,
-      "learning_rate": 2.3925049309664694e-05,
-      "loss": 1.5788,
+      "epoch": 17.22,
+      "learning_rate": 6.55625e-05,
+      "loss": 1.0684,
       "step": 3857
     },
     {
-      "epoch": 11.41,
-      "learning_rate": 2.3905325443786983e-05,
-      "loss": 1.5739,
+      "epoch": 17.22,
+      "learning_rate": 6.555357142857144e-05,
+      "loss": 1.0735,
       "step": 3858
     },
     {
-      "epoch": 11.42,
-      "learning_rate": 2.3885601577909272e-05,
-      "loss": 1.5237,
+      "epoch": 17.23,
+      "learning_rate": 6.554464285714285e-05,
+      "loss": 1.0682,
       "step": 3859
     },
     {
-      "epoch": 11.42,
-      "learning_rate": 2.386587771203156e-05,
-      "loss": 1.5577,
+      "epoch": 17.23,
+      "learning_rate": 6.553571428571429e-05,
+      "loss": 1.0357,
       "step": 3860
     },
     {
-      "epoch": 11.42,
-      "learning_rate": 2.384615384615385e-05,
-      "loss": 1.4586,
+      "epoch": 17.24,
+      "learning_rate": 6.552678571428572e-05,
+      "loss": 1.0318,
       "step": 3861
     },
     {
-      "epoch": 11.43,
-      "learning_rate": 2.3826429980276135e-05,
-      "loss": 1.4176,
+      "epoch": 17.24,
+      "learning_rate": 6.551785714285714e-05,
+      "loss": 1.053,
       "step": 3862
     },
     {
-      "epoch": 11.43,
-      "learning_rate": 2.3806706114398423e-05,
-      "loss": 1.4003,
+      "epoch": 17.25,
+      "learning_rate": 6.550892857142857e-05,
+      "loss": 1.1235,
       "step": 3863
     },
     {
-      "epoch": 11.43,
-      "learning_rate": 2.3786982248520712e-05,
-      "loss": 1.4539,
+      "epoch": 17.25,
+      "learning_rate": 6.55e-05,
+      "loss": 1.0619,
       "step": 3864
     },
     {
-      "epoch": 11.43,
-      "learning_rate": 2.3767258382642997e-05,
-      "loss": 1.5073,
+      "epoch": 17.25,
+      "learning_rate": 6.549107142857143e-05,
+      "loss": 1.1138,
       "step": 3865
     },
     {
-      "epoch": 11.44,
-      "learning_rate": 2.374753451676529e-05,
-      "loss": 1.511,
+      "epoch": 17.26,
+      "learning_rate": 6.548214285714286e-05,
+      "loss": 1.1066,
       "step": 3866
     },
     {
-      "epoch": 11.44,
-      "learning_rate": 2.3727810650887575e-05,
-      "loss": 1.5356,
+      "epoch": 17.26,
+      "learning_rate": 6.54732142857143e-05,
+      "loss": 1.0969,
       "step": 3867
     },
     {
-      "epoch": 11.44,
-      "learning_rate": 2.3708086785009864e-05,
-      "loss": 1.5218,
+      "epoch": 17.27,
+      "learning_rate": 6.546428571428572e-05,
+      "loss": 1.0658,
       "step": 3868
     },
     {
-      "epoch": 11.45,
-      "learning_rate": 2.3688362919132152e-05,
-      "loss": 1.4847,
+      "epoch": 17.27,
+      "learning_rate": 6.545535714285715e-05,
+      "loss": 1.171,
       "step": 3869
     },
     {
-      "epoch": 11.45,
-      "learning_rate": 2.3668639053254438e-05,
-      "loss": 1.5406,
+      "epoch": 17.28,
+      "learning_rate": 6.544642857142857e-05,
+      "loss": 1.0527,
       "step": 3870
     },
     {
-      "epoch": 11.45,
-      "learning_rate": 2.3648915187376726e-05,
-      "loss": 1.6504,
+      "epoch": 17.28,
+      "learning_rate": 6.543750000000001e-05,
+      "loss": 1.0583,
       "step": 3871
     },
     {
-      "epoch": 11.46,
-      "learning_rate": 2.3629191321499015e-05,
-      "loss": 1.4019,
+      "epoch": 17.29,
+      "learning_rate": 6.542857142857142e-05,
+      "loss": 0.9918,
       "step": 3872
     },
     {
-      "epoch": 11.46,
-      "learning_rate": 2.3609467455621304e-05,
-      "loss": 1.5051,
+      "epoch": 17.29,
+      "learning_rate": 6.541964285714286e-05,
+      "loss": 1.1323,
       "step": 3873
     },
     {
-      "epoch": 11.46,
-      "learning_rate": 2.358974358974359e-05,
-      "loss": 1.5612,
+      "epoch": 17.29,
+      "learning_rate": 6.541071428571429e-05,
+      "loss": 1.0946,
       "step": 3874
     },
     {
-      "epoch": 11.46,
-      "learning_rate": 2.357001972386588e-05,
-      "loss": 1.5369,
+      "epoch": 17.3,
+      "learning_rate": 6.540178571428571e-05,
+      "loss": 1.1032,
       "step": 3875
     },
     {
-      "epoch": 11.47,
-      "learning_rate": 2.3550295857988167e-05,
-      "loss": 1.4753,
+      "epoch": 17.3,
+      "learning_rate": 6.539285714285714e-05,
+      "loss": 1.143,
       "step": 3876
     },
     {
-      "epoch": 11.47,
-      "learning_rate": 2.3530571992110455e-05,
-      "loss": 1.6106,
+      "epoch": 17.31,
+      "learning_rate": 6.538392857142858e-05,
+      "loss": 1.1227,
       "step": 3877
     },
     {
-      "epoch": 11.47,
-      "learning_rate": 2.3510848126232744e-05,
-      "loss": 1.5731,
+      "epoch": 17.31,
+      "learning_rate": 6.5375e-05,
+      "loss": 1.1216,
       "step": 3878
     },
     {
-      "epoch": 11.48,
-      "learning_rate": 2.349112426035503e-05,
-      "loss": 1.5377,
+      "epoch": 17.32,
+      "learning_rate": 6.536607142857143e-05,
+      "loss": 1.1036,
       "step": 3879
     },
     {
-      "epoch": 11.48,
-      "learning_rate": 2.3471400394477318e-05,
-      "loss": 1.5032,
+      "epoch": 17.32,
+      "learning_rate": 6.535714285714287e-05,
+      "loss": 1.051,
       "step": 3880
     },
     {
-      "epoch": 11.48,
-      "learning_rate": 2.3451676528599607e-05,
-      "loss": 1.4935,
+      "epoch": 17.33,
+      "learning_rate": 6.53482142857143e-05,
+      "loss": 1.0645,
       "step": 3881
     },
     {
-      "epoch": 11.49,
-      "learning_rate": 2.3431952662721896e-05,
-      "loss": 1.4418,
+      "epoch": 17.33,
+      "learning_rate": 6.533928571428572e-05,
+      "loss": 1.1823,
       "step": 3882
     },
     {
-      "epoch": 11.49,
-      "learning_rate": 2.341222879684418e-05,
-      "loss": 1.5347,
+      "epoch": 17.33,
+      "learning_rate": 6.533035714285714e-05,
+      "loss": 1.1111,
       "step": 3883
     },
     {
-      "epoch": 11.49,
-      "learning_rate": 2.3392504930966473e-05,
-      "loss": 1.5239,
+      "epoch": 17.34,
+      "learning_rate": 6.532142857142858e-05,
+      "loss": 1.1075,
       "step": 3884
     },
     {
-      "epoch": 11.49,
-      "learning_rate": 2.337278106508876e-05,
-      "loss": 1.4942,
+      "epoch": 17.34,
+      "learning_rate": 6.53125e-05,
+      "loss": 1.1138,
       "step": 3885
     },
     {
-      "epoch": 11.5,
-      "learning_rate": 2.3353057199211044e-05,
-      "loss": 1.4287,
+      "epoch": 17.35,
+      "learning_rate": 6.530357142857143e-05,
+      "loss": 1.0891,
       "step": 3886
     },
     {
-      "epoch": 11.5,
-      "learning_rate": 2.3333333333333336e-05,
-      "loss": 1.555,
+      "epoch": 17.35,
+      "learning_rate": 6.529464285714286e-05,
+      "loss": 1.1397,
       "step": 3887
     },
     {
-      "epoch": 11.5,
-      "learning_rate": 2.331360946745562e-05,
-      "loss": 1.4773,
+      "epoch": 17.36,
+      "learning_rate": 6.528571428571429e-05,
+      "loss": 1.119,
       "step": 3888
     },
     {
-      "epoch": 11.51,
-      "learning_rate": 2.329388560157791e-05,
-      "loss": 1.5395,
+      "epoch": 17.36,
+      "learning_rate": 6.527678571428571e-05,
+      "loss": 1.0427,
       "step": 3889
     },
     {
-      "epoch": 11.51,
-      "learning_rate": 2.32741617357002e-05,
-      "loss": 1.4844,
+      "epoch": 17.37,
+      "learning_rate": 6.526785714285715e-05,
+      "loss": 1.15,
       "step": 3890
     },
     {
-      "epoch": 11.51,
-      "learning_rate": 2.3254437869822487e-05,
-      "loss": 1.4484,
+      "epoch": 17.37,
+      "learning_rate": 6.525892857142857e-05,
+      "loss": 1.0748,
       "step": 3891
     },
     {
-      "epoch": 11.51,
-      "learning_rate": 2.3234714003944773e-05,
-      "loss": 1.605,
+      "epoch": 17.38,
+      "learning_rate": 6.525e-05,
+      "loss": 1.0667,
       "step": 3892
     },
     {
-      "epoch": 11.52,
-      "learning_rate": 2.321499013806706e-05,
-      "loss": 1.5195,
+      "epoch": 17.38,
+      "learning_rate": 6.524107142857143e-05,
+      "loss": 1.0895,
       "step": 3893
     },
     {
-      "epoch": 11.52,
-      "learning_rate": 2.319526627218935e-05,
-      "loss": 1.497,
+      "epoch": 17.38,
+      "learning_rate": 6.523214285714286e-05,
+      "loss": 1.0532,
       "step": 3894
     },
     {
-      "epoch": 11.52,
-      "learning_rate": 2.317554240631164e-05,
-      "loss": 1.4831,
+      "epoch": 17.39,
+      "learning_rate": 6.522321428571429e-05,
+      "loss": 1.1959,
       "step": 3895
     },
     {
-      "epoch": 11.53,
-      "learning_rate": 2.3155818540433928e-05,
-      "loss": 1.484,
+      "epoch": 17.39,
+      "learning_rate": 6.521428571428572e-05,
+      "loss": 1.1088,
       "step": 3896
     },
     {
-      "epoch": 11.53,
-      "learning_rate": 2.3136094674556213e-05,
-      "loss": 1.5322,
+      "epoch": 17.4,
+      "learning_rate": 6.520535714285715e-05,
+      "loss": 1.0274,
       "step": 3897
     },
     {
-      "epoch": 11.53,
-      "learning_rate": 2.31163708086785e-05,
-      "loss": 1.5574,
+      "epoch": 17.4,
+      "learning_rate": 6.519642857142857e-05,
+      "loss": 1.1049,
       "step": 3898
     },
     {
-      "epoch": 11.54,
-      "learning_rate": 2.309664694280079e-05,
-      "loss": 1.4984,
+      "epoch": 17.41,
+      "learning_rate": 6.51875e-05,
+      "loss": 1.1999,
       "step": 3899
     },
     {
-      "epoch": 11.54,
-      "learning_rate": 2.307692307692308e-05,
-      "loss": 1.4911,
+      "epoch": 17.41,
+      "learning_rate": 6.517857142857143e-05,
+      "loss": 1.114,
       "step": 3900
     },
     {
-      "epoch": 11.54,
-      "learning_rate": 2.3057199211045364e-05,
-      "loss": 1.5551,
+      "epoch": 17.41,
+      "eval_loss": 1.1413170099258423,
+      "eval_runtime": 44.7492,
+      "eval_samples_per_second": 292.609,
+      "eval_steps_per_second": 1.542,
+      "step": 3900
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 6.516964285714287e-05,
+      "loss": 1.1063,
       "step": 3901
     },
     {
-      "epoch": 11.54,
-      "learning_rate": 2.3037475345167653e-05,
-      "loss": 1.5136,
+      "epoch": 17.42,
+      "learning_rate": 6.516071428571428e-05,
+      "loss": 1.1295,
       "step": 3902
     },
     {
-      "epoch": 11.55,
-      "learning_rate": 2.3017751479289942e-05,
-      "loss": 1.6196,
+      "epoch": 17.42,
+      "learning_rate": 6.515178571428572e-05,
+      "loss": 1.0899,
       "step": 3903
     },
     {
-      "epoch": 11.55,
-      "learning_rate": 2.299802761341223e-05,
-      "loss": 1.4466,
+      "epoch": 17.43,
+      "learning_rate": 6.514285714285715e-05,
+      "loss": 1.1566,
       "step": 3904
     },
     {
-      "epoch": 11.55,
-      "learning_rate": 2.297830374753452e-05,
-      "loss": 1.4943,
+      "epoch": 17.43,
+      "learning_rate": 6.513392857142857e-05,
+      "loss": 1.0116,
       "step": 3905
     },
     {
-      "epoch": 11.56,
-      "learning_rate": 2.2958579881656805e-05,
-      "loss": 1.6048,
+      "epoch": 17.44,
+      "learning_rate": 6.5125e-05,
+      "loss": 1.1541,
       "step": 3906
     },
     {
-      "epoch": 11.56,
-      "learning_rate": 2.2938856015779093e-05,
-      "loss": 1.4924,
+      "epoch": 17.44,
+      "learning_rate": 6.511607142857144e-05,
+      "loss": 1.0738,
       "step": 3907
     },
     {
-      "epoch": 11.56,
-      "learning_rate": 2.2919132149901382e-05,
-      "loss": 1.5339,
+      "epoch": 17.45,
+      "learning_rate": 6.510714285714286e-05,
+      "loss": 1.1022,
       "step": 3908
     },
     {
-      "epoch": 11.57,
-      "learning_rate": 2.2899408284023667e-05,
-      "loss": 1.4731,
+      "epoch": 17.45,
+      "learning_rate": 6.509821428571429e-05,
+      "loss": 1.0834,
       "step": 3909
     },
     {
-      "epoch": 11.57,
-      "learning_rate": 2.287968441814596e-05,
-      "loss": 1.5362,
+      "epoch": 17.46,
+      "learning_rate": 6.508928571428571e-05,
+      "loss": 1.1574,
       "step": 3910
     },
     {
-      "epoch": 11.57,
-      "learning_rate": 2.2859960552268245e-05,
-      "loss": 1.4232,
+      "epoch": 17.46,
+      "learning_rate": 6.508035714285715e-05,
+      "loss": 1.1083,
       "step": 3911
     },
     {
-      "epoch": 11.57,
-      "learning_rate": 2.2840236686390534e-05,
-      "loss": 1.5834,
+      "epoch": 17.46,
+      "learning_rate": 6.507142857142858e-05,
+      "loss": 1.089,
       "step": 3912
     },
     {
-      "epoch": 11.58,
-      "learning_rate": 2.2820512820512822e-05,
-      "loss": 1.52,
+      "epoch": 17.47,
+      "learning_rate": 6.50625e-05,
+      "loss": 1.087,
       "step": 3913
     },
     {
-      "epoch": 11.58,
-      "learning_rate": 2.280078895463511e-05,
-      "loss": 1.538,
+      "epoch": 17.47,
+      "learning_rate": 6.505357142857144e-05,
+      "loss": 1.0931,
       "step": 3914
     },
     {
-      "epoch": 11.58,
-      "learning_rate": 2.2781065088757396e-05,
-      "loss": 1.4782,
+      "epoch": 17.48,
+      "learning_rate": 6.504464285714285e-05,
+      "loss": 1.231,
       "step": 3915
     },
     {
-      "epoch": 11.59,
-      "learning_rate": 2.276134122287969e-05,
-      "loss": 1.5761,
+      "epoch": 17.48,
+      "learning_rate": 6.503571428571429e-05,
+      "loss": 1.0979,
       "step": 3916
     },
     {
-      "epoch": 11.59,
-      "learning_rate": 2.2741617357001974e-05,
-      "loss": 1.5675,
+      "epoch": 17.49,
+      "learning_rate": 6.502678571428572e-05,
+      "loss": 1.0432,
       "step": 3917
     },
     {
-      "epoch": 11.59,
-      "learning_rate": 2.272189349112426e-05,
-      "loss": 1.4933,
+      "epoch": 17.49,
+      "learning_rate": 6.501785714285714e-05,
+      "loss": 1.1732,
       "step": 3918
     },
     {
-      "epoch": 11.59,
-      "learning_rate": 2.270216962524655e-05,
-      "loss": 1.5684,
+      "epoch": 17.5,
+      "learning_rate": 6.500892857142857e-05,
+      "loss": 1.0802,
       "step": 3919
     },
     {
-      "epoch": 11.6,
-      "learning_rate": 2.2682445759368837e-05,
-      "loss": 1.4698,
+      "epoch": 17.5,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.0751,
       "step": 3920
     },
     {
-      "epoch": 11.6,
-      "learning_rate": 2.2662721893491125e-05,
-      "loss": 1.5726,
+      "epoch": 17.5,
+      "learning_rate": 6.499107142857142e-05,
+      "loss": 1.1534,
       "step": 3921
     },
     {
-      "epoch": 11.6,
-      "learning_rate": 2.2642998027613414e-05,
-      "loss": 1.5573,
+      "epoch": 17.51,
+      "learning_rate": 6.498214285714286e-05,
+      "loss": 1.0418,
       "step": 3922
     },
     {
-      "epoch": 11.61,
-      "learning_rate": 2.2623274161735703e-05,
-      "loss": 1.4633,
+      "epoch": 17.51,
+      "learning_rate": 6.497321428571428e-05,
+      "loss": 1.1336,
       "step": 3923
     },
     {
-      "epoch": 11.61,
-      "learning_rate": 2.2603550295857988e-05,
-      "loss": 1.61,
+      "epoch": 17.52,
+      "learning_rate": 6.496428571428572e-05,
+      "loss": 1.0554,
       "step": 3924
     },
     {
-      "epoch": 11.61,
-      "learning_rate": 2.2583826429980277e-05,
-      "loss": 1.5585,
+      "epoch": 17.52,
+      "learning_rate": 6.495535714285715e-05,
+      "loss": 1.1025,
       "step": 3925
     },
     {
-      "epoch": 11.62,
-      "learning_rate": 2.2564102564102566e-05,
-      "loss": 1.4016,
+      "epoch": 17.53,
+      "learning_rate": 6.494642857142857e-05,
+      "loss": 1.172,
       "step": 3926
     },
     {
-      "epoch": 11.62,
-      "learning_rate": 2.254437869822485e-05,
-      "loss": 1.566,
+      "epoch": 17.53,
+      "learning_rate": 6.493750000000001e-05,
+      "loss": 1.1497,
       "step": 3927
     },
     {
-      "epoch": 11.62,
-      "learning_rate": 2.2524654832347143e-05,
-      "loss": 1.5971,
+      "epoch": 17.54,
+      "learning_rate": 6.492857142857143e-05,
+      "loss": 1.1061,
       "step": 3928
     },
     {
-      "epoch": 11.62,
-      "learning_rate": 2.250493096646943e-05,
-      "loss": 1.5449,
+      "epoch": 17.54,
+      "learning_rate": 6.491964285714286e-05,
+      "loss": 1.1635,
       "step": 3929
     },
     {
-      "epoch": 11.63,
-      "learning_rate": 2.2485207100591717e-05,
-      "loss": 1.5265,
+      "epoch": 17.54,
+      "learning_rate": 6.491071428571429e-05,
+      "loss": 1.1402,
       "step": 3930
     },
     {
-      "epoch": 11.63,
-      "learning_rate": 2.2465483234714006e-05,
-      "loss": 1.483,
+      "epoch": 17.55,
+      "learning_rate": 6.490178571428571e-05,
+      "loss": 1.057,
       "step": 3931
     },
     {
-      "epoch": 11.63,
-      "learning_rate": 2.244575936883629e-05,
-      "loss": 1.4567,
+      "epoch": 17.55,
+      "learning_rate": 6.489285714285714e-05,
+      "loss": 1.1446,
       "step": 3932
     },
     {
-      "epoch": 11.64,
-      "learning_rate": 2.242603550295858e-05,
-      "loss": 1.6027,
+      "epoch": 17.56,
+      "learning_rate": 6.488392857142858e-05,
+      "loss": 1.0981,
       "step": 3933
     },
     {
-      "epoch": 11.64,
-      "learning_rate": 2.240631163708087e-05,
-      "loss": 1.4606,
+      "epoch": 17.56,
+      "learning_rate": 6.4875e-05,
+      "loss": 1.0217,
       "step": 3934
     },
     {
-      "epoch": 11.64,
-      "learning_rate": 2.2386587771203157e-05,
-      "loss": 1.5066,
+      "epoch": 17.57,
+      "learning_rate": 6.486607142857143e-05,
+      "loss": 1.0876,
       "step": 3935
     },
     {
-      "epoch": 11.64,
-      "learning_rate": 2.2366863905325443e-05,
-      "loss": 1.533,
+      "epoch": 17.57,
+      "learning_rate": 6.485714285714286e-05,
+      "loss": 1.1008,
       "step": 3936
     },
     {
-      "epoch": 11.65,
-      "learning_rate": 2.2347140039447735e-05,
-      "loss": 1.5199,
+      "epoch": 17.58,
+      "learning_rate": 6.48482142857143e-05,
+      "loss": 1.0472,
       "step": 3937
     },
     {
-      "epoch": 11.65,
-      "learning_rate": 2.232741617357002e-05,
-      "loss": 1.5797,
+      "epoch": 17.58,
+      "learning_rate": 6.483928571428572e-05,
+      "loss": 1.1076,
       "step": 3938
     },
     {
-      "epoch": 11.65,
-      "learning_rate": 2.230769230769231e-05,
-      "loss": 1.5237,
+      "epoch": 17.58,
+      "learning_rate": 6.483035714285715e-05,
+      "loss": 1.0594,
       "step": 3939
     },
     {
-      "epoch": 11.66,
-      "learning_rate": 2.2287968441814598e-05,
-      "loss": 1.5205,
+      "epoch": 17.59,
+      "learning_rate": 6.482142857142857e-05,
+      "loss": 1.0996,
       "step": 3940
     },
     {
-      "epoch": 11.66,
-      "learning_rate": 2.2268244575936883e-05,
-      "loss": 1.4477,
+      "epoch": 17.59,
+      "learning_rate": 6.48125e-05,
+      "loss": 1.1658,
       "step": 3941
     },
     {
-      "epoch": 11.66,
-      "learning_rate": 2.224852071005917e-05,
-      "loss": 1.5671,
+      "epoch": 17.6,
+      "learning_rate": 6.480357142857144e-05,
+      "loss": 1.1053,
       "step": 3942
     },
     {
-      "epoch": 11.67,
-      "learning_rate": 2.222879684418146e-05,
-      "loss": 1.4821,
+      "epoch": 17.6,
+      "learning_rate": 6.479464285714286e-05,
+      "loss": 1.1035,
       "step": 3943
     },
     {
-      "epoch": 11.67,
-      "learning_rate": 2.220907297830375e-05,
-      "loss": 1.4893,
+      "epoch": 17.61,
+      "learning_rate": 6.47857142857143e-05,
+      "loss": 1.2014,
       "step": 3944
     },
     {
-      "epoch": 11.67,
-      "learning_rate": 2.2189349112426034e-05,
-      "loss": 1.5215,
+      "epoch": 17.61,
+      "learning_rate": 6.477678571428571e-05,
+      "loss": 1.136,
       "step": 3945
     },
     {
-      "epoch": 11.67,
-      "learning_rate": 2.2169625246548327e-05,
-      "loss": 1.5608,
+      "epoch": 17.62,
+      "learning_rate": 6.476785714285715e-05,
+      "loss": 1.0852,
       "step": 3946
     },
     {
-      "epoch": 11.68,
-      "learning_rate": 2.2149901380670612e-05,
-      "loss": 1.5188,
+      "epoch": 17.62,
+      "learning_rate": 6.475892857142858e-05,
+      "loss": 1.1168,
       "step": 3947
     },
     {
-      "epoch": 11.68,
-      "learning_rate": 2.21301775147929e-05,
-      "loss": 1.5168,
+      "epoch": 17.62,
+      "learning_rate": 6.475e-05,
+      "loss": 1.0747,
       "step": 3948
     },
     {
-      "epoch": 11.68,
-      "learning_rate": 2.211045364891519e-05,
-      "loss": 1.5293,
+      "epoch": 17.63,
+      "learning_rate": 6.474107142857143e-05,
+      "loss": 1.1004,
       "step": 3949
     },
     {
-      "epoch": 11.69,
-      "learning_rate": 2.2090729783037475e-05,
-      "loss": 1.4956,
+      "epoch": 17.63,
+      "learning_rate": 6.473214285714287e-05,
+      "loss": 1.1401,
       "step": 3950
     },
     {
-      "epoch": 11.69,
-      "learning_rate": 2.2071005917159763e-05,
-      "loss": 1.4969,
+      "epoch": 17.64,
+      "learning_rate": 6.472321428571428e-05,
+      "loss": 1.2034,
       "step": 3951
     },
     {
-      "epoch": 11.69,
-      "learning_rate": 2.2051282051282052e-05,
-      "loss": 1.5545,
+      "epoch": 17.64,
+      "learning_rate": 6.471428571428572e-05,
+      "loss": 1.0846,
       "step": 3952
     },
     {
-      "epoch": 11.7,
-      "learning_rate": 2.203155818540434e-05,
-      "loss": 1.5051,
+      "epoch": 17.65,
+      "learning_rate": 6.470535714285714e-05,
+      "loss": 1.0366,
       "step": 3953
     },
     {
-      "epoch": 11.7,
-      "learning_rate": 2.201183431952663e-05,
-      "loss": 1.5447,
+      "epoch": 17.65,
+      "learning_rate": 6.469642857142858e-05,
+      "loss": 1.1295,
       "step": 3954
     },
     {
-      "epoch": 11.7,
-      "learning_rate": 2.199211045364892e-05,
-      "loss": 1.5402,
+      "epoch": 17.66,
+      "learning_rate": 6.468750000000001e-05,
+      "loss": 1.0924,
       "step": 3955
     },
     {
-      "epoch": 11.7,
-      "learning_rate": 2.1972386587771204e-05,
-      "loss": 1.5198,
+      "epoch": 17.66,
+      "learning_rate": 6.467857142857143e-05,
+      "loss": 1.0528,
       "step": 3956
     },
     {
-      "epoch": 11.71,
-      "learning_rate": 2.1952662721893492e-05,
-      "loss": 1.5177,
+      "epoch": 17.67,
+      "learning_rate": 6.466964285714287e-05,
+      "loss": 1.0997,
       "step": 3957
     },
     {
-      "epoch": 11.71,
-      "learning_rate": 2.193293885601578e-05,
-      "loss": 1.4708,
+      "epoch": 17.67,
+      "learning_rate": 6.466071428571428e-05,
+      "loss": 1.0921,
       "step": 3958
     },
     {
-      "epoch": 11.71,
-      "learning_rate": 2.1913214990138066e-05,
-      "loss": 1.5379,
+      "epoch": 17.67,
+      "learning_rate": 6.465178571428572e-05,
+      "loss": 1.1502,
       "step": 3959
     },
     {
-      "epoch": 11.72,
-      "learning_rate": 2.189349112426036e-05,
-      "loss": 1.5549,
+      "epoch": 17.68,
+      "learning_rate": 6.464285714285715e-05,
+      "loss": 1.1246,
       "step": 3960
     },
     {
-      "epoch": 11.72,
-      "learning_rate": 2.1873767258382644e-05,
-      "loss": 1.5236,
+      "epoch": 17.68,
+      "learning_rate": 6.463392857142857e-05,
+      "loss": 1.1606,
       "step": 3961
     },
     {
-      "epoch": 11.72,
-      "learning_rate": 2.1854043392504933e-05,
-      "loss": 1.535,
+      "epoch": 17.69,
+      "learning_rate": 6.4625e-05,
+      "loss": 1.1508,
       "step": 3962
     },
     {
-      "epoch": 11.72,
-      "learning_rate": 2.183431952662722e-05,
-      "loss": 1.5194,
+      "epoch": 17.69,
+      "learning_rate": 6.461607142857144e-05,
+      "loss": 1.0459,
       "step": 3963
     },
     {
-      "epoch": 11.73,
-      "learning_rate": 2.1814595660749507e-05,
-      "loss": 1.5852,
+      "epoch": 17.7,
+      "learning_rate": 6.460714285714285e-05,
+      "loss": 1.097,
       "step": 3964
     },
     {
-      "epoch": 11.73,
-      "learning_rate": 2.1794871794871795e-05,
-      "loss": 1.4721,
+      "epoch": 17.7,
+      "learning_rate": 6.459821428571429e-05,
+      "loss": 1.0905,
       "step": 3965
     },
     {
-      "epoch": 11.73,
-      "learning_rate": 2.1775147928994084e-05,
-      "loss": 1.625,
+      "epoch": 17.71,
+      "learning_rate": 6.458928571428571e-05,
+      "loss": 1.0977,
       "step": 3966
     },
     {
-      "epoch": 11.74,
-      "learning_rate": 2.1755424063116373e-05,
-      "loss": 1.4772,
+      "epoch": 17.71,
+      "learning_rate": 6.458035714285715e-05,
+      "loss": 1.0416,
       "step": 3967
     },
     {
-      "epoch": 11.74,
-      "learning_rate": 2.1735700197238658e-05,
-      "loss": 1.5934,
+      "epoch": 17.71,
+      "learning_rate": 6.457142857142856e-05,
+      "loss": 1.1174,
       "step": 3968
     },
     {
-      "epoch": 11.74,
-      "learning_rate": 2.171597633136095e-05,
-      "loss": 1.4686,
+      "epoch": 17.72,
+      "learning_rate": 6.45625e-05,
+      "loss": 1.1846,
       "step": 3969
     },
     {
-      "epoch": 11.75,
-      "learning_rate": 2.1696252465483236e-05,
-      "loss": 1.433,
+      "epoch": 17.72,
+      "learning_rate": 6.455357142857143e-05,
+      "loss": 1.0784,
       "step": 3970
     },
     {
-      "epoch": 11.75,
-      "learning_rate": 2.1676528599605524e-05,
-      "loss": 1.5019,
+      "epoch": 17.73,
+      "learning_rate": 6.454464285714285e-05,
+      "loss": 1.0164,
       "step": 3971
     },
     {
-      "epoch": 11.75,
-      "learning_rate": 2.1656804733727813e-05,
-      "loss": 1.5283,
+      "epoch": 17.73,
+      "learning_rate": 6.45357142857143e-05,
+      "loss": 1.1076,
       "step": 3972
     },
     {
-      "epoch": 11.75,
-      "learning_rate": 2.16370808678501e-05,
-      "loss": 1.5143,
+      "epoch": 17.74,
+      "learning_rate": 6.452678571428572e-05,
+      "loss": 1.1995,
       "step": 3973
     },
     {
-      "epoch": 11.76,
-      "learning_rate": 2.1617357001972387e-05,
-      "loss": 1.5033,
+      "epoch": 17.74,
+      "learning_rate": 6.451785714285714e-05,
+      "loss": 1.0684,
       "step": 3974
     },
     {
-      "epoch": 11.76,
-      "learning_rate": 2.1597633136094676e-05,
-      "loss": 1.5821,
+      "epoch": 17.75,
+      "learning_rate": 6.450892857142857e-05,
+      "loss": 1.0471,
       "step": 3975
     },
     {
-      "epoch": 11.76,
-      "learning_rate": 2.1577909270216965e-05,
-      "loss": 1.4746,
+      "epoch": 17.75,
+      "learning_rate": 6.450000000000001e-05,
+      "loss": 1.1258,
       "step": 3976
     },
     {
-      "epoch": 11.77,
-      "learning_rate": 2.155818540433925e-05,
-      "loss": 1.4856,
+      "epoch": 17.75,
+      "learning_rate": 6.449107142857143e-05,
+      "loss": 1.1579,
       "step": 3977
     },
     {
-      "epoch": 11.77,
-      "learning_rate": 2.1538461538461542e-05,
-      "loss": 1.5431,
+      "epoch": 17.76,
+      "learning_rate": 6.448214285714286e-05,
+      "loss": 1.088,
       "step": 3978
     },
     {
-      "epoch": 11.77,
-      "learning_rate": 2.1518737672583827e-05,
-      "loss": 1.5486,
+      "epoch": 17.76,
+      "learning_rate": 6.447321428571429e-05,
+      "loss": 1.0464,
       "step": 3979
     },
     {
-      "epoch": 11.78,
-      "learning_rate": 2.1499013806706113e-05,
-      "loss": 1.4281,
+      "epoch": 17.77,
+      "learning_rate": 6.446428571428572e-05,
+      "loss": 1.1045,
       "step": 3980
     },
     {
-      "epoch": 11.78,
-      "learning_rate": 2.1479289940828405e-05,
-      "loss": 1.4135,
+      "epoch": 17.77,
+      "learning_rate": 6.445535714285714e-05,
+      "loss": 1.0433,
       "step": 3981
     },
     {
-      "epoch": 11.78,
-      "learning_rate": 2.145956607495069e-05,
-      "loss": 1.5298,
+      "epoch": 17.78,
+      "learning_rate": 6.444642857142858e-05,
+      "loss": 1.099,
       "step": 3982
     },
     {
-      "epoch": 11.78,
-      "learning_rate": 2.143984220907298e-05,
-      "loss": 1.5711,
+      "epoch": 17.78,
+      "learning_rate": 6.44375e-05,
+      "loss": 1.0612,
       "step": 3983
     },
     {
-      "epoch": 11.79,
-      "learning_rate": 2.1420118343195268e-05,
-      "loss": 1.4044,
+      "epoch": 17.79,
+      "learning_rate": 6.442857142857143e-05,
+      "loss": 1.0865,
       "step": 3984
     },
     {
-      "epoch": 11.79,
-      "learning_rate": 2.1400394477317556e-05,
-      "loss": 1.4524,
+      "epoch": 17.79,
+      "learning_rate": 6.441964285714287e-05,
+      "loss": 1.0644,
       "step": 3985
     },
     {
-      "epoch": 11.79,
-      "learning_rate": 2.138067061143984e-05,
-      "loss": 1.4978,
+      "epoch": 17.79,
+      "learning_rate": 6.441071428571429e-05,
+      "loss": 1.1389,
       "step": 3986
     },
     {
-      "epoch": 11.8,
-      "learning_rate": 2.136094674556213e-05,
-      "loss": 1.609,
+      "epoch": 17.8,
+      "learning_rate": 6.440178571428572e-05,
+      "loss": 1.0838,
       "step": 3987
     },
     {
-      "epoch": 11.8,
-      "learning_rate": 2.134122287968442e-05,
-      "loss": 1.5264,
+      "epoch": 17.8,
+      "learning_rate": 6.439285714285714e-05,
+      "loss": 1.1137,
       "step": 3988
     },
     {
-      "epoch": 11.8,
-      "learning_rate": 2.1321499013806704e-05,
-      "loss": 1.5291,
+      "epoch": 17.81,
+      "learning_rate": 6.438392857142858e-05,
+      "loss": 1.0848,
       "step": 3989
     },
     {
-      "epoch": 11.8,
-      "learning_rate": 2.1301775147928997e-05,
-      "loss": 1.4846,
+      "epoch": 17.81,
+      "learning_rate": 6.4375e-05,
+      "loss": 1.171,
       "step": 3990
     },
     {
-      "epoch": 11.81,
-      "learning_rate": 2.1282051282051282e-05,
-      "loss": 1.5262,
+      "epoch": 17.82,
+      "learning_rate": 6.436607142857143e-05,
+      "loss": 1.0883,
       "step": 3991
     },
     {
-      "epoch": 11.81,
-      "learning_rate": 2.126232741617357e-05,
-      "loss": 1.5409,
+      "epoch": 17.82,
+      "learning_rate": 6.435714285714286e-05,
+      "loss": 1.122,
       "step": 3992
     },
     {
-      "epoch": 11.81,
-      "learning_rate": 2.124260355029586e-05,
-      "loss": 1.4966,
+      "epoch": 17.83,
+      "learning_rate": 6.43482142857143e-05,
+      "loss": 1.1018,
       "step": 3993
     },
     {
-      "epoch": 11.82,
-      "learning_rate": 2.1222879684418148e-05,
-      "loss": 1.4439,
+      "epoch": 17.83,
+      "learning_rate": 6.433928571428571e-05,
+      "loss": 1.0625,
       "step": 3994
     },
     {
-      "epoch": 11.82,
-      "learning_rate": 2.1203155818540433e-05,
-      "loss": 1.5454,
+      "epoch": 17.83,
+      "learning_rate": 6.433035714285715e-05,
+      "loss": 1.0724,
       "step": 3995
     },
     {
-      "epoch": 11.82,
-      "learning_rate": 2.1183431952662722e-05,
-      "loss": 1.5864,
+      "epoch": 17.84,
+      "learning_rate": 6.432142857142857e-05,
+      "loss": 1.1054,
       "step": 3996
     },
     {
-      "epoch": 11.83,
-      "learning_rate": 2.116370808678501e-05,
-      "loss": 1.5029,
+      "epoch": 17.84,
+      "learning_rate": 6.43125e-05,
+      "loss": 1.074,
       "step": 3997
     },
     {
-      "epoch": 11.83,
-      "learning_rate": 2.11439842209073e-05,
-      "loss": 1.5343,
+      "epoch": 17.85,
+      "learning_rate": 6.430357142857142e-05,
+      "loss": 1.039,
       "step": 3998
     },
     {
-      "epoch": 11.83,
-      "learning_rate": 2.112426035502959e-05,
-      "loss": 1.4932,
+      "epoch": 17.85,
+      "learning_rate": 6.429464285714286e-05,
+      "loss": 1.086,
       "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,
+      "epoch": 17.86,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 1.1802,
       "step": 4000
     },
     {
-      "epoch": 11.84,
-      "learning_rate": 2.1084812623274162e-05,
-      "loss": 1.5637,
+      "epoch": 17.86,
+      "learning_rate": 6.427678571428571e-05,
+      "loss": 1.1104,
       "step": 4001
     },
     {
-      "epoch": 11.84,
-      "learning_rate": 2.106508875739645e-05,
-      "loss": 1.5078,
+      "epoch": 17.87,
+      "learning_rate": 6.426785714285715e-05,
+      "loss": 1.1565,
       "step": 4002
     },
     {
-      "epoch": 11.84,
-      "learning_rate": 2.1045364891518736e-05,
-      "loss": 1.5403,
+      "epoch": 17.87,
+      "learning_rate": 6.425892857142858e-05,
+      "loss": 1.1047,
       "step": 4003
     },
     {
-      "epoch": 11.85,
-      "learning_rate": 2.102564102564103e-05,
-      "loss": 1.5753,
+      "epoch": 17.88,
+      "learning_rate": 6.425e-05,
+      "loss": 1.1056,
       "step": 4004
     },
     {
-      "epoch": 11.85,
-      "learning_rate": 2.1005917159763314e-05,
-      "loss": 1.5425,
+      "epoch": 17.88,
+      "learning_rate": 6.424107142857143e-05,
+      "loss": 1.1669,
       "step": 4005
     },
     {
-      "epoch": 11.85,
-      "learning_rate": 2.0986193293885603e-05,
-      "loss": 1.4366,
+      "epoch": 17.88,
+      "learning_rate": 6.423214285714287e-05,
+      "loss": 0.9843,
       "step": 4006
     },
     {
-      "epoch": 11.86,
-      "learning_rate": 2.096646942800789e-05,
-      "loss": 1.4725,
+      "epoch": 17.89,
+      "learning_rate": 6.422321428571428e-05,
+      "loss": 1.1293,
       "step": 4007
     },
     {
-      "epoch": 11.86,
-      "learning_rate": 2.094674556213018e-05,
-      "loss": 1.4372,
+      "epoch": 17.89,
+      "learning_rate": 6.421428571428572e-05,
+      "loss": 1.0625,
       "step": 4008
     },
     {
-      "epoch": 11.86,
-      "learning_rate": 2.0927021696252465e-05,
-      "loss": 1.5661,
+      "epoch": 17.9,
+      "learning_rate": 6.420535714285714e-05,
+      "loss": 1.2013,
       "step": 4009
     },
     {
-      "epoch": 11.86,
-      "learning_rate": 2.0907297830374754e-05,
-      "loss": 1.6106,
+      "epoch": 17.9,
+      "learning_rate": 6.419642857142858e-05,
+      "loss": 1.1798,
       "step": 4010
     },
     {
-      "epoch": 11.87,
-      "learning_rate": 2.0887573964497043e-05,
-      "loss": 1.5462,
+      "epoch": 17.91,
+      "learning_rate": 6.41875e-05,
+      "loss": 1.2092,
       "step": 4011
     },
     {
-      "epoch": 11.87,
-      "learning_rate": 2.0867850098619328e-05,
-      "loss": 1.4595,
+      "epoch": 17.91,
+      "learning_rate": 6.417857142857143e-05,
+      "loss": 1.1179,
       "step": 4012
     },
     {
-      "epoch": 11.87,
-      "learning_rate": 2.084812623274162e-05,
-      "loss": 1.5283,
+      "epoch": 17.92,
+      "learning_rate": 6.416964285714286e-05,
+      "loss": 1.0513,
       "step": 4013
     },
     {
-      "epoch": 11.88,
-      "learning_rate": 2.0828402366863906e-05,
-      "loss": 1.5471,
+      "epoch": 17.92,
+      "learning_rate": 6.416071428571428e-05,
+      "loss": 1.1249,
       "step": 4014
     },
     {
-      "epoch": 11.88,
-      "learning_rate": 2.0808678500986194e-05,
-      "loss": 1.4812,
+      "epoch": 17.92,
+      "learning_rate": 6.415178571428572e-05,
+      "loss": 1.0506,
       "step": 4015
     },
     {
-      "epoch": 11.88,
-      "learning_rate": 2.0788954635108483e-05,
-      "loss": 1.4654,
+      "epoch": 17.93,
+      "learning_rate": 6.414285714285715e-05,
+      "loss": 1.0531,
       "step": 4016
     },
     {
-      "epoch": 11.88,
-      "learning_rate": 2.0769230769230772e-05,
-      "loss": 1.505,
+      "epoch": 17.93,
+      "learning_rate": 6.413392857142857e-05,
+      "loss": 1.0406,
       "step": 4017
     },
     {
-      "epoch": 11.89,
-      "learning_rate": 2.0749506903353057e-05,
-      "loss": 1.5331,
+      "epoch": 17.94,
+      "learning_rate": 6.4125e-05,
+      "loss": 1.1354,
       "step": 4018
     },
     {
-      "epoch": 11.89,
-      "learning_rate": 2.0729783037475346e-05,
-      "loss": 1.5755,
+      "epoch": 17.94,
+      "learning_rate": 6.411607142857144e-05,
+      "loss": 1.1128,
       "step": 4019
     },
     {
-      "epoch": 11.89,
-      "learning_rate": 2.0710059171597635e-05,
-      "loss": 1.5443,
+      "epoch": 17.95,
+      "learning_rate": 6.410714285714286e-05,
+      "loss": 1.0831,
       "step": 4020
     },
     {
-      "epoch": 11.9,
-      "learning_rate": 2.069033530571992e-05,
-      "loss": 1.5015,
+      "epoch": 17.95,
+      "learning_rate": 6.409821428571429e-05,
+      "loss": 1.1208,
       "step": 4021
     },
     {
-      "epoch": 11.9,
-      "learning_rate": 2.0670611439842212e-05,
-      "loss": 1.5939,
+      "epoch": 17.96,
+      "learning_rate": 6.408928571428572e-05,
+      "loss": 1.0403,
       "step": 4022
     },
     {
-      "epoch": 11.9,
-      "learning_rate": 2.0650887573964497e-05,
-      "loss": 1.4551,
+      "epoch": 17.96,
+      "learning_rate": 6.408035714285715e-05,
+      "loss": 1.0834,
       "step": 4023
     },
     {
-      "epoch": 11.91,
-      "learning_rate": 2.0631163708086786e-05,
-      "loss": 1.5079,
+      "epoch": 17.96,
+      "learning_rate": 6.407142857142857e-05,
+      "loss": 1.1466,
       "step": 4024
     },
     {
-      "epoch": 11.91,
-      "learning_rate": 2.0611439842209075e-05,
-      "loss": 1.4947,
+      "epoch": 17.97,
+      "learning_rate": 6.40625e-05,
+      "loss": 1.1627,
       "step": 4025
     },
     {
-      "epoch": 11.91,
-      "learning_rate": 2.0591715976331364e-05,
-      "loss": 1.4686,
+      "epoch": 17.97,
+      "learning_rate": 6.405357142857143e-05,
+      "loss": 0.9805,
       "step": 4026
     },
     {
-      "epoch": 11.91,
-      "learning_rate": 2.057199211045365e-05,
-      "loss": 1.4927,
+      "epoch": 17.98,
+      "learning_rate": 6.404464285714286e-05,
+      "loss": 1.01,
       "step": 4027
     },
     {
-      "epoch": 11.92,
-      "learning_rate": 2.0552268244575938e-05,
-      "loss": 1.5401,
+      "epoch": 17.98,
+      "learning_rate": 6.403571428571428e-05,
+      "loss": 1.0826,
       "step": 4028
     },
     {
-      "epoch": 11.92,
-      "learning_rate": 2.0532544378698226e-05,
-      "loss": 1.5739,
+      "epoch": 17.99,
+      "learning_rate": 6.402678571428572e-05,
+      "loss": 1.0792,
       "step": 4029
     },
     {
-      "epoch": 11.92,
-      "learning_rate": 2.0512820512820512e-05,
-      "loss": 1.538,
+      "epoch": 17.99,
+      "learning_rate": 6.401785714285715e-05,
+      "loss": 1.1203,
       "step": 4030
     },
     {
-      "epoch": 11.93,
-      "learning_rate": 2.0493096646942804e-05,
-      "loss": 1.4578,
+      "epoch": 18.0,
+      "learning_rate": 6.400892857142857e-05,
+      "loss": 1.0486,
       "step": 4031
     },
     {
-      "epoch": 11.93,
-      "learning_rate": 2.047337278106509e-05,
-      "loss": 1.5104,
+      "epoch": 18.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.195,
       "step": 4032
     },
     {
-      "epoch": 11.93,
-      "learning_rate": 2.0453648915187378e-05,
-      "loss": 1.4663,
+      "epoch": 18.0,
+      "learning_rate": 6.399107142857144e-05,
+      "loss": 1.1468,
       "step": 4033
     },
     {
-      "epoch": 11.93,
-      "learning_rate": 2.0433925049309667e-05,
-      "loss": 1.4785,
+      "epoch": 18.01,
+      "learning_rate": 6.398214285714286e-05,
+      "loss": 1.1098,
       "step": 4034
     },
     {
-      "epoch": 11.94,
-      "learning_rate": 2.0414201183431952e-05,
-      "loss": 1.4676,
+      "epoch": 18.01,
+      "learning_rate": 6.397321428571429e-05,
+      "loss": 1.0224,
       "step": 4035
     },
     {
-      "epoch": 11.94,
-      "learning_rate": 2.039447731755424e-05,
-      "loss": 1.6178,
+      "epoch": 18.02,
+      "learning_rate": 6.396428571428573e-05,
+      "loss": 1.1301,
       "step": 4036
     },
     {
-      "epoch": 11.94,
-      "learning_rate": 2.037475345167653e-05,
-      "loss": 1.4598,
+      "epoch": 18.02,
+      "learning_rate": 6.395535714285714e-05,
+      "loss": 1.0217,
       "step": 4037
     },
     {
-      "epoch": 11.95,
-      "learning_rate": 2.0355029585798818e-05,
-      "loss": 1.4459,
+      "epoch": 18.03,
+      "learning_rate": 6.394642857142858e-05,
+      "loss": 1.1227,
       "step": 4038
     },
     {
-      "epoch": 11.95,
-      "learning_rate": 2.0335305719921103e-05,
-      "loss": 1.4983,
+      "epoch": 18.03,
+      "learning_rate": 6.39375e-05,
+      "loss": 1.0781,
       "step": 4039
     },
     {
-      "epoch": 11.95,
-      "learning_rate": 2.0315581854043396e-05,
-      "loss": 1.4677,
+      "epoch": 18.04,
+      "learning_rate": 6.392857142857143e-05,
+      "loss": 1.0688,
       "step": 4040
     },
     {
-      "epoch": 11.96,
-      "learning_rate": 2.029585798816568e-05,
-      "loss": 1.5457,
+      "epoch": 18.04,
+      "learning_rate": 6.391964285714285e-05,
+      "loss": 1.0926,
       "step": 4041
     },
     {
-      "epoch": 11.96,
-      "learning_rate": 2.027613412228797e-05,
-      "loss": 1.4478,
+      "epoch": 18.04,
+      "learning_rate": 6.391071428571429e-05,
+      "loss": 1.1436,
       "step": 4042
     },
     {
-      "epoch": 11.96,
-      "learning_rate": 2.025641025641026e-05,
-      "loss": 1.5569,
+      "epoch": 18.05,
+      "learning_rate": 6.390178571428572e-05,
+      "loss": 1.1426,
       "step": 4043
     },
     {
-      "epoch": 11.96,
-      "learning_rate": 2.0236686390532544e-05,
-      "loss": 1.5118,
+      "epoch": 18.05,
+      "learning_rate": 6.389285714285714e-05,
+      "loss": 1.0384,
       "step": 4044
     },
     {
-      "epoch": 11.97,
-      "learning_rate": 2.0216962524654832e-05,
-      "loss": 1.455,
+      "epoch": 18.06,
+      "learning_rate": 6.388392857142857e-05,
+      "loss": 1.0772,
       "step": 4045
     },
     {
-      "epoch": 11.97,
-      "learning_rate": 2.019723865877712e-05,
-      "loss": 1.5226,
+      "epoch": 18.06,
+      "learning_rate": 6.387500000000001e-05,
+      "loss": 1.1195,
       "step": 4046
     },
     {
-      "epoch": 11.97,
-      "learning_rate": 2.017751479289941e-05,
-      "loss": 1.4823,
+      "epoch": 18.07,
+      "learning_rate": 6.386607142857143e-05,
+      "loss": 1.0855,
       "step": 4047
     },
     {
-      "epoch": 11.98,
-      "learning_rate": 2.0157790927021695e-05,
-      "loss": 1.4724,
+      "epoch": 18.07,
+      "learning_rate": 6.385714285714286e-05,
+      "loss": 1.0597,
       "step": 4048
     },
     {
-      "epoch": 11.98,
-      "learning_rate": 2.0138067061143987e-05,
-      "loss": 1.4857,
+      "epoch": 18.08,
+      "learning_rate": 6.38482142857143e-05,
+      "loss": 1.1286,
       "step": 4049
     },
     {
-      "epoch": 11.98,
-      "learning_rate": 2.0118343195266273e-05,
-      "loss": 1.6352,
+      "epoch": 18.08,
+      "learning_rate": 6.383928571428571e-05,
+      "loss": 1.0497,
       "step": 4050
     },
     {
-      "epoch": 11.99,
-      "learning_rate": 2.009861932938856e-05,
-      "loss": 1.4974,
+      "epoch": 18.08,
+      "learning_rate": 6.383035714285715e-05,
+      "loss": 1.1613,
       "step": 4051
     },
     {
-      "epoch": 11.99,
-      "learning_rate": 2.007889546351085e-05,
-      "loss": 1.4888,
+      "epoch": 18.09,
+      "learning_rate": 6.382142857142857e-05,
+      "loss": 1.1024,
       "step": 4052
     },
     {
-      "epoch": 11.99,
-      "learning_rate": 2.0059171597633135e-05,
-      "loss": 1.5282,
+      "epoch": 18.09,
+      "learning_rate": 6.381250000000001e-05,
+      "loss": 1.0826,
       "step": 4053
     },
     {
-      "epoch": 11.99,
-      "learning_rate": 2.0039447731755424e-05,
-      "loss": 1.5485,
+      "epoch": 18.1,
+      "learning_rate": 6.380357142857142e-05,
+      "loss": 1.0152,
       "step": 4054
     },
     {
-      "epoch": 12.0,
-      "learning_rate": 2.0019723865877713e-05,
-      "loss": 1.6068,
+      "epoch": 18.1,
+      "learning_rate": 6.379464285714286e-05,
+      "loss": 0.9967,
       "step": 4055
     },
     {
-      "epoch": 12.0,
-      "learning_rate": 2e-05,
-      "loss": 1.5936,
+      "epoch": 18.11,
+      "learning_rate": 6.378571428571429e-05,
+      "loss": 1.1444,
       "step": 4056
     },
     {
-      "epoch": 12.0,
-      "learning_rate": 1.998027613412229e-05,
-      "loss": 1.5364,
+      "epoch": 18.11,
+      "learning_rate": 6.377678571428571e-05,
+      "loss": 1.1173,
       "step": 4057
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 1.9960552268244576e-05,
-      "loss": 1.438,
+      "epoch": 18.12,
+      "learning_rate": 6.376785714285714e-05,
+      "loss": 1.0825,
       "step": 4058
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 1.9940828402366864e-05,
-      "loss": 1.5224,
+      "epoch": 18.12,
+      "learning_rate": 6.375892857142858e-05,
+      "loss": 1.0962,
       "step": 4059
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 1.9921104536489153e-05,
-      "loss": 1.5321,
+      "epoch": 18.12,
+      "learning_rate": 6.375e-05,
+      "loss": 1.0742,
       "step": 4060
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 1.9901380670611442e-05,
-      "loss": 1.5962,
+      "epoch": 18.13,
+      "learning_rate": 6.374107142857143e-05,
+      "loss": 1.1047,
       "step": 4061
     },
     {
-      "epoch": 12.02,
-      "learning_rate": 1.9881656804733727e-05,
-      "loss": 1.3947,
+      "epoch": 18.13,
+      "learning_rate": 6.373214285714287e-05,
+      "loss": 1.1307,
       "step": 4062
     },
     {
-      "epoch": 12.02,
-      "learning_rate": 1.986193293885602e-05,
-      "loss": 1.481,
+      "epoch": 18.14,
+      "learning_rate": 6.372321428571428e-05,
+      "loss": 1.0675,
       "step": 4063
     },
     {
-      "epoch": 12.02,
-      "learning_rate": 1.9842209072978305e-05,
-      "loss": 1.6088,
+      "epoch": 18.14,
+      "learning_rate": 6.371428571428572e-05,
+      "loss": 1.0532,
       "step": 4064
     },
     {
-      "epoch": 12.03,
-      "learning_rate": 1.9822485207100593e-05,
-      "loss": 1.566,
+      "epoch": 18.15,
+      "learning_rate": 6.370535714285715e-05,
+      "loss": 1.1075,
       "step": 4065
     },
     {
-      "epoch": 12.03,
-      "learning_rate": 1.9802761341222882e-05,
-      "loss": 1.4306,
+      "epoch": 18.15,
+      "learning_rate": 6.369642857142858e-05,
+      "loss": 1.054,
       "step": 4066
     },
     {
-      "epoch": 12.03,
-      "learning_rate": 1.9783037475345167e-05,
-      "loss": 1.4425,
+      "epoch": 18.16,
+      "learning_rate": 6.36875e-05,
+      "loss": 1.1584,
       "step": 4067
     },
     {
-      "epoch": 12.04,
-      "learning_rate": 1.9763313609467456e-05,
-      "loss": 1.5176,
+      "epoch": 18.16,
+      "learning_rate": 6.367857142857144e-05,
+      "loss": 1.1031,
       "step": 4068
     },
     {
-      "epoch": 12.04,
-      "learning_rate": 1.9743589743589745e-05,
-      "loss": 1.5632,
+      "epoch": 18.17,
+      "learning_rate": 6.366964285714286e-05,
+      "loss": 1.1358,
       "step": 4069
     },
     {
-      "epoch": 12.04,
-      "learning_rate": 1.9723865877712034e-05,
-      "loss": 1.4995,
+      "epoch": 18.17,
+      "learning_rate": 6.366071428571429e-05,
+      "loss": 1.069,
       "step": 4070
     },
     {
-      "epoch": 12.04,
-      "learning_rate": 1.970414201183432e-05,
-      "loss": 1.4757,
+      "epoch": 18.17,
+      "learning_rate": 6.365178571428571e-05,
+      "loss": 1.0678,
       "step": 4071
     },
     {
-      "epoch": 12.05,
-      "learning_rate": 1.968441814595661e-05,
-      "loss": 1.5045,
+      "epoch": 18.18,
+      "learning_rate": 6.364285714285715e-05,
+      "loss": 1.0537,
       "step": 4072
     },
     {
-      "epoch": 12.05,
-      "learning_rate": 1.9664694280078896e-05,
-      "loss": 1.5206,
+      "epoch": 18.18,
+      "learning_rate": 6.363392857142858e-05,
+      "loss": 1.1358,
       "step": 4073
     },
     {
-      "epoch": 12.05,
-      "learning_rate": 1.9644970414201182e-05,
-      "loss": 1.4875,
+      "epoch": 18.19,
+      "learning_rate": 6.3625e-05,
+      "loss": 1.0898,
       "step": 4074
     },
     {
-      "epoch": 12.06,
-      "learning_rate": 1.9625246548323474e-05,
-      "loss": 1.5377,
+      "epoch": 18.19,
+      "learning_rate": 6.361607142857143e-05,
+      "loss": 1.0881,
       "step": 4075
     },
     {
-      "epoch": 12.06,
-      "learning_rate": 1.960552268244576e-05,
-      "loss": 1.4142,
+      "epoch": 18.2,
+      "learning_rate": 6.360714285714287e-05,
+      "loss": 1.0534,
       "step": 4076
     },
     {
-      "epoch": 12.06,
-      "learning_rate": 1.9585798816568048e-05,
-      "loss": 1.4396,
+      "epoch": 18.2,
+      "learning_rate": 6.359821428571429e-05,
+      "loss": 1.0969,
       "step": 4077
     },
     {
-      "epoch": 12.07,
-      "learning_rate": 1.9566074950690337e-05,
-      "loss": 1.4574,
+      "epoch": 18.21,
+      "learning_rate": 6.358928571428572e-05,
+      "loss": 1.0667,
       "step": 4078
     },
     {
-      "epoch": 12.07,
-      "learning_rate": 1.9546351084812625e-05,
-      "loss": 1.521,
+      "epoch": 18.21,
+      "learning_rate": 6.358035714285716e-05,
+      "loss": 1.1621,
       "step": 4079
     },
     {
-      "epoch": 12.07,
-      "learning_rate": 1.952662721893491e-05,
-      "loss": 1.548,
+      "epoch": 18.21,
+      "learning_rate": 6.357142857142857e-05,
+      "loss": 0.9702,
       "step": 4080
     },
     {
-      "epoch": 12.07,
-      "learning_rate": 1.9506903353057203e-05,
-      "loss": 1.5249,
+      "epoch": 18.22,
+      "learning_rate": 6.356250000000001e-05,
+      "loss": 1.0638,
       "step": 4081
     },
     {
-      "epoch": 12.08,
-      "learning_rate": 1.9487179487179488e-05,
-      "loss": 1.5541,
+      "epoch": 18.22,
+      "learning_rate": 6.355357142857143e-05,
+      "loss": 1.0788,
       "step": 4082
     },
     {
-      "epoch": 12.08,
-      "learning_rate": 1.9467455621301774e-05,
-      "loss": 1.4202,
+      "epoch": 18.23,
+      "learning_rate": 6.354464285714286e-05,
+      "loss": 1.0381,
       "step": 4083
     },
     {
-      "epoch": 12.08,
-      "learning_rate": 1.9447731755424066e-05,
-      "loss": 1.4028,
+      "epoch": 18.23,
+      "learning_rate": 6.353571428571428e-05,
+      "loss": 1.0403,
       "step": 4084
     },
     {
-      "epoch": 12.09,
-      "learning_rate": 1.942800788954635e-05,
-      "loss": 1.5512,
+      "epoch": 18.24,
+      "learning_rate": 6.352678571428572e-05,
+      "loss": 1.074,
       "step": 4085
     },
     {
-      "epoch": 12.09,
-      "learning_rate": 1.940828402366864e-05,
-      "loss": 1.4373,
+      "epoch": 18.24,
+      "learning_rate": 6.351785714285715e-05,
+      "loss": 1.0826,
       "step": 4086
     },
     {
-      "epoch": 12.09,
-      "learning_rate": 1.938856015779093e-05,
-      "loss": 1.4749,
+      "epoch": 18.25,
+      "learning_rate": 6.350892857142857e-05,
+      "loss": 1.1594,
       "step": 4087
     },
     {
-      "epoch": 12.09,
-      "learning_rate": 1.9368836291913217e-05,
-      "loss": 1.4923,
+      "epoch": 18.25,
+      "learning_rate": 6.35e-05,
+      "loss": 1.0959,
       "step": 4088
     },
     {
-      "epoch": 12.1,
-      "learning_rate": 1.9349112426035502e-05,
-      "loss": 1.495,
+      "epoch": 18.25,
+      "learning_rate": 6.349107142857144e-05,
+      "loss": 1.0389,
       "step": 4089
     },
     {
-      "epoch": 12.1,
-      "learning_rate": 1.932938856015779e-05,
-      "loss": 1.5964,
+      "epoch": 18.26,
+      "learning_rate": 6.348214285714286e-05,
+      "loss": 1.1089,
       "step": 4090
     },
     {
-      "epoch": 12.1,
-      "learning_rate": 1.930966469428008e-05,
-      "loss": 1.5687,
+      "epoch": 18.26,
+      "learning_rate": 6.347321428571429e-05,
+      "loss": 1.0784,
       "step": 4091
     },
     {
-      "epoch": 12.11,
-      "learning_rate": 1.9289940828402365e-05,
-      "loss": 1.4688,
+      "epoch": 18.27,
+      "learning_rate": 6.346428571428571e-05,
+      "loss": 1.0723,
       "step": 4092
     },
     {
-      "epoch": 12.11,
-      "learning_rate": 1.9270216962524657e-05,
-      "loss": 1.4821,
+      "epoch": 18.27,
+      "learning_rate": 6.345535714285714e-05,
+      "loss": 1.1133,
       "step": 4093
     },
     {
-      "epoch": 12.11,
-      "learning_rate": 1.9250493096646943e-05,
-      "loss": 1.5323,
+      "epoch": 18.28,
+      "learning_rate": 6.344642857142858e-05,
+      "loss": 1.0581,
       "step": 4094
     },
     {
-      "epoch": 12.12,
-      "learning_rate": 1.923076923076923e-05,
-      "loss": 1.5593,
+      "epoch": 18.28,
+      "learning_rate": 6.34375e-05,
+      "loss": 1.1328,
       "step": 4095
     },
     {
-      "epoch": 12.12,
-      "learning_rate": 1.921104536489152e-05,
-      "loss": 1.5117,
+      "epoch": 18.29,
+      "learning_rate": 6.342857142857143e-05,
+      "loss": 1.0827,
       "step": 4096
     },
     {
-      "epoch": 12.12,
-      "learning_rate": 1.9191321499013805e-05,
-      "loss": 1.5514,
+      "epoch": 18.29,
+      "learning_rate": 6.341964285714285e-05,
+      "loss": 1.0637,
       "step": 4097
     },
     {
-      "epoch": 12.12,
-      "learning_rate": 1.9171597633136094e-05,
-      "loss": 1.534,
+      "epoch": 18.29,
+      "learning_rate": 6.34107142857143e-05,
+      "loss": 1.0901,
       "step": 4098
     },
     {
-      "epoch": 12.13,
-      "learning_rate": 1.9151873767258383e-05,
-      "loss": 1.4936,
+      "epoch": 18.3,
+      "learning_rate": 6.340178571428572e-05,
+      "loss": 1.0997,
       "step": 4099
     },
     {
-      "epoch": 12.13,
-      "learning_rate": 1.913214990138067e-05,
-      "loss": 1.4988,
+      "epoch": 18.3,
+      "learning_rate": 6.339285714285714e-05,
+      "loss": 1.1484,
       "step": 4100
     },
     {
-      "epoch": 12.13,
-      "learning_rate": 1.911242603550296e-05,
-      "loss": 1.5185,
+      "epoch": 18.31,
+      "learning_rate": 6.338392857142857e-05,
+      "loss": 1.1245,
       "step": 4101
     },
     {
-      "epoch": 12.14,
-      "learning_rate": 1.909270216962525e-05,
-      "loss": 1.5722,
+      "epoch": 18.31,
+      "learning_rate": 6.337500000000001e-05,
+      "loss": 1.0712,
       "step": 4102
     },
     {
-      "epoch": 12.14,
-      "learning_rate": 1.9072978303747534e-05,
-      "loss": 1.597,
+      "epoch": 18.32,
+      "learning_rate": 6.336607142857142e-05,
+      "loss": 1.1362,
       "step": 4103
     },
     {
-      "epoch": 12.14,
-      "learning_rate": 1.9053254437869823e-05,
-      "loss": 1.4537,
+      "epoch": 18.32,
+      "learning_rate": 6.335714285714286e-05,
+      "loss": 1.0353,
       "step": 4104
     },
     {
-      "epoch": 12.14,
-      "learning_rate": 1.9033530571992112e-05,
-      "loss": 1.5169,
+      "epoch": 18.33,
+      "learning_rate": 6.334821428571429e-05,
+      "loss": 1.1183,
       "step": 4105
     },
     {
-      "epoch": 12.15,
-      "learning_rate": 1.9013806706114397e-05,
-      "loss": 1.5272,
+      "epoch": 18.33,
+      "learning_rate": 6.333928571428571e-05,
+      "loss": 1.0752,
       "step": 4106
     },
     {
-      "epoch": 12.15,
-      "learning_rate": 1.899408284023669e-05,
-      "loss": 1.5185,
+      "epoch": 18.33,
+      "learning_rate": 6.333035714285715e-05,
+      "loss": 1.1885,
       "step": 4107
     },
     {
-      "epoch": 12.15,
-      "learning_rate": 1.8974358974358975e-05,
-      "loss": 1.4632,
+      "epoch": 18.34,
+      "learning_rate": 6.332142857142858e-05,
+      "loss": 1.1038,
       "step": 4108
     },
     {
-      "epoch": 12.16,
-      "learning_rate": 1.8954635108481263e-05,
-      "loss": 1.5662,
+      "epoch": 18.34,
+      "learning_rate": 6.331250000000001e-05,
+      "loss": 1.2189,
       "step": 4109
     },
     {
-      "epoch": 12.16,
-      "learning_rate": 1.8934911242603552e-05,
-      "loss": 1.5147,
+      "epoch": 18.35,
+      "learning_rate": 6.330357142857143e-05,
+      "loss": 1.069,
       "step": 4110
     },
     {
-      "epoch": 12.16,
-      "learning_rate": 1.891518737672584e-05,
-      "loss": 1.5018,
+      "epoch": 18.35,
+      "learning_rate": 6.329464285714287e-05,
+      "loss": 1.1233,
       "step": 4111
     },
     {
-      "epoch": 12.17,
-      "learning_rate": 1.8895463510848126e-05,
-      "loss": 1.4931,
+      "epoch": 18.36,
+      "learning_rate": 6.328571428571429e-05,
+      "loss": 1.0793,
       "step": 4112
     },
     {
-      "epoch": 12.17,
-      "learning_rate": 1.8875739644970415e-05,
-      "loss": 1.595,
+      "epoch": 18.36,
+      "learning_rate": 6.327678571428572e-05,
+      "loss": 1.0026,
       "step": 4113
     },
     {
-      "epoch": 12.17,
-      "learning_rate": 1.8856015779092704e-05,
-      "loss": 1.5018,
+      "epoch": 18.37,
+      "learning_rate": 6.326785714285714e-05,
+      "loss": 1.072,
       "step": 4114
     },
     {
-      "epoch": 12.17,
-      "learning_rate": 1.883629191321499e-05,
-      "loss": 1.5057,
+      "epoch": 18.37,
+      "learning_rate": 6.325892857142858e-05,
+      "loss": 1.1419,
       "step": 4115
     },
     {
-      "epoch": 12.18,
-      "learning_rate": 1.881656804733728e-05,
-      "loss": 1.5517,
+      "epoch": 18.38,
+      "learning_rate": 6.324999999999999e-05,
+      "loss": 0.9762,
       "step": 4116
     },
     {
-      "epoch": 12.18,
-      "learning_rate": 1.8796844181459566e-05,
-      "loss": 1.5801,
+      "epoch": 18.38,
+      "learning_rate": 6.324107142857143e-05,
+      "loss": 1.1314,
       "step": 4117
     },
     {
-      "epoch": 12.18,
-      "learning_rate": 1.8777120315581855e-05,
-      "loss": 1.6221,
+      "epoch": 18.38,
+      "learning_rate": 6.323214285714286e-05,
+      "loss": 1.0852,
       "step": 4118
     },
     {
-      "epoch": 12.19,
-      "learning_rate": 1.8757396449704144e-05,
-      "loss": 1.496,
+      "epoch": 18.39,
+      "learning_rate": 6.32232142857143e-05,
+      "loss": 1.0709,
       "step": 4119
     },
     {
-      "epoch": 12.19,
-      "learning_rate": 1.8737672583826433e-05,
-      "loss": 1.5247,
+      "epoch": 18.39,
+      "learning_rate": 6.321428571428572e-05,
+      "loss": 1.1099,
       "step": 4120
     },
     {
-      "epoch": 12.19,
-      "learning_rate": 1.8717948717948718e-05,
-      "loss": 1.5099,
+      "epoch": 18.4,
+      "learning_rate": 6.320535714285715e-05,
+      "loss": 1.0907,
       "step": 4121
     },
     {
-      "epoch": 12.2,
-      "learning_rate": 1.8698224852071007e-05,
-      "loss": 1.5412,
+      "epoch": 18.4,
+      "learning_rate": 6.319642857142857e-05,
+      "loss": 1.1225,
       "step": 4122
     },
     {
-      "epoch": 12.2,
-      "learning_rate": 1.8678500986193295e-05,
-      "loss": 1.5734,
+      "epoch": 18.41,
+      "learning_rate": 6.31875e-05,
+      "loss": 1.0495,
       "step": 4123
     },
     {
-      "epoch": 12.2,
-      "learning_rate": 1.865877712031558e-05,
-      "loss": 1.5339,
+      "epoch": 18.41,
+      "learning_rate": 6.317857142857144e-05,
+      "loss": 1.0343,
       "step": 4124
     },
     {
-      "epoch": 12.2,
-      "learning_rate": 1.8639053254437873e-05,
-      "loss": 1.5401,
+      "epoch": 18.42,
+      "learning_rate": 6.316964285714286e-05,
+      "loss": 1.106,
       "step": 4125
     },
     {
-      "epoch": 12.21,
-      "learning_rate": 1.8619329388560158e-05,
-      "loss": 1.5508,
+      "epoch": 18.42,
+      "learning_rate": 6.316071428571429e-05,
+      "loss": 1.0849,
       "step": 4126
     },
     {
-      "epoch": 12.21,
-      "learning_rate": 1.8599605522682447e-05,
-      "loss": 1.5531,
+      "epoch": 18.42,
+      "learning_rate": 6.315178571428571e-05,
+      "loss": 1.0641,
       "step": 4127
     },
     {
-      "epoch": 12.21,
-      "learning_rate": 1.8579881656804736e-05,
-      "loss": 1.4241,
+      "epoch": 18.43,
+      "learning_rate": 6.314285714285715e-05,
+      "loss": 1.1234,
       "step": 4128
     },
     {
-      "epoch": 12.22,
-      "learning_rate": 1.856015779092702e-05,
-      "loss": 1.5448,
+      "epoch": 18.43,
+      "learning_rate": 6.313392857142858e-05,
+      "loss": 1.0796,
       "step": 4129
     },
     {
-      "epoch": 12.22,
-      "learning_rate": 1.854043392504931e-05,
-      "loss": 1.5039,
+      "epoch": 18.44,
+      "learning_rate": 6.3125e-05,
+      "loss": 1.1021,
       "step": 4130
     },
     {
-      "epoch": 12.22,
-      "learning_rate": 1.85207100591716e-05,
-      "loss": 1.5151,
+      "epoch": 18.44,
+      "learning_rate": 6.311607142857143e-05,
+      "loss": 1.1126,
       "step": 4131
     },
     {
-      "epoch": 12.22,
-      "learning_rate": 1.8500986193293887e-05,
-      "loss": 1.6209,
+      "epoch": 18.45,
+      "learning_rate": 6.310714285714287e-05,
+      "loss": 1.0949,
       "step": 4132
     },
     {
-      "epoch": 12.23,
-      "learning_rate": 1.8481262327416172e-05,
-      "loss": 1.5379,
+      "epoch": 18.45,
+      "learning_rate": 6.309821428571428e-05,
+      "loss": 1.1123,
       "step": 4133
     },
     {
-      "epoch": 12.23,
-      "learning_rate": 1.8461538461538465e-05,
-      "loss": 1.5718,
+      "epoch": 18.46,
+      "learning_rate": 6.308928571428572e-05,
+      "loss": 1.1136,
       "step": 4134
     },
     {
-      "epoch": 12.23,
-      "learning_rate": 1.844181459566075e-05,
-      "loss": 1.5115,
+      "epoch": 18.46,
+      "learning_rate": 6.308035714285714e-05,
+      "loss": 1.1409,
       "step": 4135
     },
     {
-      "epoch": 12.24,
-      "learning_rate": 1.842209072978304e-05,
-      "loss": 1.6294,
+      "epoch": 18.46,
+      "learning_rate": 6.307142857142857e-05,
+      "loss": 1.0432,
       "step": 4136
     },
     {
-      "epoch": 12.24,
-      "learning_rate": 1.8402366863905327e-05,
-      "loss": 1.5502,
+      "epoch": 18.47,
+      "learning_rate": 6.306250000000001e-05,
+      "loss": 1.1394,
       "step": 4137
     },
     {
-      "epoch": 12.24,
-      "learning_rate": 1.8382642998027613e-05,
-      "loss": 1.484,
+      "epoch": 18.47,
+      "learning_rate": 6.305357142857143e-05,
+      "loss": 1.0699,
       "step": 4138
     },
     {
-      "epoch": 12.25,
-      "learning_rate": 1.83629191321499e-05,
-      "loss": 1.4592,
+      "epoch": 18.48,
+      "learning_rate": 6.304464285714286e-05,
+      "loss": 1.1182,
       "step": 4139
     },
     {
-      "epoch": 12.25,
-      "learning_rate": 1.834319526627219e-05,
-      "loss": 1.5247,
+      "epoch": 18.48,
+      "learning_rate": 6.303571428571428e-05,
+      "loss": 1.1432,
       "step": 4140
     },
     {
-      "epoch": 12.25,
-      "learning_rate": 1.832347140039448e-05,
-      "loss": 1.5581,
+      "epoch": 18.49,
+      "learning_rate": 6.302678571428572e-05,
+      "loss": 1.2301,
       "step": 4141
     },
     {
-      "epoch": 12.25,
-      "learning_rate": 1.8303747534516764e-05,
-      "loss": 1.5239,
+      "epoch": 18.49,
+      "learning_rate": 6.301785714285715e-05,
+      "loss": 0.9947,
       "step": 4142
     },
     {
-      "epoch": 12.26,
-      "learning_rate": 1.8284023668639056e-05,
-      "loss": 1.5446,
+      "epoch": 18.5,
+      "learning_rate": 6.300892857142857e-05,
+      "loss": 1.0805,
       "step": 4143
     },
     {
-      "epoch": 12.26,
-      "learning_rate": 1.8264299802761342e-05,
-      "loss": 1.552,
+      "epoch": 18.5,
+      "learning_rate": 6.3e-05,
+      "loss": 1.156,
       "step": 4144
     },
     {
-      "epoch": 12.26,
-      "learning_rate": 1.824457593688363e-05,
-      "loss": 1.5326,
+      "epoch": 18.5,
+      "learning_rate": 6.299107142857144e-05,
+      "loss": 1.1047,
       "step": 4145
     },
     {
-      "epoch": 12.27,
-      "learning_rate": 1.822485207100592e-05,
-      "loss": 1.4967,
+      "epoch": 18.51,
+      "learning_rate": 6.298214285714285e-05,
+      "loss": 1.1204,
       "step": 4146
     },
     {
-      "epoch": 12.27,
-      "learning_rate": 1.8205128205128204e-05,
-      "loss": 1.5451,
+      "epoch": 18.51,
+      "learning_rate": 6.297321428571429e-05,
+      "loss": 1.056,
       "step": 4147
     },
     {
-      "epoch": 12.27,
-      "learning_rate": 1.8185404339250493e-05,
-      "loss": 1.5132,
+      "epoch": 18.52,
+      "learning_rate": 6.296428571428572e-05,
+      "loss": 1.1038,
       "step": 4148
     },
     {
-      "epoch": 12.28,
-      "learning_rate": 1.8165680473372782e-05,
-      "loss": 1.5085,
+      "epoch": 18.52,
+      "learning_rate": 6.295535714285714e-05,
+      "loss": 1.1616,
       "step": 4149
     },
     {
-      "epoch": 12.28,
-      "learning_rate": 1.814595660749507e-05,
-      "loss": 1.5028,
+      "epoch": 18.53,
+      "learning_rate": 6.294642857142857e-05,
+      "loss": 1.1718,
       "step": 4150
     },
     {
-      "epoch": 12.28,
-      "learning_rate": 1.812623274161736e-05,
-      "loss": 1.5593,
+      "epoch": 18.53,
+      "learning_rate": 6.29375e-05,
+      "loss": 1.1045,
       "step": 4151
     },
     {
-      "epoch": 12.28,
-      "learning_rate": 1.8106508875739645e-05,
-      "loss": 1.5202,
+      "epoch": 18.54,
+      "learning_rate": 6.292857142857143e-05,
+      "loss": 1.0761,
       "step": 4152
     },
     {
-      "epoch": 12.29,
-      "learning_rate": 1.8086785009861933e-05,
-      "loss": 1.4873,
+      "epoch": 18.54,
+      "learning_rate": 6.291964285714286e-05,
+      "loss": 1.0604,
       "step": 4153
     },
     {
-      "epoch": 12.29,
-      "learning_rate": 1.8067061143984222e-05,
-      "loss": 1.599,
+      "epoch": 18.54,
+      "learning_rate": 6.29107142857143e-05,
+      "loss": 1.0973,
       "step": 4154
     },
     {
-      "epoch": 12.29,
-      "learning_rate": 1.804733727810651e-05,
-      "loss": 1.4669,
+      "epoch": 18.55,
+      "learning_rate": 6.290178571428572e-05,
+      "loss": 1.0546,
       "step": 4155
     },
     {
-      "epoch": 12.3,
-      "learning_rate": 1.8027613412228796e-05,
-      "loss": 1.5187,
+      "epoch": 18.55,
+      "learning_rate": 6.289285714285715e-05,
+      "loss": 1.0999,
       "step": 4156
     },
     {
-      "epoch": 12.3,
-      "learning_rate": 1.8007889546351085e-05,
-      "loss": 1.4357,
+      "epoch": 18.56,
+      "learning_rate": 6.288392857142857e-05,
+      "loss": 1.1094,
       "step": 4157
     },
     {
-      "epoch": 12.3,
-      "learning_rate": 1.7988165680473374e-05,
-      "loss": 1.4705,
+      "epoch": 18.56,
+      "learning_rate": 6.287500000000001e-05,
+      "loss": 1.069,
       "step": 4158
     },
     {
-      "epoch": 12.3,
-      "learning_rate": 1.7968441814595662e-05,
-      "loss": 1.4793,
+      "epoch": 18.57,
+      "learning_rate": 6.286607142857142e-05,
+      "loss": 1.1489,
       "step": 4159
     },
     {
-      "epoch": 12.31,
-      "learning_rate": 1.794871794871795e-05,
-      "loss": 1.5264,
+      "epoch": 18.57,
+      "learning_rate": 6.285714285714286e-05,
+      "loss": 1.0555,
       "step": 4160
     },
     {
-      "epoch": 12.31,
-      "learning_rate": 1.7928994082840236e-05,
-      "loss": 1.5674,
+      "epoch": 18.58,
+      "learning_rate": 6.284821428571429e-05,
+      "loss": 1.0839,
       "step": 4161
     },
     {
-      "epoch": 12.31,
-      "learning_rate": 1.7909270216962525e-05,
-      "loss": 1.4315,
+      "epoch": 18.58,
+      "learning_rate": 6.283928571428573e-05,
+      "loss": 1.0501,
       "step": 4162
     },
     {
-      "epoch": 12.32,
-      "learning_rate": 1.7889546351084814e-05,
-      "loss": 1.5988,
+      "epoch": 18.58,
+      "learning_rate": 6.283035714285714e-05,
+      "loss": 1.0656,
       "step": 4163
     },
     {
-      "epoch": 12.32,
-      "learning_rate": 1.7869822485207103e-05,
-      "loss": 1.4935,
+      "epoch": 18.59,
+      "learning_rate": 6.282142857142858e-05,
+      "loss": 1.1196,
       "step": 4164
     },
     {
-      "epoch": 12.32,
-      "learning_rate": 1.7850098619329388e-05,
-      "loss": 1.4646,
+      "epoch": 18.59,
+      "learning_rate": 6.28125e-05,
+      "loss": 1.1133,
       "step": 4165
     },
     {
-      "epoch": 12.33,
-      "learning_rate": 1.783037475345168e-05,
-      "loss": 1.4999,
+      "epoch": 18.6,
+      "learning_rate": 6.280357142857143e-05,
+      "loss": 1.0622,
       "step": 4166
     },
     {
-      "epoch": 12.33,
-      "learning_rate": 1.7810650887573965e-05,
-      "loss": 1.5417,
+      "epoch": 18.6,
+      "learning_rate": 6.279464285714287e-05,
+      "loss": 1.0793,
       "step": 4167
     },
     {
-      "epoch": 12.33,
-      "learning_rate": 1.779092702169625e-05,
-      "loss": 1.4964,
+      "epoch": 18.61,
+      "learning_rate": 6.278571428571429e-05,
+      "loss": 1.1041,
       "step": 4168
     },
     {
-      "epoch": 12.33,
-      "learning_rate": 1.7771203155818543e-05,
-      "loss": 1.5509,
+      "epoch": 18.61,
+      "learning_rate": 6.277678571428572e-05,
+      "loss": 1.069,
       "step": 4169
     },
     {
-      "epoch": 12.34,
-      "learning_rate": 1.7751479289940828e-05,
-      "loss": 1.4779,
+      "epoch": 18.62,
+      "learning_rate": 6.276785714285714e-05,
+      "loss": 1.1123,
       "step": 4170
     },
     {
-      "epoch": 12.34,
-      "learning_rate": 1.7731755424063117e-05,
-      "loss": 1.4014,
+      "epoch": 18.62,
+      "learning_rate": 6.275892857142858e-05,
+      "loss": 1.1254,
       "step": 4171
     },
     {
-      "epoch": 12.34,
-      "learning_rate": 1.7712031558185406e-05,
-      "loss": 1.5449,
+      "epoch": 18.62,
+      "learning_rate": 6.275e-05,
+      "loss": 1.1767,
       "step": 4172
     },
     {
-      "epoch": 12.35,
-      "learning_rate": 1.7692307692307694e-05,
-      "loss": 1.5072,
+      "epoch": 18.63,
+      "learning_rate": 6.274107142857143e-05,
+      "loss": 1.0511,
       "step": 4173
     },
     {
-      "epoch": 12.35,
-      "learning_rate": 1.767258382642998e-05,
-      "loss": 1.498,
+      "epoch": 18.63,
+      "learning_rate": 6.273214285714286e-05,
+      "loss": 1.0662,
       "step": 4174
     },
     {
-      "epoch": 12.35,
-      "learning_rate": 1.7652859960552272e-05,
-      "loss": 1.6421,
+      "epoch": 18.64,
+      "learning_rate": 6.27232142857143e-05,
+      "loss": 1.0553,
       "step": 4175
     },
     {
-      "epoch": 12.36,
-      "learning_rate": 1.7633136094674557e-05,
-      "loss": 1.6284,
+      "epoch": 18.64,
+      "learning_rate": 6.271428571428571e-05,
+      "loss": 1.1044,
       "step": 4176
     },
     {
-      "epoch": 12.36,
-      "learning_rate": 1.7613412228796843e-05,
-      "loss": 1.5651,
+      "epoch": 18.65,
+      "learning_rate": 6.270535714285715e-05,
+      "loss": 1.1359,
       "step": 4177
     },
     {
-      "epoch": 12.36,
-      "learning_rate": 1.7593688362919135e-05,
-      "loss": 1.5575,
+      "epoch": 18.65,
+      "learning_rate": 6.269642857142857e-05,
+      "loss": 1.0222,
       "step": 4178
     },
     {
-      "epoch": 12.36,
-      "learning_rate": 1.757396449704142e-05,
-      "loss": 1.4848,
+      "epoch": 18.66,
+      "learning_rate": 6.26875e-05,
+      "loss": 1.0311,
       "step": 4179
     },
     {
-      "epoch": 12.37,
-      "learning_rate": 1.755424063116371e-05,
-      "loss": 1.4944,
+      "epoch": 18.66,
+      "learning_rate": 6.267857142857142e-05,
+      "loss": 1.1603,
       "step": 4180
     },
     {
-      "epoch": 12.37,
-      "learning_rate": 1.7534516765285997e-05,
-      "loss": 1.4636,
+      "epoch": 18.67,
+      "learning_rate": 6.266964285714286e-05,
+      "loss": 1.1152,
       "step": 4181
     },
     {
-      "epoch": 12.37,
-      "learning_rate": 1.7514792899408286e-05,
-      "loss": 1.5047,
+      "epoch": 18.67,
+      "learning_rate": 6.266071428571429e-05,
+      "loss": 1.0468,
       "step": 4182
     },
     {
-      "epoch": 12.38,
-      "learning_rate": 1.749506903353057e-05,
-      "loss": 1.4512,
+      "epoch": 18.67,
+      "learning_rate": 6.265178571428571e-05,
+      "loss": 1.0619,
       "step": 4183
     },
     {
-      "epoch": 12.38,
-      "learning_rate": 1.747534516765286e-05,
-      "loss": 1.596,
+      "epoch": 18.68,
+      "learning_rate": 6.264285714285715e-05,
+      "loss": 1.1214,
       "step": 4184
     },
     {
-      "epoch": 12.38,
-      "learning_rate": 1.745562130177515e-05,
-      "loss": 1.4302,
+      "epoch": 18.68,
+      "learning_rate": 6.263392857142858e-05,
+      "loss": 1.1155,
       "step": 4185
     },
     {
-      "epoch": 12.38,
-      "learning_rate": 1.7435897435897434e-05,
-      "loss": 1.4777,
+      "epoch": 18.69,
+      "learning_rate": 6.2625e-05,
+      "loss": 1.1156,
       "step": 4186
     },
     {
-      "epoch": 12.39,
-      "learning_rate": 1.7416173570019726e-05,
-      "loss": 1.4566,
+      "epoch": 18.69,
+      "learning_rate": 6.261607142857143e-05,
+      "loss": 1.0713,
       "step": 4187
     },
     {
-      "epoch": 12.39,
-      "learning_rate": 1.7396449704142012e-05,
-      "loss": 1.5828,
+      "epoch": 18.7,
+      "learning_rate": 6.260714285714287e-05,
+      "loss": 1.0888,
       "step": 4188
     },
     {
-      "epoch": 12.39,
-      "learning_rate": 1.73767258382643e-05,
-      "loss": 1.4574,
+      "epoch": 18.7,
+      "learning_rate": 6.259821428571428e-05,
+      "loss": 1.0717,
       "step": 4189
     },
     {
-      "epoch": 12.4,
-      "learning_rate": 1.735700197238659e-05,
-      "loss": 1.4208,
+      "epoch": 18.71,
+      "learning_rate": 6.258928571428572e-05,
+      "loss": 1.0394,
       "step": 4190
     },
     {
-      "epoch": 12.4,
-      "learning_rate": 1.7337278106508875e-05,
-      "loss": 1.4113,
+      "epoch": 18.71,
+      "learning_rate": 6.258035714285715e-05,
+      "loss": 1.0563,
       "step": 4191
     },
     {
-      "epoch": 12.4,
-      "learning_rate": 1.7317554240631163e-05,
-      "loss": 1.5423,
+      "epoch": 18.71,
+      "learning_rate": 6.257142857142857e-05,
+      "loss": 1.1496,
       "step": 4192
     },
     {
-      "epoch": 12.41,
-      "learning_rate": 1.7297830374753452e-05,
-      "loss": 1.4927,
+      "epoch": 18.72,
+      "learning_rate": 6.25625e-05,
+      "loss": 1.1537,
       "step": 4193
     },
     {
-      "epoch": 12.41,
-      "learning_rate": 1.727810650887574e-05,
-      "loss": 1.515,
+      "epoch": 18.72,
+      "learning_rate": 6.255357142857144e-05,
+      "loss": 1.0676,
       "step": 4194
     },
     {
-      "epoch": 12.41,
-      "learning_rate": 1.725838264299803e-05,
-      "loss": 1.5461,
+      "epoch": 18.73,
+      "learning_rate": 6.254464285714286e-05,
+      "loss": 1.0446,
       "step": 4195
     },
     {
-      "epoch": 12.41,
-      "learning_rate": 1.7238658777120318e-05,
-      "loss": 1.4103,
+      "epoch": 18.73,
+      "learning_rate": 6.253571428571429e-05,
+      "loss": 1.0166,
       "step": 4196
     },
     {
-      "epoch": 12.42,
-      "learning_rate": 1.7218934911242603e-05,
-      "loss": 1.541,
+      "epoch": 18.74,
+      "learning_rate": 6.252678571428572e-05,
+      "loss": 1.0822,
       "step": 4197
     },
     {
-      "epoch": 12.42,
-      "learning_rate": 1.7199211045364892e-05,
-      "loss": 1.5678,
+      "epoch": 18.74,
+      "learning_rate": 6.251785714285715e-05,
+      "loss": 1.0472,
       "step": 4198
     },
     {
-      "epoch": 12.42,
-      "learning_rate": 1.717948717948718e-05,
-      "loss": 1.4378,
+      "epoch": 18.75,
+      "learning_rate": 6.250892857142858e-05,
+      "loss": 1.0725,
       "step": 4199
     },
     {
-      "epoch": 12.43,
-      "learning_rate": 1.7159763313609466e-05,
-      "loss": 1.3859,
+      "epoch": 18.75,
+      "learning_rate": 6.25e-05,
+      "loss": 1.0875,
       "step": 4200
     },
     {
-      "epoch": 12.43,
-      "learning_rate": 1.7140039447731755e-05,
-      "loss": 1.5438,
+      "epoch": 18.75,
+      "eval_loss": 1.137090802192688,
+      "eval_runtime": 45.1945,
+      "eval_samples_per_second": 289.725,
+      "eval_steps_per_second": 1.527,
+      "step": 4200
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.249107142857144e-05,
+      "loss": 1.0596,
       "step": 4201
     },
     {
-      "epoch": 12.43,
-      "learning_rate": 1.7120315581854044e-05,
-      "loss": 1.6406,
+      "epoch": 18.76,
+      "learning_rate": 6.248214285714285e-05,
+      "loss": 1.0203,
       "step": 4202
     },
     {
-      "epoch": 12.43,
-      "learning_rate": 1.7100591715976332e-05,
-      "loss": 1.4286,
+      "epoch": 18.76,
+      "learning_rate": 6.247321428571429e-05,
+      "loss": 1.1037,
       "step": 4203
     },
     {
-      "epoch": 12.44,
-      "learning_rate": 1.708086785009862e-05,
-      "loss": 1.6236,
+      "epoch": 18.77,
+      "learning_rate": 6.246428571428572e-05,
+      "loss": 1.08,
       "step": 4204
     },
     {
-      "epoch": 12.44,
-      "learning_rate": 1.706114398422091e-05,
-      "loss": 1.5294,
+      "epoch": 18.77,
+      "learning_rate": 6.245535714285714e-05,
+      "loss": 1.0686,
       "step": 4205
     },
     {
-      "epoch": 12.44,
-      "learning_rate": 1.7041420118343195e-05,
-      "loss": 1.5139,
+      "epoch": 18.78,
+      "learning_rate": 6.244642857142857e-05,
+      "loss": 1.075,
       "step": 4206
     },
     {
-      "epoch": 12.45,
-      "learning_rate": 1.7021696252465484e-05,
-      "loss": 1.5072,
+      "epoch": 18.78,
+      "learning_rate": 6.24375e-05,
+      "loss": 1.0194,
       "step": 4207
     },
     {
-      "epoch": 12.45,
-      "learning_rate": 1.7001972386587773e-05,
-      "loss": 1.5861,
+      "epoch": 18.79,
+      "learning_rate": 6.242857142857143e-05,
+      "loss": 1.098,
       "step": 4208
     },
     {
-      "epoch": 12.45,
-      "learning_rate": 1.6982248520710058e-05,
-      "loss": 1.4916,
+      "epoch": 18.79,
+      "learning_rate": 6.241964285714286e-05,
+      "loss": 1.1655,
       "step": 4209
     },
     {
-      "epoch": 12.46,
-      "learning_rate": 1.696252465483235e-05,
-      "loss": 1.5846,
+      "epoch": 18.79,
+      "learning_rate": 6.241071428571428e-05,
+      "loss": 1.0665,
       "step": 4210
     },
     {
-      "epoch": 12.46,
-      "learning_rate": 1.6942800788954635e-05,
-      "loss": 1.5076,
+      "epoch": 18.8,
+      "learning_rate": 6.240178571428572e-05,
+      "loss": 1.0464,
       "step": 4211
     },
     {
-      "epoch": 12.46,
-      "learning_rate": 1.6923076923076924e-05,
-      "loss": 1.6003,
+      "epoch": 18.8,
+      "learning_rate": 6.239285714285715e-05,
+      "loss": 1.0628,
       "step": 4212
     },
     {
-      "epoch": 12.46,
-      "learning_rate": 1.6903353057199213e-05,
-      "loss": 1.4954,
+      "epoch": 18.81,
+      "learning_rate": 6.238392857142857e-05,
+      "loss": 1.0457,
       "step": 4213
     },
     {
-      "epoch": 12.47,
-      "learning_rate": 1.68836291913215e-05,
-      "loss": 1.5283,
+      "epoch": 18.81,
+      "learning_rate": 6.237500000000001e-05,
+      "loss": 1.1253,
       "step": 4214
     },
     {
-      "epoch": 12.47,
-      "learning_rate": 1.6863905325443787e-05,
-      "loss": 1.5473,
+      "epoch": 18.82,
+      "learning_rate": 6.236607142857142e-05,
+      "loss": 1.1387,
       "step": 4215
     },
     {
-      "epoch": 12.47,
-      "learning_rate": 1.6844181459566076e-05,
-      "loss": 1.4538,
+      "epoch": 18.82,
+      "learning_rate": 6.235714285714286e-05,
+      "loss": 1.0924,
       "step": 4216
     },
     {
-      "epoch": 12.48,
-      "learning_rate": 1.6824457593688364e-05,
-      "loss": 1.4968,
+      "epoch": 18.83,
+      "learning_rate": 6.234821428571429e-05,
+      "loss": 1.0582,
       "step": 4217
     },
     {
-      "epoch": 12.48,
-      "learning_rate": 1.680473372781065e-05,
-      "loss": 1.5019,
+      "epoch": 18.83,
+      "learning_rate": 6.233928571428573e-05,
+      "loss": 1.1174,
       "step": 4218
     },
     {
-      "epoch": 12.48,
-      "learning_rate": 1.6785009861932942e-05,
-      "loss": 1.5245,
+      "epoch": 18.83,
+      "learning_rate": 6.233035714285714e-05,
+      "loss": 1.0679,
       "step": 4219
     },
     {
-      "epoch": 12.49,
-      "learning_rate": 1.6765285996055227e-05,
-      "loss": 1.5367,
+      "epoch": 18.84,
+      "learning_rate": 6.232142857142858e-05,
+      "loss": 1.1442,
       "step": 4220
     },
     {
-      "epoch": 12.49,
-      "learning_rate": 1.6745562130177516e-05,
-      "loss": 1.356,
+      "epoch": 18.84,
+      "learning_rate": 6.23125e-05,
+      "loss": 1.0932,
       "step": 4221
     },
     {
-      "epoch": 12.49,
-      "learning_rate": 1.6725838264299805e-05,
-      "loss": 1.5689,
+      "epoch": 18.85,
+      "learning_rate": 6.230357142857143e-05,
+      "loss": 1.054,
       "step": 4222
     },
     {
-      "epoch": 12.49,
-      "learning_rate": 1.670611439842209e-05,
-      "loss": 1.5526,
+      "epoch": 18.85,
+      "learning_rate": 6.229464285714285e-05,
+      "loss": 1.0617,
       "step": 4223
     },
     {
-      "epoch": 12.5,
-      "learning_rate": 1.668639053254438e-05,
-      "loss": 1.5399,
+      "epoch": 18.86,
+      "learning_rate": 6.22857142857143e-05,
+      "loss": 1.1193,
       "step": 4224
     },
     {
-      "epoch": 12.5,
-      "learning_rate": 1.6666666666666667e-05,
-      "loss": 1.5434,
+      "epoch": 18.86,
+      "learning_rate": 6.227678571428572e-05,
+      "loss": 1.0671,
       "step": 4225
     },
     {
-      "epoch": 12.5,
-      "learning_rate": 1.6646942800788956e-05,
-      "loss": 1.4228,
+      "epoch": 18.87,
+      "learning_rate": 6.226785714285714e-05,
+      "loss": 1.0213,
       "step": 4226
     },
     {
-      "epoch": 12.51,
-      "learning_rate": 1.662721893491124e-05,
-      "loss": 1.5076,
+      "epoch": 18.87,
+      "learning_rate": 6.225892857142857e-05,
+      "loss": 1.0212,
       "step": 4227
     },
     {
-      "epoch": 12.51,
-      "learning_rate": 1.6607495069033534e-05,
-      "loss": 1.521,
+      "epoch": 18.88,
+      "learning_rate": 6.225000000000001e-05,
+      "loss": 1.0833,
       "step": 4228
     },
     {
-      "epoch": 12.51,
-      "learning_rate": 1.658777120315582e-05,
-      "loss": 1.4152,
+      "epoch": 18.88,
+      "learning_rate": 6.224107142857143e-05,
+      "loss": 1.1142,
       "step": 4229
     },
     {
-      "epoch": 12.51,
-      "learning_rate": 1.6568047337278108e-05,
-      "loss": 1.4537,
+      "epoch": 18.88,
+      "learning_rate": 6.223214285714286e-05,
+      "loss": 1.0606,
       "step": 4230
     },
     {
-      "epoch": 12.52,
-      "learning_rate": 1.6548323471400396e-05,
-      "loss": 1.4465,
+      "epoch": 18.89,
+      "learning_rate": 6.22232142857143e-05,
+      "loss": 1.1476,
       "step": 4231
     },
     {
-      "epoch": 12.52,
-      "learning_rate": 1.6528599605522682e-05,
-      "loss": 1.3849,
+      "epoch": 18.89,
+      "learning_rate": 6.221428571428571e-05,
+      "loss": 1.0491,
       "step": 4232
     },
     {
-      "epoch": 12.52,
-      "learning_rate": 1.650887573964497e-05,
-      "loss": 1.5079,
+      "epoch": 18.9,
+      "learning_rate": 6.220535714285715e-05,
+      "loss": 1.0802,
       "step": 4233
     },
     {
-      "epoch": 12.53,
-      "learning_rate": 1.648915187376726e-05,
-      "loss": 1.5592,
+      "epoch": 18.9,
+      "learning_rate": 6.219642857142858e-05,
+      "loss": 1.0253,
       "step": 4234
     },
     {
-      "epoch": 12.53,
-      "learning_rate": 1.6469428007889548e-05,
-      "loss": 1.5028,
+      "epoch": 18.91,
+      "learning_rate": 6.21875e-05,
+      "loss": 1.1093,
       "step": 4235
     },
     {
-      "epoch": 12.53,
-      "learning_rate": 1.6449704142011833e-05,
-      "loss": 1.5578,
+      "epoch": 18.91,
+      "learning_rate": 6.217857142857143e-05,
+      "loss": 1.047,
       "step": 4236
     },
     {
-      "epoch": 12.54,
-      "learning_rate": 1.6429980276134125e-05,
-      "loss": 1.5864,
+      "epoch": 18.92,
+      "learning_rate": 6.216964285714286e-05,
+      "loss": 1.0598,
       "step": 4237
     },
     {
-      "epoch": 12.54,
-      "learning_rate": 1.641025641025641e-05,
-      "loss": 1.5559,
+      "epoch": 18.92,
+      "learning_rate": 6.216071428571428e-05,
+      "loss": 1.0374,
       "step": 4238
     },
     {
-      "epoch": 12.54,
-      "learning_rate": 1.63905325443787e-05,
-      "loss": 1.5133,
+      "epoch": 18.92,
+      "learning_rate": 6.215178571428572e-05,
+      "loss": 1.1851,
       "step": 4239
     },
     {
-      "epoch": 12.54,
-      "learning_rate": 1.6370808678500988e-05,
-      "loss": 1.5003,
+      "epoch": 18.93,
+      "learning_rate": 6.214285714285714e-05,
+      "loss": 1.0852,
       "step": 4240
     },
     {
-      "epoch": 12.55,
-      "learning_rate": 1.6351084812623274e-05,
-      "loss": 1.5315,
+      "epoch": 18.93,
+      "learning_rate": 6.213392857142858e-05,
+      "loss": 1.0941,
       "step": 4241
     },
     {
-      "epoch": 12.55,
-      "learning_rate": 1.6331360946745562e-05,
-      "loss": 1.4813,
+      "epoch": 18.94,
+      "learning_rate": 6.2125e-05,
+      "loss": 1.0924,
       "step": 4242
     },
     {
-      "epoch": 12.55,
-      "learning_rate": 1.631163708086785e-05,
-      "loss": 1.5168,
+      "epoch": 18.94,
+      "learning_rate": 6.211607142857143e-05,
+      "loss": 1.1193,
       "step": 4243
     },
     {
-      "epoch": 12.56,
-      "learning_rate": 1.629191321499014e-05,
-      "loss": 1.5972,
+      "epoch": 18.95,
+      "learning_rate": 6.210714285714287e-05,
+      "loss": 1.0808,
       "step": 4244
     },
     {
-      "epoch": 12.56,
-      "learning_rate": 1.6272189349112425e-05,
-      "loss": 1.5016,
+      "epoch": 18.95,
+      "learning_rate": 6.209821428571428e-05,
+      "loss": 1.1589,
       "step": 4245
     },
     {
-      "epoch": 12.56,
-      "learning_rate": 1.6252465483234714e-05,
-      "loss": 1.4523,
+      "epoch": 18.96,
+      "learning_rate": 6.208928571428572e-05,
+      "loss": 1.0466,
       "step": 4246
     },
     {
-      "epoch": 12.57,
-      "learning_rate": 1.6232741617357002e-05,
-      "loss": 1.5243,
+      "epoch": 18.96,
+      "learning_rate": 6.208035714285715e-05,
+      "loss": 1.0777,
       "step": 4247
     },
     {
-      "epoch": 12.57,
-      "learning_rate": 1.621301775147929e-05,
-      "loss": 1.6179,
+      "epoch": 18.96,
+      "learning_rate": 6.207142857142857e-05,
+      "loss": 1.1607,
       "step": 4248
     },
     {
-      "epoch": 12.57,
-      "learning_rate": 1.619329388560158e-05,
-      "loss": 1.6028,
+      "epoch": 18.97,
+      "learning_rate": 6.20625e-05,
+      "loss": 1.0185,
       "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,
+      "epoch": 18.97,
+      "learning_rate": 6.205357142857144e-05,
+      "loss": 1.1709,
       "step": 4250
     },
     {
-      "epoch": 12.58,
-      "learning_rate": 1.6153846153846154e-05,
-      "loss": 1.4997,
+      "epoch": 18.98,
+      "learning_rate": 6.204464285714286e-05,
+      "loss": 1.1262,
       "step": 4251
     },
     {
-      "epoch": 12.58,
-      "learning_rate": 1.6134122287968443e-05,
-      "loss": 1.4373,
+      "epoch": 18.98,
+      "learning_rate": 6.203571428571429e-05,
+      "loss": 1.0757,
       "step": 4252
     },
     {
-      "epoch": 12.58,
-      "learning_rate": 1.611439842209073e-05,
-      "loss": 1.4793,
+      "epoch": 18.99,
+      "learning_rate": 6.202678571428571e-05,
+      "loss": 1.149,
       "step": 4253
     },
     {
-      "epoch": 12.59,
-      "learning_rate": 1.609467455621302e-05,
-      "loss": 1.4426,
+      "epoch": 18.99,
+      "learning_rate": 6.201785714285715e-05,
+      "loss": 1.0491,
       "step": 4254
     },
     {
-      "epoch": 12.59,
-      "learning_rate": 1.6074950690335306e-05,
-      "loss": 1.4371,
+      "epoch": 19.0,
+      "learning_rate": 6.200892857142858e-05,
+      "loss": 1.1193,
       "step": 4255
     },
     {
-      "epoch": 12.59,
-      "learning_rate": 1.6055226824457594e-05,
-      "loss": 1.4883,
+      "epoch": 19.0,
+      "learning_rate": 6.2e-05,
+      "loss": 1.1715,
       "step": 4256
     },
     {
-      "epoch": 12.59,
-      "learning_rate": 1.6035502958579883e-05,
-      "loss": 1.4728,
+      "epoch": 19.0,
+      "learning_rate": 6.199107142857143e-05,
+      "loss": 1.0644,
       "step": 4257
     },
     {
-      "epoch": 12.6,
-      "learning_rate": 1.601577909270217e-05,
-      "loss": 1.4832,
+      "epoch": 19.01,
+      "learning_rate": 6.198214285714285e-05,
+      "loss": 1.0923,
       "step": 4258
     },
     {
-      "epoch": 12.6,
-      "learning_rate": 1.5996055226824457e-05,
-      "loss": 1.4007,
+      "epoch": 19.01,
+      "learning_rate": 6.197321428571429e-05,
+      "loss": 1.019,
       "step": 4259
     },
     {
-      "epoch": 12.6,
-      "learning_rate": 1.5976331360946746e-05,
-      "loss": 1.4617,
+      "epoch": 19.02,
+      "learning_rate": 6.196428571428572e-05,
+      "loss": 1.0737,
       "step": 4260
     },
     {
-      "epoch": 12.61,
-      "learning_rate": 1.5956607495069034e-05,
-      "loss": 1.4971,
+      "epoch": 19.02,
+      "learning_rate": 6.195535714285716e-05,
+      "loss": 1.0443,
       "step": 4261
     },
     {
-      "epoch": 12.61,
-      "learning_rate": 1.593688362919132e-05,
-      "loss": 1.4827,
+      "epoch": 19.03,
+      "learning_rate": 6.194642857142857e-05,
+      "loss": 1.0379,
       "step": 4262
     },
     {
-      "epoch": 12.61,
-      "learning_rate": 1.5917159763313612e-05,
-      "loss": 1.5495,
+      "epoch": 19.03,
+      "learning_rate": 6.193750000000001e-05,
+      "loss": 1.0952,
       "step": 4263
     },
     {
-      "epoch": 12.62,
-      "learning_rate": 1.5897435897435897e-05,
-      "loss": 1.5454,
+      "epoch": 19.04,
+      "learning_rate": 6.192857142857143e-05,
+      "loss": 1.0259,
       "step": 4264
     },
     {
-      "epoch": 12.62,
-      "learning_rate": 1.5877712031558186e-05,
-      "loss": 1.5385,
+      "epoch": 19.04,
+      "learning_rate": 6.191964285714286e-05,
+      "loss": 1.0544,
       "step": 4265
     },
     {
-      "epoch": 12.62,
-      "learning_rate": 1.5857988165680475e-05,
-      "loss": 1.4699,
+      "epoch": 19.04,
+      "learning_rate": 6.191071428571428e-05,
+      "loss": 1.0718,
       "step": 4266
     },
     {
-      "epoch": 12.62,
-      "learning_rate": 1.5838264299802763e-05,
-      "loss": 1.4406,
+      "epoch": 19.05,
+      "learning_rate": 6.190178571428572e-05,
+      "loss": 1.0589,
       "step": 4267
     },
     {
-      "epoch": 12.63,
-      "learning_rate": 1.581854043392505e-05,
-      "loss": 1.5118,
+      "epoch": 19.05,
+      "learning_rate": 6.189285714285714e-05,
+      "loss": 1.085,
       "step": 4268
     },
     {
-      "epoch": 12.63,
-      "learning_rate": 1.579881656804734e-05,
-      "loss": 1.5187,
+      "epoch": 19.06,
+      "learning_rate": 6.188392857142857e-05,
+      "loss": 1.0377,
       "step": 4269
     },
     {
-      "epoch": 12.63,
-      "learning_rate": 1.5779092702169626e-05,
-      "loss": 1.503,
+      "epoch": 19.06,
+      "learning_rate": 6.1875e-05,
+      "loss": 1.0791,
       "step": 4270
     },
     {
-      "epoch": 12.64,
-      "learning_rate": 1.575936883629191e-05,
-      "loss": 1.5519,
+      "epoch": 19.07,
+      "learning_rate": 6.186607142857143e-05,
+      "loss": 1.1403,
       "step": 4271
     },
     {
-      "epoch": 12.64,
-      "learning_rate": 1.5739644970414204e-05,
-      "loss": 1.4622,
+      "epoch": 19.07,
+      "learning_rate": 6.185714285714286e-05,
+      "loss": 1.0928,
       "step": 4272
     },
     {
-      "epoch": 12.64,
-      "learning_rate": 1.571992110453649e-05,
-      "loss": 1.4861,
+      "epoch": 19.08,
+      "learning_rate": 6.184821428571429e-05,
+      "loss": 1.0938,
       "step": 4273
     },
     {
-      "epoch": 12.64,
-      "learning_rate": 1.5700197238658778e-05,
-      "loss": 1.5439,
+      "epoch": 19.08,
+      "learning_rate": 6.183928571428573e-05,
+      "loss": 1.1105,
       "step": 4274
     },
     {
-      "epoch": 12.65,
-      "learning_rate": 1.5680473372781066e-05,
-      "loss": 1.5291,
+      "epoch": 19.08,
+      "learning_rate": 6.183035714285714e-05,
+      "loss": 1.0584,
       "step": 4275
     },
     {
-      "epoch": 12.65,
-      "learning_rate": 1.5660749506903355e-05,
-      "loss": 1.5214,
+      "epoch": 19.09,
+      "learning_rate": 6.182142857142858e-05,
+      "loss": 1.0636,
       "step": 4276
     },
     {
-      "epoch": 12.65,
-      "learning_rate": 1.564102564102564e-05,
-      "loss": 1.4889,
+      "epoch": 19.09,
+      "learning_rate": 6.18125e-05,
+      "loss": 1.0904,
       "step": 4277
     },
     {
-      "epoch": 12.66,
-      "learning_rate": 1.562130177514793e-05,
-      "loss": 1.4854,
+      "epoch": 19.1,
+      "learning_rate": 6.180357142857143e-05,
+      "loss": 1.0474,
       "step": 4278
     },
     {
-      "epoch": 12.66,
-      "learning_rate": 1.5601577909270218e-05,
-      "loss": 1.5007,
+      "epoch": 19.1,
+      "learning_rate": 6.179464285714286e-05,
+      "loss": 1.0972,
       "step": 4279
     },
     {
-      "epoch": 12.66,
-      "learning_rate": 1.5581854043392503e-05,
-      "loss": 1.3762,
+      "epoch": 19.11,
+      "learning_rate": 6.17857142857143e-05,
+      "loss": 1.1222,
       "step": 4280
     },
     {
-      "epoch": 12.67,
-      "learning_rate": 1.5562130177514795e-05,
-      "loss": 1.4904,
+      "epoch": 19.11,
+      "learning_rate": 6.17767857142857e-05,
+      "loss": 1.0634,
       "step": 4281
     },
     {
-      "epoch": 12.67,
-      "learning_rate": 1.554240631163708e-05,
-      "loss": 1.5093,
+      "epoch": 19.12,
+      "learning_rate": 6.176785714285715e-05,
+      "loss": 1.0963,
       "step": 4282
     },
     {
-      "epoch": 12.67,
-      "learning_rate": 1.552268244575937e-05,
-      "loss": 1.478,
+      "epoch": 19.12,
+      "learning_rate": 6.175892857142857e-05,
+      "loss": 1.1116,
       "step": 4283
     },
     {
-      "epoch": 12.67,
-      "learning_rate": 1.5502958579881658e-05,
-      "loss": 1.4785,
+      "epoch": 19.12,
+      "learning_rate": 6.175000000000001e-05,
+      "loss": 1.1061,
       "step": 4284
     },
     {
-      "epoch": 12.68,
-      "learning_rate": 1.5483234714003947e-05,
-      "loss": 1.5555,
+      "epoch": 19.13,
+      "learning_rate": 6.174107142857142e-05,
+      "loss": 1.0291,
       "step": 4285
     },
     {
-      "epoch": 12.68,
-      "learning_rate": 1.5463510848126232e-05,
-      "loss": 1.5286,
+      "epoch": 19.13,
+      "learning_rate": 6.173214285714286e-05,
+      "loss": 1.0824,
       "step": 4286
     },
     {
-      "epoch": 12.68,
-      "learning_rate": 1.544378698224852e-05,
-      "loss": 1.4648,
+      "epoch": 19.14,
+      "learning_rate": 6.172321428571429e-05,
+      "loss": 1.0175,
       "step": 4287
     },
     {
-      "epoch": 12.69,
-      "learning_rate": 1.542406311637081e-05,
-      "loss": 1.4908,
+      "epoch": 19.14,
+      "learning_rate": 6.171428571428571e-05,
+      "loss": 1.0729,
       "step": 4288
     },
     {
-      "epoch": 12.69,
-      "learning_rate": 1.5404339250493095e-05,
-      "loss": 1.4917,
+      "epoch": 19.15,
+      "learning_rate": 6.170535714285715e-05,
+      "loss": 1.0608,
       "step": 4289
     },
     {
-      "epoch": 12.69,
-      "learning_rate": 1.5384615384615387e-05,
-      "loss": 1.525,
+      "epoch": 19.15,
+      "learning_rate": 6.169642857142858e-05,
+      "loss": 1.0978,
       "step": 4290
     },
     {
-      "epoch": 12.7,
-      "learning_rate": 1.5364891518737673e-05,
-      "loss": 1.4672,
+      "epoch": 19.16,
+      "learning_rate": 6.16875e-05,
+      "loss": 0.984,
       "step": 4291
     },
     {
-      "epoch": 12.7,
-      "learning_rate": 1.534516765285996e-05,
-      "loss": 1.4288,
+      "epoch": 19.16,
+      "learning_rate": 6.167857142857143e-05,
+      "loss": 1.1101,
       "step": 4292
     },
     {
-      "epoch": 12.7,
-      "learning_rate": 1.532544378698225e-05,
-      "loss": 1.5937,
+      "epoch": 19.17,
+      "learning_rate": 6.166964285714287e-05,
+      "loss": 1.1314,
       "step": 4293
     },
     {
-      "epoch": 12.7,
-      "learning_rate": 1.5305719921104535e-05,
-      "loss": 1.5354,
+      "epoch": 19.17,
+      "learning_rate": 6.166071428571429e-05,
+      "loss": 1.0854,
       "step": 4294
     },
     {
-      "epoch": 12.71,
-      "learning_rate": 1.5285996055226824e-05,
-      "loss": 1.5589,
+      "epoch": 19.17,
+      "learning_rate": 6.165178571428572e-05,
+      "loss": 1.035,
       "step": 4295
     },
     {
-      "epoch": 12.71,
-      "learning_rate": 1.5266272189349113e-05,
-      "loss": 1.5225,
+      "epoch": 19.18,
+      "learning_rate": 6.164285714285714e-05,
+      "loss": 1.1343,
       "step": 4296
     },
     {
-      "epoch": 12.71,
-      "learning_rate": 1.5246548323471401e-05,
-      "loss": 1.4931,
+      "epoch": 19.18,
+      "learning_rate": 6.163392857142858e-05,
+      "loss": 1.0571,
       "step": 4297
     },
     {
-      "epoch": 12.72,
-      "learning_rate": 1.5226824457593689e-05,
-      "loss": 1.5184,
+      "epoch": 19.19,
+      "learning_rate": 6.1625e-05,
+      "loss": 1.108,
       "step": 4298
     },
     {
-      "epoch": 12.72,
-      "learning_rate": 1.5207100591715979e-05,
-      "loss": 1.4811,
+      "epoch": 19.19,
+      "learning_rate": 6.161607142857143e-05,
+      "loss": 1.1107,
       "step": 4299
     },
     {
-      "epoch": 12.72,
-      "learning_rate": 1.5187376725838264e-05,
-      "loss": 1.3628,
+      "epoch": 19.2,
+      "learning_rate": 6.160714285714286e-05,
+      "loss": 1.0448,
       "step": 4300
     },
     {
-      "epoch": 12.72,
-      "learning_rate": 1.5167652859960551e-05,
-      "loss": 1.4928,
+      "epoch": 19.2,
+      "learning_rate": 6.159821428571428e-05,
+      "loss": 1.0472,
       "step": 4301
     },
     {
-      "epoch": 12.73,
-      "learning_rate": 1.5147928994082842e-05,
-      "loss": 1.5553,
+      "epoch": 19.21,
+      "learning_rate": 6.158928571428572e-05,
+      "loss": 1.1758,
       "step": 4302
     },
     {
-      "epoch": 12.73,
-      "learning_rate": 1.5128205128205129e-05,
-      "loss": 1.5047,
+      "epoch": 19.21,
+      "learning_rate": 6.158035714285715e-05,
+      "loss": 1.1364,
       "step": 4303
     },
     {
-      "epoch": 12.73,
-      "learning_rate": 1.5108481262327417e-05,
-      "loss": 1.5047,
+      "epoch": 19.21,
+      "learning_rate": 6.157142857142857e-05,
+      "loss": 1.0875,
       "step": 4304
     },
     {
-      "epoch": 12.74,
-      "learning_rate": 1.5088757396449705e-05,
-      "loss": 1.5257,
+      "epoch": 19.22,
+      "learning_rate": 6.15625e-05,
+      "loss": 1.1133,
       "step": 4305
     },
     {
-      "epoch": 12.74,
-      "learning_rate": 1.5069033530571993e-05,
-      "loss": 1.4726,
+      "epoch": 19.22,
+      "learning_rate": 6.155357142857144e-05,
+      "loss": 1.1154,
       "step": 4306
     },
     {
-      "epoch": 12.74,
-      "learning_rate": 1.504930966469428e-05,
-      "loss": 1.6115,
+      "epoch": 19.23,
+      "learning_rate": 6.154464285714286e-05,
+      "loss": 1.0467,
       "step": 4307
     },
     {
-      "epoch": 12.75,
-      "learning_rate": 1.502958579881657e-05,
-      "loss": 1.4472,
+      "epoch": 19.23,
+      "learning_rate": 6.153571428571429e-05,
+      "loss": 1.0763,
       "step": 4308
     },
     {
-      "epoch": 12.75,
-      "learning_rate": 1.5009861932938856e-05,
-      "loss": 1.5375,
+      "epoch": 19.24,
+      "learning_rate": 6.152678571428571e-05,
+      "loss": 1.0594,
       "step": 4309
     },
     {
-      "epoch": 12.75,
-      "learning_rate": 1.4990138067061143e-05,
-      "loss": 1.3987,
+      "epoch": 19.24,
+      "learning_rate": 6.151785714285715e-05,
+      "loss": 1.0427,
       "step": 4310
     },
     {
-      "epoch": 12.75,
-      "learning_rate": 1.4970414201183433e-05,
-      "loss": 1.4239,
+      "epoch": 19.25,
+      "learning_rate": 6.150892857142857e-05,
+      "loss": 1.0553,
       "step": 4311
     },
     {
-      "epoch": 12.76,
-      "learning_rate": 1.495069033530572e-05,
-      "loss": 1.5228,
+      "epoch": 19.25,
+      "learning_rate": 6.15e-05,
+      "loss": 1.0762,
       "step": 4312
     },
     {
-      "epoch": 12.76,
-      "learning_rate": 1.493096646942801e-05,
-      "loss": 1.5467,
+      "epoch": 19.25,
+      "learning_rate": 6.149107142857143e-05,
+      "loss": 1.0409,
       "step": 4313
     },
     {
-      "epoch": 12.76,
-      "learning_rate": 1.4911242603550296e-05,
-      "loss": 1.5023,
+      "epoch": 19.26,
+      "learning_rate": 6.148214285714285e-05,
+      "loss": 1.0663,
       "step": 4314
     },
     {
-      "epoch": 12.77,
-      "learning_rate": 1.4891518737672585e-05,
-      "loss": 1.4661,
+      "epoch": 19.26,
+      "learning_rate": 6.147321428571428e-05,
+      "loss": 1.0258,
       "step": 4315
     },
     {
-      "epoch": 12.77,
-      "learning_rate": 1.4871794871794872e-05,
-      "loss": 1.5034,
+      "epoch": 19.27,
+      "learning_rate": 6.146428571428572e-05,
+      "loss": 1.0319,
       "step": 4316
     },
     {
-      "epoch": 12.77,
-      "learning_rate": 1.4852071005917159e-05,
-      "loss": 1.4954,
+      "epoch": 19.27,
+      "learning_rate": 6.145535714285714e-05,
+      "loss": 1.1162,
       "step": 4317
     },
     {
-      "epoch": 12.78,
-      "learning_rate": 1.483234714003945e-05,
-      "loss": 1.498,
+      "epoch": 19.28,
+      "learning_rate": 6.144642857142857e-05,
+      "loss": 1.0633,
       "step": 4318
     },
     {
-      "epoch": 12.78,
-      "learning_rate": 1.4812623274161735e-05,
-      "loss": 1.4703,
+      "epoch": 19.28,
+      "learning_rate": 6.143750000000001e-05,
+      "loss": 1.1523,
       "step": 4319
     },
     {
-      "epoch": 12.78,
-      "learning_rate": 1.4792899408284025e-05,
-      "loss": 1.5557,
+      "epoch": 19.29,
+      "learning_rate": 6.142857142857143e-05,
+      "loss": 1.02,
       "step": 4320
     },
     {
-      "epoch": 12.78,
-      "learning_rate": 1.4773175542406312e-05,
-      "loss": 1.4528,
+      "epoch": 19.29,
+      "learning_rate": 6.141964285714286e-05,
+      "loss": 0.973,
       "step": 4321
     },
     {
-      "epoch": 12.79,
-      "learning_rate": 1.4753451676528601e-05,
-      "loss": 1.4365,
+      "epoch": 19.29,
+      "learning_rate": 6.141071428571429e-05,
+      "loss": 0.9741,
       "step": 4322
     },
     {
-      "epoch": 12.79,
-      "learning_rate": 1.4733727810650888e-05,
-      "loss": 1.429,
+      "epoch": 19.3,
+      "learning_rate": 6.140178571428572e-05,
+      "loss": 1.1001,
       "step": 4323
     },
     {
-      "epoch": 12.79,
-      "learning_rate": 1.4714003944773177e-05,
-      "loss": 1.4518,
+      "epoch": 19.3,
+      "learning_rate": 6.139285714285714e-05,
+      "loss": 1.0815,
       "step": 4324
     },
     {
-      "epoch": 12.8,
-      "learning_rate": 1.4694280078895464e-05,
-      "loss": 1.4352,
+      "epoch": 19.31,
+      "learning_rate": 6.138392857142858e-05,
+      "loss": 1.0752,
       "step": 4325
     },
     {
-      "epoch": 12.8,
-      "learning_rate": 1.467455621301775e-05,
-      "loss": 1.5836,
+      "epoch": 19.31,
+      "learning_rate": 6.1375e-05,
+      "loss": 1.0757,
       "step": 4326
     },
     {
-      "epoch": 12.8,
-      "learning_rate": 1.4654832347140041e-05,
-      "loss": 1.5365,
+      "epoch": 19.32,
+      "learning_rate": 6.136607142857144e-05,
+      "loss": 1.0256,
       "step": 4327
     },
     {
-      "epoch": 12.8,
-      "learning_rate": 1.4635108481262327e-05,
-      "loss": 1.5222,
+      "epoch": 19.32,
+      "learning_rate": 6.135714285714285e-05,
+      "loss": 1.0258,
       "step": 4328
     },
     {
-      "epoch": 12.81,
-      "learning_rate": 1.4615384615384617e-05,
-      "loss": 1.4345,
+      "epoch": 19.33,
+      "learning_rate": 6.134821428571429e-05,
+      "loss": 1.0843,
       "step": 4329
     },
     {
-      "epoch": 12.81,
-      "learning_rate": 1.4595660749506904e-05,
-      "loss": 1.5458,
+      "epoch": 19.33,
+      "learning_rate": 6.133928571428572e-05,
+      "loss": 1.1042,
       "step": 4330
     },
     {
-      "epoch": 12.81,
-      "learning_rate": 1.4575936883629193e-05,
-      "loss": 1.4795,
+      "epoch": 19.33,
+      "learning_rate": 6.133035714285714e-05,
+      "loss": 1.0193,
       "step": 4331
     },
     {
-      "epoch": 12.82,
-      "learning_rate": 1.455621301775148e-05,
-      "loss": 1.4989,
+      "epoch": 19.34,
+      "learning_rate": 6.132142857142858e-05,
+      "loss": 1.0743,
       "step": 4332
     },
     {
-      "epoch": 12.82,
-      "learning_rate": 1.4536489151873767e-05,
-      "loss": 1.4408,
+      "epoch": 19.34,
+      "learning_rate": 6.13125e-05,
+      "loss": 1.0819,
       "step": 4333
     },
     {
-      "epoch": 12.82,
-      "learning_rate": 1.4516765285996056e-05,
-      "loss": 1.4602,
+      "epoch": 19.35,
+      "learning_rate": 6.130357142857143e-05,
+      "loss": 1.0718,
       "step": 4334
     },
     {
-      "epoch": 12.83,
-      "learning_rate": 1.4497041420118343e-05,
-      "loss": 1.5116,
+      "epoch": 19.35,
+      "learning_rate": 6.129464285714286e-05,
+      "loss": 1.1875,
       "step": 4335
     },
     {
-      "epoch": 12.83,
-      "learning_rate": 1.4477317554240633e-05,
-      "loss": 1.4874,
+      "epoch": 19.36,
+      "learning_rate": 6.12857142857143e-05,
+      "loss": 1.105,
       "step": 4336
     },
     {
-      "epoch": 12.83,
-      "learning_rate": 1.445759368836292e-05,
-      "loss": 1.5332,
+      "epoch": 19.36,
+      "learning_rate": 6.127678571428572e-05,
+      "loss": 1.045,
       "step": 4337
     },
     {
-      "epoch": 12.83,
-      "learning_rate": 1.4437869822485209e-05,
-      "loss": 1.5278,
+      "epoch": 19.37,
+      "learning_rate": 6.126785714285715e-05,
+      "loss": 1.0943,
       "step": 4338
     },
     {
-      "epoch": 12.84,
-      "learning_rate": 1.4418145956607496e-05,
-      "loss": 1.4024,
+      "epoch": 19.37,
+      "learning_rate": 6.125892857142857e-05,
+      "loss": 1.0252,
       "step": 4339
     },
     {
-      "epoch": 12.84,
-      "learning_rate": 1.4398422090729784e-05,
-      "loss": 1.4589,
+      "epoch": 19.38,
+      "learning_rate": 6.125000000000001e-05,
+      "loss": 1.1317,
       "step": 4340
     },
     {
-      "epoch": 12.84,
-      "learning_rate": 1.4378698224852072e-05,
-      "loss": 1.5949,
+      "epoch": 19.38,
+      "learning_rate": 6.124107142857142e-05,
+      "loss": 1.043,
       "step": 4341
     },
     {
-      "epoch": 12.85,
-      "learning_rate": 1.4358974358974359e-05,
-      "loss": 1.3883,
+      "epoch": 19.38,
+      "learning_rate": 6.123214285714286e-05,
+      "loss": 1.1117,
       "step": 4342
     },
     {
-      "epoch": 12.85,
-      "learning_rate": 1.4339250493096649e-05,
-      "loss": 1.6363,
+      "epoch": 19.39,
+      "learning_rate": 6.122321428571429e-05,
+      "loss": 1.0643,
       "step": 4343
     },
     {
-      "epoch": 12.85,
-      "learning_rate": 1.4319526627218934e-05,
-      "loss": 1.5038,
+      "epoch": 19.39,
+      "learning_rate": 6.121428571428571e-05,
+      "loss": 1.1182,
       "step": 4344
     },
     {
-      "epoch": 12.86,
-      "learning_rate": 1.4299802761341225e-05,
-      "loss": 1.5245,
+      "epoch": 19.4,
+      "learning_rate": 6.120535714285714e-05,
+      "loss": 1.1041,
       "step": 4345
     },
     {
-      "epoch": 12.86,
-      "learning_rate": 1.4280078895463512e-05,
-      "loss": 1.5139,
+      "epoch": 19.4,
+      "learning_rate": 6.119642857142858e-05,
+      "loss": 1.0227,
       "step": 4346
     },
     {
-      "epoch": 12.86,
-      "learning_rate": 1.42603550295858e-05,
-      "loss": 1.4926,
+      "epoch": 19.41,
+      "learning_rate": 6.11875e-05,
+      "loss": 1.069,
       "step": 4347
     },
     {
-      "epoch": 12.86,
-      "learning_rate": 1.4240631163708088e-05,
-      "loss": 1.4609,
+      "epoch": 19.41,
+      "learning_rate": 6.117857142857143e-05,
+      "loss": 1.1119,
       "step": 4348
     },
     {
-      "epoch": 12.87,
-      "learning_rate": 1.4220907297830375e-05,
-      "loss": 1.5655,
+      "epoch": 19.42,
+      "learning_rate": 6.116964285714287e-05,
+      "loss": 1.0902,
       "step": 4349
     },
     {
-      "epoch": 12.87,
-      "learning_rate": 1.4201183431952663e-05,
-      "loss": 1.5513,
+      "epoch": 19.42,
+      "learning_rate": 6.116071428571429e-05,
+      "loss": 1.1168,
       "step": 4350
     },
     {
-      "epoch": 12.87,
-      "learning_rate": 1.418145956607495e-05,
-      "loss": 1.537,
+      "epoch": 19.42,
+      "learning_rate": 6.115178571428572e-05,
+      "loss": 1.1161,
       "step": 4351
     },
     {
-      "epoch": 12.88,
-      "learning_rate": 1.416173570019724e-05,
-      "loss": 1.5518,
+      "epoch": 19.43,
+      "learning_rate": 6.114285714285714e-05,
+      "loss": 1.0906,
       "step": 4352
     },
     {
-      "epoch": 12.88,
-      "learning_rate": 1.4142011834319526e-05,
-      "loss": 1.5066,
+      "epoch": 19.43,
+      "learning_rate": 6.113392857142858e-05,
+      "loss": 1.1508,
       "step": 4353
     },
     {
-      "epoch": 12.88,
-      "learning_rate": 1.4122287968441816e-05,
-      "loss": 1.554,
+      "epoch": 19.44,
+      "learning_rate": 6.1125e-05,
+      "loss": 1.0511,
       "step": 4354
     },
     {
-      "epoch": 12.88,
-      "learning_rate": 1.4102564102564104e-05,
-      "loss": 1.5868,
+      "epoch": 19.44,
+      "learning_rate": 6.111607142857143e-05,
+      "loss": 1.0166,
       "step": 4355
     },
     {
-      "epoch": 12.89,
-      "learning_rate": 1.408284023668639e-05,
-      "loss": 1.3856,
+      "epoch": 19.45,
+      "learning_rate": 6.110714285714286e-05,
+      "loss": 1.0857,
       "step": 4356
     },
     {
-      "epoch": 12.89,
-      "learning_rate": 1.406311637080868e-05,
-      "loss": 1.5059,
+      "epoch": 19.45,
+      "learning_rate": 6.109821428571428e-05,
+      "loss": 1.03,
       "step": 4357
     },
     {
-      "epoch": 12.89,
-      "learning_rate": 1.4043392504930966e-05,
-      "loss": 1.5634,
+      "epoch": 19.46,
+      "learning_rate": 6.108928571428571e-05,
+      "loss": 1.068,
       "step": 4358
     },
     {
-      "epoch": 12.9,
-      "learning_rate": 1.4023668639053255e-05,
-      "loss": 1.6097,
+      "epoch": 19.46,
+      "learning_rate": 6.108035714285715e-05,
+      "loss": 1.1845,
       "step": 4359
     },
     {
-      "epoch": 12.9,
-      "learning_rate": 1.4003944773175542e-05,
-      "loss": 1.4543,
+      "epoch": 19.46,
+      "learning_rate": 6.107142857142857e-05,
+      "loss": 1.0728,
       "step": 4360
     },
     {
-      "epoch": 12.9,
-      "learning_rate": 1.3984220907297832e-05,
-      "loss": 1.482,
+      "epoch": 19.47,
+      "learning_rate": 6.10625e-05,
+      "loss": 1.0908,
       "step": 4361
     },
     {
-      "epoch": 12.91,
-      "learning_rate": 1.396449704142012e-05,
-      "loss": 1.4936,
+      "epoch": 19.47,
+      "learning_rate": 6.105357142857143e-05,
+      "loss": 1.1497,
       "step": 4362
     },
     {
-      "epoch": 12.91,
-      "learning_rate": 1.3944773175542408e-05,
-      "loss": 1.658,
+      "epoch": 19.48,
+      "learning_rate": 6.104464285714286e-05,
+      "loss": 1.1612,
       "step": 4363
     },
     {
-      "epoch": 12.91,
-      "learning_rate": 1.3925049309664695e-05,
-      "loss": 1.4461,
+      "epoch": 19.48,
+      "learning_rate": 6.103571428571429e-05,
+      "loss": 1.0578,
       "step": 4364
     },
     {
-      "epoch": 12.91,
-      "learning_rate": 1.3905325443786982e-05,
-      "loss": 1.4767,
+      "epoch": 19.49,
+      "learning_rate": 6.1026785714285716e-05,
+      "loss": 1.0557,
       "step": 4365
     },
     {
-      "epoch": 12.92,
-      "learning_rate": 1.3885601577909271e-05,
-      "loss": 1.5614,
+      "epoch": 19.49,
+      "learning_rate": 6.101785714285715e-05,
+      "loss": 1.0723,
       "step": 4366
     },
     {
-      "epoch": 12.92,
-      "learning_rate": 1.3865877712031558e-05,
-      "loss": 1.4945,
+      "epoch": 19.5,
+      "learning_rate": 6.1008928571428566e-05,
+      "loss": 1.0533,
       "step": 4367
     },
     {
-      "epoch": 12.92,
-      "learning_rate": 1.3846153846153847e-05,
-      "loss": 1.6189,
+      "epoch": 19.5,
+      "learning_rate": 6.1e-05,
+      "loss": 1.0365,
       "step": 4368
     },
     {
-      "epoch": 12.93,
-      "learning_rate": 1.3826429980276134e-05,
-      "loss": 1.5819,
+      "epoch": 19.5,
+      "learning_rate": 6.099107142857143e-05,
+      "loss": 1.0246,
       "step": 4369
     },
     {
-      "epoch": 12.93,
-      "learning_rate": 1.3806706114398424e-05,
-      "loss": 1.5447,
+      "epoch": 19.51,
+      "learning_rate": 6.098214285714286e-05,
+      "loss": 1.064,
       "step": 4370
     },
     {
-      "epoch": 12.93,
-      "learning_rate": 1.3786982248520711e-05,
-      "loss": 1.4642,
+      "epoch": 19.51,
+      "learning_rate": 6.097321428571429e-05,
+      "loss": 1.089,
       "step": 4371
     },
     {
-      "epoch": 12.93,
-      "learning_rate": 1.3767258382642997e-05,
-      "loss": 1.483,
+      "epoch": 19.52,
+      "learning_rate": 6.096428571428572e-05,
+      "loss": 1.0191,
       "step": 4372
     },
     {
-      "epoch": 12.94,
-      "learning_rate": 1.3747534516765287e-05,
-      "loss": 1.5282,
+      "epoch": 19.52,
+      "learning_rate": 6.095535714285715e-05,
+      "loss": 1.0505,
       "step": 4373
     },
     {
-      "epoch": 12.94,
-      "learning_rate": 1.3727810650887574e-05,
-      "loss": 1.5253,
+      "epoch": 19.53,
+      "learning_rate": 6.094642857142857e-05,
+      "loss": 1.0634,
       "step": 4374
     },
     {
-      "epoch": 12.94,
-      "learning_rate": 1.3708086785009863e-05,
-      "loss": 1.4755,
+      "epoch": 19.53,
+      "learning_rate": 6.0937500000000004e-05,
+      "loss": 1.0817,
       "step": 4375
     },
     {
-      "epoch": 12.95,
-      "learning_rate": 1.368836291913215e-05,
-      "loss": 1.4912,
+      "epoch": 19.54,
+      "learning_rate": 6.0928571428571436e-05,
+      "loss": 1.1372,
       "step": 4376
     },
     {
-      "epoch": 12.95,
-      "learning_rate": 1.366863905325444e-05,
-      "loss": 1.5332,
+      "epoch": 19.54,
+      "learning_rate": 6.0919642857142855e-05,
+      "loss": 1.1187,
       "step": 4377
     },
     {
-      "epoch": 12.95,
-      "learning_rate": 1.3648915187376726e-05,
-      "loss": 1.5044,
+      "epoch": 19.54,
+      "learning_rate": 6.091071428571429e-05,
+      "loss": 1.0415,
       "step": 4378
     },
     {
-      "epoch": 12.96,
-      "learning_rate": 1.3629191321499016e-05,
-      "loss": 1.4001,
+      "epoch": 19.55,
+      "learning_rate": 6.090178571428572e-05,
+      "loss": 1.1091,
       "step": 4379
     },
     {
-      "epoch": 12.96,
-      "learning_rate": 1.3609467455621303e-05,
-      "loss": 1.4888,
+      "epoch": 19.55,
+      "learning_rate": 6.089285714285714e-05,
+      "loss": 1.1101,
       "step": 4380
     },
     {
-      "epoch": 12.96,
-      "learning_rate": 1.358974358974359e-05,
-      "loss": 1.6381,
+      "epoch": 19.56,
+      "learning_rate": 6.088392857142857e-05,
+      "loss": 1.056,
       "step": 4381
     },
     {
-      "epoch": 12.96,
-      "learning_rate": 1.3570019723865879e-05,
-      "loss": 1.5089,
+      "epoch": 19.56,
+      "learning_rate": 6.0875e-05,
+      "loss": 1.0744,
       "step": 4382
     },
     {
-      "epoch": 12.97,
-      "learning_rate": 1.3550295857988166e-05,
-      "loss": 1.4796,
+      "epoch": 19.57,
+      "learning_rate": 6.0866071428571435e-05,
+      "loss": 1.0197,
       "step": 4383
     },
     {
-      "epoch": 12.97,
-      "learning_rate": 1.3530571992110455e-05,
-      "loss": 1.5374,
+      "epoch": 19.57,
+      "learning_rate": 6.085714285714286e-05,
+      "loss": 1.1714,
       "step": 4384
     },
     {
-      "epoch": 12.97,
-      "learning_rate": 1.3510848126232742e-05,
-      "loss": 1.4941,
+      "epoch": 19.58,
+      "learning_rate": 6.0848214285714286e-05,
+      "loss": 1.0502,
       "step": 4385
     },
     {
-      "epoch": 12.98,
-      "learning_rate": 1.3491124260355032e-05,
-      "loss": 1.6142,
+      "epoch": 19.58,
+      "learning_rate": 6.083928571428572e-05,
+      "loss": 1.1102,
       "step": 4386
     },
     {
-      "epoch": 12.98,
-      "learning_rate": 1.3471400394477319e-05,
-      "loss": 1.4876,
+      "epoch": 19.58,
+      "learning_rate": 6.083035714285714e-05,
+      "loss": 0.9402,
       "step": 4387
     },
     {
-      "epoch": 12.98,
-      "learning_rate": 1.3451676528599604e-05,
-      "loss": 1.5567,
+      "epoch": 19.59,
+      "learning_rate": 6.0821428571428575e-05,
+      "loss": 1.0452,
       "step": 4388
     },
     {
-      "epoch": 12.99,
-      "learning_rate": 1.3431952662721895e-05,
-      "loss": 1.5013,
+      "epoch": 19.59,
+      "learning_rate": 6.081250000000001e-05,
+      "loss": 1.1297,
       "step": 4389
     },
     {
-      "epoch": 12.99,
-      "learning_rate": 1.3412228796844182e-05,
-      "loss": 1.5129,
+      "epoch": 19.6,
+      "learning_rate": 6.0803571428571426e-05,
+      "loss": 1.1092,
       "step": 4390
     },
     {
-      "epoch": 12.99,
-      "learning_rate": 1.339250493096647e-05,
-      "loss": 1.4824,
+      "epoch": 19.6,
+      "learning_rate": 6.079464285714286e-05,
+      "loss": 1.0158,
       "step": 4391
     },
     {
-      "epoch": 12.99,
-      "learning_rate": 1.3372781065088758e-05,
-      "loss": 1.4687,
+      "epoch": 19.61,
+      "learning_rate": 6.078571428571429e-05,
+      "loss": 1.0328,
       "step": 4392
     },
     {
-      "epoch": 13.0,
-      "learning_rate": 1.3353057199211046e-05,
-      "loss": 1.5359,
+      "epoch": 19.61,
+      "learning_rate": 6.077678571428572e-05,
+      "loss": 1.1179,
       "step": 4393
     },
     {
-      "epoch": 13.0,
-      "learning_rate": 1.3333333333333333e-05,
-      "loss": 1.4585,
+      "epoch": 19.62,
+      "learning_rate": 6.076785714285714e-05,
+      "loss": 1.1504,
       "step": 4394
     },
     {
-      "epoch": 13.0,
-      "learning_rate": 1.3313609467455624e-05,
-      "loss": 1.408,
+      "epoch": 19.62,
+      "learning_rate": 6.0758928571428574e-05,
+      "loss": 1.1349,
       "step": 4395
     },
     {
-      "epoch": 13.01,
-      "learning_rate": 1.329388560157791e-05,
-      "loss": 1.5439,
+      "epoch": 19.62,
+      "learning_rate": 6.0750000000000006e-05,
+      "loss": 1.0631,
       "step": 4396
     },
     {
-      "epoch": 13.01,
-      "learning_rate": 1.3274161735700196e-05,
-      "loss": 1.5025,
+      "epoch": 19.63,
+      "learning_rate": 6.0741071428571425e-05,
+      "loss": 1.0958,
       "step": 4397
     },
     {
-      "epoch": 13.01,
-      "learning_rate": 1.3254437869822487e-05,
-      "loss": 1.4338,
+      "epoch": 19.63,
+      "learning_rate": 6.073214285714286e-05,
+      "loss": 1.1092,
       "step": 4398
     },
     {
-      "epoch": 13.01,
-      "learning_rate": 1.3234714003944774e-05,
-      "loss": 1.5286,
+      "epoch": 19.64,
+      "learning_rate": 6.072321428571429e-05,
+      "loss": 1.1014,
       "step": 4399
     },
     {
-      "epoch": 13.02,
-      "learning_rate": 1.3214990138067062e-05,
-      "loss": 1.4862,
+      "epoch": 19.64,
+      "learning_rate": 6.0714285714285715e-05,
+      "loss": 1.0617,
       "step": 4400
     },
     {
-      "epoch": 13.02,
-      "learning_rate": 1.319526627218935e-05,
-      "loss": 1.5425,
+      "epoch": 19.65,
+      "learning_rate": 6.070535714285715e-05,
+      "loss": 1.131,
       "step": 4401
     },
     {
-      "epoch": 13.02,
-      "learning_rate": 1.317554240631164e-05,
-      "loss": 1.482,
+      "epoch": 19.65,
+      "learning_rate": 6.069642857142858e-05,
+      "loss": 1.0582,
       "step": 4402
     },
     {
-      "epoch": 13.03,
-      "learning_rate": 1.3155818540433925e-05,
-      "loss": 1.5337,
+      "epoch": 19.66,
+      "learning_rate": 6.068750000000001e-05,
+      "loss": 1.0269,
       "step": 4403
     },
     {
-      "epoch": 13.03,
-      "learning_rate": 1.3136094674556212e-05,
-      "loss": 1.682,
+      "epoch": 19.66,
+      "learning_rate": 6.067857142857143e-05,
+      "loss": 1.0038,
       "step": 4404
     },
     {
-      "epoch": 13.03,
-      "learning_rate": 1.3116370808678502e-05,
-      "loss": 1.4914,
+      "epoch": 19.67,
+      "learning_rate": 6.066964285714286e-05,
+      "loss": 1.1206,
       "step": 4405
     },
     {
-      "epoch": 13.04,
-      "learning_rate": 1.309664694280079e-05,
-      "loss": 1.4931,
+      "epoch": 19.67,
+      "learning_rate": 6.0660714285714295e-05,
+      "loss": 1.0429,
       "step": 4406
     },
     {
-      "epoch": 13.04,
-      "learning_rate": 1.3076923076923078e-05,
-      "loss": 1.4782,
+      "epoch": 19.67,
+      "learning_rate": 6.065178571428571e-05,
+      "loss": 1.0639,
       "step": 4407
     },
     {
-      "epoch": 13.04,
-      "learning_rate": 1.3057199211045365e-05,
-      "loss": 1.5732,
+      "epoch": 19.68,
+      "learning_rate": 6.0642857142857145e-05,
+      "loss": 1.1319,
       "step": 4408
     },
     {
-      "epoch": 13.04,
-      "learning_rate": 1.3037475345167654e-05,
-      "loss": 1.5009,
+      "epoch": 19.68,
+      "learning_rate": 6.063392857142858e-05,
+      "loss": 1.0726,
       "step": 4409
     },
     {
-      "epoch": 13.05,
-      "learning_rate": 1.3017751479289941e-05,
-      "loss": 1.5962,
+      "epoch": 19.69,
+      "learning_rate": 6.0624999999999996e-05,
+      "loss": 1.0671,
       "step": 4410
     },
     {
-      "epoch": 13.05,
-      "learning_rate": 1.2998027613412228e-05,
-      "loss": 1.5356,
+      "epoch": 19.69,
+      "learning_rate": 6.061607142857143e-05,
+      "loss": 1.1012,
       "step": 4411
     },
     {
-      "epoch": 13.05,
-      "learning_rate": 1.2978303747534517e-05,
-      "loss": 1.5665,
+      "epoch": 19.7,
+      "learning_rate": 6.060714285714286e-05,
+      "loss": 1.0727,
       "step": 4412
     },
     {
-      "epoch": 13.06,
-      "learning_rate": 1.2958579881656804e-05,
-      "loss": 1.4855,
+      "epoch": 19.7,
+      "learning_rate": 6.0598214285714286e-05,
+      "loss": 1.088,
       "step": 4413
     },
     {
-      "epoch": 13.06,
-      "learning_rate": 1.2938856015779094e-05,
-      "loss": 1.61,
+      "epoch": 19.71,
+      "learning_rate": 6.058928571428571e-05,
+      "loss": 1.1952,
       "step": 4414
     },
     {
-      "epoch": 13.06,
-      "learning_rate": 1.2919132149901381e-05,
-      "loss": 1.5796,
+      "epoch": 19.71,
+      "learning_rate": 6.0580357142857144e-05,
+      "loss": 1.0734,
       "step": 4415
     },
     {
-      "epoch": 13.07,
-      "learning_rate": 1.289940828402367e-05,
-      "loss": 1.5221,
+      "epoch": 19.71,
+      "learning_rate": 6.0571428571428576e-05,
+      "loss": 1.0414,
       "step": 4416
     },
     {
-      "epoch": 13.07,
-      "learning_rate": 1.2879684418145957e-05,
-      "loss": 1.4913,
+      "epoch": 19.72,
+      "learning_rate": 6.05625e-05,
+      "loss": 1.0918,
       "step": 4417
     },
     {
-      "epoch": 13.07,
-      "learning_rate": 1.2859960552268246e-05,
-      "loss": 1.5283,
+      "epoch": 19.72,
+      "learning_rate": 6.0553571428571434e-05,
+      "loss": 1.056,
       "step": 4418
     },
     {
-      "epoch": 13.07,
-      "learning_rate": 1.2840236686390533e-05,
-      "loss": 1.5149,
+      "epoch": 19.73,
+      "learning_rate": 6.0544642857142866e-05,
+      "loss": 1.0726,
       "step": 4419
     },
     {
-      "epoch": 13.08,
-      "learning_rate": 1.282051282051282e-05,
-      "loss": 1.6281,
+      "epoch": 19.73,
+      "learning_rate": 6.0535714285714285e-05,
+      "loss": 1.0427,
       "step": 4420
     },
     {
-      "epoch": 13.08,
-      "learning_rate": 1.280078895463511e-05,
-      "loss": 1.5614,
+      "epoch": 19.74,
+      "learning_rate": 6.052678571428572e-05,
+      "loss": 1.1177,
       "step": 4421
     },
     {
-      "epoch": 13.08,
-      "learning_rate": 1.2781065088757396e-05,
-      "loss": 1.4296,
+      "epoch": 19.74,
+      "learning_rate": 6.051785714285715e-05,
+      "loss": 1.0793,
       "step": 4422
     },
     {
-      "epoch": 13.09,
-      "learning_rate": 1.2761341222879686e-05,
-      "loss": 1.4523,
+      "epoch": 19.75,
+      "learning_rate": 6.050892857142857e-05,
+      "loss": 1.0205,
       "step": 4423
     },
     {
-      "epoch": 13.09,
-      "learning_rate": 1.2741617357001973e-05,
-      "loss": 1.4713,
+      "epoch": 19.75,
+      "learning_rate": 6.05e-05,
+      "loss": 1.1019,
       "step": 4424
     },
     {
-      "epoch": 13.09,
-      "learning_rate": 1.2721893491124262e-05,
-      "loss": 1.3836,
+      "epoch": 19.75,
+      "learning_rate": 6.049107142857143e-05,
+      "loss": 1.0417,
       "step": 4425
     },
     {
-      "epoch": 13.09,
-      "learning_rate": 1.2702169625246549e-05,
-      "loss": 1.5866,
+      "epoch": 19.76,
+      "learning_rate": 6.0482142857142865e-05,
+      "loss": 1.0663,
       "step": 4426
     },
     {
-      "epoch": 13.1,
-      "learning_rate": 1.2682445759368836e-05,
-      "loss": 1.5002,
+      "epoch": 19.76,
+      "learning_rate": 6.047321428571428e-05,
+      "loss": 1.1216,
       "step": 4427
     },
     {
-      "epoch": 13.1,
-      "learning_rate": 1.2662721893491125e-05,
-      "loss": 1.5102,
+      "epoch": 19.77,
+      "learning_rate": 6.0464285714285715e-05,
+      "loss": 1.0414,
       "step": 4428
     },
     {
-      "epoch": 13.1,
-      "learning_rate": 1.2642998027613412e-05,
-      "loss": 1.5146,
+      "epoch": 19.77,
+      "learning_rate": 6.045535714285715e-05,
+      "loss": 1.0705,
       "step": 4429
     },
     {
-      "epoch": 13.11,
-      "learning_rate": 1.2623274161735702e-05,
-      "loss": 1.5274,
+      "epoch": 19.78,
+      "learning_rate": 6.044642857142857e-05,
+      "loss": 1.0607,
       "step": 4430
     },
     {
-      "epoch": 13.11,
-      "learning_rate": 1.2603550295857989e-05,
-      "loss": 1.5167,
+      "epoch": 19.78,
+      "learning_rate": 6.0437500000000005e-05,
+      "loss": 1.0529,
       "step": 4431
     },
     {
-      "epoch": 13.11,
-      "learning_rate": 1.2583826429980278e-05,
-      "loss": 1.588,
+      "epoch": 19.79,
+      "learning_rate": 6.042857142857144e-05,
+      "loss": 1.1125,
       "step": 4432
     },
     {
-      "epoch": 13.12,
-      "learning_rate": 1.2564102564102565e-05,
-      "loss": 1.4729,
+      "epoch": 19.79,
+      "learning_rate": 6.0419642857142856e-05,
+      "loss": 1.0194,
       "step": 4433
     },
     {
-      "epoch": 13.12,
-      "learning_rate": 1.2544378698224854e-05,
-      "loss": 1.585,
+      "epoch": 19.79,
+      "learning_rate": 6.041071428571429e-05,
+      "loss": 1.0768,
       "step": 4434
     },
     {
-      "epoch": 13.12,
-      "learning_rate": 1.252465483234714e-05,
-      "loss": 1.4174,
+      "epoch": 19.8,
+      "learning_rate": 6.040178571428572e-05,
+      "loss": 1.0209,
       "step": 4435
     },
     {
-      "epoch": 13.12,
-      "learning_rate": 1.2504930966469428e-05,
-      "loss": 1.5354,
+      "epoch": 19.8,
+      "learning_rate": 6.039285714285715e-05,
+      "loss": 1.0548,
       "step": 4436
     },
     {
-      "epoch": 13.13,
-      "learning_rate": 1.2485207100591716e-05,
-      "loss": 1.4194,
+      "epoch": 19.81,
+      "learning_rate": 6.038392857142857e-05,
+      "loss": 1.0796,
       "step": 4437
     },
     {
-      "epoch": 13.13,
-      "learning_rate": 1.2465483234714005e-05,
-      "loss": 1.468,
+      "epoch": 19.81,
+      "learning_rate": 6.0375000000000004e-05,
+      "loss": 1.1481,
       "step": 4438
     },
     {
-      "epoch": 13.13,
-      "learning_rate": 1.2445759368836292e-05,
-      "loss": 1.5739,
+      "epoch": 19.82,
+      "learning_rate": 6.0366071428571436e-05,
+      "loss": 1.0749,
       "step": 4439
     },
     {
-      "epoch": 13.14,
-      "learning_rate": 1.242603550295858e-05,
-      "loss": 1.5346,
+      "epoch": 19.82,
+      "learning_rate": 6.0357142857142855e-05,
+      "loss": 1.0682,
       "step": 4440
     },
     {
-      "epoch": 13.14,
-      "learning_rate": 1.2406311637080868e-05,
-      "loss": 1.5406,
+      "epoch": 19.83,
+      "learning_rate": 6.034821428571429e-05,
+      "loss": 1.0757,
       "step": 4441
     },
     {
-      "epoch": 13.14,
-      "learning_rate": 1.2386587771203157e-05,
-      "loss": 1.3586,
+      "epoch": 19.83,
+      "learning_rate": 6.033928571428572e-05,
+      "loss": 1.0749,
       "step": 4442
     },
     {
-      "epoch": 13.14,
-      "learning_rate": 1.2366863905325445e-05,
-      "loss": 1.5062,
+      "epoch": 19.83,
+      "learning_rate": 6.033035714285714e-05,
+      "loss": 1.0767,
       "step": 4443
     },
     {
-      "epoch": 13.15,
-      "learning_rate": 1.2347140039447732e-05,
-      "loss": 1.5051,
+      "epoch": 19.84,
+      "learning_rate": 6.032142857142857e-05,
+      "loss": 1.0962,
       "step": 4444
     },
     {
-      "epoch": 13.15,
-      "learning_rate": 1.2327416173570021e-05,
-      "loss": 1.5155,
+      "epoch": 19.84,
+      "learning_rate": 6.03125e-05,
+      "loss": 1.0438,
       "step": 4445
     },
     {
-      "epoch": 13.15,
-      "learning_rate": 1.230769230769231e-05,
-      "loss": 1.4375,
+      "epoch": 19.85,
+      "learning_rate": 6.030357142857143e-05,
+      "loss": 1.0475,
       "step": 4446
     },
     {
-      "epoch": 13.16,
-      "learning_rate": 1.2287968441814595e-05,
-      "loss": 1.4258,
+      "epoch": 19.85,
+      "learning_rate": 6.029464285714286e-05,
+      "loss": 1.0488,
       "step": 4447
     },
     {
-      "epoch": 13.16,
-      "learning_rate": 1.2268244575936884e-05,
-      "loss": 1.5,
+      "epoch": 19.86,
+      "learning_rate": 6.028571428571429e-05,
+      "loss": 1.0666,
       "step": 4448
     },
     {
-      "epoch": 13.16,
-      "learning_rate": 1.2248520710059173e-05,
-      "loss": 1.5402,
+      "epoch": 19.86,
+      "learning_rate": 6.0276785714285724e-05,
+      "loss": 1.0253,
       "step": 4449
     },
     {
-      "epoch": 13.17,
-      "learning_rate": 1.222879684418146e-05,
-      "loss": 1.5751,
+      "epoch": 19.87,
+      "learning_rate": 6.026785714285714e-05,
+      "loss": 1.1376,
       "step": 4450
     },
     {
-      "epoch": 13.17,
-      "learning_rate": 1.2209072978303748e-05,
-      "loss": 1.4664,
+      "epoch": 19.87,
+      "learning_rate": 6.0258928571428575e-05,
+      "loss": 1.067,
       "step": 4451
     },
     {
-      "epoch": 13.17,
-      "learning_rate": 1.2189349112426037e-05,
-      "loss": 1.5469,
+      "epoch": 19.88,
+      "learning_rate": 6.025000000000001e-05,
+      "loss": 1.0699,
       "step": 4452
     },
     {
-      "epoch": 13.17,
-      "learning_rate": 1.2169625246548324e-05,
-      "loss": 1.5547,
+      "epoch": 19.88,
+      "learning_rate": 6.0241071428571426e-05,
+      "loss": 1.1847,
       "step": 4453
     },
     {
-      "epoch": 13.18,
-      "learning_rate": 1.2149901380670611e-05,
-      "loss": 1.5317,
+      "epoch": 19.88,
+      "learning_rate": 6.023214285714286e-05,
+      "loss": 1.0679,
       "step": 4454
     },
     {
-      "epoch": 13.18,
-      "learning_rate": 1.21301775147929e-05,
-      "loss": 1.5794,
+      "epoch": 19.89,
+      "learning_rate": 6.022321428571429e-05,
+      "loss": 1.1411,
       "step": 4455
     },
     {
-      "epoch": 13.18,
-      "learning_rate": 1.2110453648915187e-05,
-      "loss": 1.4884,
+      "epoch": 19.89,
+      "learning_rate": 6.021428571428571e-05,
+      "loss": 1.1019,
       "step": 4456
     },
     {
-      "epoch": 13.19,
-      "learning_rate": 1.2090729783037476e-05,
-      "loss": 1.51,
+      "epoch": 19.9,
+      "learning_rate": 6.020535714285714e-05,
+      "loss": 1.106,
       "step": 4457
     },
     {
-      "epoch": 13.19,
-      "learning_rate": 1.2071005917159764e-05,
-      "loss": 1.4951,
+      "epoch": 19.9,
+      "learning_rate": 6.0196428571428574e-05,
+      "loss": 1.1493,
       "step": 4458
     },
     {
-      "epoch": 13.19,
-      "learning_rate": 1.2051282051282051e-05,
-      "loss": 1.4528,
+      "epoch": 19.91,
+      "learning_rate": 6.0187500000000006e-05,
+      "loss": 1.1474,
       "step": 4459
     },
     {
-      "epoch": 13.2,
-      "learning_rate": 1.203155818540434e-05,
-      "loss": 1.4795,
+      "epoch": 19.91,
+      "learning_rate": 6.017857142857143e-05,
+      "loss": 1.0763,
       "step": 4460
     },
     {
-      "epoch": 13.2,
-      "learning_rate": 1.2011834319526629e-05,
-      "loss": 1.4589,
+      "epoch": 19.92,
+      "learning_rate": 6.0169642857142864e-05,
+      "loss": 1.1131,
       "step": 4461
     },
     {
-      "epoch": 13.2,
-      "learning_rate": 1.1992110453648916e-05,
-      "loss": 1.5372,
+      "epoch": 19.92,
+      "learning_rate": 6.016071428571429e-05,
+      "loss": 1.0791,
       "step": 4462
     },
     {
-      "epoch": 13.2,
-      "learning_rate": 1.1972386587771203e-05,
-      "loss": 1.402,
+      "epoch": 19.92,
+      "learning_rate": 6.0151785714285715e-05,
+      "loss": 1.0868,
       "step": 4463
     },
     {
-      "epoch": 13.21,
-      "learning_rate": 1.1952662721893492e-05,
-      "loss": 1.5279,
+      "epoch": 19.93,
+      "learning_rate": 6.014285714285715e-05,
+      "loss": 1.0445,
       "step": 4464
     },
     {
-      "epoch": 13.21,
-      "learning_rate": 1.193293885601578e-05,
-      "loss": 1.5442,
+      "epoch": 19.93,
+      "learning_rate": 6.013392857142858e-05,
+      "loss": 1.0724,
       "step": 4465
     },
     {
-      "epoch": 13.21,
-      "learning_rate": 1.1913214990138067e-05,
-      "loss": 1.5387,
+      "epoch": 19.94,
+      "learning_rate": 6.0125e-05,
+      "loss": 1.0919,
       "step": 4466
     },
     {
-      "epoch": 13.22,
-      "learning_rate": 1.1893491124260356e-05,
-      "loss": 1.5206,
+      "epoch": 19.94,
+      "learning_rate": 6.011607142857143e-05,
+      "loss": 1.1493,
       "step": 4467
     },
     {
-      "epoch": 13.22,
-      "learning_rate": 1.1873767258382645e-05,
-      "loss": 1.5165,
+      "epoch": 19.95,
+      "learning_rate": 6.010714285714286e-05,
+      "loss": 1.0947,
       "step": 4468
     },
     {
-      "epoch": 13.22,
-      "learning_rate": 1.1854043392504932e-05,
-      "loss": 1.5633,
+      "epoch": 19.95,
+      "learning_rate": 6.0098214285714294e-05,
+      "loss": 1.1212,
       "step": 4469
     },
     {
-      "epoch": 13.22,
-      "learning_rate": 1.1834319526627219e-05,
-      "loss": 1.4873,
+      "epoch": 19.96,
+      "learning_rate": 6.008928571428571e-05,
+      "loss": 1.1228,
       "step": 4470
     },
     {
-      "epoch": 13.23,
-      "learning_rate": 1.1814595660749508e-05,
-      "loss": 1.4568,
+      "epoch": 19.96,
+      "learning_rate": 6.0080357142857145e-05,
+      "loss": 1.059,
       "step": 4471
     },
     {
-      "epoch": 13.23,
-      "learning_rate": 1.1794871794871795e-05,
-      "loss": 1.5036,
+      "epoch": 19.96,
+      "learning_rate": 6.007142857142858e-05,
+      "loss": 1.0468,
       "step": 4472
     },
     {
-      "epoch": 13.23,
-      "learning_rate": 1.1775147928994083e-05,
-      "loss": 1.5199,
+      "epoch": 19.97,
+      "learning_rate": 6.0062499999999996e-05,
+      "loss": 1.0861,
       "step": 4473
     },
     {
-      "epoch": 13.24,
-      "learning_rate": 1.1755424063116372e-05,
-      "loss": 1.4326,
+      "epoch": 19.97,
+      "learning_rate": 6.005357142857143e-05,
+      "loss": 1.0808,
       "step": 4474
     },
     {
-      "epoch": 13.24,
-      "learning_rate": 1.1735700197238659e-05,
-      "loss": 1.5925,
+      "epoch": 19.98,
+      "learning_rate": 6.004464285714286e-05,
+      "loss": 1.0623,
       "step": 4475
     },
     {
-      "epoch": 13.24,
-      "learning_rate": 1.1715976331360948e-05,
-      "loss": 1.4351,
+      "epoch": 19.98,
+      "learning_rate": 6.0035714285714286e-05,
+      "loss": 1.1046,
       "step": 4476
     },
     {
-      "epoch": 13.25,
-      "learning_rate": 1.1696252465483237e-05,
-      "loss": 1.572,
+      "epoch": 19.99,
+      "learning_rate": 6.002678571428572e-05,
+      "loss": 1.0987,
       "step": 4477
     },
     {
-      "epoch": 13.25,
-      "learning_rate": 1.1676528599605522e-05,
-      "loss": 1.4266,
+      "epoch": 19.99,
+      "learning_rate": 6.001785714285715e-05,
+      "loss": 1.0607,
       "step": 4478
     },
     {
-      "epoch": 13.25,
-      "learning_rate": 1.165680473372781e-05,
-      "loss": 1.5375,
+      "epoch": 20.0,
+      "learning_rate": 6.000892857142858e-05,
+      "loss": 1.1205,
       "step": 4479
     },
     {
-      "epoch": 13.25,
-      "learning_rate": 1.16370808678501e-05,
-      "loss": 1.3816,
+      "epoch": 20.0,
+      "learning_rate": 6e-05,
+      "loss": 0.9208,
       "step": 4480
     },
     {
-      "epoch": 13.26,
-      "learning_rate": 1.1617357001972386e-05,
-      "loss": 1.5579,
+      "epoch": 20.0,
+      "learning_rate": 5.9991071428571434e-05,
+      "loss": 1.0962,
       "step": 4481
     },
     {
-      "epoch": 13.26,
-      "learning_rate": 1.1597633136094675e-05,
-      "loss": 1.4604,
+      "epoch": 20.01,
+      "learning_rate": 5.9982142857142866e-05,
+      "loss": 1.0946,
       "step": 4482
     },
     {
-      "epoch": 13.26,
-      "learning_rate": 1.1577909270216964e-05,
-      "loss": 1.4528,
+      "epoch": 20.01,
+      "learning_rate": 5.9973214285714285e-05,
+      "loss": 1.0334,
       "step": 4483
     },
     {
-      "epoch": 13.27,
-      "learning_rate": 1.155818540433925e-05,
-      "loss": 1.4225,
+      "epoch": 20.02,
+      "learning_rate": 5.996428571428572e-05,
+      "loss": 1.0869,
       "step": 4484
     },
     {
-      "epoch": 13.27,
-      "learning_rate": 1.153846153846154e-05,
-      "loss": 1.4957,
+      "epoch": 20.02,
+      "learning_rate": 5.995535714285715e-05,
+      "loss": 1.0249,
       "step": 4485
     },
     {
-      "epoch": 13.27,
-      "learning_rate": 1.1518737672583827e-05,
-      "loss": 1.3856,
+      "epoch": 20.03,
+      "learning_rate": 5.994642857142857e-05,
+      "loss": 1.0623,
       "step": 4486
     },
     {
-      "epoch": 13.28,
-      "learning_rate": 1.1499013806706115e-05,
-      "loss": 1.5849,
+      "epoch": 20.03,
+      "learning_rate": 5.99375e-05,
+      "loss": 1.0549,
       "step": 4487
     },
     {
-      "epoch": 13.28,
-      "learning_rate": 1.1479289940828402e-05,
-      "loss": 1.5095,
+      "epoch": 20.04,
+      "learning_rate": 5.992857142857143e-05,
+      "loss": 1.0886,
       "step": 4488
     },
     {
-      "epoch": 13.28,
-      "learning_rate": 1.1459566074950691e-05,
-      "loss": 1.6027,
+      "epoch": 20.04,
+      "learning_rate": 5.991964285714286e-05,
+      "loss": 1.047,
       "step": 4489
     },
     {
-      "epoch": 13.28,
-      "learning_rate": 1.143984220907298e-05,
-      "loss": 1.5778,
+      "epoch": 20.04,
+      "learning_rate": 5.991071428571429e-05,
+      "loss": 1.0234,
       "step": 4490
     },
     {
-      "epoch": 13.29,
-      "learning_rate": 1.1420118343195267e-05,
-      "loss": 1.5374,
+      "epoch": 20.05,
+      "learning_rate": 5.9901785714285715e-05,
+      "loss": 1.0412,
       "step": 4491
     },
     {
-      "epoch": 13.29,
-      "learning_rate": 1.1400394477317556e-05,
-      "loss": 1.5647,
+      "epoch": 20.05,
+      "learning_rate": 5.989285714285715e-05,
+      "loss": 1.059,
       "step": 4492
     },
     {
-      "epoch": 13.29,
-      "learning_rate": 1.1380670611439844e-05,
-      "loss": 1.5798,
+      "epoch": 20.06,
+      "learning_rate": 5.988392857142857e-05,
+      "loss": 1.1057,
       "step": 4493
     },
     {
-      "epoch": 13.3,
-      "learning_rate": 1.136094674556213e-05,
-      "loss": 1.5163,
+      "epoch": 20.06,
+      "learning_rate": 5.9875000000000005e-05,
+      "loss": 1.0794,
       "step": 4494
     },
     {
-      "epoch": 13.3,
-      "learning_rate": 1.1341222879684418e-05,
-      "loss": 1.4523,
+      "epoch": 20.07,
+      "learning_rate": 5.986607142857144e-05,
+      "loss": 1.0585,
       "step": 4495
     },
     {
-      "epoch": 13.3,
-      "learning_rate": 1.1321499013806707e-05,
-      "loss": 1.5288,
+      "epoch": 20.07,
+      "learning_rate": 5.9857142857142856e-05,
+      "loss": 1.1472,
       "step": 4496
     },
     {
-      "epoch": 13.3,
-      "learning_rate": 1.1301775147928994e-05,
-      "loss": 1.4928,
+      "epoch": 20.08,
+      "learning_rate": 5.984821428571429e-05,
+      "loss": 1.0528,
       "step": 4497
     },
     {
-      "epoch": 13.31,
-      "learning_rate": 1.1282051282051283e-05,
-      "loss": 1.4881,
+      "epoch": 20.08,
+      "learning_rate": 5.983928571428572e-05,
+      "loss": 1.0105,
       "step": 4498
     },
     {
-      "epoch": 13.31,
-      "learning_rate": 1.1262327416173572e-05,
-      "loss": 1.6119,
+      "epoch": 20.08,
+      "learning_rate": 5.983035714285714e-05,
+      "loss": 1.0794,
       "step": 4499
     },
     {
-      "epoch": 13.31,
-      "learning_rate": 1.1242603550295859e-05,
-      "loss": 1.4719,
+      "epoch": 20.09,
+      "learning_rate": 5.982142857142857e-05,
+      "loss": 1.1051,
       "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,
+      "epoch": 20.09,
+      "eval_loss": 1.1328850984573364,
+      "eval_runtime": 45.0698,
+      "eval_samples_per_second": 290.527,
+      "eval_steps_per_second": 1.531,
       "step": 4500
     },
     {
-      "epoch": 13.32,
-      "learning_rate": 1.1222879684418146e-05,
-      "loss": 1.5319,
+      "epoch": 20.09,
+      "learning_rate": 5.9812500000000004e-05,
+      "loss": 1.0655,
       "step": 4501
     },
     {
-      "epoch": 13.32,
-      "learning_rate": 1.1203155818540434e-05,
-      "loss": 1.5758,
+      "epoch": 20.1,
+      "learning_rate": 5.9803571428571436e-05,
+      "loss": 1.0207,
       "step": 4502
     },
     {
-      "epoch": 13.32,
-      "learning_rate": 1.1183431952662721e-05,
-      "loss": 1.5193,
+      "epoch": 20.1,
+      "learning_rate": 5.9794642857142855e-05,
+      "loss": 1.0629,
       "step": 4503
     },
     {
-      "epoch": 13.33,
-      "learning_rate": 1.116370808678501e-05,
-      "loss": 1.3976,
+      "epoch": 20.11,
+      "learning_rate": 5.978571428571429e-05,
+      "loss": 1.0622,
       "step": 4504
     },
     {
-      "epoch": 13.33,
-      "learning_rate": 1.1143984220907299e-05,
-      "loss": 1.5623,
+      "epoch": 20.11,
+      "learning_rate": 5.977678571428572e-05,
+      "loss": 1.0118,
       "step": 4505
     },
     {
-      "epoch": 13.33,
-      "learning_rate": 1.1124260355029586e-05,
-      "loss": 1.4782,
+      "epoch": 20.12,
+      "learning_rate": 5.9767857142857145e-05,
+      "loss": 1.0341,
       "step": 4506
     },
     {
-      "epoch": 13.33,
-      "learning_rate": 1.1104536489151875e-05,
-      "loss": 1.5643,
+      "epoch": 20.12,
+      "learning_rate": 5.975892857142858e-05,
+      "loss": 1.0189,
       "step": 4507
     },
     {
-      "epoch": 13.34,
-      "learning_rate": 1.1084812623274163e-05,
-      "loss": 1.5236,
+      "epoch": 20.12,
+      "learning_rate": 5.975000000000001e-05,
+      "loss": 0.9821,
       "step": 4508
     },
     {
-      "epoch": 13.34,
-      "learning_rate": 1.106508875739645e-05,
-      "loss": 1.4397,
+      "epoch": 20.13,
+      "learning_rate": 5.974107142857143e-05,
+      "loss": 1.0742,
       "step": 4509
     },
     {
-      "epoch": 13.34,
-      "learning_rate": 1.1045364891518737e-05,
-      "loss": 1.5783,
+      "epoch": 20.13,
+      "learning_rate": 5.973214285714286e-05,
+      "loss": 1.0357,
       "step": 4510
     },
     {
-      "epoch": 13.35,
-      "learning_rate": 1.1025641025641026e-05,
-      "loss": 1.5263,
+      "epoch": 20.14,
+      "learning_rate": 5.972321428571429e-05,
+      "loss": 1.0434,
       "step": 4511
     },
     {
-      "epoch": 13.35,
-      "learning_rate": 1.1005917159763315e-05,
-      "loss": 1.4982,
+      "epoch": 20.14,
+      "learning_rate": 5.9714285714285724e-05,
+      "loss": 1.0505,
       "step": 4512
     },
     {
-      "epoch": 13.35,
-      "learning_rate": 1.0986193293885602e-05,
-      "loss": 1.5102,
+      "epoch": 20.15,
+      "learning_rate": 5.970535714285714e-05,
+      "loss": 1.0495,
       "step": 4513
     },
     {
-      "epoch": 13.36,
-      "learning_rate": 1.096646942800789e-05,
-      "loss": 1.5041,
+      "epoch": 20.15,
+      "learning_rate": 5.9696428571428575e-05,
+      "loss": 1.053,
       "step": 4514
     },
     {
-      "epoch": 13.36,
-      "learning_rate": 1.094674556213018e-05,
-      "loss": 1.5178,
+      "epoch": 20.16,
+      "learning_rate": 5.968750000000001e-05,
+      "loss": 1.065,
       "step": 4515
     },
     {
-      "epoch": 13.36,
-      "learning_rate": 1.0927021696252466e-05,
-      "loss": 1.5584,
+      "epoch": 20.16,
+      "learning_rate": 5.9678571428571426e-05,
+      "loss": 1.099,
       "step": 4516
     },
     {
-      "epoch": 13.36,
-      "learning_rate": 1.0907297830374753e-05,
-      "loss": 1.5301,
+      "epoch": 20.17,
+      "learning_rate": 5.966964285714286e-05,
+      "loss": 1.095,
       "step": 4517
     },
     {
-      "epoch": 13.37,
-      "learning_rate": 1.0887573964497042e-05,
-      "loss": 1.4868,
+      "epoch": 20.17,
+      "learning_rate": 5.966071428571429e-05,
+      "loss": 1.0557,
       "step": 4518
     },
     {
-      "epoch": 13.37,
-      "learning_rate": 1.0867850098619329e-05,
-      "loss": 1.4577,
+      "epoch": 20.17,
+      "learning_rate": 5.9651785714285716e-05,
+      "loss": 1.0054,
       "step": 4519
     },
     {
-      "epoch": 13.37,
-      "learning_rate": 1.0848126232741618e-05,
-      "loss": 1.5632,
+      "epoch": 20.18,
+      "learning_rate": 5.964285714285714e-05,
+      "loss": 1.0599,
       "step": 4520
     },
     {
-      "epoch": 13.38,
-      "learning_rate": 1.0828402366863907e-05,
-      "loss": 1.4598,
+      "epoch": 20.18,
+      "learning_rate": 5.9633928571428574e-05,
+      "loss": 1.083,
       "step": 4521
     },
     {
-      "epoch": 13.38,
-      "learning_rate": 1.0808678500986194e-05,
-      "loss": 1.5108,
+      "epoch": 20.19,
+      "learning_rate": 5.9625e-05,
+      "loss": 1.0249,
       "step": 4522
     },
     {
-      "epoch": 13.38,
-      "learning_rate": 1.0788954635108482e-05,
-      "loss": 1.5648,
+      "epoch": 20.19,
+      "learning_rate": 5.961607142857143e-05,
+      "loss": 1.0089,
       "step": 4523
     },
     {
-      "epoch": 13.38,
-      "learning_rate": 1.0769230769230771e-05,
-      "loss": 1.3821,
+      "epoch": 20.2,
+      "learning_rate": 5.9607142857142864e-05,
+      "loss": 1.0971,
       "step": 4524
     },
     {
-      "epoch": 13.39,
-      "learning_rate": 1.0749506903353056e-05,
-      "loss": 1.4677,
+      "epoch": 20.2,
+      "learning_rate": 5.9598214285714296e-05,
+      "loss": 1.0869,
       "step": 4525
     },
     {
-      "epoch": 13.39,
-      "learning_rate": 1.0729783037475345e-05,
-      "loss": 1.585,
+      "epoch": 20.21,
+      "learning_rate": 5.9589285714285715e-05,
+      "loss": 1.0855,
       "step": 4526
     },
     {
-      "epoch": 13.39,
-      "learning_rate": 1.0710059171597634e-05,
-      "loss": 1.4728,
+      "epoch": 20.21,
+      "learning_rate": 5.958035714285715e-05,
+      "loss": 1.0815,
       "step": 4527
     },
     {
-      "epoch": 13.4,
-      "learning_rate": 1.069033530571992e-05,
-      "loss": 1.4412,
+      "epoch": 20.21,
+      "learning_rate": 5.957142857142858e-05,
+      "loss": 1.1242,
       "step": 4528
     },
     {
-      "epoch": 13.4,
-      "learning_rate": 1.067061143984221e-05,
-      "loss": 1.5083,
+      "epoch": 20.22,
+      "learning_rate": 5.95625e-05,
+      "loss": 1.0985,
       "step": 4529
     },
     {
-      "epoch": 13.4,
-      "learning_rate": 1.0650887573964498e-05,
-      "loss": 1.5206,
+      "epoch": 20.22,
+      "learning_rate": 5.955357142857143e-05,
+      "loss": 1.0635,
       "step": 4530
     },
     {
-      "epoch": 13.41,
-      "learning_rate": 1.0631163708086785e-05,
-      "loss": 1.5228,
+      "epoch": 20.23,
+      "learning_rate": 5.954464285714286e-05,
+      "loss": 1.1162,
       "step": 4531
     },
     {
-      "epoch": 13.41,
-      "learning_rate": 1.0611439842209074e-05,
-      "loss": 1.5545,
+      "epoch": 20.23,
+      "learning_rate": 5.953571428571428e-05,
+      "loss": 0.9718,
       "step": 4532
     },
     {
-      "epoch": 13.41,
-      "learning_rate": 1.0591715976331361e-05,
-      "loss": 1.4624,
+      "epoch": 20.24,
+      "learning_rate": 5.952678571428571e-05,
+      "loss": 1.0707,
       "step": 4533
     },
     {
-      "epoch": 13.41,
-      "learning_rate": 1.057199211045365e-05,
-      "loss": 1.5496,
+      "epoch": 20.24,
+      "learning_rate": 5.9517857142857145e-05,
+      "loss": 1.0343,
       "step": 4534
     },
     {
-      "epoch": 13.42,
-      "learning_rate": 1.0552268244575937e-05,
-      "loss": 1.5362,
+      "epoch": 20.25,
+      "learning_rate": 5.950892857142858e-05,
+      "loss": 1.1312,
       "step": 4535
     },
     {
-      "epoch": 13.42,
-      "learning_rate": 1.0532544378698226e-05,
-      "loss": 1.5171,
+      "epoch": 20.25,
+      "learning_rate": 5.95e-05,
+      "loss": 1.1381,
       "step": 4536
     },
     {
-      "epoch": 13.42,
-      "learning_rate": 1.0512820512820514e-05,
-      "loss": 1.5534,
+      "epoch": 20.25,
+      "learning_rate": 5.9491071428571435e-05,
+      "loss": 1.1191,
       "step": 4537
     },
     {
-      "epoch": 13.43,
-      "learning_rate": 1.0493096646942801e-05,
-      "loss": 1.4954,
+      "epoch": 20.26,
+      "learning_rate": 5.948214285714287e-05,
+      "loss": 1.0488,
       "step": 4538
     },
     {
-      "epoch": 13.43,
-      "learning_rate": 1.047337278106509e-05,
-      "loss": 1.4859,
+      "epoch": 20.26,
+      "learning_rate": 5.9473214285714286e-05,
+      "loss": 1.0418,
       "step": 4539
     },
     {
-      "epoch": 13.43,
-      "learning_rate": 1.0453648915187377e-05,
-      "loss": 1.4528,
+      "epoch": 20.27,
+      "learning_rate": 5.946428571428572e-05,
+      "loss": 1.0941,
       "step": 4540
     },
     {
-      "epoch": 13.43,
-      "learning_rate": 1.0433925049309664e-05,
-      "loss": 1.4255,
+      "epoch": 20.27,
+      "learning_rate": 5.945535714285715e-05,
+      "loss": 1.1062,
       "step": 4541
     },
     {
-      "epoch": 13.44,
-      "learning_rate": 1.0414201183431953e-05,
-      "loss": 1.4854,
+      "epoch": 20.28,
+      "learning_rate": 5.944642857142857e-05,
+      "loss": 1.1115,
       "step": 4542
     },
     {
-      "epoch": 13.44,
-      "learning_rate": 1.0394477317554242e-05,
-      "loss": 1.5269,
+      "epoch": 20.28,
+      "learning_rate": 5.94375e-05,
+      "loss": 1.059,
       "step": 4543
     },
     {
-      "epoch": 13.44,
-      "learning_rate": 1.0374753451676529e-05,
-      "loss": 1.5569,
+      "epoch": 20.29,
+      "learning_rate": 5.9428571428571434e-05,
+      "loss": 1.131,
       "step": 4544
     },
     {
-      "epoch": 13.45,
-      "learning_rate": 1.0355029585798817e-05,
-      "loss": 1.4417,
+      "epoch": 20.29,
+      "learning_rate": 5.9419642857142866e-05,
+      "loss": 1.0445,
       "step": 4545
     },
     {
-      "epoch": 13.45,
-      "learning_rate": 1.0335305719921106e-05,
-      "loss": 1.4147,
+      "epoch": 20.29,
+      "learning_rate": 5.9410714285714285e-05,
+      "loss": 1.0455,
       "step": 4546
     },
     {
-      "epoch": 13.45,
-      "learning_rate": 1.0315581854043393e-05,
-      "loss": 1.4721,
+      "epoch": 20.3,
+      "learning_rate": 5.940178571428572e-05,
+      "loss": 1.0759,
       "step": 4547
     },
     {
-      "epoch": 13.46,
-      "learning_rate": 1.0295857988165682e-05,
-      "loss": 1.4688,
+      "epoch": 20.3,
+      "learning_rate": 5.939285714285715e-05,
+      "loss": 1.0712,
       "step": 4548
     },
     {
-      "epoch": 13.46,
-      "learning_rate": 1.0276134122287969e-05,
-      "loss": 1.4414,
+      "epoch": 20.31,
+      "learning_rate": 5.938392857142857e-05,
+      "loss": 1.0867,
       "step": 4549
     },
     {
-      "epoch": 13.46,
-      "learning_rate": 1.0256410256410256e-05,
-      "loss": 1.5879,
+      "epoch": 20.31,
+      "learning_rate": 5.9375e-05,
+      "loss": 1.0843,
       "step": 4550
     },
     {
-      "epoch": 13.46,
-      "learning_rate": 1.0236686390532545e-05,
-      "loss": 1.5453,
+      "epoch": 20.32,
+      "learning_rate": 5.936607142857143e-05,
+      "loss": 1.1158,
       "step": 4551
     },
     {
-      "epoch": 13.47,
-      "learning_rate": 1.0216962524654833e-05,
-      "loss": 1.5071,
+      "epoch": 20.32,
+      "learning_rate": 5.935714285714286e-05,
+      "loss": 1.0554,
       "step": 4552
     },
     {
-      "epoch": 13.47,
-      "learning_rate": 1.019723865877712e-05,
-      "loss": 1.5494,
+      "epoch": 20.33,
+      "learning_rate": 5.934821428571429e-05,
+      "loss": 1.0957,
       "step": 4553
     },
     {
-      "epoch": 13.47,
-      "learning_rate": 1.0177514792899409e-05,
-      "loss": 1.5148,
+      "epoch": 20.33,
+      "learning_rate": 5.933928571428572e-05,
+      "loss": 1.0936,
       "step": 4554
     },
     {
-      "epoch": 13.48,
-      "learning_rate": 1.0157790927021698e-05,
-      "loss": 1.4966,
+      "epoch": 20.33,
+      "learning_rate": 5.933035714285714e-05,
+      "loss": 1.0778,
       "step": 4555
     },
     {
-      "epoch": 13.48,
-      "learning_rate": 1.0138067061143985e-05,
-      "loss": 1.5086,
+      "epoch": 20.34,
+      "learning_rate": 5.932142857142857e-05,
+      "loss": 1.0612,
       "step": 4556
     },
     {
-      "epoch": 13.48,
-      "learning_rate": 1.0118343195266272e-05,
-      "loss": 1.4476,
+      "epoch": 20.34,
+      "learning_rate": 5.9312500000000005e-05,
+      "loss": 1.0577,
       "step": 4557
     },
     {
-      "epoch": 13.49,
-      "learning_rate": 1.009861932938856e-05,
-      "loss": 1.5852,
+      "epoch": 20.35,
+      "learning_rate": 5.930357142857144e-05,
+      "loss": 1.0258,
       "step": 4558
     },
     {
-      "epoch": 13.49,
-      "learning_rate": 1.0078895463510848e-05,
-      "loss": 1.5884,
+      "epoch": 20.35,
+      "learning_rate": 5.9294642857142856e-05,
+      "loss": 1.0556,
       "step": 4559
     },
     {
-      "epoch": 13.49,
-      "learning_rate": 1.0059171597633136e-05,
-      "loss": 1.5983,
+      "epoch": 20.36,
+      "learning_rate": 5.928571428571429e-05,
+      "loss": 1.0664,
       "step": 4560
     },
     {
-      "epoch": 13.49,
-      "learning_rate": 1.0039447731755425e-05,
-      "loss": 1.5413,
+      "epoch": 20.36,
+      "learning_rate": 5.927678571428572e-05,
+      "loss": 1.0144,
       "step": 4561
     },
     {
-      "epoch": 13.5,
-      "learning_rate": 1.0019723865877712e-05,
-      "loss": 1.5489,
+      "epoch": 20.37,
+      "learning_rate": 5.926785714285714e-05,
+      "loss": 1.1212,
       "step": 4562
     },
     {
-      "epoch": 13.5,
-      "learning_rate": 1e-05,
-      "loss": 1.52,
+      "epoch": 20.37,
+      "learning_rate": 5.925892857142857e-05,
+      "loss": 1.1317,
       "step": 4563
     },
     {
-      "epoch": 13.5,
-      "learning_rate": 9.980276134122288e-06,
-      "loss": 1.481,
+      "epoch": 20.38,
+      "learning_rate": 5.9250000000000004e-05,
+      "loss": 1.1263,
       "step": 4564
     },
     {
-      "epoch": 13.51,
-      "learning_rate": 9.960552268244577e-06,
-      "loss": 1.515,
+      "epoch": 20.38,
+      "learning_rate": 5.924107142857143e-05,
+      "loss": 1.0541,
       "step": 4565
     },
     {
-      "epoch": 13.51,
-      "learning_rate": 9.940828402366864e-06,
-      "loss": 1.4688,
+      "epoch": 20.38,
+      "learning_rate": 5.923214285714286e-05,
+      "loss": 1.0631,
       "step": 4566
     },
     {
-      "epoch": 13.51,
-      "learning_rate": 9.921104536489152e-06,
-      "loss": 1.5169,
+      "epoch": 20.39,
+      "learning_rate": 5.9223214285714294e-05,
+      "loss": 1.0748,
       "step": 4567
     },
     {
-      "epoch": 13.51,
-      "learning_rate": 9.901380670611441e-06,
-      "loss": 1.5418,
+      "epoch": 20.39,
+      "learning_rate": 5.921428571428572e-05,
+      "loss": 1.1107,
       "step": 4568
     },
     {
-      "epoch": 13.52,
-      "learning_rate": 9.881656804733728e-06,
-      "loss": 1.5893,
+      "epoch": 20.4,
+      "learning_rate": 5.9205357142857144e-05,
+      "loss": 1.0428,
       "step": 4569
     },
     {
-      "epoch": 13.52,
-      "learning_rate": 9.861932938856017e-06,
-      "loss": 1.4654,
+      "epoch": 20.4,
+      "learning_rate": 5.919642857142858e-05,
+      "loss": 1.0198,
       "step": 4570
     },
     {
-      "epoch": 13.52,
-      "learning_rate": 9.842209072978306e-06,
-      "loss": 1.4535,
+      "epoch": 20.41,
+      "learning_rate": 5.918750000000001e-05,
+      "loss": 1.0673,
       "step": 4571
     },
     {
-      "epoch": 13.53,
-      "learning_rate": 9.822485207100591e-06,
-      "loss": 1.5634,
+      "epoch": 20.41,
+      "learning_rate": 5.917857142857143e-05,
+      "loss": 0.9683,
       "step": 4572
     },
     {
-      "epoch": 13.53,
-      "learning_rate": 9.80276134122288e-06,
-      "loss": 1.5744,
+      "epoch": 20.42,
+      "learning_rate": 5.916964285714286e-05,
+      "loss": 1.0767,
       "step": 4573
     },
     {
-      "epoch": 13.53,
-      "learning_rate": 9.783037475345168e-06,
-      "loss": 1.5237,
+      "epoch": 20.42,
+      "learning_rate": 5.916071428571429e-05,
+      "loss": 1.0007,
       "step": 4574
     },
     {
-      "epoch": 13.54,
-      "learning_rate": 9.763313609467455e-06,
-      "loss": 1.5229,
+      "epoch": 20.42,
+      "learning_rate": 5.915178571428571e-05,
+      "loss": 1.0757,
       "step": 4575
     },
     {
-      "epoch": 13.54,
-      "learning_rate": 9.743589743589744e-06,
-      "loss": 1.4538,
+      "epoch": 20.43,
+      "learning_rate": 5.914285714285714e-05,
+      "loss": 1.1346,
       "step": 4576
     },
     {
-      "epoch": 13.54,
-      "learning_rate": 9.723865877712033e-06,
-      "loss": 1.4927,
+      "epoch": 20.43,
+      "learning_rate": 5.9133928571428575e-05,
+      "loss": 1.0789,
       "step": 4577
     },
     {
-      "epoch": 13.54,
-      "learning_rate": 9.70414201183432e-06,
-      "loss": 1.5437,
+      "epoch": 20.44,
+      "learning_rate": 5.912500000000001e-05,
+      "loss": 1.0251,
       "step": 4578
     },
     {
-      "epoch": 13.55,
-      "learning_rate": 9.684418145956609e-06,
-      "loss": 1.4452,
+      "epoch": 20.44,
+      "learning_rate": 5.9116071428571426e-05,
+      "loss": 1.1408,
       "step": 4579
     },
     {
-      "epoch": 13.55,
-      "learning_rate": 9.664694280078896e-06,
-      "loss": 1.5662,
+      "epoch": 20.45,
+      "learning_rate": 5.910714285714286e-05,
+      "loss": 1.1024,
       "step": 4580
     },
     {
-      "epoch": 13.55,
-      "learning_rate": 9.644970414201183e-06,
-      "loss": 1.4411,
+      "epoch": 20.45,
+      "learning_rate": 5.909821428571429e-05,
+      "loss": 1.0405,
       "step": 4581
     },
     {
-      "epoch": 13.56,
-      "learning_rate": 9.625246548323471e-06,
-      "loss": 1.4432,
+      "epoch": 20.46,
+      "learning_rate": 5.9089285714285716e-05,
+      "loss": 1.0668,
       "step": 4582
     },
     {
-      "epoch": 13.56,
-      "learning_rate": 9.60552268244576e-06,
-      "loss": 1.4488,
+      "epoch": 20.46,
+      "learning_rate": 5.908035714285715e-05,
+      "loss": 1.101,
       "step": 4583
     },
     {
-      "epoch": 13.56,
-      "learning_rate": 9.585798816568047e-06,
-      "loss": 1.5543,
+      "epoch": 20.46,
+      "learning_rate": 5.907142857142858e-05,
+      "loss": 1.0328,
       "step": 4584
     },
     {
-      "epoch": 13.57,
-      "learning_rate": 9.566074950690336e-06,
-      "loss": 1.5046,
+      "epoch": 20.47,
+      "learning_rate": 5.90625e-05,
+      "loss": 1.0741,
       "step": 4585
     },
     {
-      "epoch": 13.57,
-      "learning_rate": 9.546351084812625e-06,
-      "loss": 1.4724,
+      "epoch": 20.47,
+      "learning_rate": 5.905357142857143e-05,
+      "loss": 1.1075,
       "step": 4586
     },
     {
-      "epoch": 13.57,
-      "learning_rate": 9.526627218934912e-06,
-      "loss": 1.5135,
+      "epoch": 20.48,
+      "learning_rate": 5.9044642857142864e-05,
+      "loss": 1.091,
       "step": 4587
     },
     {
-      "epoch": 13.57,
-      "learning_rate": 9.506903353057199e-06,
-      "loss": 1.5463,
+      "epoch": 20.48,
+      "learning_rate": 5.903571428571428e-05,
+      "loss": 1.071,
       "step": 4588
     },
     {
-      "epoch": 13.58,
-      "learning_rate": 9.487179487179487e-06,
-      "loss": 1.4529,
+      "epoch": 20.49,
+      "learning_rate": 5.9026785714285714e-05,
+      "loss": 1.0085,
       "step": 4589
     },
     {
-      "epoch": 13.58,
-      "learning_rate": 9.467455621301776e-06,
-      "loss": 1.6195,
+      "epoch": 20.49,
+      "learning_rate": 5.901785714285715e-05,
+      "loss": 1.0696,
       "step": 4590
     },
     {
-      "epoch": 13.58,
-      "learning_rate": 9.447731755424063e-06,
-      "loss": 1.6467,
+      "epoch": 20.5,
+      "learning_rate": 5.900892857142858e-05,
+      "loss": 1.0988,
       "step": 4591
     },
     {
-      "epoch": 13.59,
-      "learning_rate": 9.428007889546352e-06,
-      "loss": 1.4606,
+      "epoch": 20.5,
+      "learning_rate": 5.9e-05,
+      "loss": 1.0045,
       "step": 4592
     },
     {
-      "epoch": 13.59,
-      "learning_rate": 9.40828402366864e-06,
-      "loss": 1.4812,
+      "epoch": 20.5,
+      "learning_rate": 5.899107142857143e-05,
+      "loss": 1.0132,
       "step": 4593
     },
     {
-      "epoch": 13.59,
-      "learning_rate": 9.388560157790928e-06,
-      "loss": 1.408,
+      "epoch": 20.51,
+      "learning_rate": 5.898214285714286e-05,
+      "loss": 1.0266,
       "step": 4594
     },
     {
-      "epoch": 13.59,
-      "learning_rate": 9.368836291913216e-06,
-      "loss": 1.5475,
+      "epoch": 20.51,
+      "learning_rate": 5.897321428571429e-05,
+      "loss": 1.0893,
       "step": 4595
     },
     {
-      "epoch": 13.6,
-      "learning_rate": 9.349112426035503e-06,
-      "loss": 1.4236,
+      "epoch": 20.52,
+      "learning_rate": 5.896428571428572e-05,
+      "loss": 1.1005,
       "step": 4596
     },
     {
-      "epoch": 13.6,
-      "learning_rate": 9.32938856015779e-06,
-      "loss": 1.5355,
+      "epoch": 20.52,
+      "learning_rate": 5.8955357142857145e-05,
+      "loss": 1.1509,
       "step": 4597
     },
     {
-      "epoch": 13.6,
-      "learning_rate": 9.309664694280079e-06,
-      "loss": 1.4159,
+      "epoch": 20.53,
+      "learning_rate": 5.894642857142857e-05,
+      "loss": 1.099,
       "step": 4598
     },
     {
-      "epoch": 13.61,
-      "learning_rate": 9.289940828402368e-06,
-      "loss": 1.4632,
+      "epoch": 20.53,
+      "learning_rate": 5.89375e-05,
+      "loss": 1.1058,
       "step": 4599
     },
     {
-      "epoch": 13.61,
-      "learning_rate": 9.270216962524655e-06,
-      "loss": 1.5385,
+      "epoch": 20.54,
+      "learning_rate": 5.8928571428571435e-05,
+      "loss": 1.0601,
       "step": 4600
     },
     {
-      "epoch": 13.61,
-      "learning_rate": 9.250493096646944e-06,
-      "loss": 1.5154,
+      "epoch": 20.54,
+      "learning_rate": 5.891964285714287e-05,
+      "loss": 1.0391,
       "step": 4601
     },
     {
-      "epoch": 13.62,
-      "learning_rate": 9.230769230769232e-06,
-      "loss": 1.4405,
+      "epoch": 20.54,
+      "learning_rate": 5.8910714285714286e-05,
+      "loss": 1.0653,
       "step": 4602
     },
     {
-      "epoch": 13.62,
-      "learning_rate": 9.21104536489152e-06,
-      "loss": 1.4594,
+      "epoch": 20.55,
+      "learning_rate": 5.890178571428572e-05,
+      "loss": 1.073,
       "step": 4603
     },
     {
-      "epoch": 13.62,
-      "learning_rate": 9.191321499013806e-06,
-      "loss": 1.5343,
+      "epoch": 20.55,
+      "learning_rate": 5.889285714285715e-05,
+      "loss": 1.1074,
       "step": 4604
     },
     {
-      "epoch": 13.62,
-      "learning_rate": 9.171597633136095e-06,
-      "loss": 1.4666,
+      "epoch": 20.56,
+      "learning_rate": 5.888392857142857e-05,
+      "loss": 0.9932,
       "step": 4605
     },
     {
-      "epoch": 13.63,
-      "learning_rate": 9.151873767258382e-06,
-      "loss": 1.4538,
+      "epoch": 20.56,
+      "learning_rate": 5.8875e-05,
+      "loss": 1.0373,
       "step": 4606
     },
     {
-      "epoch": 13.63,
-      "learning_rate": 9.132149901380671e-06,
-      "loss": 1.4032,
+      "epoch": 20.57,
+      "learning_rate": 5.8866071428571434e-05,
+      "loss": 1.0766,
       "step": 4607
     },
     {
-      "epoch": 13.63,
-      "learning_rate": 9.11242603550296e-06,
-      "loss": 1.4715,
+      "epoch": 20.57,
+      "learning_rate": 5.885714285714285e-05,
+      "loss": 0.982,
       "step": 4608
     },
     {
-      "epoch": 13.64,
-      "learning_rate": 9.092702169625247e-06,
-      "loss": 1.4757,
+      "epoch": 20.58,
+      "learning_rate": 5.8848214285714284e-05,
+      "loss": 1.08,
       "step": 4609
     },
     {
-      "epoch": 13.64,
-      "learning_rate": 9.072978303747535e-06,
-      "loss": 1.5301,
+      "epoch": 20.58,
+      "learning_rate": 5.883928571428572e-05,
+      "loss": 1.1262,
       "step": 4610
     },
     {
-      "epoch": 13.64,
-      "learning_rate": 9.053254437869822e-06,
-      "loss": 1.5518,
+      "epoch": 20.58,
+      "learning_rate": 5.883035714285715e-05,
+      "loss": 1.0472,
       "step": 4611
     },
     {
-      "epoch": 13.64,
-      "learning_rate": 9.033530571992111e-06,
-      "loss": 1.4914,
+      "epoch": 20.59,
+      "learning_rate": 5.8821428571428574e-05,
+      "loss": 1.0538,
       "step": 4612
     },
     {
-      "epoch": 13.65,
-      "learning_rate": 9.013806706114398e-06,
-      "loss": 1.5893,
+      "epoch": 20.59,
+      "learning_rate": 5.8812500000000007e-05,
+      "loss": 1.1207,
       "step": 4613
     },
     {
-      "epoch": 13.65,
-      "learning_rate": 8.994082840236687e-06,
-      "loss": 1.4901,
+      "epoch": 20.6,
+      "learning_rate": 5.880357142857144e-05,
+      "loss": 1.0984,
       "step": 4614
     },
     {
-      "epoch": 13.65,
-      "learning_rate": 8.974358974358976e-06,
-      "loss": 1.431,
+      "epoch": 20.6,
+      "learning_rate": 5.879464285714286e-05,
+      "loss": 1.1267,
       "step": 4615
     },
     {
-      "epoch": 13.66,
-      "learning_rate": 8.954635108481263e-06,
-      "loss": 1.5575,
+      "epoch": 20.61,
+      "learning_rate": 5.878571428571429e-05,
+      "loss": 1.1524,
       "step": 4616
     },
     {
-      "epoch": 13.66,
-      "learning_rate": 8.934911242603551e-06,
-      "loss": 1.6481,
+      "epoch": 20.61,
+      "learning_rate": 5.877678571428572e-05,
+      "loss": 1.0482,
       "step": 4617
     },
     {
-      "epoch": 13.66,
-      "learning_rate": 8.91518737672584e-06,
-      "loss": 1.5141,
+      "epoch": 20.62,
+      "learning_rate": 5.876785714285714e-05,
+      "loss": 1.113,
       "step": 4618
     },
     {
-      "epoch": 13.67,
-      "learning_rate": 8.895463510848125e-06,
-      "loss": 1.5286,
+      "epoch": 20.62,
+      "learning_rate": 5.875892857142857e-05,
+      "loss": 1.0634,
       "step": 4619
     },
     {
-      "epoch": 13.67,
-      "learning_rate": 8.875739644970414e-06,
-      "loss": 1.4458,
+      "epoch": 20.62,
+      "learning_rate": 5.8750000000000005e-05,
+      "loss": 0.9775,
       "step": 4620
     },
     {
-      "epoch": 13.67,
-      "learning_rate": 8.856015779092703e-06,
-      "loss": 1.5639,
+      "epoch": 20.63,
+      "learning_rate": 5.8741071428571424e-05,
+      "loss": 1.1112,
       "step": 4621
     },
     {
-      "epoch": 13.67,
-      "learning_rate": 8.83629191321499e-06,
-      "loss": 1.4643,
+      "epoch": 20.63,
+      "learning_rate": 5.8732142857142856e-05,
+      "loss": 1.1259,
       "step": 4622
     },
     {
-      "epoch": 13.68,
-      "learning_rate": 8.816568047337279e-06,
-      "loss": 1.4249,
+      "epoch": 20.64,
+      "learning_rate": 5.872321428571429e-05,
+      "loss": 1.032,
       "step": 4623
     },
     {
-      "epoch": 13.68,
-      "learning_rate": 8.796844181459567e-06,
-      "loss": 1.5165,
+      "epoch": 20.64,
+      "learning_rate": 5.871428571428572e-05,
+      "loss": 1.0612,
       "step": 4624
     },
     {
-      "epoch": 13.68,
-      "learning_rate": 8.777120315581854e-06,
-      "loss": 1.441,
+      "epoch": 20.65,
+      "learning_rate": 5.8705357142857146e-05,
+      "loss": 1.1452,
       "step": 4625
     },
     {
-      "epoch": 13.69,
-      "learning_rate": 8.757396449704143e-06,
-      "loss": 1.5218,
+      "epoch": 20.65,
+      "learning_rate": 5.869642857142857e-05,
+      "loss": 1.0105,
       "step": 4626
     },
     {
-      "epoch": 13.69,
-      "learning_rate": 8.73767258382643e-06,
-      "loss": 1.4988,
+      "epoch": 20.66,
+      "learning_rate": 5.8687500000000003e-05,
+      "loss": 1.0799,
       "step": 4627
     },
     {
-      "epoch": 13.69,
-      "learning_rate": 8.717948717948717e-06,
-      "loss": 1.4345,
+      "epoch": 20.66,
+      "learning_rate": 5.867857142857143e-05,
+      "loss": 1.1586,
       "step": 4628
     },
     {
-      "epoch": 13.7,
-      "learning_rate": 8.698224852071006e-06,
-      "loss": 1.4325,
+      "epoch": 20.67,
+      "learning_rate": 5.866964285714286e-05,
+      "loss": 1.0329,
       "step": 4629
     },
     {
-      "epoch": 13.7,
-      "learning_rate": 8.678500986193295e-06,
-      "loss": 1.5832,
+      "epoch": 20.67,
+      "learning_rate": 5.8660714285714293e-05,
+      "loss": 1.0798,
       "step": 4630
     },
     {
-      "epoch": 13.7,
-      "learning_rate": 8.658777120315582e-06,
-      "loss": 1.5201,
+      "epoch": 20.67,
+      "learning_rate": 5.865178571428571e-05,
+      "loss": 1.0795,
       "step": 4631
     },
     {
-      "epoch": 13.7,
-      "learning_rate": 8.63905325443787e-06,
-      "loss": 1.4544,
+      "epoch": 20.68,
+      "learning_rate": 5.8642857142857144e-05,
+      "loss": 1.0906,
       "step": 4632
     },
     {
-      "epoch": 13.71,
-      "learning_rate": 8.619329388560159e-06,
-      "loss": 1.5565,
+      "epoch": 20.68,
+      "learning_rate": 5.8633928571428577e-05,
+      "loss": 1.1996,
       "step": 4633
     },
     {
-      "epoch": 13.71,
-      "learning_rate": 8.599605522682446e-06,
-      "loss": 1.5176,
+      "epoch": 20.69,
+      "learning_rate": 5.862500000000001e-05,
+      "loss": 0.9845,
       "step": 4634
     },
     {
-      "epoch": 13.71,
-      "learning_rate": 8.579881656804733e-06,
-      "loss": 1.4058,
+      "epoch": 20.69,
+      "learning_rate": 5.861607142857143e-05,
+      "loss": 1.0429,
       "step": 4635
     },
     {
-      "epoch": 13.72,
-      "learning_rate": 8.560157790927022e-06,
-      "loss": 1.4833,
+      "epoch": 20.7,
+      "learning_rate": 5.860714285714286e-05,
+      "loss": 1.1194,
       "step": 4636
     },
     {
-      "epoch": 13.72,
-      "learning_rate": 8.54043392504931e-06,
-      "loss": 1.515,
+      "epoch": 20.7,
+      "learning_rate": 5.859821428571429e-05,
+      "loss": 1.011,
       "step": 4637
     },
     {
-      "epoch": 13.72,
-      "learning_rate": 8.520710059171598e-06,
-      "loss": 1.5557,
+      "epoch": 20.71,
+      "learning_rate": 5.858928571428571e-05,
+      "loss": 1.116,
       "step": 4638
     },
     {
-      "epoch": 13.72,
-      "learning_rate": 8.500986193293886e-06,
-      "loss": 1.4854,
+      "epoch": 20.71,
+      "learning_rate": 5.858035714285714e-05,
+      "loss": 1.1694,
       "step": 4639
     },
     {
-      "epoch": 13.73,
-      "learning_rate": 8.481262327416175e-06,
-      "loss": 1.5795,
+      "epoch": 20.71,
+      "learning_rate": 5.8571428571428575e-05,
+      "loss": 1.1273,
       "step": 4640
     },
     {
-      "epoch": 13.73,
-      "learning_rate": 8.461538461538462e-06,
-      "loss": 1.4351,
+      "epoch": 20.72,
+      "learning_rate": 5.85625e-05,
+      "loss": 1.1444,
       "step": 4641
     },
     {
-      "epoch": 13.73,
-      "learning_rate": 8.44181459566075e-06,
-      "loss": 1.4354,
+      "epoch": 20.72,
+      "learning_rate": 5.855357142857143e-05,
+      "loss": 1.0406,
       "step": 4642
     },
     {
-      "epoch": 13.74,
-      "learning_rate": 8.422090729783038e-06,
-      "loss": 1.4783,
+      "epoch": 20.73,
+      "learning_rate": 5.8544642857142865e-05,
+      "loss": 1.072,
       "step": 4643
     },
     {
-      "epoch": 13.74,
-      "learning_rate": 8.402366863905325e-06,
-      "loss": 1.4676,
+      "epoch": 20.73,
+      "learning_rate": 5.853571428571429e-05,
+      "loss": 1.0672,
       "step": 4644
     },
     {
-      "epoch": 13.74,
-      "learning_rate": 8.382642998027614e-06,
-      "loss": 1.4841,
+      "epoch": 20.74,
+      "learning_rate": 5.8526785714285716e-05,
+      "loss": 1.1146,
       "step": 4645
     },
     {
-      "epoch": 13.75,
-      "learning_rate": 8.362919132149902e-06,
-      "loss": 1.4492,
+      "epoch": 20.74,
+      "learning_rate": 5.851785714285715e-05,
+      "loss": 1.122,
       "step": 4646
     },
     {
-      "epoch": 13.75,
-      "learning_rate": 8.34319526627219e-06,
-      "loss": 1.4598,
+      "epoch": 20.75,
+      "learning_rate": 5.850892857142858e-05,
+      "loss": 1.0847,
       "step": 4647
     },
     {
-      "epoch": 13.75,
-      "learning_rate": 8.323471400394478e-06,
-      "loss": 1.5564,
+      "epoch": 20.75,
+      "learning_rate": 5.85e-05,
+      "loss": 1.0749,
       "step": 4648
     },
     {
-      "epoch": 13.75,
-      "learning_rate": 8.303747534516767e-06,
-      "loss": 1.4876,
+      "epoch": 20.75,
+      "learning_rate": 5.849107142857143e-05,
+      "loss": 1.1299,
       "step": 4649
     },
     {
-      "epoch": 13.76,
-      "learning_rate": 8.284023668639054e-06,
-      "loss": 1.5112,
+      "epoch": 20.76,
+      "learning_rate": 5.8482142857142863e-05,
+      "loss": 1.0447,
       "step": 4650
     },
     {
-      "epoch": 13.76,
-      "learning_rate": 8.264299802761341e-06,
-      "loss": 1.5159,
+      "epoch": 20.76,
+      "learning_rate": 5.847321428571428e-05,
+      "loss": 1.0424,
       "step": 4651
     },
     {
-      "epoch": 13.76,
-      "learning_rate": 8.24457593688363e-06,
-      "loss": 1.4748,
+      "epoch": 20.77,
+      "learning_rate": 5.8464285714285714e-05,
+      "loss": 1.1245,
       "step": 4652
     },
     {
-      "epoch": 13.77,
-      "learning_rate": 8.224852071005917e-06,
-      "loss": 1.4625,
+      "epoch": 20.77,
+      "learning_rate": 5.8455357142857147e-05,
+      "loss": 1.0954,
       "step": 4653
     },
     {
-      "epoch": 13.77,
-      "learning_rate": 8.205128205128205e-06,
-      "loss": 1.5199,
+      "epoch": 20.78,
+      "learning_rate": 5.844642857142858e-05,
+      "loss": 1.1065,
       "step": 4654
     },
     {
-      "epoch": 13.77,
-      "learning_rate": 8.185404339250494e-06,
-      "loss": 1.5122,
+      "epoch": 20.78,
+      "learning_rate": 5.84375e-05,
+      "loss": 1.0314,
       "step": 4655
     },
     {
-      "epoch": 13.78,
-      "learning_rate": 8.165680473372781e-06,
-      "loss": 1.5162,
+      "epoch": 20.79,
+      "learning_rate": 5.842857142857143e-05,
+      "loss": 1.064,
       "step": 4656
     },
     {
-      "epoch": 13.78,
-      "learning_rate": 8.14595660749507e-06,
-      "loss": 1.5875,
+      "epoch": 20.79,
+      "learning_rate": 5.841964285714286e-05,
+      "loss": 1.0949,
       "step": 4657
     },
     {
-      "epoch": 13.78,
-      "learning_rate": 8.126232741617357e-06,
-      "loss": 1.4914,
+      "epoch": 20.79,
+      "learning_rate": 5.841071428571429e-05,
+      "loss": 1.0686,
       "step": 4658
     },
     {
-      "epoch": 13.78,
-      "learning_rate": 8.106508875739646e-06,
-      "loss": 1.5028,
+      "epoch": 20.8,
+      "learning_rate": 5.840178571428572e-05,
+      "loss": 1.095,
       "step": 4659
     },
     {
-      "epoch": 13.79,
-      "learning_rate": 8.086785009861933e-06,
-      "loss": 1.4772,
+      "epoch": 20.8,
+      "learning_rate": 5.839285714285715e-05,
+      "loss": 1.0085,
       "step": 4660
     },
     {
-      "epoch": 13.79,
-      "learning_rate": 8.067061143984221e-06,
-      "loss": 1.4521,
+      "epoch": 20.81,
+      "learning_rate": 5.838392857142857e-05,
+      "loss": 1.009,
       "step": 4661
     },
     {
-      "epoch": 13.79,
-      "learning_rate": 8.04733727810651e-06,
-      "loss": 1.4987,
+      "epoch": 20.81,
+      "learning_rate": 5.8375e-05,
+      "loss": 1.0501,
       "step": 4662
     },
     {
-      "epoch": 13.8,
-      "learning_rate": 8.027613412228797e-06,
-      "loss": 1.5708,
+      "epoch": 20.82,
+      "learning_rate": 5.8366071428571435e-05,
+      "loss": 1.0598,
       "step": 4663
     },
     {
-      "epoch": 13.8,
-      "learning_rate": 8.007889546351086e-06,
-      "loss": 1.4634,
+      "epoch": 20.82,
+      "learning_rate": 5.8357142857142854e-05,
+      "loss": 1.0561,
       "step": 4664
     },
     {
-      "epoch": 13.8,
-      "learning_rate": 7.988165680473373e-06,
-      "loss": 1.5043,
+      "epoch": 20.83,
+      "learning_rate": 5.8348214285714286e-05,
+      "loss": 1.1148,
       "step": 4665
     },
     {
-      "epoch": 13.8,
-      "learning_rate": 7.96844181459566e-06,
-      "loss": 1.5349,
+      "epoch": 20.83,
+      "learning_rate": 5.833928571428572e-05,
+      "loss": 1.1137,
       "step": 4666
     },
     {
-      "epoch": 13.81,
-      "learning_rate": 7.948717948717949e-06,
-      "loss": 1.4501,
+      "epoch": 20.83,
+      "learning_rate": 5.833035714285715e-05,
+      "loss": 1.0922,
       "step": 4667
     },
     {
-      "epoch": 13.81,
-      "learning_rate": 7.928994082840237e-06,
-      "loss": 1.4839,
+      "epoch": 20.84,
+      "learning_rate": 5.832142857142857e-05,
+      "loss": 1.0639,
       "step": 4668
     },
     {
-      "epoch": 13.81,
-      "learning_rate": 7.909270216962524e-06,
-      "loss": 1.4788,
+      "epoch": 20.84,
+      "learning_rate": 5.83125e-05,
+      "loss": 1.0481,
       "step": 4669
     },
     {
-      "epoch": 13.82,
-      "learning_rate": 7.889546351084813e-06,
-      "loss": 1.4325,
+      "epoch": 20.85,
+      "learning_rate": 5.830357142857143e-05,
+      "loss": 1.0869,
       "step": 4670
     },
     {
-      "epoch": 13.82,
-      "learning_rate": 7.869822485207102e-06,
-      "loss": 1.5616,
+      "epoch": 20.85,
+      "learning_rate": 5.829464285714286e-05,
+      "loss": 1.0685,
       "step": 4671
     },
     {
-      "epoch": 13.82,
-      "learning_rate": 7.850098619329389e-06,
-      "loss": 1.5384,
+      "epoch": 20.86,
+      "learning_rate": 5.828571428571429e-05,
+      "loss": 1.0663,
       "step": 4672
     },
     {
-      "epoch": 13.83,
-      "learning_rate": 7.830374753451678e-06,
-      "loss": 1.5243,
+      "epoch": 20.86,
+      "learning_rate": 5.8276785714285717e-05,
+      "loss": 1.1046,
       "step": 4673
     },
     {
-      "epoch": 13.83,
-      "learning_rate": 7.810650887573965e-06,
-      "loss": 1.5067,
+      "epoch": 20.87,
+      "learning_rate": 5.826785714285714e-05,
+      "loss": 1.0076,
       "step": 4674
     },
     {
-      "epoch": 13.83,
-      "learning_rate": 7.790927021696252e-06,
-      "loss": 1.4136,
+      "epoch": 20.87,
+      "learning_rate": 5.8258928571428574e-05,
+      "loss": 0.9798,
       "step": 4675
     },
     {
-      "epoch": 13.83,
-      "learning_rate": 7.77120315581854e-06,
-      "loss": 1.5292,
+      "epoch": 20.88,
+      "learning_rate": 5.8250000000000006e-05,
+      "loss": 1.014,
       "step": 4676
     },
     {
-      "epoch": 13.84,
-      "learning_rate": 7.751479289940829e-06,
-      "loss": 1.4561,
+      "epoch": 20.88,
+      "learning_rate": 5.824107142857144e-05,
+      "loss": 1.083,
       "step": 4677
     },
     {
-      "epoch": 13.84,
-      "learning_rate": 7.731755424063116e-06,
-      "loss": 1.5028,
+      "epoch": 20.88,
+      "learning_rate": 5.823214285714286e-05,
+      "loss": 1.0918,
       "step": 4678
     },
     {
-      "epoch": 13.84,
-      "learning_rate": 7.712031558185405e-06,
-      "loss": 1.5155,
+      "epoch": 20.89,
+      "learning_rate": 5.822321428571429e-05,
+      "loss": 1.072,
       "step": 4679
     },
     {
-      "epoch": 13.85,
-      "learning_rate": 7.692307692307694e-06,
-      "loss": 1.4598,
+      "epoch": 20.89,
+      "learning_rate": 5.821428571428572e-05,
+      "loss": 0.9667,
       "step": 4680
     },
     {
-      "epoch": 13.85,
-      "learning_rate": 7.67258382642998e-06,
-      "loss": 1.4999,
+      "epoch": 20.9,
+      "learning_rate": 5.820535714285714e-05,
+      "loss": 1.0854,
       "step": 4681
     },
     {
-      "epoch": 13.85,
-      "learning_rate": 7.652859960552268e-06,
-      "loss": 1.5121,
+      "epoch": 20.9,
+      "learning_rate": 5.819642857142857e-05,
+      "loss": 1.0906,
       "step": 4682
     },
     {
-      "epoch": 13.86,
-      "learning_rate": 7.633136094674556e-06,
-      "loss": 1.5301,
+      "epoch": 20.91,
+      "learning_rate": 5.8187500000000005e-05,
+      "loss": 1.0092,
       "step": 4683
     },
     {
-      "epoch": 13.86,
-      "learning_rate": 7.613412228796844e-06,
-      "loss": 1.5079,
+      "epoch": 20.91,
+      "learning_rate": 5.8178571428571424e-05,
+      "loss": 1.0158,
       "step": 4684
     },
     {
-      "epoch": 13.86,
-      "learning_rate": 7.593688362919132e-06,
-      "loss": 1.5107,
+      "epoch": 20.92,
+      "learning_rate": 5.8169642857142856e-05,
+      "loss": 1.1016,
       "step": 4685
     },
     {
-      "epoch": 13.86,
-      "learning_rate": 7.573964497041421e-06,
-      "loss": 1.4796,
+      "epoch": 20.92,
+      "learning_rate": 5.816071428571429e-05,
+      "loss": 1.0823,
       "step": 4686
     },
     {
-      "epoch": 13.87,
-      "learning_rate": 7.554240631163709e-06,
-      "loss": 1.528,
+      "epoch": 20.92,
+      "learning_rate": 5.815178571428572e-05,
+      "loss": 1.061,
       "step": 4687
     },
     {
-      "epoch": 13.87,
-      "learning_rate": 7.534516765285997e-06,
-      "loss": 1.4729,
+      "epoch": 20.93,
+      "learning_rate": 5.8142857142857146e-05,
+      "loss": 1.0871,
       "step": 4688
     },
     {
-      "epoch": 13.87,
-      "learning_rate": 7.514792899408285e-06,
-      "loss": 1.3823,
+      "epoch": 20.93,
+      "learning_rate": 5.813392857142858e-05,
+      "loss": 1.0271,
       "step": 4689
     },
     {
-      "epoch": 13.88,
-      "learning_rate": 7.4950690335305715e-06,
-      "loss": 1.5994,
+      "epoch": 20.94,
+      "learning_rate": 5.812500000000001e-05,
+      "loss": 1.0526,
       "step": 4690
     },
     {
-      "epoch": 13.88,
-      "learning_rate": 7.47534516765286e-06,
-      "loss": 1.567,
+      "epoch": 20.94,
+      "learning_rate": 5.811607142857143e-05,
+      "loss": 1.082,
       "step": 4691
     },
     {
-      "epoch": 13.88,
-      "learning_rate": 7.455621301775148e-06,
-      "loss": 1.5709,
+      "epoch": 20.95,
+      "learning_rate": 5.810714285714286e-05,
+      "loss": 1.105,
       "step": 4692
     },
     {
-      "epoch": 13.88,
-      "learning_rate": 7.435897435897436e-06,
-      "loss": 1.4886,
+      "epoch": 20.95,
+      "learning_rate": 5.809821428571429e-05,
+      "loss": 1.0666,
       "step": 4693
     },
     {
-      "epoch": 13.89,
-      "learning_rate": 7.416173570019725e-06,
-      "loss": 1.4158,
+      "epoch": 20.96,
+      "learning_rate": 5.808928571428571e-05,
+      "loss": 1.0748,
       "step": 4694
     },
     {
-      "epoch": 13.89,
-      "learning_rate": 7.396449704142013e-06,
-      "loss": 1.4831,
+      "epoch": 20.96,
+      "learning_rate": 5.8080357142857144e-05,
+      "loss": 0.9897,
       "step": 4695
     },
     {
-      "epoch": 13.89,
-      "learning_rate": 7.3767258382643005e-06,
-      "loss": 1.5123,
+      "epoch": 20.96,
+      "learning_rate": 5.8071428571428576e-05,
+      "loss": 1.0453,
       "step": 4696
     },
     {
-      "epoch": 13.9,
-      "learning_rate": 7.357001972386588e-06,
-      "loss": 1.5019,
+      "epoch": 20.97,
+      "learning_rate": 5.8062499999999995e-05,
+      "loss": 1.0199,
       "step": 4697
     },
     {
-      "epoch": 13.9,
-      "learning_rate": 7.337278106508875e-06,
-      "loss": 1.5614,
+      "epoch": 20.97,
+      "learning_rate": 5.805357142857143e-05,
+      "loss": 1.0253,
       "step": 4698
     },
     {
-      "epoch": 13.9,
-      "learning_rate": 7.317554240631163e-06,
-      "loss": 1.5035,
+      "epoch": 20.98,
+      "learning_rate": 5.804464285714286e-05,
+      "loss": 1.0463,
       "step": 4699
     },
     {
-      "epoch": 13.91,
-      "learning_rate": 7.297830374753452e-06,
-      "loss": 1.473,
+      "epoch": 20.98,
+      "learning_rate": 5.803571428571429e-05,
+      "loss": 1.0713,
       "step": 4700
     },
     {
-      "epoch": 13.91,
-      "learning_rate": 7.27810650887574e-06,
-      "loss": 1.4877,
+      "epoch": 20.99,
+      "learning_rate": 5.802678571428572e-05,
+      "loss": 1.0883,
       "step": 4701
     },
     {
-      "epoch": 13.91,
-      "learning_rate": 7.258382642998028e-06,
-      "loss": 1.4283,
+      "epoch": 20.99,
+      "learning_rate": 5.801785714285714e-05,
+      "loss": 1.0432,
       "step": 4702
     },
     {
-      "epoch": 13.91,
-      "learning_rate": 7.2386587771203165e-06,
-      "loss": 1.5233,
+      "epoch": 21.0,
+      "learning_rate": 5.8008928571428575e-05,
+      "loss": 1.0407,
       "step": 4703
     },
     {
-      "epoch": 13.92,
-      "learning_rate": 7.218934911242604e-06,
-      "loss": 1.464,
+      "epoch": 21.0,
+      "learning_rate": 5.8e-05,
+      "loss": 0.9636,
       "step": 4704
     },
     {
-      "epoch": 13.92,
-      "learning_rate": 7.199211045364892e-06,
-      "loss": 1.5517,
+      "epoch": 21.0,
+      "learning_rate": 5.799107142857143e-05,
+      "loss": 1.0109,
       "step": 4705
     },
     {
-      "epoch": 13.92,
-      "learning_rate": 7.179487179487179e-06,
-      "loss": 1.5264,
+      "epoch": 21.01,
+      "learning_rate": 5.7982142857142865e-05,
+      "loss": 1.1112,
       "step": 4706
     },
     {
-      "epoch": 13.93,
-      "learning_rate": 7.159763313609467e-06,
-      "loss": 1.4836,
+      "epoch": 21.01,
+      "learning_rate": 5.7973214285714283e-05,
+      "loss": 1.074,
       "step": 4707
     },
     {
-      "epoch": 13.93,
-      "learning_rate": 7.140039447731756e-06,
-      "loss": 1.5136,
+      "epoch": 21.02,
+      "learning_rate": 5.7964285714285716e-05,
+      "loss": 0.9978,
       "step": 4708
     },
     {
-      "epoch": 13.93,
-      "learning_rate": 7.120315581854044e-06,
-      "loss": 1.5405,
+      "epoch": 21.02,
+      "learning_rate": 5.795535714285715e-05,
+      "loss": 1.0262,
       "step": 4709
     },
     {
-      "epoch": 13.93,
-      "learning_rate": 7.100591715976332e-06,
-      "loss": 1.4694,
+      "epoch": 21.03,
+      "learning_rate": 5.794642857142858e-05,
+      "loss": 1.0377,
       "step": 4710
     },
     {
-      "epoch": 13.94,
-      "learning_rate": 7.08086785009862e-06,
-      "loss": 1.5195,
+      "epoch": 21.03,
+      "learning_rate": 5.79375e-05,
+      "loss": 1.1096,
       "step": 4711
     },
     {
-      "epoch": 13.94,
-      "learning_rate": 7.061143984220908e-06,
-      "loss": 1.4639,
+      "epoch": 21.04,
+      "learning_rate": 5.792857142857143e-05,
+      "loss": 1.0511,
       "step": 4712
     },
     {
-      "epoch": 13.94,
-      "learning_rate": 7.041420118343195e-06,
-      "loss": 1.5696,
+      "epoch": 21.04,
+      "learning_rate": 5.791964285714286e-05,
+      "loss": 1.0212,
       "step": 4713
     },
     {
-      "epoch": 13.95,
-      "learning_rate": 7.021696252465483e-06,
-      "loss": 1.5226,
+      "epoch": 21.04,
+      "learning_rate": 5.791071428571428e-05,
+      "loss": 1.0096,
       "step": 4714
     },
     {
-      "epoch": 13.95,
-      "learning_rate": 7.001972386587771e-06,
-      "loss": 1.4682,
+      "epoch": 21.05,
+      "learning_rate": 5.7901785714285714e-05,
+      "loss": 1.1077,
       "step": 4715
     },
     {
-      "epoch": 13.95,
-      "learning_rate": 6.98224852071006e-06,
-      "loss": 1.5026,
+      "epoch": 21.05,
+      "learning_rate": 5.7892857142857146e-05,
+      "loss": 1.015,
       "step": 4716
     },
     {
-      "epoch": 13.96,
-      "learning_rate": 6.962524654832348e-06,
-      "loss": 1.506,
+      "epoch": 21.06,
+      "learning_rate": 5.788392857142857e-05,
+      "loss": 1.061,
       "step": 4717
     },
     {
-      "epoch": 13.96,
-      "learning_rate": 6.9428007889546355e-06,
-      "loss": 1.5298,
+      "epoch": 21.06,
+      "learning_rate": 5.7875000000000004e-05,
+      "loss": 1.0936,
       "step": 4718
     },
     {
-      "epoch": 13.96,
-      "learning_rate": 6.923076923076923e-06,
-      "loss": 1.5621,
+      "epoch": 21.07,
+      "learning_rate": 5.7866071428571436e-05,
+      "loss": 1.0382,
       "step": 4719
     },
     {
-      "epoch": 13.96,
-      "learning_rate": 6.903353057199212e-06,
-      "loss": 1.4526,
+      "epoch": 21.07,
+      "learning_rate": 5.785714285714287e-05,
+      "loss": 1.0218,
       "step": 4720
     },
     {
-      "epoch": 13.97,
-      "learning_rate": 6.883629191321498e-06,
-      "loss": 1.5005,
+      "epoch": 21.08,
+      "learning_rate": 5.784821428571429e-05,
+      "loss": 1.0526,
       "step": 4721
     },
     {
-      "epoch": 13.97,
-      "learning_rate": 6.863905325443787e-06,
-      "loss": 1.4651,
+      "epoch": 21.08,
+      "learning_rate": 5.783928571428572e-05,
+      "loss": 1.0859,
       "step": 4722
     },
     {
-      "epoch": 13.97,
-      "learning_rate": 6.844181459566075e-06,
-      "loss": 1.5505,
+      "epoch": 21.08,
+      "learning_rate": 5.783035714285715e-05,
+      "loss": 1.0831,
       "step": 4723
     },
     {
-      "epoch": 13.98,
-      "learning_rate": 6.824457593688363e-06,
-      "loss": 1.4798,
+      "epoch": 21.09,
+      "learning_rate": 5.782142857142857e-05,
+      "loss": 1.0981,
       "step": 4724
     },
     {
-      "epoch": 13.98,
-      "learning_rate": 6.8047337278106515e-06,
-      "loss": 1.5577,
+      "epoch": 21.09,
+      "learning_rate": 5.78125e-05,
+      "loss": 1.0152,
       "step": 4725
     },
     {
-      "epoch": 13.98,
-      "learning_rate": 6.785009861932939e-06,
-      "loss": 1.5206,
+      "epoch": 21.1,
+      "learning_rate": 5.7803571428571435e-05,
+      "loss": 0.991,
       "step": 4726
     },
     {
-      "epoch": 13.99,
-      "learning_rate": 6.765285996055227e-06,
-      "loss": 1.5925,
+      "epoch": 21.1,
+      "learning_rate": 5.7794642857142853e-05,
+      "loss": 1.0114,
       "step": 4727
     },
     {
-      "epoch": 13.99,
-      "learning_rate": 6.745562130177516e-06,
-      "loss": 1.5099,
+      "epoch": 21.11,
+      "learning_rate": 5.7785714285714286e-05,
+      "loss": 1.0669,
       "step": 4728
     },
     {
-      "epoch": 13.99,
-      "learning_rate": 6.725838264299802e-06,
-      "loss": 1.5857,
+      "epoch": 21.11,
+      "learning_rate": 5.777678571428572e-05,
+      "loss": 1.0538,
       "step": 4729
     },
     {
-      "epoch": 13.99,
-      "learning_rate": 6.706114398422091e-06,
-      "loss": 1.5007,
+      "epoch": 21.12,
+      "learning_rate": 5.776785714285714e-05,
+      "loss": 1.0961,
       "step": 4730
     },
     {
-      "epoch": 14.0,
-      "learning_rate": 6.686390532544379e-06,
-      "loss": 1.4499,
+      "epoch": 21.12,
+      "learning_rate": 5.775892857142857e-05,
+      "loss": 1.0965,
       "step": 4731
     },
     {
-      "epoch": 14.0,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 1.4894,
+      "epoch": 21.12,
+      "learning_rate": 5.775e-05,
+      "loss": 1.0863,
       "step": 4732
     },
     {
-      "epoch": 14.0,
-      "learning_rate": 6.646942800788955e-06,
-      "loss": 1.505,
+      "epoch": 21.13,
+      "learning_rate": 5.774107142857143e-05,
+      "loss": 0.9981,
       "step": 4733
     },
     {
-      "epoch": 14.01,
-      "learning_rate": 6.627218934911243e-06,
-      "loss": 1.6157,
+      "epoch": 21.13,
+      "learning_rate": 5.773214285714286e-05,
+      "loss": 1.0935,
       "step": 4734
     },
     {
-      "epoch": 14.01,
-      "learning_rate": 6.607495069033531e-06,
-      "loss": 1.5003,
+      "epoch": 21.14,
+      "learning_rate": 5.772321428571429e-05,
+      "loss": 1.1163,
       "step": 4735
     },
     {
-      "epoch": 14.01,
-      "learning_rate": 6.58777120315582e-06,
-      "loss": 1.5386,
+      "epoch": 21.14,
+      "learning_rate": 5.771428571428572e-05,
+      "loss": 1.0292,
       "step": 4736
     },
     {
-      "epoch": 14.01,
-      "learning_rate": 6.568047337278106e-06,
-      "loss": 1.5206,
+      "epoch": 21.15,
+      "learning_rate": 5.770535714285714e-05,
+      "loss": 1.0744,
       "step": 4737
     },
     {
-      "epoch": 14.02,
-      "learning_rate": 6.548323471400395e-06,
-      "loss": 1.4733,
+      "epoch": 21.15,
+      "learning_rate": 5.7696428571428574e-05,
+      "loss": 1.0662,
       "step": 4738
     },
     {
-      "epoch": 14.02,
-      "learning_rate": 6.528599605522683e-06,
-      "loss": 1.5458,
+      "epoch": 21.16,
+      "learning_rate": 5.7687500000000006e-05,
+      "loss": 1.0467,
       "step": 4739
     },
     {
-      "epoch": 14.02,
-      "learning_rate": 6.5088757396449705e-06,
-      "loss": 1.3759,
+      "epoch": 21.16,
+      "learning_rate": 5.7678571428571425e-05,
+      "loss": 1.0745,
       "step": 4740
     },
     {
-      "epoch": 14.03,
-      "learning_rate": 6.489151873767258e-06,
-      "loss": 1.5464,
+      "epoch": 21.17,
+      "learning_rate": 5.766964285714286e-05,
+      "loss": 1.0878,
       "step": 4741
     },
     {
-      "epoch": 14.03,
-      "learning_rate": 6.469428007889547e-06,
-      "loss": 1.5275,
+      "epoch": 21.17,
+      "learning_rate": 5.766071428571429e-05,
+      "loss": 1.0464,
       "step": 4742
     },
     {
-      "epoch": 14.03,
-      "learning_rate": 6.449704142011835e-06,
-      "loss": 1.5618,
+      "epoch": 21.17,
+      "learning_rate": 5.765178571428572e-05,
+      "loss": 1.0229,
       "step": 4743
     },
     {
-      "epoch": 14.04,
-      "learning_rate": 6.429980276134123e-06,
-      "loss": 1.5161,
+      "epoch": 21.18,
+      "learning_rate": 5.764285714285714e-05,
+      "loss": 1.0015,
       "step": 4744
     },
     {
-      "epoch": 14.04,
-      "learning_rate": 6.41025641025641e-06,
-      "loss": 1.5037,
+      "epoch": 21.18,
+      "learning_rate": 5.763392857142857e-05,
+      "loss": 1.0518,
       "step": 4745
     },
     {
-      "epoch": 14.04,
-      "learning_rate": 6.390532544378698e-06,
-      "loss": 1.4424,
+      "epoch": 21.19,
+      "learning_rate": 5.7625000000000005e-05,
+      "loss": 0.9818,
       "step": 4746
     },
     {
-      "epoch": 14.04,
-      "learning_rate": 6.3708086785009865e-06,
-      "loss": 1.4973,
+      "epoch": 21.19,
+      "learning_rate": 5.761607142857143e-05,
+      "loss": 1.0134,
       "step": 4747
     },
     {
-      "epoch": 14.05,
-      "learning_rate": 6.351084812623274e-06,
-      "loss": 1.5167,
+      "epoch": 21.2,
+      "learning_rate": 5.760714285714286e-05,
+      "loss": 1.0801,
       "step": 4748
     },
     {
-      "epoch": 14.05,
-      "learning_rate": 6.331360946745562e-06,
-      "loss": 1.4974,
+      "epoch": 21.2,
+      "learning_rate": 5.7598214285714295e-05,
+      "loss": 1.1148,
       "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,
+      "epoch": 21.21,
+      "learning_rate": 5.758928571428571e-05,
+      "loss": 1.0032,
       "step": 4750
     },
     {
-      "epoch": 14.06,
-      "learning_rate": 6.291913214990139e-06,
-      "loss": 1.5984,
+      "epoch": 21.21,
+      "learning_rate": 5.7580357142857146e-05,
+      "loss": 1.0685,
       "step": 4751
     },
     {
-      "epoch": 14.06,
-      "learning_rate": 6.272189349112427e-06,
-      "loss": 1.405,
+      "epoch": 21.21,
+      "learning_rate": 5.757142857142858e-05,
+      "loss": 1.0907,
       "step": 4752
     },
     {
-      "epoch": 14.06,
-      "learning_rate": 6.252465483234714e-06,
-      "loss": 1.5788,
+      "epoch": 21.22,
+      "learning_rate": 5.756250000000001e-05,
+      "loss": 1.0393,
       "step": 4753
     },
     {
-      "epoch": 14.07,
-      "learning_rate": 6.2327416173570025e-06,
-      "loss": 1.4384,
+      "epoch": 21.22,
+      "learning_rate": 5.755357142857143e-05,
+      "loss": 1.0818,
       "step": 4754
     },
     {
-      "epoch": 14.07,
-      "learning_rate": 6.21301775147929e-06,
-      "loss": 1.5015,
+      "epoch": 21.23,
+      "learning_rate": 5.754464285714286e-05,
+      "loss": 1.0615,
       "step": 4755
     },
     {
-      "epoch": 14.07,
-      "learning_rate": 6.193293885601578e-06,
-      "loss": 1.5134,
+      "epoch": 21.23,
+      "learning_rate": 5.753571428571429e-05,
+      "loss": 1.133,
       "step": 4756
     },
     {
-      "epoch": 14.07,
-      "learning_rate": 6.173570019723866e-06,
-      "loss": 1.4876,
+      "epoch": 21.24,
+      "learning_rate": 5.752678571428571e-05,
+      "loss": 1.0978,
       "step": 4757
     },
     {
-      "epoch": 14.08,
-      "learning_rate": 6.153846153846155e-06,
-      "loss": 1.5256,
+      "epoch": 21.24,
+      "learning_rate": 5.7517857142857144e-05,
+      "loss": 1.0832,
       "step": 4758
     },
     {
-      "epoch": 14.08,
-      "learning_rate": 6.134122287968442e-06,
-      "loss": 1.5403,
+      "epoch": 21.25,
+      "learning_rate": 5.7508928571428576e-05,
+      "loss": 1.0592,
       "step": 4759
     },
     {
-      "epoch": 14.08,
-      "learning_rate": 6.11439842209073e-06,
-      "loss": 1.5524,
+      "epoch": 21.25,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 1.1046,
       "step": 4760
     },
     {
-      "epoch": 14.09,
-      "learning_rate": 6.0946745562130185e-06,
-      "loss": 1.5337,
+      "epoch": 21.25,
+      "learning_rate": 5.749107142857143e-05,
+      "loss": 1.0814,
       "step": 4761
     },
     {
-      "epoch": 14.09,
-      "learning_rate": 6.0749506903353055e-06,
-      "loss": 1.5065,
+      "epoch": 21.26,
+      "learning_rate": 5.748214285714286e-05,
+      "loss": 1.0751,
       "step": 4762
     },
     {
-      "epoch": 14.09,
-      "learning_rate": 6.055226824457593e-06,
-      "loss": 1.5805,
+      "epoch": 21.26,
+      "learning_rate": 5.7473214285714285e-05,
+      "loss": 1.0343,
       "step": 4763
     },
     {
-      "epoch": 14.09,
-      "learning_rate": 6.035502958579882e-06,
-      "loss": 1.6528,
+      "epoch": 21.27,
+      "learning_rate": 5.746428571428572e-05,
+      "loss": 1.0948,
       "step": 4764
     },
     {
-      "epoch": 14.1,
-      "learning_rate": 6.01577909270217e-06,
-      "loss": 1.4761,
+      "epoch": 21.27,
+      "learning_rate": 5.745535714285715e-05,
+      "loss": 1.0594,
       "step": 4765
     },
     {
-      "epoch": 14.1,
-      "learning_rate": 5.996055226824458e-06,
-      "loss": 1.564,
+      "epoch": 21.28,
+      "learning_rate": 5.744642857142858e-05,
+      "loss": 1.0431,
       "step": 4766
     },
     {
-      "epoch": 14.1,
-      "learning_rate": 5.976331360946746e-06,
-      "loss": 1.5328,
+      "epoch": 21.28,
+      "learning_rate": 5.74375e-05,
+      "loss": 1.0025,
       "step": 4767
     },
     {
-      "epoch": 14.11,
-      "learning_rate": 5.956607495069034e-06,
-      "loss": 1.5358,
+      "epoch": 21.29,
+      "learning_rate": 5.742857142857143e-05,
+      "loss": 1.0337,
       "step": 4768
     },
     {
-      "epoch": 14.11,
-      "learning_rate": 5.936883629191322e-06,
-      "loss": 1.6012,
+      "epoch": 21.29,
+      "learning_rate": 5.7419642857142865e-05,
+      "loss": 1.0222,
       "step": 4769
     },
     {
-      "epoch": 14.11,
-      "learning_rate": 5.917159763313609e-06,
-      "loss": 1.5372,
+      "epoch": 21.29,
+      "learning_rate": 5.741071428571428e-05,
+      "loss": 1.1126,
       "step": 4770
     },
     {
-      "epoch": 14.12,
-      "learning_rate": 5.897435897435897e-06,
-      "loss": 1.4105,
+      "epoch": 21.3,
+      "learning_rate": 5.7401785714285716e-05,
+      "loss": 0.9922,
       "step": 4771
     },
     {
-      "epoch": 14.12,
-      "learning_rate": 5.877712031558186e-06,
-      "loss": 1.4671,
+      "epoch": 21.3,
+      "learning_rate": 5.739285714285715e-05,
+      "loss": 1.0536,
       "step": 4772
     },
     {
-      "epoch": 14.12,
-      "learning_rate": 5.857988165680474e-06,
-      "loss": 1.467,
+      "epoch": 21.31,
+      "learning_rate": 5.7383928571428566e-05,
+      "loss": 0.9841,
       "step": 4773
     },
     {
-      "epoch": 14.12,
-      "learning_rate": 5.838264299802761e-06,
-      "loss": 1.4765,
+      "epoch": 21.31,
+      "learning_rate": 5.7375e-05,
+      "loss": 1.066,
       "step": 4774
     },
     {
-      "epoch": 14.13,
-      "learning_rate": 5.81854043392505e-06,
-      "loss": 1.4821,
+      "epoch": 21.32,
+      "learning_rate": 5.736607142857143e-05,
+      "loss": 1.0888,
       "step": 4775
     },
     {
-      "epoch": 14.13,
-      "learning_rate": 5.7988165680473375e-06,
-      "loss": 1.5277,
+      "epoch": 21.32,
+      "learning_rate": 5.735714285714286e-05,
+      "loss": 1.0995,
       "step": 4776
     },
     {
-      "epoch": 14.13,
-      "learning_rate": 5.779092702169625e-06,
-      "loss": 1.4904,
+      "epoch": 21.33,
+      "learning_rate": 5.734821428571429e-05,
+      "loss": 0.977,
       "step": 4777
     },
     {
-      "epoch": 14.14,
-      "learning_rate": 5.759368836291913e-06,
-      "loss": 1.4022,
+      "epoch": 21.33,
+      "learning_rate": 5.733928571428572e-05,
+      "loss": 1.0239,
       "step": 4778
     },
     {
-      "epoch": 14.14,
-      "learning_rate": 5.739644970414201e-06,
-      "loss": 1.4523,
+      "epoch": 21.33,
+      "learning_rate": 5.7330357142857146e-05,
+      "loss": 1.0737,
       "step": 4779
     },
     {
-      "epoch": 14.14,
-      "learning_rate": 5.71992110453649e-06,
-      "loss": 1.4252,
+      "epoch": 21.34,
+      "learning_rate": 5.732142857142857e-05,
+      "loss": 1.0345,
       "step": 4780
     },
     {
-      "epoch": 14.14,
-      "learning_rate": 5.700197238658778e-06,
-      "loss": 1.5071,
+      "epoch": 21.34,
+      "learning_rate": 5.7312500000000004e-05,
+      "loss": 1.0313,
       "step": 4781
     },
     {
-      "epoch": 14.15,
-      "learning_rate": 5.680473372781065e-06,
-      "loss": 1.4713,
+      "epoch": 21.35,
+      "learning_rate": 5.7303571428571436e-05,
+      "loss": 1.0865,
       "step": 4782
     },
     {
-      "epoch": 14.15,
-      "learning_rate": 5.6607495069033535e-06,
-      "loss": 1.5751,
+      "epoch": 21.35,
+      "learning_rate": 5.7294642857142855e-05,
+      "loss": 1.0395,
       "step": 4783
     },
     {
-      "epoch": 14.15,
-      "learning_rate": 5.641025641025641e-06,
-      "loss": 1.467,
+      "epoch": 21.36,
+      "learning_rate": 5.728571428571429e-05,
+      "loss": 1.0398,
       "step": 4784
     },
     {
-      "epoch": 14.16,
-      "learning_rate": 5.621301775147929e-06,
-      "loss": 1.4518,
+      "epoch": 21.36,
+      "learning_rate": 5.727678571428572e-05,
+      "loss": 1.1384,
       "step": 4785
     },
     {
-      "epoch": 14.16,
-      "learning_rate": 5.601577909270217e-06,
-      "loss": 1.5738,
+      "epoch": 21.37,
+      "learning_rate": 5.726785714285715e-05,
+      "loss": 1.0789,
       "step": 4786
     },
     {
-      "epoch": 14.16,
-      "learning_rate": 5.581854043392505e-06,
-      "loss": 1.5338,
+      "epoch": 21.37,
+      "learning_rate": 5.725892857142857e-05,
+      "loss": 1.0538,
       "step": 4787
     },
     {
-      "epoch": 14.17,
-      "learning_rate": 5.562130177514793e-06,
-      "loss": 1.5466,
+      "epoch": 21.38,
+      "learning_rate": 5.725e-05,
+      "loss": 1.0471,
       "step": 4788
     },
     {
-      "epoch": 14.17,
-      "learning_rate": 5.542406311637082e-06,
-      "loss": 1.5026,
+      "epoch": 21.38,
+      "learning_rate": 5.7241071428571435e-05,
+      "loss": 1.0213,
       "step": 4789
     },
     {
-      "epoch": 14.17,
-      "learning_rate": 5.522682445759369e-06,
-      "loss": 1.5082,
+      "epoch": 21.38,
+      "learning_rate": 5.723214285714285e-05,
+      "loss": 1.0626,
       "step": 4790
     },
     {
-      "epoch": 14.17,
-      "learning_rate": 5.502958579881657e-06,
-      "loss": 1.5016,
+      "epoch": 21.39,
+      "learning_rate": 5.7223214285714286e-05,
+      "loss": 1.0131,
       "step": 4791
     },
     {
-      "epoch": 14.18,
-      "learning_rate": 5.483234714003945e-06,
-      "loss": 1.5297,
+      "epoch": 21.39,
+      "learning_rate": 5.721428571428572e-05,
+      "loss": 1.0358,
       "step": 4792
     },
     {
-      "epoch": 14.18,
-      "learning_rate": 5.463510848126233e-06,
-      "loss": 1.5414,
+      "epoch": 21.4,
+      "learning_rate": 5.720535714285714e-05,
+      "loss": 1.0425,
       "step": 4793
     },
     {
-      "epoch": 14.18,
-      "learning_rate": 5.443786982248521e-06,
-      "loss": 1.5132,
+      "epoch": 21.4,
+      "learning_rate": 5.7196428571428575e-05,
+      "loss": 1.0205,
       "step": 4794
     },
     {
-      "epoch": 14.19,
-      "learning_rate": 5.424063116370809e-06,
-      "loss": 1.4368,
+      "epoch": 21.41,
+      "learning_rate": 5.718750000000001e-05,
+      "loss": 0.9917,
       "step": 4795
     },
     {
-      "epoch": 14.19,
-      "learning_rate": 5.404339250493097e-06,
-      "loss": 1.5607,
+      "epoch": 21.41,
+      "learning_rate": 5.7178571428571426e-05,
+      "loss": 1.0617,
       "step": 4796
     },
     {
-      "epoch": 14.19,
-      "learning_rate": 5.3846153846153855e-06,
-      "loss": 1.5111,
+      "epoch": 21.42,
+      "learning_rate": 5.716964285714286e-05,
+      "loss": 1.093,
       "step": 4797
     },
     {
-      "epoch": 14.2,
-      "learning_rate": 5.3648915187376725e-06,
-      "loss": 1.4841,
+      "epoch": 21.42,
+      "learning_rate": 5.716071428571429e-05,
+      "loss": 0.9918,
       "step": 4798
     },
     {
-      "epoch": 14.2,
-      "learning_rate": 5.34516765285996e-06,
-      "loss": 1.5806,
+      "epoch": 21.42,
+      "learning_rate": 5.715178571428572e-05,
+      "loss": 0.9986,
       "step": 4799
     },
     {
-      "epoch": 14.2,
-      "learning_rate": 5.325443786982249e-06,
-      "loss": 1.5408,
+      "epoch": 21.43,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 1.0534,
       "step": 4800
     },
     {
-      "epoch": 14.2,
-      "learning_rate": 5.305719921104537e-06,
-      "loss": 1.4758,
+      "epoch": 21.43,
+      "eval_loss": 1.1253058910369873,
+      "eval_runtime": 44.9964,
+      "eval_samples_per_second": 291.001,
+      "eval_steps_per_second": 1.533,
+      "step": 4800
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 5.7133928571428574e-05,
+      "loss": 1.0576,
       "step": 4801
     },
     {
-      "epoch": 14.21,
-      "learning_rate": 5.285996055226825e-06,
-      "loss": 1.5556,
+      "epoch": 21.44,
+      "learning_rate": 5.7125000000000006e-05,
+      "loss": 1.0181,
       "step": 4802
     },
     {
-      "epoch": 14.21,
-      "learning_rate": 5.266272189349113e-06,
-      "loss": 1.5019,
+      "epoch": 21.44,
+      "learning_rate": 5.7116071428571425e-05,
+      "loss": 1.0864,
       "step": 4803
     },
     {
-      "epoch": 14.21,
-      "learning_rate": 5.246548323471401e-06,
-      "loss": 1.4918,
+      "epoch": 21.45,
+      "learning_rate": 5.710714285714286e-05,
+      "loss": 1.0932,
       "step": 4804
     },
     {
-      "epoch": 14.22,
-      "learning_rate": 5.2268244575936885e-06,
-      "loss": 1.5137,
+      "epoch": 21.45,
+      "learning_rate": 5.709821428571429e-05,
+      "loss": 1.0149,
       "step": 4805
     },
     {
-      "epoch": 14.22,
-      "learning_rate": 5.207100591715976e-06,
-      "loss": 1.5335,
+      "epoch": 21.46,
+      "learning_rate": 5.7089285714285715e-05,
+      "loss": 1.0443,
       "step": 4806
     },
     {
-      "epoch": 14.22,
-      "learning_rate": 5.187376725838264e-06,
-      "loss": 1.4084,
+      "epoch": 21.46,
+      "learning_rate": 5.708035714285715e-05,
+      "loss": 1.0592,
       "step": 4807
     },
     {
-      "epoch": 14.22,
-      "learning_rate": 5.167652859960553e-06,
-      "loss": 1.487,
+      "epoch": 21.46,
+      "learning_rate": 5.707142857142857e-05,
+      "loss": 1.0007,
       "step": 4808
     },
     {
-      "epoch": 14.23,
-      "learning_rate": 5.147928994082841e-06,
-      "loss": 1.6162,
+      "epoch": 21.47,
+      "learning_rate": 5.7062500000000005e-05,
+      "loss": 1.0324,
       "step": 4809
     },
     {
-      "epoch": 14.23,
-      "learning_rate": 5.128205128205128e-06,
-      "loss": 1.4465,
+      "epoch": 21.47,
+      "learning_rate": 5.705357142857143e-05,
+      "loss": 1.0645,
       "step": 4810
     },
     {
-      "epoch": 14.23,
-      "learning_rate": 5.108481262327417e-06,
-      "loss": 1.4373,
+      "epoch": 21.48,
+      "learning_rate": 5.704464285714286e-05,
+      "loss": 1.0128,
       "step": 4811
     },
     {
-      "epoch": 14.24,
-      "learning_rate": 5.0887573964497045e-06,
-      "loss": 1.4168,
+      "epoch": 21.48,
+      "learning_rate": 5.7035714285714295e-05,
+      "loss": 0.9511,
       "step": 4812
     },
     {
-      "epoch": 14.24,
-      "learning_rate": 5.069033530571992e-06,
-      "loss": 1.5013,
+      "epoch": 21.49,
+      "learning_rate": 5.702678571428571e-05,
+      "loss": 1.0244,
       "step": 4813
     },
     {
-      "epoch": 14.24,
-      "learning_rate": 5.04930966469428e-06,
-      "loss": 1.3873,
+      "epoch": 21.49,
+      "learning_rate": 5.7017857142857145e-05,
+      "loss": 0.9958,
       "step": 4814
     },
     {
-      "epoch": 14.25,
-      "learning_rate": 5.029585798816568e-06,
-      "loss": 1.4884,
+      "epoch": 21.5,
+      "learning_rate": 5.700892857142858e-05,
+      "loss": 1.165,
       "step": 4815
     },
     {
-      "epoch": 14.25,
-      "learning_rate": 5.009861932938856e-06,
-      "loss": 1.4456,
+      "epoch": 21.5,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.0459,
       "step": 4816
     },
     {
-      "epoch": 14.25,
-      "learning_rate": 4.990138067061144e-06,
-      "loss": 1.4467,
+      "epoch": 21.5,
+      "learning_rate": 5.699107142857143e-05,
+      "loss": 1.0807,
       "step": 4817
     },
     {
-      "epoch": 14.25,
-      "learning_rate": 4.970414201183432e-06,
-      "loss": 1.535,
+      "epoch": 21.51,
+      "learning_rate": 5.698214285714286e-05,
+      "loss": 1.1182,
       "step": 4818
     },
     {
-      "epoch": 14.26,
-      "learning_rate": 4.9506903353057205e-06,
-      "loss": 1.4916,
+      "epoch": 21.51,
+      "learning_rate": 5.697321428571429e-05,
+      "loss": 1.0465,
       "step": 4819
     },
     {
-      "epoch": 14.26,
-      "learning_rate": 4.930966469428008e-06,
-      "loss": 1.4436,
+      "epoch": 21.52,
+      "learning_rate": 5.696428571428571e-05,
+      "loss": 1.1208,
       "step": 4820
     },
     {
-      "epoch": 14.26,
-      "learning_rate": 4.9112426035502954e-06,
-      "loss": 1.4457,
+      "epoch": 21.52,
+      "learning_rate": 5.6955357142857144e-05,
+      "loss": 1.0891,
       "step": 4821
     },
     {
-      "epoch": 14.27,
-      "learning_rate": 4.891518737672584e-06,
-      "loss": 1.4539,
+      "epoch": 21.53,
+      "learning_rate": 5.6946428571428576e-05,
+      "loss": 1.0036,
       "step": 4822
     },
     {
-      "epoch": 14.27,
-      "learning_rate": 4.871794871794872e-06,
-      "loss": 1.4705,
+      "epoch": 21.53,
+      "learning_rate": 5.69375e-05,
+      "loss": 1.0245,
       "step": 4823
     },
     {
-      "epoch": 14.27,
-      "learning_rate": 4.85207100591716e-06,
-      "loss": 1.4756,
+      "epoch": 21.54,
+      "learning_rate": 5.6928571428571434e-05,
+      "loss": 1.0753,
       "step": 4824
     },
     {
-      "epoch": 14.28,
-      "learning_rate": 4.832347140039448e-06,
-      "loss": 1.4614,
+      "epoch": 21.54,
+      "learning_rate": 5.6919642857142866e-05,
+      "loss": 1.1147,
       "step": 4825
     },
     {
-      "epoch": 14.28,
-      "learning_rate": 4.812623274161736e-06,
-      "loss": 1.4738,
+      "epoch": 21.54,
+      "learning_rate": 5.6910714285714285e-05,
+      "loss": 1.0327,
       "step": 4826
     },
     {
-      "epoch": 14.28,
-      "learning_rate": 4.7928994082840236e-06,
-      "loss": 1.5051,
+      "epoch": 21.55,
+      "learning_rate": 5.690178571428572e-05,
+      "loss": 1.1557,
       "step": 4827
     },
     {
-      "epoch": 14.28,
-      "learning_rate": 4.773175542406312e-06,
-      "loss": 1.3982,
+      "epoch": 21.55,
+      "learning_rate": 5.689285714285715e-05,
+      "loss": 1.0282,
       "step": 4828
     },
     {
-      "epoch": 14.29,
-      "learning_rate": 4.753451676528599e-06,
-      "loss": 1.5996,
+      "epoch": 21.56,
+      "learning_rate": 5.688392857142858e-05,
+      "loss": 1.068,
       "step": 4829
     },
     {
-      "epoch": 14.29,
-      "learning_rate": 4.733727810650888e-06,
-      "loss": 1.419,
+      "epoch": 21.56,
+      "learning_rate": 5.6875e-05,
+      "loss": 1.0756,
       "step": 4830
     },
     {
-      "epoch": 14.29,
-      "learning_rate": 4.714003944773176e-06,
-      "loss": 1.6281,
+      "epoch": 21.57,
+      "learning_rate": 5.686607142857143e-05,
+      "loss": 1.0608,
       "step": 4831
     },
     {
-      "epoch": 14.3,
-      "learning_rate": 4.694280078895464e-06,
-      "loss": 1.5096,
+      "epoch": 21.57,
+      "learning_rate": 5.6857142857142865e-05,
+      "loss": 1.1118,
       "step": 4832
     },
     {
-      "epoch": 14.3,
-      "learning_rate": 4.674556213017752e-06,
-      "loss": 1.4688,
+      "epoch": 21.58,
+      "learning_rate": 5.684821428571428e-05,
+      "loss": 1.0883,
       "step": 4833
     },
     {
-      "epoch": 14.3,
-      "learning_rate": 4.6548323471400395e-06,
-      "loss": 1.4721,
+      "epoch": 21.58,
+      "learning_rate": 5.6839285714285715e-05,
+      "loss": 1.0117,
       "step": 4834
     },
     {
-      "epoch": 14.3,
-      "learning_rate": 4.6351084812623274e-06,
-      "loss": 1.5087,
+      "epoch": 21.58,
+      "learning_rate": 5.683035714285715e-05,
+      "loss": 1.0595,
       "step": 4835
     },
     {
-      "epoch": 14.31,
-      "learning_rate": 4.615384615384616e-06,
-      "loss": 1.4755,
+      "epoch": 21.59,
+      "learning_rate": 5.682142857142857e-05,
+      "loss": 1.0937,
       "step": 4836
     },
     {
-      "epoch": 14.31,
-      "learning_rate": 4.595660749506903e-06,
-      "loss": 1.4728,
+      "epoch": 21.59,
+      "learning_rate": 5.68125e-05,
+      "loss": 1.0465,
       "step": 4837
     },
     {
-      "epoch": 14.31,
-      "learning_rate": 4.575936883629191e-06,
-      "loss": 1.5102,
+      "epoch": 21.6,
+      "learning_rate": 5.680357142857143e-05,
+      "loss": 1.0607,
       "step": 4838
     },
     {
-      "epoch": 14.32,
-      "learning_rate": 4.55621301775148e-06,
-      "loss": 1.6027,
+      "epoch": 21.6,
+      "learning_rate": 5.6794642857142856e-05,
+      "loss": 1.0404,
       "step": 4839
     },
     {
-      "epoch": 14.32,
-      "learning_rate": 4.536489151873768e-06,
-      "loss": 1.4724,
+      "epoch": 21.61,
+      "learning_rate": 5.678571428571429e-05,
+      "loss": 1.0778,
       "step": 4840
     },
     {
-      "epoch": 14.32,
-      "learning_rate": 4.5167652859960555e-06,
-      "loss": 1.5002,
+      "epoch": 21.61,
+      "learning_rate": 5.677678571428572e-05,
+      "loss": 1.0709,
       "step": 4841
     },
     {
-      "epoch": 14.33,
-      "learning_rate": 4.497041420118343e-06,
-      "loss": 1.4557,
+      "epoch": 21.62,
+      "learning_rate": 5.676785714285715e-05,
+      "loss": 1.0694,
       "step": 4842
     },
     {
-      "epoch": 14.33,
-      "learning_rate": 4.477317554240631e-06,
-      "loss": 1.5635,
+      "epoch": 21.62,
+      "learning_rate": 5.675892857142857e-05,
+      "loss": 0.9792,
       "step": 4843
     },
     {
-      "epoch": 14.33,
-      "learning_rate": 4.45759368836292e-06,
-      "loss": 1.46,
+      "epoch": 21.62,
+      "learning_rate": 5.6750000000000004e-05,
+      "loss": 1.1097,
       "step": 4844
     },
     {
-      "epoch": 14.33,
-      "learning_rate": 4.437869822485207e-06,
-      "loss": 1.5329,
+      "epoch": 21.63,
+      "learning_rate": 5.6741071428571436e-05,
+      "loss": 1.1865,
       "step": 4845
     },
     {
-      "epoch": 14.34,
-      "learning_rate": 4.418145956607495e-06,
-      "loss": 1.517,
+      "epoch": 21.63,
+      "learning_rate": 5.6732142857142855e-05,
+      "loss": 1.1138,
       "step": 4846
     },
     {
-      "epoch": 14.34,
-      "learning_rate": 4.398422090729784e-06,
-      "loss": 1.5802,
+      "epoch": 21.64,
+      "learning_rate": 5.672321428571429e-05,
+      "loss": 1.0653,
       "step": 4847
     },
     {
-      "epoch": 14.34,
-      "learning_rate": 4.3786982248520715e-06,
-      "loss": 1.4897,
+      "epoch": 21.64,
+      "learning_rate": 5.671428571428572e-05,
+      "loss": 1.0358,
       "step": 4848
     },
     {
-      "epoch": 14.35,
-      "learning_rate": 4.3589743589743586e-06,
-      "loss": 1.5879,
+      "epoch": 21.65,
+      "learning_rate": 5.670535714285714e-05,
+      "loss": 1.0404,
       "step": 4849
     },
     {
-      "epoch": 14.35,
-      "learning_rate": 4.339250493096647e-06,
-      "loss": 1.4931,
+      "epoch": 21.65,
+      "learning_rate": 5.669642857142857e-05,
+      "loss": 1.0488,
       "step": 4850
     },
     {
-      "epoch": 14.35,
-      "learning_rate": 4.319526627218935e-06,
-      "loss": 1.5616,
+      "epoch": 21.66,
+      "learning_rate": 5.66875e-05,
+      "loss": 1.0892,
       "step": 4851
     },
     {
-      "epoch": 14.36,
-      "learning_rate": 4.299802761341223e-06,
-      "loss": 1.509,
+      "epoch": 21.66,
+      "learning_rate": 5.6678571428571435e-05,
+      "loss": 1.0932,
       "step": 4852
     },
     {
-      "epoch": 14.36,
-      "learning_rate": 4.280078895463511e-06,
-      "loss": 1.5095,
+      "epoch": 21.67,
+      "learning_rate": 5.666964285714286e-05,
+      "loss": 1.1439,
       "step": 4853
     },
     {
-      "epoch": 14.36,
-      "learning_rate": 4.260355029585799e-06,
-      "loss": 1.4551,
+      "epoch": 21.67,
+      "learning_rate": 5.666071428571429e-05,
+      "loss": 1.082,
       "step": 4854
     },
     {
-      "epoch": 14.36,
-      "learning_rate": 4.2406311637080875e-06,
-      "loss": 1.4874,
+      "epoch": 21.67,
+      "learning_rate": 5.6651785714285724e-05,
+      "loss": 1.0475,
       "step": 4855
     },
     {
-      "epoch": 14.37,
-      "learning_rate": 4.220907297830375e-06,
-      "loss": 1.5681,
+      "epoch": 21.68,
+      "learning_rate": 5.664285714285714e-05,
+      "loss": 1.0371,
       "step": 4856
     },
     {
-      "epoch": 14.37,
-      "learning_rate": 4.2011834319526624e-06,
-      "loss": 1.5489,
+      "epoch": 21.68,
+      "learning_rate": 5.6633928571428575e-05,
+      "loss": 0.9673,
       "step": 4857
     },
     {
-      "epoch": 14.37,
-      "learning_rate": 4.181459566074951e-06,
-      "loss": 1.464,
+      "epoch": 21.69,
+      "learning_rate": 5.662500000000001e-05,
+      "loss": 1.0619,
       "step": 4858
     },
     {
-      "epoch": 14.38,
-      "learning_rate": 4.161735700197239e-06,
-      "loss": 1.5587,
+      "epoch": 21.69,
+      "learning_rate": 5.6616071428571426e-05,
+      "loss": 1.0445,
       "step": 4859
     },
     {
-      "epoch": 14.38,
-      "learning_rate": 4.142011834319527e-06,
-      "loss": 1.4997,
+      "epoch": 21.7,
+      "learning_rate": 5.660714285714286e-05,
+      "loss": 1.0834,
       "step": 4860
     },
     {
-      "epoch": 14.38,
-      "learning_rate": 4.122287968441815e-06,
-      "loss": 1.5652,
+      "epoch": 21.7,
+      "learning_rate": 5.659821428571429e-05,
+      "loss": 1.0055,
       "step": 4861
     },
     {
-      "epoch": 14.38,
-      "learning_rate": 4.102564102564103e-06,
-      "loss": 1.5012,
+      "epoch": 21.71,
+      "learning_rate": 5.658928571428572e-05,
+      "loss": 1.079,
       "step": 4862
     },
     {
-      "epoch": 14.39,
-      "learning_rate": 4.0828402366863906e-06,
-      "loss": 1.5211,
+      "epoch": 21.71,
+      "learning_rate": 5.658035714285714e-05,
+      "loss": 1.0379,
       "step": 4863
     },
     {
-      "epoch": 14.39,
-      "learning_rate": 4.0631163708086784e-06,
-      "loss": 1.5765,
+      "epoch": 21.71,
+      "learning_rate": 5.6571428571428574e-05,
+      "loss": 1.0195,
       "step": 4864
     },
     {
-      "epoch": 14.39,
-      "learning_rate": 4.043392504930966e-06,
-      "loss": 1.5418,
+      "epoch": 21.72,
+      "learning_rate": 5.6562500000000006e-05,
+      "loss": 1.1171,
       "step": 4865
     },
     {
-      "epoch": 14.4,
-      "learning_rate": 4.023668639053255e-06,
-      "loss": 1.421,
+      "epoch": 21.72,
+      "learning_rate": 5.6553571428571425e-05,
+      "loss": 1.0471,
       "step": 4866
     },
     {
-      "epoch": 14.4,
-      "learning_rate": 4.003944773175543e-06,
-      "loss": 1.5385,
+      "epoch": 21.73,
+      "learning_rate": 5.654464285714286e-05,
+      "loss": 1.0569,
       "step": 4867
     },
     {
-      "epoch": 14.4,
-      "learning_rate": 3.98422090729783e-06,
-      "loss": 1.5389,
+      "epoch": 21.73,
+      "learning_rate": 5.653571428571429e-05,
+      "loss": 1.0641,
       "step": 4868
     },
     {
-      "epoch": 14.41,
-      "learning_rate": 3.964497041420119e-06,
-      "loss": 1.4719,
+      "epoch": 21.74,
+      "learning_rate": 5.6526785714285715e-05,
+      "loss": 1.0532,
       "step": 4869
     },
     {
-      "epoch": 14.41,
-      "learning_rate": 3.9447731755424066e-06,
-      "loss": 1.3723,
+      "epoch": 21.74,
+      "learning_rate": 5.651785714285715e-05,
+      "loss": 1.1075,
       "step": 4870
     },
     {
-      "epoch": 14.41,
-      "learning_rate": 3.9250493096646944e-06,
-      "loss": 1.5203,
+      "epoch": 21.75,
+      "learning_rate": 5.650892857142858e-05,
+      "loss": 1.0935,
       "step": 4871
     },
     {
-      "epoch": 14.41,
-      "learning_rate": 3.905325443786982e-06,
-      "loss": 1.5167,
+      "epoch": 21.75,
+      "learning_rate": 5.65e-05,
+      "loss": 0.9839,
       "step": 4872
     },
     {
-      "epoch": 14.42,
-      "learning_rate": 3.88560157790927e-06,
-      "loss": 1.4802,
+      "epoch": 21.75,
+      "learning_rate": 5.649107142857143e-05,
+      "loss": 1.0259,
       "step": 4873
     },
     {
-      "epoch": 14.42,
-      "learning_rate": 3.865877712031558e-06,
-      "loss": 1.5624,
+      "epoch": 21.76,
+      "learning_rate": 5.648214285714286e-05,
+      "loss": 1.0258,
       "step": 4874
     },
     {
-      "epoch": 14.42,
-      "learning_rate": 3.846153846153847e-06,
-      "loss": 1.4021,
+      "epoch": 21.76,
+      "learning_rate": 5.6473214285714294e-05,
+      "loss": 1.0597,
       "step": 4875
     },
     {
-      "epoch": 14.43,
-      "learning_rate": 3.826429980276134e-06,
-      "loss": 1.5516,
+      "epoch": 21.77,
+      "learning_rate": 5.646428571428571e-05,
+      "loss": 1.0706,
       "step": 4876
     },
     {
-      "epoch": 14.43,
-      "learning_rate": 3.806706114398422e-06,
-      "loss": 1.5251,
+      "epoch": 21.77,
+      "learning_rate": 5.6455357142857145e-05,
+      "loss": 1.1524,
       "step": 4877
     },
     {
-      "epoch": 14.43,
-      "learning_rate": 3.7869822485207104e-06,
-      "loss": 1.5286,
+      "epoch": 21.78,
+      "learning_rate": 5.644642857142858e-05,
+      "loss": 1.0448,
       "step": 4878
     },
     {
-      "epoch": 14.43,
-      "learning_rate": 3.7672583826429983e-06,
-      "loss": 1.5264,
+      "epoch": 21.78,
+      "learning_rate": 5.6437499999999996e-05,
+      "loss": 1.0945,
       "step": 4879
     },
     {
-      "epoch": 14.44,
-      "learning_rate": 3.7475345167652858e-06,
-      "loss": 1.5508,
+      "epoch": 21.79,
+      "learning_rate": 5.642857142857143e-05,
+      "loss": 0.9857,
       "step": 4880
     },
     {
-      "epoch": 14.44,
-      "learning_rate": 3.727810650887574e-06,
-      "loss": 1.5248,
+      "epoch": 21.79,
+      "learning_rate": 5.641964285714286e-05,
+      "loss": 1.053,
       "step": 4881
     },
     {
-      "epoch": 14.44,
-      "learning_rate": 3.7080867850098624e-06,
-      "loss": 1.4757,
+      "epoch": 21.79,
+      "learning_rate": 5.6410714285714286e-05,
+      "loss": 1.1624,
       "step": 4882
     },
     {
-      "epoch": 14.45,
-      "learning_rate": 3.6883629191321502e-06,
-      "loss": 1.5019,
+      "epoch": 21.8,
+      "learning_rate": 5.640178571428572e-05,
+      "loss": 1.1036,
       "step": 4883
     },
     {
-      "epoch": 14.45,
-      "learning_rate": 3.6686390532544377e-06,
-      "loss": 1.5014,
+      "epoch": 21.8,
+      "learning_rate": 5.639285714285715e-05,
+      "loss": 1.0688,
       "step": 4884
     },
     {
-      "epoch": 14.45,
-      "learning_rate": 3.648915187376726e-06,
-      "loss": 1.4407,
+      "epoch": 21.81,
+      "learning_rate": 5.6383928571428576e-05,
+      "loss": 1.0741,
       "step": 4885
     },
     {
-      "epoch": 14.46,
-      "learning_rate": 3.629191321499014e-06,
-      "loss": 1.4362,
+      "epoch": 21.81,
+      "learning_rate": 5.6375e-05,
+      "loss": 1.0888,
       "step": 4886
     },
     {
-      "epoch": 14.46,
-      "learning_rate": 3.609467455621302e-06,
-      "loss": 1.4353,
+      "epoch": 21.82,
+      "learning_rate": 5.6366071428571434e-05,
+      "loss": 1.0256,
       "step": 4887
     },
     {
-      "epoch": 14.46,
-      "learning_rate": 3.5897435897435896e-06,
-      "loss": 1.5225,
+      "epoch": 21.82,
+      "learning_rate": 5.6357142857142866e-05,
+      "loss": 1.1027,
       "step": 4888
     },
     {
-      "epoch": 14.46,
-      "learning_rate": 3.570019723865878e-06,
-      "loss": 1.4921,
+      "epoch": 21.83,
+      "learning_rate": 5.6348214285714285e-05,
+      "loss": 1.1212,
       "step": 4889
     },
     {
-      "epoch": 14.47,
-      "learning_rate": 3.550295857988166e-06,
-      "loss": 1.5012,
+      "epoch": 21.83,
+      "learning_rate": 5.633928571428572e-05,
+      "loss": 1.1035,
       "step": 4890
     },
     {
-      "epoch": 14.47,
-      "learning_rate": 3.530571992110454e-06,
-      "loss": 1.4543,
+      "epoch": 21.83,
+      "learning_rate": 5.633035714285715e-05,
+      "loss": 1.1693,
       "step": 4891
     },
     {
-      "epoch": 14.47,
-      "learning_rate": 3.5108481262327416e-06,
-      "loss": 1.481,
+      "epoch": 21.84,
+      "learning_rate": 5.632142857142857e-05,
+      "loss": 1.0229,
       "step": 4892
     },
     {
-      "epoch": 14.48,
-      "learning_rate": 3.49112426035503e-06,
-      "loss": 1.5091,
+      "epoch": 21.84,
+      "learning_rate": 5.63125e-05,
+      "loss": 1.0946,
       "step": 4893
     },
     {
-      "epoch": 14.48,
-      "learning_rate": 3.4714003944773178e-06,
-      "loss": 1.4999,
+      "epoch": 21.85,
+      "learning_rate": 5.630357142857143e-05,
+      "loss": 1.0941,
       "step": 4894
     },
     {
-      "epoch": 14.48,
-      "learning_rate": 3.451676528599606e-06,
-      "loss": 1.3969,
+      "epoch": 21.85,
+      "learning_rate": 5.6294642857142864e-05,
+      "loss": 1.064,
       "step": 4895
     },
     {
-      "epoch": 14.49,
-      "learning_rate": 3.4319526627218935e-06,
-      "loss": 1.3679,
+      "epoch": 21.86,
+      "learning_rate": 5.628571428571428e-05,
+      "loss": 1.0601,
       "step": 4896
     },
     {
-      "epoch": 14.49,
-      "learning_rate": 3.4122287968441814e-06,
-      "loss": 1.4016,
+      "epoch": 21.86,
+      "learning_rate": 5.6276785714285715e-05,
+      "loss": 1.0791,
       "step": 4897
     },
     {
-      "epoch": 14.49,
-      "learning_rate": 3.3925049309664697e-06,
-      "loss": 1.476,
+      "epoch": 21.87,
+      "learning_rate": 5.626785714285715e-05,
+      "loss": 1.0691,
       "step": 4898
     },
     {
-      "epoch": 14.49,
-      "learning_rate": 3.372781065088758e-06,
-      "loss": 1.4645,
+      "epoch": 21.87,
+      "learning_rate": 5.625892857142857e-05,
+      "loss": 1.101,
       "step": 4899
     },
     {
-      "epoch": 14.5,
-      "learning_rate": 3.3530571992110454e-06,
-      "loss": 1.5035,
+      "epoch": 21.88,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 1.0802,
       "step": 4900
     },
     {
-      "epoch": 14.5,
-      "learning_rate": 3.3333333333333333e-06,
-      "loss": 1.452,
+      "epoch": 21.88,
+      "learning_rate": 5.624107142857144e-05,
+      "loss": 1.1201,
       "step": 4901
     },
     {
-      "epoch": 14.5,
-      "learning_rate": 3.3136094674556216e-06,
-      "loss": 1.4936,
+      "epoch": 21.88,
+      "learning_rate": 5.6232142857142856e-05,
+      "loss": 0.998,
       "step": 4902
     },
     {
-      "epoch": 14.51,
-      "learning_rate": 3.29388560157791e-06,
-      "loss": 1.4109,
+      "epoch": 21.89,
+      "learning_rate": 5.622321428571429e-05,
+      "loss": 1.0764,
       "step": 4903
     },
     {
-      "epoch": 14.51,
-      "learning_rate": 3.2741617357001974e-06,
-      "loss": 1.4575,
+      "epoch": 21.89,
+      "learning_rate": 5.621428571428572e-05,
+      "loss": 1.1051,
       "step": 4904
     },
     {
-      "epoch": 14.51,
-      "learning_rate": 3.2544378698224853e-06,
-      "loss": 1.4934,
+      "epoch": 21.9,
+      "learning_rate": 5.620535714285714e-05,
+      "loss": 1.0258,
       "step": 4905
     },
     {
-      "epoch": 14.51,
-      "learning_rate": 3.2347140039447736e-06,
-      "loss": 1.6156,
+      "epoch": 21.9,
+      "learning_rate": 5.619642857142857e-05,
+      "loss": 1.0591,
       "step": 4906
     },
     {
-      "epoch": 14.52,
-      "learning_rate": 3.2149901380670614e-06,
-      "loss": 1.4454,
+      "epoch": 21.91,
+      "learning_rate": 5.6187500000000004e-05,
+      "loss": 0.9998,
       "step": 4907
     },
     {
-      "epoch": 14.52,
-      "learning_rate": 3.195266272189349e-06,
-      "loss": 1.5062,
+      "epoch": 21.91,
+      "learning_rate": 5.6178571428571436e-05,
+      "loss": 0.9759,
       "step": 4908
     },
     {
-      "epoch": 14.52,
-      "learning_rate": 3.175542406311637e-06,
-      "loss": 1.4824,
+      "epoch": 21.92,
+      "learning_rate": 5.6169642857142855e-05,
+      "loss": 1.0082,
       "step": 4909
     },
     {
-      "epoch": 14.53,
-      "learning_rate": 3.1558185404339255e-06,
-      "loss": 1.5364,
+      "epoch": 21.92,
+      "learning_rate": 5.616071428571429e-05,
+      "loss": 1.019,
       "step": 4910
     },
     {
-      "epoch": 14.53,
-      "learning_rate": 3.1360946745562134e-06,
-      "loss": 1.4698,
+      "epoch": 21.92,
+      "learning_rate": 5.615178571428572e-05,
+      "loss": 1.0243,
       "step": 4911
     },
     {
-      "epoch": 14.53,
-      "learning_rate": 3.1163708086785013e-06,
-      "loss": 1.4314,
+      "epoch": 21.93,
+      "learning_rate": 5.6142857142857145e-05,
+      "loss": 0.9858,
       "step": 4912
     },
     {
-      "epoch": 14.54,
-      "learning_rate": 3.096646942800789e-06,
-      "loss": 1.4891,
+      "epoch": 21.93,
+      "learning_rate": 5.613392857142858e-05,
+      "loss": 1.0852,
       "step": 4913
     },
     {
-      "epoch": 14.54,
-      "learning_rate": 3.0769230769230774e-06,
-      "loss": 1.4364,
+      "epoch": 21.94,
+      "learning_rate": 5.6125e-05,
+      "loss": 1.0474,
       "step": 4914
     },
     {
-      "epoch": 14.54,
-      "learning_rate": 3.057199211045365e-06,
-      "loss": 1.3672,
+      "epoch": 21.94,
+      "learning_rate": 5.611607142857143e-05,
+      "loss": 1.043,
       "step": 4915
     },
     {
-      "epoch": 14.54,
-      "learning_rate": 3.0374753451676528e-06,
-      "loss": 1.544,
+      "epoch": 21.95,
+      "learning_rate": 5.610714285714286e-05,
+      "loss": 1.0483,
       "step": 4916
     },
     {
-      "epoch": 14.55,
-      "learning_rate": 3.017751479289941e-06,
-      "loss": 1.6116,
+      "epoch": 21.95,
+      "learning_rate": 5.609821428571429e-05,
+      "loss": 1.0242,
       "step": 4917
     },
     {
-      "epoch": 14.55,
-      "learning_rate": 2.998027613412229e-06,
-      "loss": 1.5827,
+      "epoch": 21.96,
+      "learning_rate": 5.6089285714285724e-05,
+      "loss": 1.0066,
       "step": 4918
     },
     {
-      "epoch": 14.55,
-      "learning_rate": 2.978303747534517e-06,
-      "loss": 1.5072,
+      "epoch": 21.96,
+      "learning_rate": 5.608035714285714e-05,
+      "loss": 1.027,
       "step": 4919
     },
     {
-      "epoch": 14.56,
-      "learning_rate": 2.9585798816568047e-06,
-      "loss": 1.5127,
+      "epoch": 21.96,
+      "learning_rate": 5.6071428571428575e-05,
+      "loss": 1.0451,
       "step": 4920
     },
     {
-      "epoch": 14.56,
-      "learning_rate": 2.938856015779093e-06,
-      "loss": 1.5171,
+      "epoch": 21.97,
+      "learning_rate": 5.606250000000001e-05,
+      "loss": 1.1106,
       "step": 4921
     },
     {
-      "epoch": 14.56,
-      "learning_rate": 2.9191321499013805e-06,
-      "loss": 1.54,
+      "epoch": 21.97,
+      "learning_rate": 5.6053571428571426e-05,
+      "loss": 1.0799,
       "step": 4922
     },
     {
-      "epoch": 14.57,
-      "learning_rate": 2.8994082840236688e-06,
-      "loss": 1.5738,
+      "epoch": 21.98,
+      "learning_rate": 5.604464285714286e-05,
+      "loss": 1.0363,
       "step": 4923
     },
     {
-      "epoch": 14.57,
-      "learning_rate": 2.8796844181459566e-06,
-      "loss": 1.4842,
+      "epoch": 21.98,
+      "learning_rate": 5.603571428571429e-05,
+      "loss": 1.1509,
       "step": 4924
     },
     {
-      "epoch": 14.57,
-      "learning_rate": 2.859960552268245e-06,
-      "loss": 1.4526,
+      "epoch": 21.99,
+      "learning_rate": 5.602678571428571e-05,
+      "loss": 1.017,
       "step": 4925
     },
     {
-      "epoch": 14.57,
-      "learning_rate": 2.8402366863905324e-06,
-      "loss": 1.5009,
+      "epoch": 21.99,
+      "learning_rate": 5.601785714285714e-05,
+      "loss": 1.1303,
       "step": 4926
     },
     {
-      "epoch": 14.58,
-      "learning_rate": 2.8205128205128207e-06,
-      "loss": 1.5843,
+      "epoch": 22.0,
+      "learning_rate": 5.6008928571428574e-05,
+      "loss": 1.0987,
       "step": 4927
     },
     {
-      "epoch": 14.58,
-      "learning_rate": 2.8007889546351086e-06,
-      "loss": 1.5028,
+      "epoch": 22.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.974,
       "step": 4928
     },
     {
-      "epoch": 14.58,
-      "learning_rate": 2.7810650887573965e-06,
-      "loss": 1.5506,
+      "epoch": 22.0,
+      "learning_rate": 5.599107142857143e-05,
+      "loss": 0.9759,
       "step": 4929
     },
     {
-      "epoch": 14.59,
-      "learning_rate": 2.7613412228796843e-06,
-      "loss": 1.511,
+      "epoch": 22.01,
+      "learning_rate": 5.5982142857142864e-05,
+      "loss": 1.0248,
       "step": 4930
     },
     {
-      "epoch": 14.59,
-      "learning_rate": 2.7416173570019726e-06,
-      "loss": 1.4684,
+      "epoch": 22.01,
+      "learning_rate": 5.5973214285714296e-05,
+      "loss": 1.081,
       "step": 4931
     },
     {
-      "epoch": 14.59,
-      "learning_rate": 2.7218934911242605e-06,
-      "loss": 1.4829,
+      "epoch": 22.02,
+      "learning_rate": 5.5964285714285715e-05,
+      "loss": 1.0183,
       "step": 4932
     },
     {
-      "epoch": 14.59,
-      "learning_rate": 2.7021696252465484e-06,
-      "loss": 1.4769,
+      "epoch": 22.02,
+      "learning_rate": 5.595535714285715e-05,
+      "loss": 1.0051,
       "step": 4933
     },
     {
-      "epoch": 14.6,
-      "learning_rate": 2.6824457593688363e-06,
-      "loss": 1.4692,
+      "epoch": 22.03,
+      "learning_rate": 5.594642857142858e-05,
+      "loss": 0.9773,
       "step": 4934
     },
     {
-      "epoch": 14.6,
-      "learning_rate": 2.6627218934911246e-06,
-      "loss": 1.4715,
+      "epoch": 22.03,
+      "learning_rate": 5.59375e-05,
+      "loss": 1.0196,
       "step": 4935
     },
     {
-      "epoch": 14.6,
-      "learning_rate": 2.6429980276134125e-06,
-      "loss": 1.4625,
+      "epoch": 22.04,
+      "learning_rate": 5.592857142857143e-05,
+      "loss": 1.0007,
       "step": 4936
     },
     {
-      "epoch": 14.61,
-      "learning_rate": 2.6232741617357003e-06,
-      "loss": 1.4575,
+      "epoch": 22.04,
+      "learning_rate": 5.591964285714286e-05,
+      "loss": 1.0057,
       "step": 4937
     },
     {
-      "epoch": 14.61,
-      "learning_rate": 2.603550295857988e-06,
-      "loss": 1.4557,
+      "epoch": 22.04,
+      "learning_rate": 5.591071428571428e-05,
+      "loss": 1.0031,
       "step": 4938
     },
     {
-      "epoch": 14.61,
-      "learning_rate": 2.5838264299802765e-06,
-      "loss": 1.431,
+      "epoch": 22.05,
+      "learning_rate": 5.590178571428571e-05,
+      "loss": 1.135,
       "step": 4939
     },
     {
-      "epoch": 14.62,
-      "learning_rate": 2.564102564102564e-06,
-      "loss": 1.4767,
+      "epoch": 22.05,
+      "learning_rate": 5.5892857142857145e-05,
+      "loss": 0.9817,
       "step": 4940
     },
     {
-      "epoch": 14.62,
-      "learning_rate": 2.5443786982248523e-06,
-      "loss": 1.564,
+      "epoch": 22.06,
+      "learning_rate": 5.588392857142858e-05,
+      "loss": 1.0195,
       "step": 4941
     },
     {
-      "epoch": 14.62,
-      "learning_rate": 2.52465483234714e-06,
-      "loss": 1.4597,
+      "epoch": 22.06,
+      "learning_rate": 5.5875e-05,
+      "loss": 1.0178,
       "step": 4942
     },
     {
-      "epoch": 14.62,
-      "learning_rate": 2.504930966469428e-06,
-      "loss": 1.5002,
+      "epoch": 22.07,
+      "learning_rate": 5.586607142857143e-05,
+      "loss": 1.0889,
       "step": 4943
     },
     {
-      "epoch": 14.63,
-      "learning_rate": 2.485207100591716e-06,
-      "loss": 1.527,
+      "epoch": 22.07,
+      "learning_rate": 5.585714285714286e-05,
+      "loss": 1.1304,
       "step": 4944
     },
     {
-      "epoch": 14.63,
-      "learning_rate": 2.465483234714004e-06,
-      "loss": 1.5678,
+      "epoch": 22.08,
+      "learning_rate": 5.5848214285714286e-05,
+      "loss": 1.0375,
       "step": 4945
     },
     {
-      "epoch": 14.63,
-      "learning_rate": 2.445759368836292e-06,
-      "loss": 1.5493,
+      "epoch": 22.08,
+      "learning_rate": 5.583928571428572e-05,
+      "loss": 1.0038,
       "step": 4946
     },
     {
-      "epoch": 14.64,
-      "learning_rate": 2.42603550295858e-06,
-      "loss": 1.4934,
+      "epoch": 22.08,
+      "learning_rate": 5.583035714285715e-05,
+      "loss": 1.0451,
       "step": 4947
     },
     {
-      "epoch": 14.64,
-      "learning_rate": 2.406311637080868e-06,
-      "loss": 1.5408,
+      "epoch": 22.09,
+      "learning_rate": 5.582142857142857e-05,
+      "loss": 1.1091,
       "step": 4948
     },
     {
-      "epoch": 14.64,
-      "learning_rate": 2.386587771203156e-06,
-      "loss": 1.6009,
+      "epoch": 22.09,
+      "learning_rate": 5.58125e-05,
+      "loss": 1.0596,
       "step": 4949
     },
     {
-      "epoch": 14.64,
-      "learning_rate": 2.366863905325444e-06,
-      "loss": 1.5665,
+      "epoch": 22.1,
+      "learning_rate": 5.5803571428571434e-05,
+      "loss": 1.0515,
       "step": 4950
     },
     {
-      "epoch": 14.65,
-      "learning_rate": 2.347140039447732e-06,
-      "loss": 1.5641,
+      "epoch": 22.1,
+      "learning_rate": 5.5794642857142866e-05,
+      "loss": 1.0379,
       "step": 4951
     },
     {
-      "epoch": 14.65,
-      "learning_rate": 2.3274161735700198e-06,
-      "loss": 1.4012,
+      "epoch": 22.11,
+      "learning_rate": 5.5785714285714285e-05,
+      "loss": 1.0386,
       "step": 4952
     },
     {
-      "epoch": 14.65,
-      "learning_rate": 2.307692307692308e-06,
-      "loss": 1.4796,
+      "epoch": 22.11,
+      "learning_rate": 5.577678571428572e-05,
+      "loss": 1.0645,
       "step": 4953
     },
     {
-      "epoch": 14.66,
-      "learning_rate": 2.2879684418145955e-06,
-      "loss": 1.5148,
+      "epoch": 22.12,
+      "learning_rate": 5.576785714285715e-05,
+      "loss": 1.0458,
       "step": 4954
     },
     {
-      "epoch": 14.66,
-      "learning_rate": 2.268244575936884e-06,
-      "loss": 1.6046,
+      "epoch": 22.12,
+      "learning_rate": 5.575892857142857e-05,
+      "loss": 1.039,
       "step": 4955
     },
     {
-      "epoch": 14.66,
-      "learning_rate": 2.2485207100591717e-06,
-      "loss": 1.5349,
+      "epoch": 22.12,
+      "learning_rate": 5.575e-05,
+      "loss": 1.0859,
       "step": 4956
     },
     {
-      "epoch": 14.67,
-      "learning_rate": 2.22879684418146e-06,
-      "loss": 1.6172,
+      "epoch": 22.13,
+      "learning_rate": 5.574107142857143e-05,
+      "loss": 1.0564,
       "step": 4957
     },
     {
-      "epoch": 14.67,
-      "learning_rate": 2.2090729783037475e-06,
-      "loss": 1.5022,
+      "epoch": 22.13,
+      "learning_rate": 5.573214285714286e-05,
+      "loss": 1.1287,
       "step": 4958
     },
     {
-      "epoch": 14.67,
-      "learning_rate": 2.1893491124260358e-06,
-      "loss": 1.4739,
+      "epoch": 22.14,
+      "learning_rate": 5.572321428571429e-05,
+      "loss": 1.0134,
       "step": 4959
     },
     {
-      "epoch": 14.67,
-      "learning_rate": 2.1696252465483236e-06,
-      "loss": 1.5705,
+      "epoch": 22.14,
+      "learning_rate": 5.571428571428572e-05,
+      "loss": 1.002,
       "step": 4960
     },
     {
-      "epoch": 14.68,
-      "learning_rate": 2.1499013806706115e-06,
-      "loss": 1.4976,
+      "epoch": 22.15,
+      "learning_rate": 5.570535714285715e-05,
+      "loss": 1.0262,
       "step": 4961
     },
     {
-      "epoch": 14.68,
-      "learning_rate": 2.1301775147928994e-06,
-      "loss": 1.4873,
+      "epoch": 22.15,
+      "learning_rate": 5.569642857142857e-05,
+      "loss": 1.0733,
       "step": 4962
     },
     {
-      "epoch": 14.68,
-      "learning_rate": 2.1104536489151877e-06,
-      "loss": 1.5008,
+      "epoch": 22.16,
+      "learning_rate": 5.5687500000000005e-05,
+      "loss": 1.0276,
       "step": 4963
     },
     {
-      "epoch": 14.69,
-      "learning_rate": 2.0907297830374756e-06,
-      "loss": 1.5401,
+      "epoch": 22.16,
+      "learning_rate": 5.567857142857144e-05,
+      "loss": 1.0578,
       "step": 4964
     },
     {
-      "epoch": 14.69,
-      "learning_rate": 2.0710059171597635e-06,
-      "loss": 1.5599,
+      "epoch": 22.17,
+      "learning_rate": 5.5669642857142856e-05,
+      "loss": 1.0735,
       "step": 4965
     },
     {
-      "epoch": 14.69,
-      "learning_rate": 2.0512820512820513e-06,
-      "loss": 1.4222,
+      "epoch": 22.17,
+      "learning_rate": 5.566071428571429e-05,
+      "loss": 1.0381,
       "step": 4966
     },
     {
-      "epoch": 14.7,
-      "learning_rate": 2.0315581854043392e-06,
-      "loss": 1.5207,
+      "epoch": 22.17,
+      "learning_rate": 5.565178571428572e-05,
+      "loss": 1.1135,
       "step": 4967
     },
     {
-      "epoch": 14.7,
-      "learning_rate": 2.0118343195266275e-06,
-      "loss": 1.445,
+      "epoch": 22.18,
+      "learning_rate": 5.564285714285714e-05,
+      "loss": 1.0741,
       "step": 4968
     },
     {
-      "epoch": 14.7,
-      "learning_rate": 1.992110453648915e-06,
-      "loss": 1.5467,
+      "epoch": 22.18,
+      "learning_rate": 5.563392857142857e-05,
+      "loss": 1.0753,
       "step": 4969
     },
     {
-      "epoch": 14.7,
-      "learning_rate": 1.9723865877712033e-06,
-      "loss": 1.511,
+      "epoch": 22.19,
+      "learning_rate": 5.5625000000000004e-05,
+      "loss": 1.0987,
       "step": 4970
     },
     {
-      "epoch": 14.71,
-      "learning_rate": 1.952662721893491e-06,
-      "loss": 1.5844,
+      "epoch": 22.19,
+      "learning_rate": 5.561607142857143e-05,
+      "loss": 0.9627,
       "step": 4971
     },
     {
-      "epoch": 14.71,
-      "learning_rate": 1.932938856015779e-06,
-      "loss": 1.4944,
+      "epoch": 22.2,
+      "learning_rate": 5.5607142857142854e-05,
+      "loss": 1.0644,
       "step": 4972
     },
     {
-      "epoch": 14.71,
-      "learning_rate": 1.913214990138067e-06,
-      "loss": 1.5418,
+      "epoch": 22.2,
+      "learning_rate": 5.559821428571429e-05,
+      "loss": 1.0179,
       "step": 4973
     },
     {
-      "epoch": 14.72,
-      "learning_rate": 1.8934911242603552e-06,
-      "loss": 1.4706,
+      "epoch": 22.21,
+      "learning_rate": 5.558928571428572e-05,
+      "loss": 0.9514,
       "step": 4974
     },
     {
-      "epoch": 14.72,
-      "learning_rate": 1.8737672583826429e-06,
-      "loss": 1.4544,
+      "epoch": 22.21,
+      "learning_rate": 5.5580357142857144e-05,
+      "loss": 1.0392,
       "step": 4975
     },
     {
-      "epoch": 14.72,
-      "learning_rate": 1.8540433925049312e-06,
-      "loss": 1.4864,
+      "epoch": 22.21,
+      "learning_rate": 5.557142857142858e-05,
+      "loss": 1.1497,
       "step": 4976
     },
     {
-      "epoch": 14.72,
-      "learning_rate": 1.8343195266272188e-06,
-      "loss": 1.5465,
+      "epoch": 22.22,
+      "learning_rate": 5.556250000000001e-05,
+      "loss": 1.0652,
       "step": 4977
     },
     {
-      "epoch": 14.73,
-      "learning_rate": 1.814595660749507e-06,
-      "loss": 1.5149,
+      "epoch": 22.22,
+      "learning_rate": 5.555357142857143e-05,
+      "loss": 0.973,
       "step": 4978
     },
     {
-      "epoch": 14.73,
-      "learning_rate": 1.7948717948717948e-06,
-      "loss": 1.4297,
+      "epoch": 22.23,
+      "learning_rate": 5.554464285714286e-05,
+      "loss": 1.0443,
       "step": 4979
     },
     {
-      "epoch": 14.73,
-      "learning_rate": 1.775147928994083e-06,
-      "loss": 1.4647,
+      "epoch": 22.23,
+      "learning_rate": 5.553571428571429e-05,
+      "loss": 1.139,
       "step": 4980
     },
     {
-      "epoch": 14.74,
-      "learning_rate": 1.7554240631163708e-06,
-      "loss": 1.6061,
+      "epoch": 22.24,
+      "learning_rate": 5.552678571428571e-05,
+      "loss": 1.0373,
       "step": 4981
     },
     {
-      "epoch": 14.74,
-      "learning_rate": 1.7357001972386589e-06,
-      "loss": 1.4876,
+      "epoch": 22.24,
+      "learning_rate": 5.551785714285714e-05,
+      "loss": 1.0883,
       "step": 4982
     },
     {
-      "epoch": 14.74,
-      "learning_rate": 1.7159763313609468e-06,
-      "loss": 1.5115,
+      "epoch": 22.25,
+      "learning_rate": 5.5508928571428575e-05,
+      "loss": 1.0834,
       "step": 4983
     },
     {
-      "epoch": 14.75,
-      "learning_rate": 1.6962524654832348e-06,
-      "loss": 1.4207,
+      "epoch": 22.25,
+      "learning_rate": 5.550000000000001e-05,
+      "loss": 1.0715,
       "step": 4984
     },
     {
-      "epoch": 14.75,
-      "learning_rate": 1.6765285996055227e-06,
-      "loss": 1.4745,
+      "epoch": 22.25,
+      "learning_rate": 5.5491071428571426e-05,
+      "loss": 1.1168,
       "step": 4985
     },
     {
-      "epoch": 14.75,
-      "learning_rate": 1.6568047337278108e-06,
-      "loss": 1.5263,
+      "epoch": 22.26,
+      "learning_rate": 5.548214285714286e-05,
+      "loss": 1.0565,
       "step": 4986
     },
     {
-      "epoch": 14.75,
-      "learning_rate": 1.6370808678500987e-06,
-      "loss": 1.4185,
+      "epoch": 22.26,
+      "learning_rate": 5.547321428571429e-05,
+      "loss": 1.0397,
       "step": 4987
     },
     {
-      "epoch": 14.76,
-      "learning_rate": 1.6173570019723868e-06,
-      "loss": 1.5049,
+      "epoch": 22.27,
+      "learning_rate": 5.5464285714285716e-05,
+      "loss": 1.02,
       "step": 4988
     },
     {
-      "epoch": 14.76,
-      "learning_rate": 1.5976331360946744e-06,
-      "loss": 1.5205,
+      "epoch": 22.27,
+      "learning_rate": 5.545535714285715e-05,
+      "loss": 1.0977,
       "step": 4989
     },
     {
-      "epoch": 14.76,
-      "learning_rate": 1.5779092702169627e-06,
-      "loss": 1.5299,
+      "epoch": 22.28,
+      "learning_rate": 5.5446428571428574e-05,
+      "loss": 1.0646,
       "step": 4990
     },
     {
-      "epoch": 14.77,
-      "learning_rate": 1.5581854043392506e-06,
-      "loss": 1.5808,
+      "epoch": 22.28,
+      "learning_rate": 5.54375e-05,
+      "loss": 0.9788,
       "step": 4991
     },
     {
-      "epoch": 14.77,
-      "learning_rate": 1.5384615384615387e-06,
-      "loss": 1.621,
+      "epoch": 22.29,
+      "learning_rate": 5.542857142857143e-05,
+      "loss": 1.0713,
       "step": 4992
     },
     {
-      "epoch": 14.77,
-      "learning_rate": 1.5187376725838264e-06,
-      "loss": 1.4845,
+      "epoch": 22.29,
+      "learning_rate": 5.5419642857142863e-05,
+      "loss": 1.0409,
       "step": 4993
     },
     {
-      "epoch": 14.78,
-      "learning_rate": 1.4990138067061145e-06,
-      "loss": 1.4918,
+      "epoch": 22.29,
+      "learning_rate": 5.5410714285714296e-05,
+      "loss": 1.1248,
       "step": 4994
     },
     {
-      "epoch": 14.78,
-      "learning_rate": 1.4792899408284024e-06,
-      "loss": 1.3984,
+      "epoch": 22.3,
+      "learning_rate": 5.5401785714285714e-05,
+      "loss": 1.0067,
       "step": 4995
     },
     {
-      "epoch": 14.78,
-      "learning_rate": 1.4595660749506902e-06,
-      "loss": 1.5662,
+      "epoch": 22.3,
+      "learning_rate": 5.539285714285715e-05,
+      "loss": 1.0192,
       "step": 4996
     },
     {
-      "epoch": 14.78,
-      "learning_rate": 1.4398422090729783e-06,
-      "loss": 1.4453,
+      "epoch": 22.31,
+      "learning_rate": 5.538392857142858e-05,
+      "loss": 1.1358,
       "step": 4997
     },
     {
-      "epoch": 14.79,
-      "learning_rate": 1.4201183431952662e-06,
-      "loss": 1.4596,
+      "epoch": 22.31,
+      "learning_rate": 5.5375e-05,
+      "loss": 1.0081,
       "step": 4998
     },
     {
-      "epoch": 14.79,
-      "learning_rate": 1.4003944773175543e-06,
-      "loss": 1.497,
+      "epoch": 22.32,
+      "learning_rate": 5.536607142857143e-05,
+      "loss": 1.1008,
       "step": 4999
     },
     {
-      "epoch": 14.79,
-      "learning_rate": 1.3806706114398422e-06,
-      "loss": 1.5055,
+      "epoch": 22.32,
+      "learning_rate": 5.535714285714286e-05,
+      "loss": 1.0503,
       "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
+      "epoch": 22.33,
+      "learning_rate": 5.534821428571428e-05,
+      "loss": 1.0727,
+      "step": 5001
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 5.533928571428571e-05,
+      "loss": 1.0386,
+      "step": 5002
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 5.5330357142857145e-05,
+      "loss": 1.0326,
+      "step": 5003
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 5.532142857142858e-05,
+      "loss": 1.0068,
+      "step": 5004
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 5.53125e-05,
+      "loss": 1.0388,
+      "step": 5005
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 5.5303571428571435e-05,
+      "loss": 0.9833,
+      "step": 5006
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 5.529464285714287e-05,
+      "loss": 1.0403,
+      "step": 5007
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 5.5285714285714286e-05,
+      "loss": 1.0346,
+      "step": 5008
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 5.527678571428572e-05,
+      "loss": 1.0662,
+      "step": 5009
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 5.526785714285715e-05,
+      "loss": 1.082,
+      "step": 5010
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 5.525892857142857e-05,
+      "loss": 1.1357,
+      "step": 5011
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.525e-05,
+      "loss": 1.0503,
+      "step": 5012
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.5241071428571433e-05,
+      "loss": 1.0008,
+      "step": 5013
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.523214285714285e-05,
+      "loss": 1.0792,
+      "step": 5014
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 5.5223214285714284e-05,
+      "loss": 1.0769,
+      "step": 5015
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 5.521428571428572e-05,
+      "loss": 1.0616,
+      "step": 5016
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 5.520535714285715e-05,
+      "loss": 1.0468,
+      "step": 5017
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 5.5196428571428574e-05,
+      "loss": 1.0244,
+      "step": 5018
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 5.51875e-05,
+      "loss": 1.114,
+      "step": 5019
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 5.517857142857143e-05,
+      "loss": 0.9771,
+      "step": 5020
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.516964285714286e-05,
+      "loss": 1.0541,
+      "step": 5021
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.516071428571429e-05,
+      "loss": 1.0634,
+      "step": 5022
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.515178571428572e-05,
+      "loss": 1.0485,
+      "step": 5023
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 5.514285714285714e-05,
+      "loss": 1.0422,
+      "step": 5024
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 5.513392857142857e-05,
+      "loss": 1.0372,
+      "step": 5025
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 5.5125000000000005e-05,
+      "loss": 1.0106,
+      "step": 5026
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 5.511607142857144e-05,
+      "loss": 1.0207,
+      "step": 5027
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 5.5107142857142856e-05,
+      "loss": 1.0228,
+      "step": 5028
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 5.509821428571429e-05,
+      "loss": 0.9966,
+      "step": 5029
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 5.508928571428572e-05,
+      "loss": 1.0269,
+      "step": 5030
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 5.508035714285714e-05,
+      "loss": 1.0602,
+      "step": 5031
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 5.507142857142857e-05,
+      "loss": 1.0694,
+      "step": 5032
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 5.5062500000000003e-05,
+      "loss": 1.1025,
+      "step": 5033
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 5.505357142857143e-05,
+      "loss": 1.0918,
+      "step": 5034
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 5.504464285714286e-05,
+      "loss": 1.0209,
+      "step": 5035
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 5.503571428571429e-05,
+      "loss": 1.056,
+      "step": 5036
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 5.5026785714285726e-05,
+      "loss": 1.0496,
+      "step": 5037
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 5.5017857142857144e-05,
+      "loss": 1.0631,
+      "step": 5038
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 5.5008928571428577e-05,
+      "loss": 1.0185,
+      "step": 5039
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.0299,
+      "step": 5040
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 5.499107142857143e-05,
+      "loss": 1.0061,
+      "step": 5041
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 5.498214285714286e-05,
+      "loss": 1.0325,
+      "step": 5042
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 5.497321428571429e-05,
+      "loss": 1.0717,
+      "step": 5043
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 5.496428571428571e-05,
+      "loss": 1.0524,
+      "step": 5044
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 5.495535714285714e-05,
+      "loss": 1.0704,
+      "step": 5045
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 5.4946428571428575e-05,
+      "loss": 0.9988,
+      "step": 5046
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 5.49375e-05,
+      "loss": 1.0857,
+      "step": 5047
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 5.4928571428571426e-05,
+      "loss": 1.0139,
+      "step": 5048
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 5.491964285714286e-05,
+      "loss": 1.0191,
+      "step": 5049
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 5.491071428571429e-05,
+      "loss": 1.0146,
+      "step": 5050
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 5.4901785714285716e-05,
+      "loss": 1.0172,
+      "step": 5051
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 5.489285714285715e-05,
+      "loss": 1.0154,
+      "step": 5052
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 5.488392857142858e-05,
+      "loss": 1.0998,
+      "step": 5053
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 5.4875e-05,
+      "loss": 1.0624,
+      "step": 5054
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 5.486607142857143e-05,
+      "loss": 1.0483,
+      "step": 5055
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 5.485714285714286e-05,
+      "loss": 1.0135,
+      "step": 5056
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 5.484821428571428e-05,
+      "loss": 1.0539,
+      "step": 5057
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 5.4839285714285714e-05,
+      "loss": 1.1445,
+      "step": 5058
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 5.4830357142857146e-05,
+      "loss": 1.0356,
+      "step": 5059
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 5.482142857142858e-05,
+      "loss": 1.1066,
+      "step": 5060
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 5.48125e-05,
+      "loss": 1.0838,
+      "step": 5061
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 5.480357142857143e-05,
+      "loss": 1.0638,
+      "step": 5062
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 5.479464285714286e-05,
+      "loss": 1.0865,
+      "step": 5063
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 5.478571428571429e-05,
+      "loss": 1.0102,
+      "step": 5064
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 5.477678571428572e-05,
+      "loss": 1.0369,
+      "step": 5065
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 5.476785714285715e-05,
+      "loss": 1.0391,
+      "step": 5066
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 5.475892857142857e-05,
+      "loss": 1.0677,
+      "step": 5067
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 5.475e-05,
+      "loss": 1.0373,
+      "step": 5068
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 5.4741071428571435e-05,
+      "loss": 1.0081,
+      "step": 5069
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 5.473214285714287e-05,
+      "loss": 1.0838,
+      "step": 5070
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 5.4723214285714286e-05,
+      "loss": 0.9944,
+      "step": 5071
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 5.471428571428572e-05,
+      "loss": 1.0691,
+      "step": 5072
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 5.470535714285715e-05,
+      "loss": 1.0878,
+      "step": 5073
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 5.469642857142857e-05,
+      "loss": 1.0297,
+      "step": 5074
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 5.46875e-05,
+      "loss": 1.0754,
+      "step": 5075
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 5.467857142857143e-05,
+      "loss": 1.0099,
+      "step": 5076
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 5.466964285714285e-05,
+      "loss": 1.1372,
+      "step": 5077
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 5.4660714285714284e-05,
+      "loss": 1.0435,
+      "step": 5078
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 5.4651785714285716e-05,
+      "loss": 1.0634,
+      "step": 5079
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 5.464285714285714e-05,
+      "loss": 1.0632,
+      "step": 5080
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 5.4633928571428574e-05,
+      "loss": 1.0812,
+      "step": 5081
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 5.4625000000000006e-05,
+      "loss": 1.0884,
+      "step": 5082
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 5.461607142857144e-05,
+      "loss": 1.0564,
+      "step": 5083
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 5.460714285714286e-05,
+      "loss": 1.0891,
+      "step": 5084
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 5.459821428571429e-05,
+      "loss": 1.102,
+      "step": 5085
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 5.458928571428572e-05,
+      "loss": 1.0471,
+      "step": 5086
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 5.458035714285714e-05,
+      "loss": 1.0615,
+      "step": 5087
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 5.457142857142857e-05,
+      "loss": 1.1546,
+      "step": 5088
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 5.4562500000000005e-05,
+      "loss": 1.0098,
+      "step": 5089
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 5.4553571428571424e-05,
+      "loss": 1.0323,
+      "step": 5090
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 5.4544642857142856e-05,
+      "loss": 0.9186,
+      "step": 5091
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 5.453571428571429e-05,
+      "loss": 0.989,
+      "step": 5092
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 5.452678571428572e-05,
+      "loss": 1.0028,
+      "step": 5093
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 5.4517857142857146e-05,
+      "loss": 1.0196,
+      "step": 5094
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 5.450892857142858e-05,
+      "loss": 1.0622,
+      "step": 5095
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 5.45e-05,
+      "loss": 1.1215,
+      "step": 5096
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 5.449107142857143e-05,
+      "loss": 1.0329,
+      "step": 5097
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 5.448214285714286e-05,
+      "loss": 0.975,
+      "step": 5098
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 5.447321428571429e-05,
+      "loss": 1.0238,
+      "step": 5099
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 5.446428571428571e-05,
+      "loss": 1.1073,
+      "step": 5100
+    },
+    {
+      "epoch": 22.77,
+      "eval_loss": 1.1218454837799072,
+      "eval_runtime": 45.2411,
+      "eval_samples_per_second": 289.427,
+      "eval_steps_per_second": 1.525,
+      "step": 5100
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 5.4455357142857144e-05,
+      "loss": 1.0636,
+      "step": 5101
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 5.4446428571428576e-05,
+      "loss": 1.0573,
+      "step": 5102
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 5.443750000000001e-05,
+      "loss": 1.0528,
+      "step": 5103
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 5.442857142857143e-05,
+      "loss": 1.0389,
+      "step": 5104
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 5.441964285714286e-05,
+      "loss": 1.0232,
+      "step": 5105
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 5.441071428571429e-05,
+      "loss": 1.0966,
+      "step": 5106
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 5.440178571428571e-05,
+      "loss": 1.0279,
+      "step": 5107
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 5.439285714285714e-05,
+      "loss": 1.1034,
+      "step": 5108
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 5.4383928571428575e-05,
+      "loss": 1.1095,
+      "step": 5109
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 5.4375e-05,
+      "loss": 0.9918,
+      "step": 5110
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 5.436607142857143e-05,
+      "loss": 1.031,
+      "step": 5111
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 5.4357142857142865e-05,
+      "loss": 1.1557,
+      "step": 5112
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 5.4348214285714283e-05,
+      "loss": 1.0068,
+      "step": 5113
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 5.4339285714285716e-05,
+      "loss": 1.0597,
+      "step": 5114
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 5.433035714285715e-05,
+      "loss": 1.0739,
+      "step": 5115
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 5.432142857142858e-05,
+      "loss": 1.0449,
+      "step": 5116
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 5.43125e-05,
+      "loss": 1.0406,
+      "step": 5117
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 5.430357142857143e-05,
+      "loss": 1.1209,
+      "step": 5118
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 5.429464285714286e-05,
+      "loss": 1.0888,
+      "step": 5119
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 5.428571428571428e-05,
+      "loss": 1.0618,
+      "step": 5120
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 5.4276785714285714e-05,
+      "loss": 1.0194,
+      "step": 5121
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 5.4267857142857146e-05,
+      "loss": 0.989,
+      "step": 5122
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 5.425892857142857e-05,
+      "loss": 1.0446,
+      "step": 5123
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 5.4250000000000004e-05,
+      "loss": 1.0413,
+      "step": 5124
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 5.424107142857143e-05,
+      "loss": 1.1389,
+      "step": 5125
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 5.423214285714286e-05,
+      "loss": 1.0182,
+      "step": 5126
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 5.422321428571429e-05,
+      "loss": 1.0681,
+      "step": 5127
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 5.421428571428572e-05,
+      "loss": 1.0363,
+      "step": 5128
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 5.420535714285715e-05,
+      "loss": 1.1195,
+      "step": 5129
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 5.419642857142857e-05,
+      "loss": 1.0609,
+      "step": 5130
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 5.41875e-05,
+      "loss": 1.0145,
+      "step": 5131
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 5.4178571428571435e-05,
+      "loss": 1.1462,
+      "step": 5132
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 5.4169642857142853e-05,
+      "loss": 1.1401,
+      "step": 5133
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 5.4160714285714286e-05,
+      "loss": 1.0382,
+      "step": 5134
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 5.415178571428572e-05,
+      "loss": 1.0239,
+      "step": 5135
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 5.414285714285715e-05,
+      "loss": 1.0934,
+      "step": 5136
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 5.413392857142857e-05,
+      "loss": 1.0116,
+      "step": 5137
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 5.4125e-05,
+      "loss": 1.0049,
+      "step": 5138
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 5.411607142857143e-05,
+      "loss": 0.9523,
+      "step": 5139
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 5.410714285714286e-05,
+      "loss": 1.0764,
+      "step": 5140
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 5.409821428571429e-05,
+      "loss": 0.9863,
+      "step": 5141
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 5.408928571428572e-05,
+      "loss": 1.0662,
+      "step": 5142
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 5.408035714285714e-05,
+      "loss": 1.076,
+      "step": 5143
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 5.4071428571428574e-05,
+      "loss": 1.0718,
+      "step": 5144
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 5.4062500000000006e-05,
+      "loss": 0.9982,
+      "step": 5145
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 5.4053571428571425e-05,
+      "loss": 1.0203,
+      "step": 5146
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 5.404464285714286e-05,
+      "loss": 0.968,
+      "step": 5147
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 5.403571428571429e-05,
+      "loss": 1.0528,
+      "step": 5148
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 5.402678571428572e-05,
+      "loss": 1.0781,
+      "step": 5149
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 5.401785714285714e-05,
+      "loss": 1.076,
+      "step": 5150
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 5.400892857142857e-05,
+      "loss": 1.1177,
+      "step": 5151
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.9677,
+      "step": 5152
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 5.399107142857143e-05,
+      "loss": 1.0364,
+      "step": 5153
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 5.3982142857142856e-05,
+      "loss": 0.9699,
+      "step": 5154
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 5.397321428571429e-05,
+      "loss": 1.0546,
+      "step": 5155
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 5.396428571428571e-05,
+      "loss": 1.0602,
+      "step": 5156
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 5.3955357142857146e-05,
+      "loss": 0.9947,
+      "step": 5157
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 5.394642857142858e-05,
+      "loss": 1.0631,
+      "step": 5158
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 5.393750000000001e-05,
+      "loss": 1.1145,
+      "step": 5159
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 5.392857142857143e-05,
+      "loss": 1.055,
+      "step": 5160
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 5.391964285714286e-05,
+      "loss": 1.1154,
+      "step": 5161
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 5.391071428571429e-05,
+      "loss": 1.0694,
+      "step": 5162
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 5.390178571428571e-05,
+      "loss": 1.0437,
+      "step": 5163
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 5.3892857142857144e-05,
+      "loss": 1.0523,
+      "step": 5164
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 5.3883928571428576e-05,
+      "loss": 1.0405,
+      "step": 5165
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 5.3874999999999995e-05,
+      "loss": 1.0696,
+      "step": 5166
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 5.386607142857143e-05,
+      "loss": 0.9981,
+      "step": 5167
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 5.385714285714286e-05,
+      "loss": 1.0662,
+      "step": 5168
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 5.384821428571429e-05,
+      "loss": 1.0158,
+      "step": 5169
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 5.383928571428572e-05,
+      "loss": 0.974,
+      "step": 5170
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 5.383035714285715e-05,
+      "loss": 1.0049,
+      "step": 5171
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 5.382142857142858e-05,
+      "loss": 1.0436,
+      "step": 5172
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 5.38125e-05,
+      "loss": 1.0533,
+      "step": 5173
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 5.380357142857143e-05,
+      "loss": 1.0323,
+      "step": 5174
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 5.3794642857142865e-05,
+      "loss": 1.0033,
+      "step": 5175
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 5.378571428571428e-05,
+      "loss": 1.0993,
+      "step": 5176
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 5.3776785714285716e-05,
+      "loss": 1.1041,
+      "step": 5177
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 5.376785714285715e-05,
+      "loss": 1.086,
+      "step": 5178
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 5.375892857142858e-05,
+      "loss": 1.0093,
+      "step": 5179
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 5.375e-05,
+      "loss": 1.0238,
+      "step": 5180
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 5.374107142857143e-05,
+      "loss": 1.0526,
+      "step": 5181
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 5.373214285714286e-05,
+      "loss": 0.9433,
+      "step": 5182
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 5.372321428571428e-05,
+      "loss": 1.0562,
+      "step": 5183
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 5.3714285714285714e-05,
+      "loss": 1.0549,
+      "step": 5184
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 5.3705357142857146e-05,
+      "loss": 1.0156,
+      "step": 5185
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 5.369642857142857e-05,
+      "loss": 1.0315,
+      "step": 5186
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 5.3687500000000004e-05,
+      "loss": 1.0939,
+      "step": 5187
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 5.3678571428571436e-05,
+      "loss": 1.0362,
+      "step": 5188
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 5.3669642857142855e-05,
+      "loss": 1.0584,
+      "step": 5189
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 5.366071428571429e-05,
+      "loss": 1.0101,
+      "step": 5190
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 5.365178571428572e-05,
+      "loss": 1.0647,
+      "step": 5191
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 5.364285714285715e-05,
+      "loss": 1.0215,
+      "step": 5192
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 5.363392857142857e-05,
+      "loss": 1.0373,
+      "step": 5193
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 5.3625e-05,
+      "loss": 1.088,
+      "step": 5194
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 5.3616071428571435e-05,
+      "loss": 1.0546,
+      "step": 5195
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 5.360714285714285e-05,
+      "loss": 1.0928,
+      "step": 5196
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 5.3598214285714286e-05,
+      "loss": 1.0557,
+      "step": 5197
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 5.358928571428572e-05,
+      "loss": 1.0597,
+      "step": 5198
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 5.358035714285714e-05,
+      "loss": 0.9737,
+      "step": 5199
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 5.3571428571428575e-05,
+      "loss": 0.9976,
+      "step": 5200
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 5.356250000000001e-05,
+      "loss": 1.0095,
+      "step": 5201
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 5.355357142857143e-05,
+      "loss": 1.0702,
+      "step": 5202
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 5.354464285714286e-05,
+      "loss": 1.0117,
+      "step": 5203
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 5.353571428571429e-05,
+      "loss": 0.9891,
+      "step": 5204
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 5.352678571428572e-05,
+      "loss": 1.0632,
+      "step": 5205
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 5.351785714285714e-05,
+      "loss": 0.9723,
+      "step": 5206
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 5.3508928571428574e-05,
+      "loss": 1.0378,
+      "step": 5207
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 5.3500000000000006e-05,
+      "loss": 1.051,
+      "step": 5208
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 5.3491071428571425e-05,
+      "loss": 0.9979,
+      "step": 5209
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 5.348214285714286e-05,
+      "loss": 1.0521,
+      "step": 5210
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 5.347321428571429e-05,
+      "loss": 0.9819,
+      "step": 5211
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 5.346428571428572e-05,
+      "loss": 1.0972,
+      "step": 5212
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 5.345535714285714e-05,
+      "loss": 1.0347,
+      "step": 5213
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 5.344642857142857e-05,
+      "loss": 1.0322,
+      "step": 5214
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 5.3437500000000005e-05,
+      "loss": 1.0807,
+      "step": 5215
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 5.342857142857143e-05,
+      "loss": 0.9746,
+      "step": 5216
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 5.341964285714286e-05,
+      "loss": 1.1,
+      "step": 5217
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 5.3410714285714295e-05,
+      "loss": 1.0297,
+      "step": 5218
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 5.340178571428571e-05,
+      "loss": 1.0842,
+      "step": 5219
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 5.3392857142857145e-05,
+      "loss": 1.0368,
+      "step": 5220
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 5.338392857142858e-05,
+      "loss": 0.9518,
+      "step": 5221
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 5.3374999999999996e-05,
+      "loss": 1.0747,
+      "step": 5222
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 5.336607142857143e-05,
+      "loss": 0.9931,
+      "step": 5223
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 5.335714285714286e-05,
+      "loss": 0.9782,
+      "step": 5224
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 5.334821428571429e-05,
+      "loss": 1.1198,
+      "step": 5225
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 5.333928571428571e-05,
+      "loss": 1.0448,
+      "step": 5226
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 5.3330357142857144e-05,
+      "loss": 1.1075,
+      "step": 5227
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 5.3321428571428576e-05,
+      "loss": 1.1327,
+      "step": 5228
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 5.33125e-05,
+      "loss": 1.0927,
+      "step": 5229
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 5.3303571428571434e-05,
+      "loss": 1.0248,
+      "step": 5230
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 5.329464285714286e-05,
+      "loss": 1.0175,
+      "step": 5231
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 5.3285714285714285e-05,
+      "loss": 1.0591,
+      "step": 5232
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 5.327678571428572e-05,
+      "loss": 0.9889,
+      "step": 5233
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 5.326785714285715e-05,
+      "loss": 1.0468,
+      "step": 5234
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 5.325892857142858e-05,
+      "loss": 0.9933,
+      "step": 5235
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 5.325e-05,
+      "loss": 1.0133,
+      "step": 5236
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 5.324107142857143e-05,
+      "loss": 1.1044,
+      "step": 5237
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 5.3232142857142865e-05,
+      "loss": 1.0146,
+      "step": 5238
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 5.322321428571428e-05,
+      "loss": 1.0111,
+      "step": 5239
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 5.3214285714285715e-05,
+      "loss": 1.136,
+      "step": 5240
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 5.320535714285715e-05,
+      "loss": 0.9981,
+      "step": 5241
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 5.3196428571428566e-05,
+      "loss": 1.053,
+      "step": 5242
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 5.31875e-05,
+      "loss": 1.0016,
+      "step": 5243
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 5.317857142857143e-05,
+      "loss": 0.9903,
+      "step": 5244
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 5.316964285714286e-05,
+      "loss": 1.0607,
+      "step": 5245
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 5.316071428571429e-05,
+      "loss": 1.039,
+      "step": 5246
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 5.315178571428572e-05,
+      "loss": 1.0165,
+      "step": 5247
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 5.314285714285715e-05,
+      "loss": 1.0812,
+      "step": 5248
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 5.313392857142857e-05,
+      "loss": 1.0169,
+      "step": 5249
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 1.0554,
+      "step": 5250
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 5.3116071428571436e-05,
+      "loss": 1.067,
+      "step": 5251
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 5.3107142857142855e-05,
+      "loss": 1.0817,
+      "step": 5252
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 5.309821428571429e-05,
+      "loss": 1.0143,
+      "step": 5253
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 5.308928571428572e-05,
+      "loss": 1.0523,
+      "step": 5254
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 5.308035714285714e-05,
+      "loss": 1.0318,
+      "step": 5255
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 5.307142857142857e-05,
+      "loss": 1.0453,
+      "step": 5256
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 5.30625e-05,
+      "loss": 0.9671,
+      "step": 5257
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 5.3053571428571435e-05,
+      "loss": 0.9857,
+      "step": 5258
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 5.304464285714286e-05,
+      "loss": 0.9699,
+      "step": 5259
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 5.3035714285714285e-05,
+      "loss": 1.0927,
+      "step": 5260
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 5.302678571428572e-05,
+      "loss": 1.0451,
+      "step": 5261
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 5.301785714285714e-05,
+      "loss": 1.0503,
+      "step": 5262
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 5.3008928571428575e-05,
+      "loss": 1.062,
+      "step": 5263
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 1.0118,
+      "step": 5264
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 5.2991071428571426e-05,
+      "loss": 1.0208,
+      "step": 5265
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 5.298214285714286e-05,
+      "loss": 1.052,
+      "step": 5266
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 5.297321428571429e-05,
+      "loss": 1.0216,
+      "step": 5267
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 5.296428571428572e-05,
+      "loss": 1.1342,
+      "step": 5268
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 5.295535714285714e-05,
+      "loss": 1.0665,
+      "step": 5269
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 5.2946428571428574e-05,
+      "loss": 1.0043,
+      "step": 5270
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 5.2937500000000006e-05,
+      "loss": 0.9974,
+      "step": 5271
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 5.2928571428571425e-05,
+      "loss": 1.0282,
+      "step": 5272
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 5.291964285714286e-05,
+      "loss": 1.0547,
+      "step": 5273
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 5.291071428571429e-05,
+      "loss": 1.0449,
+      "step": 5274
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 5.2901785714285715e-05,
+      "loss": 1.0252,
+      "step": 5275
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 5.289285714285715e-05,
+      "loss": 1.0477,
+      "step": 5276
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 5.288392857142858e-05,
+      "loss": 0.9549,
+      "step": 5277
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 5.2875000000000005e-05,
+      "loss": 0.9641,
+      "step": 5278
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 5.286607142857143e-05,
+      "loss": 1.0996,
+      "step": 5279
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 5.285714285714286e-05,
+      "loss": 1.0075,
+      "step": 5280
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 5.2848214285714294e-05,
+      "loss": 1.0216,
+      "step": 5281
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 5.283928571428571e-05,
+      "loss": 1.091,
+      "step": 5282
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 5.2830357142857145e-05,
+      "loss": 1.0271,
+      "step": 5283
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 5.282142857142858e-05,
+      "loss": 1.0126,
+      "step": 5284
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 5.2812499999999996e-05,
+      "loss": 0.9712,
+      "step": 5285
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 5.280357142857143e-05,
+      "loss": 1.0716,
+      "step": 5286
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 5.279464285714286e-05,
+      "loss": 1.075,
+      "step": 5287
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 5.2785714285714286e-05,
+      "loss": 1.0293,
+      "step": 5288
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 5.277678571428571e-05,
+      "loss": 1.0923,
+      "step": 5289
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 5.2767857142857144e-05,
+      "loss": 1.0231,
+      "step": 5290
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 5.2758928571428576e-05,
+      "loss": 1.1045,
+      "step": 5291
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 5.275e-05,
+      "loss": 1.1165,
+      "step": 5292
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 5.2741071428571434e-05,
+      "loss": 1.029,
+      "step": 5293
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 5.2732142857142866e-05,
+      "loss": 1.11,
+      "step": 5294
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 5.2723214285714285e-05,
+      "loss": 1.0751,
+      "step": 5295
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 5.271428571428572e-05,
+      "loss": 1.0806,
+      "step": 5296
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 5.270535714285715e-05,
+      "loss": 1.081,
+      "step": 5297
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 5.269642857142857e-05,
+      "loss": 1.0152,
+      "step": 5298
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 5.26875e-05,
+      "loss": 1.0207,
+      "step": 5299
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 5.267857142857143e-05,
+      "loss": 1.0388,
+      "step": 5300
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 5.2669642857142864e-05,
+      "loss": 1.0345,
+      "step": 5301
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 5.266071428571428e-05,
+      "loss": 1.0455,
+      "step": 5302
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 5.2651785714285715e-05,
+      "loss": 1.0645,
+      "step": 5303
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 5.264285714285715e-05,
+      "loss": 0.9617,
+      "step": 5304
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 5.263392857142857e-05,
+      "loss": 1.0677,
+      "step": 5305
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 5.2625000000000005e-05,
+      "loss": 1.0126,
+      "step": 5306
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 5.261607142857143e-05,
+      "loss": 1.0358,
+      "step": 5307
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 5.2607142857142856e-05,
+      "loss": 1.0715,
+      "step": 5308
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 5.259821428571429e-05,
+      "loss": 1.0547,
+      "step": 5309
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 5.258928571428572e-05,
+      "loss": 0.9726,
+      "step": 5310
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 5.258035714285715e-05,
+      "loss": 1.0793,
+      "step": 5311
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 5.257142857142857e-05,
+      "loss": 1.0639,
+      "step": 5312
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 5.2562500000000004e-05,
+      "loss": 1.0364,
+      "step": 5313
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 5.2553571428571436e-05,
+      "loss": 1.0508,
+      "step": 5314
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 5.2544642857142855e-05,
+      "loss": 0.9716,
+      "step": 5315
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 5.253571428571429e-05,
+      "loss": 1.0585,
+      "step": 5316
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 5.252678571428572e-05,
+      "loss": 1.0472,
+      "step": 5317
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 5.251785714285714e-05,
+      "loss": 0.9993,
+      "step": 5318
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 5.250892857142857e-05,
+      "loss": 1.1004,
+      "step": 5319
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 5.25e-05,
+      "loss": 1.0221,
+      "step": 5320
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 5.249107142857143e-05,
+      "loss": 0.9926,
+      "step": 5321
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 5.248214285714286e-05,
+      "loss": 1.0395,
+      "step": 5322
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 5.247321428571429e-05,
+      "loss": 1.0129,
+      "step": 5323
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 5.2464285714285724e-05,
+      "loss": 0.9965,
+      "step": 5324
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 5.245535714285714e-05,
+      "loss": 1.0649,
+      "step": 5325
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 5.2446428571428575e-05,
+      "loss": 1.033,
+      "step": 5326
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 5.243750000000001e-05,
+      "loss": 1.0949,
+      "step": 5327
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 5.2428571428571426e-05,
+      "loss": 0.9753,
+      "step": 5328
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 5.241964285714286e-05,
+      "loss": 1.0694,
+      "step": 5329
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 5.241071428571429e-05,
+      "loss": 1.0502,
+      "step": 5330
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 5.240178571428571e-05,
+      "loss": 1.0357,
+      "step": 5331
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 5.239285714285714e-05,
+      "loss": 1.0999,
+      "step": 5332
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 5.2383928571428574e-05,
+      "loss": 1.047,
+      "step": 5333
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 5.2375000000000006e-05,
+      "loss": 1.0274,
+      "step": 5334
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 5.236607142857143e-05,
+      "loss": 1.053,
+      "step": 5335
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 5.235714285714286e-05,
+      "loss": 1.0131,
+      "step": 5336
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 5.234821428571429e-05,
+      "loss": 1.0237,
+      "step": 5337
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 5.2339285714285714e-05,
+      "loss": 1.1025,
+      "step": 5338
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 5.233035714285715e-05,
+      "loss": 1.0576,
+      "step": 5339
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 5.232142857142858e-05,
+      "loss": 1.018,
+      "step": 5340
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 5.23125e-05,
+      "loss": 1.0343,
+      "step": 5341
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 5.230357142857143e-05,
+      "loss": 1.0631,
+      "step": 5342
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 5.229464285714286e-05,
+      "loss": 1.0678,
+      "step": 5343
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 5.2285714285714294e-05,
+      "loss": 1.088,
+      "step": 5344
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 5.227678571428571e-05,
+      "loss": 1.0341,
+      "step": 5345
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 5.2267857142857145e-05,
+      "loss": 1.0824,
+      "step": 5346
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 5.225892857142858e-05,
+      "loss": 1.0346,
+      "step": 5347
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 5.2249999999999996e-05,
+      "loss": 1.103,
+      "step": 5348
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 5.224107142857143e-05,
+      "loss": 1.0505,
+      "step": 5349
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 5.223214285714286e-05,
+      "loss": 1.0158,
+      "step": 5350
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 5.2223214285714286e-05,
+      "loss": 1.0308,
+      "step": 5351
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 5.221428571428572e-05,
+      "loss": 1.0107,
+      "step": 5352
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 5.220535714285715e-05,
+      "loss": 1.0002,
+      "step": 5353
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 5.219642857142858e-05,
+      "loss": 1.0906,
+      "step": 5354
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 5.21875e-05,
+      "loss": 1.1234,
+      "step": 5355
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 5.2178571428571434e-05,
+      "loss": 0.9462,
+      "step": 5356
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 5.2169642857142866e-05,
+      "loss": 1.0869,
+      "step": 5357
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 5.2160714285714284e-05,
+      "loss": 1.0176,
+      "step": 5358
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 5.215178571428572e-05,
+      "loss": 1.0365,
+      "step": 5359
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 5.214285714285715e-05,
+      "loss": 1.0412,
+      "step": 5360
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 5.213392857142857e-05,
+      "loss": 1.0774,
+      "step": 5361
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 5.2125e-05,
+      "loss": 1.0651,
+      "step": 5362
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 5.211607142857143e-05,
+      "loss": 1.0686,
+      "step": 5363
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 5.210714285714286e-05,
+      "loss": 1.0336,
+      "step": 5364
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 5.209821428571428e-05,
+      "loss": 0.9812,
+      "step": 5365
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 5.2089285714285715e-05,
+      "loss": 1.0633,
+      "step": 5366
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 5.208035714285715e-05,
+      "loss": 1.0783,
+      "step": 5367
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 5.207142857142857e-05,
+      "loss": 1.07,
+      "step": 5368
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 5.2062500000000005e-05,
+      "loss": 1.0052,
+      "step": 5369
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 5.205357142857144e-05,
+      "loss": 1.034,
+      "step": 5370
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 5.2044642857142856e-05,
+      "loss": 1.0,
+      "step": 5371
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 5.203571428571429e-05,
+      "loss": 1.1031,
+      "step": 5372
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 5.202678571428572e-05,
+      "loss": 1.0336,
+      "step": 5373
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 5.201785714285714e-05,
+      "loss": 0.9816,
+      "step": 5374
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.200892857142857e-05,
+      "loss": 1.1128,
+      "step": 5375
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.9358,
+      "step": 5376
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.1991071428571436e-05,
+      "loss": 1.0529,
+      "step": 5377
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.1982142857142854e-05,
+      "loss": 1.029,
+      "step": 5378
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.197321428571429e-05,
+      "loss": 0.9197,
+      "step": 5379
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 5.196428571428572e-05,
+      "loss": 1.026,
+      "step": 5380
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 5.1955357142857144e-05,
+      "loss": 1.0787,
+      "step": 5381
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.1946428571428577e-05,
+      "loss": 0.9842,
+      "step": 5382
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.193750000000001e-05,
+      "loss": 1.038,
+      "step": 5383
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.192857142857143e-05,
+      "loss": 0.9957,
+      "step": 5384
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.191964285714286e-05,
+      "loss": 1.0627,
+      "step": 5385
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.191071428571429e-05,
+      "loss": 1.0209,
+      "step": 5386
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 5.1901785714285724e-05,
+      "loss": 0.9705,
+      "step": 5387
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 5.189285714285714e-05,
+      "loss": 0.9809,
+      "step": 5388
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.1883928571428575e-05,
+      "loss": 1.0695,
+      "step": 5389
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.187500000000001e-05,
+      "loss": 0.9348,
+      "step": 5390
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 5.1866071428571426e-05,
+      "loss": 1.0661,
+      "step": 5391
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 5.185714285714286e-05,
+      "loss": 1.0946,
+      "step": 5392
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 5.184821428571429e-05,
+      "loss": 1.0576,
+      "step": 5393
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 5.183928571428571e-05,
+      "loss": 0.9852,
+      "step": 5394
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 5.183035714285714e-05,
+      "loss": 1.0302,
+      "step": 5395
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.1821428571428574e-05,
+      "loss": 1.0558,
+      "step": 5396
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.18125e-05,
+      "loss": 1.0987,
+      "step": 5397
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 5.180357142857143e-05,
+      "loss": 1.0731,
+      "step": 5398
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 5.1794642857142863e-05,
+      "loss": 1.1308,
+      "step": 5399
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 5.1785714285714296e-05,
+      "loss": 1.0399,
+      "step": 5400
+    },
+    {
+      "epoch": 24.11,
+      "eval_loss": 1.1246278285980225,
+      "eval_runtime": 45.1889,
+      "eval_samples_per_second": 289.762,
+      "eval_steps_per_second": 1.527,
+      "step": 5400
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 5.1776785714285714e-05,
+      "loss": 1.03,
+      "step": 5401
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.1767857142857147e-05,
+      "loss": 1.047,
+      "step": 5402
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.175892857142858e-05,
+      "loss": 1.0897,
+      "step": 5403
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.175e-05,
+      "loss": 1.0499,
+      "step": 5404
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.174107142857143e-05,
+      "loss": 1.0566,
+      "step": 5405
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.173214285714286e-05,
+      "loss": 1.0122,
+      "step": 5406
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 5.172321428571428e-05,
+      "loss": 1.0118,
+      "step": 5407
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 5.171428571428571e-05,
+      "loss": 1.0317,
+      "step": 5408
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.1705357142857145e-05,
+      "loss": 0.9975,
+      "step": 5409
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.169642857142858e-05,
+      "loss": 1.0378,
+      "step": 5410
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.16875e-05,
+      "loss": 0.9827,
+      "step": 5411
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.1678571428571435e-05,
+      "loss": 1.09,
+      "step": 5412
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 5.166964285714286e-05,
+      "loss": 1.0344,
+      "step": 5413
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 5.1660714285714286e-05,
+      "loss": 1.0629,
+      "step": 5414
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 5.165178571428572e-05,
+      "loss": 1.008,
+      "step": 5415
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.164285714285715e-05,
+      "loss": 1.0179,
+      "step": 5416
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.163392857142857e-05,
+      "loss": 0.992,
+      "step": 5417
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.1625e-05,
+      "loss": 1.0559,
+      "step": 5418
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.1616071428571433e-05,
+      "loss": 0.9947,
+      "step": 5419
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 5.1607142857142866e-05,
+      "loss": 1.0471,
+      "step": 5420
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 5.1598214285714284e-05,
+      "loss": 1.0478,
+      "step": 5421
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 5.1589285714285717e-05,
+      "loss": 1.0139,
+      "step": 5422
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 5.158035714285715e-05,
+      "loss": 0.9856,
+      "step": 5423
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 5.157142857142857e-05,
+      "loss": 1.0349,
+      "step": 5424
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.15625e-05,
+      "loss": 0.9845,
+      "step": 5425
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.155357142857143e-05,
+      "loss": 1.019,
+      "step": 5426
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 5.154464285714286e-05,
+      "loss": 1.0802,
+      "step": 5427
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 5.153571428571429e-05,
+      "loss": 1.0003,
+      "step": 5428
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 5.152678571428572e-05,
+      "loss": 0.9405,
+      "step": 5429
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 5.151785714285714e-05,
+      "loss": 0.9678,
+      "step": 5430
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.150892857142857e-05,
+      "loss": 1.1277,
+      "step": 5431
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.1500000000000005e-05,
+      "loss": 1.0089,
+      "step": 5432
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.149107142857144e-05,
+      "loss": 1.0275,
+      "step": 5433
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 5.1482142857142856e-05,
+      "loss": 1.0146,
+      "step": 5434
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 5.147321428571429e-05,
+      "loss": 1.0311,
+      "step": 5435
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.146428571428572e-05,
+      "loss": 0.956,
+      "step": 5436
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.145535714285714e-05,
+      "loss": 1.0777,
+      "step": 5437
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 5.144642857142857e-05,
+      "loss": 1.0368,
+      "step": 5438
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 5.1437500000000003e-05,
+      "loss": 0.9272,
+      "step": 5439
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 5.142857142857143e-05,
+      "loss": 1.1315,
+      "step": 5440
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 5.141964285714286e-05,
+      "loss": 1.0628,
+      "step": 5441
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 5.1410714285714287e-05,
+      "loss": 1.0265,
+      "step": 5442
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.140178571428572e-05,
+      "loss": 1.0356,
+      "step": 5443
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.1392857142857144e-05,
+      "loss": 0.944,
+      "step": 5444
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 5.1383928571428576e-05,
+      "loss": 1.0323,
+      "step": 5445
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 5.137500000000001e-05,
+      "loss": 1.0252,
+      "step": 5446
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 5.136607142857143e-05,
+      "loss": 0.9444,
+      "step": 5447
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 5.135714285714286e-05,
+      "loss": 1.0738,
+      "step": 5448
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.134821428571429e-05,
+      "loss": 1.0239,
+      "step": 5449
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.133928571428571e-05,
+      "loss": 1.118,
+      "step": 5450
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.133035714285714e-05,
+      "loss": 1.0256,
+      "step": 5451
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.1321428571428575e-05,
+      "loss": 0.9918,
+      "step": 5452
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.131250000000001e-05,
+      "loss": 1.1223,
+      "step": 5453
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 5.1303571428571426e-05,
+      "loss": 0.9419,
+      "step": 5454
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 5.129464285714286e-05,
+      "loss": 0.9749,
+      "step": 5455
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.128571428571429e-05,
+      "loss": 1.0445,
+      "step": 5456
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.1276785714285716e-05,
+      "loss": 1.0238,
+      "step": 5457
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.126785714285715e-05,
+      "loss": 1.0612,
+      "step": 5458
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.125892857142858e-05,
+      "loss": 0.9541,
+      "step": 5459
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 5.125e-05,
+      "loss": 1.0215,
+      "step": 5460
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 5.124107142857143e-05,
+      "loss": 1.0586,
+      "step": 5461
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 5.123214285714286e-05,
+      "loss": 0.9935,
+      "step": 5462
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.122321428571428e-05,
+      "loss": 1.0308,
+      "step": 5463
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.1214285714285714e-05,
+      "loss": 1.0069,
+      "step": 5464
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.1205357142857146e-05,
+      "loss": 0.9991,
+      "step": 5465
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.119642857142858e-05,
+      "loss": 1.0644,
+      "step": 5466
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 5.11875e-05,
+      "loss": 1.0123,
+      "step": 5467
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 5.117857142857143e-05,
+      "loss": 1.047,
+      "step": 5468
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 5.116964285714286e-05,
+      "loss": 1.02,
+      "step": 5469
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 5.116071428571429e-05,
+      "loss": 1.0576,
+      "step": 5470
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 5.115178571428571e-05,
+      "loss": 1.0088,
+      "step": 5471
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.1142857142857145e-05,
+      "loss": 1.097,
+      "step": 5472
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.113392857142857e-05,
+      "loss": 1.0053,
+      "step": 5473
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 5.1125e-05,
+      "loss": 1.0734,
+      "step": 5474
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 5.1116071428571435e-05,
+      "loss": 0.9718,
+      "step": 5475
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 5.110714285714287e-05,
+      "loss": 1.0587,
+      "step": 5476
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 5.1098214285714286e-05,
+      "loss": 1.0287,
+      "step": 5477
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.108928571428572e-05,
+      "loss": 1.0482,
+      "step": 5478
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.108035714285715e-05,
+      "loss": 1.0698,
+      "step": 5479
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.107142857142857e-05,
+      "loss": 0.9992,
+      "step": 5480
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 5.10625e-05,
+      "loss": 1.0813,
+      "step": 5481
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 5.105357142857143e-05,
+      "loss": 1.0626,
+      "step": 5482
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.104464285714285e-05,
+      "loss": 1.0908,
+      "step": 5483
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.1035714285714284e-05,
+      "loss": 1.0994,
+      "step": 5484
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 5.1026785714285716e-05,
+      "loss": 1.0099,
+      "step": 5485
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 5.101785714285715e-05,
+      "loss": 1.0707,
+      "step": 5486
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 5.1008928571428574e-05,
+      "loss": 1.0658,
+      "step": 5487
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.0304,
+      "step": 5488
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 5.099107142857144e-05,
+      "loss": 1.0477,
+      "step": 5489
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.098214285714286e-05,
+      "loss": 1.005,
+      "step": 5490
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.097321428571429e-05,
+      "loss": 1.0574,
+      "step": 5491
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 5.096428571428572e-05,
+      "loss": 1.039,
+      "step": 5492
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 5.095535714285714e-05,
+      "loss": 1.027,
+      "step": 5493
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 5.094642857142857e-05,
+      "loss": 1.0575,
+      "step": 5494
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 5.0937500000000005e-05,
+      "loss": 1.1041,
+      "step": 5495
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.0928571428571424e-05,
+      "loss": 1.0523,
+      "step": 5496
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.0919642857142856e-05,
+      "loss": 1.0715,
+      "step": 5497
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.091071428571429e-05,
+      "loss": 1.0451,
+      "step": 5498
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.090178571428572e-05,
+      "loss": 0.9328,
+      "step": 5499
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.089285714285714e-05,
+      "loss": 1.0264,
+      "step": 5500
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 5.088392857142857e-05,
+      "loss": 1.0003,
+      "step": 5501
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 5.0875e-05,
+      "loss": 1.0287,
+      "step": 5502
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.086607142857143e-05,
+      "loss": 0.987,
+      "step": 5503
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.085714285714286e-05,
+      "loss": 1.0485,
+      "step": 5504
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.084821428571429e-05,
+      "loss": 1.0462,
+      "step": 5505
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.083928571428571e-05,
+      "loss": 1.0134,
+      "step": 5506
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.0830357142857144e-05,
+      "loss": 1.0258,
+      "step": 5507
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 5.0821428571428576e-05,
+      "loss": 0.984,
+      "step": 5508
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 5.081250000000001e-05,
+      "loss": 1.0392,
+      "step": 5509
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.080357142857143e-05,
+      "loss": 1.0658,
+      "step": 5510
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.079464285714286e-05,
+      "loss": 1.105,
+      "step": 5511
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.078571428571429e-05,
+      "loss": 0.9923,
+      "step": 5512
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.077678571428571e-05,
+      "loss": 1.0643,
+      "step": 5513
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 5.076785714285714e-05,
+      "loss": 0.9866,
+      "step": 5514
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 5.0758928571428575e-05,
+      "loss": 1.0131,
+      "step": 5515
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 5.075e-05,
+      "loss": 1.0465,
+      "step": 5516
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.074107142857143e-05,
+      "loss": 1.0179,
+      "step": 5517
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.0732142857142865e-05,
+      "loss": 1.0099,
+      "step": 5518
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.072321428571429e-05,
+      "loss": 1.0542,
+      "step": 5519
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.0714285714285716e-05,
+      "loss": 0.9877,
+      "step": 5520
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 5.070535714285715e-05,
+      "loss": 1.1194,
+      "step": 5521
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 5.069642857142858e-05,
+      "loss": 1.0618,
+      "step": 5522
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 5.06875e-05,
+      "loss": 1.0133,
+      "step": 5523
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 5.067857142857143e-05,
+      "loss": 1.0785,
+      "step": 5524
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.066964285714286e-05,
+      "loss": 1.1237,
+      "step": 5525
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.066071428571428e-05,
+      "loss": 0.9918,
+      "step": 5526
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.0651785714285714e-05,
+      "loss": 1.0328,
+      "step": 5527
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 5.0642857142857146e-05,
+      "loss": 1.0215,
+      "step": 5528
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 5.063392857142858e-05,
+      "loss": 0.9498,
+      "step": 5529
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 5.0625e-05,
+      "loss": 1.0239,
+      "step": 5530
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 5.061607142857143e-05,
+      "loss": 0.9758,
+      "step": 5531
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.060714285714286e-05,
+      "loss": 1.0545,
+      "step": 5532
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.059821428571429e-05,
+      "loss": 1.0446,
+      "step": 5533
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.058928571428572e-05,
+      "loss": 1.0165,
+      "step": 5534
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.058035714285715e-05,
+      "loss": 1.0502,
+      "step": 5535
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.057142857142857e-05,
+      "loss": 1.049,
+      "step": 5536
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.05625e-05,
+      "loss": 0.9996,
+      "step": 5537
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.0553571428571435e-05,
+      "loss": 1.0745,
+      "step": 5538
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.0544642857142853e-05,
+      "loss": 1.0663,
+      "step": 5539
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.0535714285714286e-05,
+      "loss": 1.0275,
+      "step": 5540
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.052678571428572e-05,
+      "loss": 1.05,
+      "step": 5541
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.051785714285715e-05,
+      "loss": 1.0135,
+      "step": 5542
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.050892857142857e-05,
+      "loss": 1.0533,
+      "step": 5543
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.05e-05,
+      "loss": 1.044,
+      "step": 5544
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.049107142857143e-05,
+      "loss": 1.0862,
+      "step": 5545
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 5.048214285714286e-05,
+      "loss": 1.0146,
+      "step": 5546
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 5.047321428571429e-05,
+      "loss": 1.0571,
+      "step": 5547
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 5.0464285714285716e-05,
+      "loss": 1.0302,
+      "step": 5548
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 5.045535714285714e-05,
+      "loss": 1.073,
+      "step": 5549
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.0446428571428574e-05,
+      "loss": 1.0051,
+      "step": 5550
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.0437500000000006e-05,
+      "loss": 1.0207,
+      "step": 5551
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.042857142857144e-05,
+      "loss": 1.0263,
+      "step": 5552
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.041964285714286e-05,
+      "loss": 1.0284,
+      "step": 5553
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.041071428571429e-05,
+      "loss": 0.9793,
+      "step": 5554
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 5.040178571428572e-05,
+      "loss": 1.0182,
+      "step": 5555
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 5.039285714285714e-05,
+      "loss": 1.0324,
+      "step": 5556
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.038392857142857e-05,
+      "loss": 1.009,
+      "step": 5557
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.0375000000000005e-05,
+      "loss": 1.0316,
+      "step": 5558
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.0366071428571423e-05,
+      "loss": 1.1009,
+      "step": 5559
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.0357142857142856e-05,
+      "loss": 1.0219,
+      "step": 5560
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 5.034821428571429e-05,
+      "loss": 1.0513,
+      "step": 5561
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 5.033928571428572e-05,
+      "loss": 0.9788,
+      "step": 5562
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 5.0330357142857146e-05,
+      "loss": 1.0164,
+      "step": 5563
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.032142857142858e-05,
+      "loss": 1.1062,
+      "step": 5564
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.031250000000001e-05,
+      "loss": 1.045,
+      "step": 5565
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.030357142857143e-05,
+      "loss": 1.0355,
+      "step": 5566
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.029464285714286e-05,
+      "loss": 1.0903,
+      "step": 5567
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.028571428571429e-05,
+      "loss": 1.0676,
+      "step": 5568
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.027678571428571e-05,
+      "loss": 1.0382,
+      "step": 5569
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 5.0267857142857144e-05,
+      "loss": 1.036,
+      "step": 5570
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 5.0258928571428576e-05,
+      "loss": 1.0274,
+      "step": 5571
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.0249999999999995e-05,
+      "loss": 0.9963,
+      "step": 5572
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.024107142857143e-05,
+      "loss": 1.0591,
+      "step": 5573
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.023214285714286e-05,
+      "loss": 1.0136,
+      "step": 5574
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.022321428571429e-05,
+      "loss": 0.9775,
+      "step": 5575
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.021428571428572e-05,
+      "loss": 1.0142,
+      "step": 5576
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 5.020535714285714e-05,
+      "loss": 1.017,
+      "step": 5577
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 5.0196428571428575e-05,
+      "loss": 1.0209,
+      "step": 5578
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.01875e-05,
+      "loss": 1.1052,
+      "step": 5579
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.017857142857143e-05,
+      "loss": 1.0264,
+      "step": 5580
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.0169642857142865e-05,
+      "loss": 1.0885,
+      "step": 5581
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.016071428571428e-05,
+      "loss": 0.988,
+      "step": 5582
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.0151785714285716e-05,
+      "loss": 1.0331,
+      "step": 5583
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 5.014285714285715e-05,
+      "loss": 1.025,
+      "step": 5584
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 5.013392857142858e-05,
+      "loss": 1.0801,
+      "step": 5585
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 5.0125e-05,
+      "loss": 0.9573,
+      "step": 5586
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 5.011607142857143e-05,
+      "loss": 1.0355,
+      "step": 5587
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.010714285714286e-05,
+      "loss": 1.0491,
+      "step": 5588
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.009821428571428e-05,
+      "loss": 1.1031,
+      "step": 5589
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 5.0089285714285714e-05,
+      "loss": 1.0199,
+      "step": 5590
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 5.0080357142857146e-05,
+      "loss": 1.0087,
+      "step": 5591
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 5.007142857142857e-05,
+      "loss": 1.0794,
+      "step": 5592
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.0062500000000004e-05,
+      "loss": 1.0787,
+      "step": 5593
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.0053571428571436e-05,
+      "loss": 0.9593,
+      "step": 5594
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.004464285714286e-05,
+      "loss": 1.0169,
+      "step": 5595
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.003571428571429e-05,
+      "loss": 1.0213,
+      "step": 5596
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 5.002678571428572e-05,
+      "loss": 0.998,
+      "step": 5597
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 5.001785714285715e-05,
+      "loss": 1.0418,
+      "step": 5598
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.000892857142857e-05,
+      "loss": 1.0645,
+      "step": 5599
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5e-05,
+      "loss": 1.0648,
+      "step": 5600
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 4.999107142857143e-05,
+      "loss": 1.0052,
+      "step": 5601
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.998214285714286e-05,
+      "loss": 1.0667,
+      "step": 5602
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.9973214285714286e-05,
+      "loss": 1.0207,
+      "step": 5603
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 4.996428571428572e-05,
+      "loss": 1.0223,
+      "step": 5604
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 4.995535714285714e-05,
+      "loss": 1.0086,
+      "step": 5605
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.994642857142857e-05,
+      "loss": 1.0449,
+      "step": 5606
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.99375e-05,
+      "loss": 0.9493,
+      "step": 5607
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 4.992857142857143e-05,
+      "loss": 1.017,
+      "step": 5608
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 4.9919642857142865e-05,
+      "loss": 0.9211,
+      "step": 5609
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 4.991071428571429e-05,
+      "loss": 0.9783,
+      "step": 5610
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 4.9901785714285716e-05,
+      "loss": 1.0299,
+      "step": 5611
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 4.989285714285715e-05,
+      "loss": 1.0278,
+      "step": 5612
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.9883928571428574e-05,
+      "loss": 1.0324,
+      "step": 5613
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.9875000000000006e-05,
+      "loss": 1.0173,
+      "step": 5614
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.986607142857143e-05,
+      "loss": 1.015,
+      "step": 5615
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.985714285714286e-05,
+      "loss": 1.026,
+      "step": 5616
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 4.984821428571429e-05,
+      "loss": 0.9702,
+      "step": 5617
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 4.9839285714285715e-05,
+      "loss": 0.9603,
+      "step": 5618
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 4.983035714285714e-05,
+      "loss": 1.0936,
+      "step": 5619
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.982142857142857e-05,
+      "loss": 1.0292,
+      "step": 5620
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.98125e-05,
+      "loss": 1.0317,
+      "step": 5621
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.980357142857143e-05,
+      "loss": 0.9756,
+      "step": 5622
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.979464285714286e-05,
+      "loss": 1.04,
+      "step": 5623
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 4.978571428571429e-05,
+      "loss": 0.9798,
+      "step": 5624
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 4.977678571428572e-05,
+      "loss": 1.0543,
+      "step": 5625
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.9767857142857145e-05,
+      "loss": 0.9916,
+      "step": 5626
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.975892857142858e-05,
+      "loss": 1.0593,
+      "step": 5627
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.975e-05,
+      "loss": 1.0353,
+      "step": 5628
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.974107142857143e-05,
+      "loss": 1.0508,
+      "step": 5629
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.973214285714286e-05,
+      "loss": 0.9985,
+      "step": 5630
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 4.9723214285714286e-05,
+      "loss": 1.0523,
+      "step": 5631
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 4.971428571428572e-05,
+      "loss": 0.9844,
+      "step": 5632
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 4.9705357142857144e-05,
+      "loss": 1.0491,
+      "step": 5633
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 4.969642857142857e-05,
+      "loss": 1.0245,
+      "step": 5634
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.96875e-05,
+      "loss": 0.9931,
+      "step": 5635
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.967857142857143e-05,
+      "loss": 1.0403,
+      "step": 5636
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.966964285714286e-05,
+      "loss": 1.0617,
+      "step": 5637
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.966071428571429e-05,
+      "loss": 0.9616,
+      "step": 5638
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.965178571428572e-05,
+      "loss": 1.0647,
+      "step": 5639
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 4.964285714285715e-05,
+      "loss": 1.0142,
+      "step": 5640
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 4.9633928571428575e-05,
+      "loss": 0.9889,
+      "step": 5641
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.962500000000001e-05,
+      "loss": 1.0645,
+      "step": 5642
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.961607142857143e-05,
+      "loss": 1.1337,
+      "step": 5643
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 4.960714285714286e-05,
+      "loss": 1.0065,
+      "step": 5644
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 4.959821428571429e-05,
+      "loss": 0.9643,
+      "step": 5645
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.9589285714285715e-05,
+      "loss": 0.9249,
+      "step": 5646
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.958035714285715e-05,
+      "loss": 0.9991,
+      "step": 5647
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.957142857142857e-05,
+      "loss": 1.0783,
+      "step": 5648
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 4.95625e-05,
+      "loss": 1.0315,
+      "step": 5649
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 4.955357142857143e-05,
+      "loss": 1.0614,
+      "step": 5650
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 4.9544642857142856e-05,
+      "loss": 1.0535,
+      "step": 5651
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 4.953571428571429e-05,
+      "loss": 0.9579,
+      "step": 5652
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.9526785714285714e-05,
+      "loss": 1.0516,
+      "step": 5653
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.9517857142857146e-05,
+      "loss": 1.011,
+      "step": 5654
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 4.950892857142858e-05,
+      "loss": 0.9778,
+      "step": 5655
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.014,
+      "step": 5656
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 4.949107142857143e-05,
+      "loss": 1.0478,
+      "step": 5657
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 4.948214285714286e-05,
+      "loss": 1.0416,
+      "step": 5658
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 4.947321428571429e-05,
+      "loss": 0.9883,
+      "step": 5659
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.946428571428572e-05,
+      "loss": 1.0464,
+      "step": 5660
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.9455357142857145e-05,
+      "loss": 1.0739,
+      "step": 5661
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.944642857142857e-05,
+      "loss": 0.9905,
+      "step": 5662
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.94375e-05,
+      "loss": 1.0148,
+      "step": 5663
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 4.942857142857143e-05,
+      "loss": 0.9947,
+      "step": 5664
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 4.941964285714286e-05,
+      "loss": 1.0317,
+      "step": 5665
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 4.9410714285714285e-05,
+      "loss": 1.0792,
+      "step": 5666
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.940178571428572e-05,
+      "loss": 1.0477,
+      "step": 5667
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.939285714285714e-05,
+      "loss": 1.0514,
+      "step": 5668
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.9383928571428575e-05,
+      "loss": 1.0741,
+      "step": 5669
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.937500000000001e-05,
+      "loss": 0.955,
+      "step": 5670
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 4.936607142857143e-05,
+      "loss": 1.1338,
+      "step": 5671
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 4.935714285714286e-05,
+      "loss": 1.0589,
+      "step": 5672
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.934821428571429e-05,
+      "loss": 1.0915,
+      "step": 5673
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.9339285714285716e-05,
+      "loss": 1.0326,
+      "step": 5674
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.933035714285715e-05,
+      "loss": 1.0008,
+      "step": 5675
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.9321428571428574e-05,
+      "loss": 1.0745,
+      "step": 5676
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.93125e-05,
+      "loss": 0.9684,
+      "step": 5677
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 4.930357142857143e-05,
+      "loss": 0.9341,
+      "step": 5678
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 4.929464285714286e-05,
+      "loss": 0.9939,
+      "step": 5679
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 4.928571428571429e-05,
+      "loss": 1.0384,
+      "step": 5680
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 4.9276785714285715e-05,
+      "loss": 0.9782,
+      "step": 5681
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.926785714285714e-05,
+      "loss": 1.1385,
+      "step": 5682
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.925892857142857e-05,
+      "loss": 1.0695,
+      "step": 5683
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 0.9873,
+      "step": 5684
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 4.924107142857144e-05,
+      "loss": 1.0297,
+      "step": 5685
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 4.923214285714286e-05,
+      "loss": 1.0078,
+      "step": 5686
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 4.922321428571429e-05,
+      "loss": 0.9348,
+      "step": 5687
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 4.921428571428572e-05,
+      "loss": 1.0816,
+      "step": 5688
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.9205357142857145e-05,
+      "loss": 1.0014,
+      "step": 5689
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.919642857142857e-05,
+      "loss": 1.057,
+      "step": 5690
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 4.91875e-05,
+      "loss": 1.0122,
+      "step": 5691
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 4.917857142857143e-05,
+      "loss": 1.003,
+      "step": 5692
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.916964285714286e-05,
+      "loss": 0.9718,
+      "step": 5693
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.9160714285714286e-05,
+      "loss": 1.0596,
+      "step": 5694
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.915178571428571e-05,
+      "loss": 0.9849,
+      "step": 5695
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 4.9142857142857144e-05,
+      "loss": 0.9769,
+      "step": 5696
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 4.913392857142857e-05,
+      "loss": 0.9988,
+      "step": 5697
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 4.9125e-05,
+      "loss": 1.0519,
+      "step": 5698
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 4.9116071428571434e-05,
+      "loss": 1.079,
+      "step": 5699
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.910714285714286e-05,
+      "loss": 1.0702,
+      "step": 5700
+    },
+    {
+      "epoch": 25.45,
+      "eval_loss": 1.1184215545654297,
+      "eval_runtime": 44.8359,
+      "eval_samples_per_second": 292.043,
+      "eval_steps_per_second": 1.539,
+      "step": 5700
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.909821428571429e-05,
+      "loss": 1.0433,
+      "step": 5701
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 4.908928571428572e-05,
+      "loss": 0.9806,
+      "step": 5702
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 4.908035714285715e-05,
+      "loss": 1.0708,
+      "step": 5703
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 4.9071428571428574e-05,
+      "loss": 0.9911,
+      "step": 5704
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 4.90625e-05,
+      "loss": 1.0295,
+      "step": 5705
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 4.905357142857143e-05,
+      "loss": 0.998,
+      "step": 5706
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.904464285714286e-05,
+      "loss": 1.0185,
+      "step": 5707
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.903571428571429e-05,
+      "loss": 0.972,
+      "step": 5708
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.9026785714285715e-05,
+      "loss": 1.0783,
+      "step": 5709
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.901785714285714e-05,
+      "loss": 1.0802,
+      "step": 5710
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 4.900892857142857e-05,
+      "loss": 0.9986,
+      "step": 5711
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 4.9e-05,
+      "loss": 0.9932,
+      "step": 5712
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 4.899107142857143e-05,
+      "loss": 1.0974,
+      "step": 5713
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.898214285714286e-05,
+      "loss": 1.0047,
+      "step": 5714
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.897321428571429e-05,
+      "loss": 0.9953,
+      "step": 5715
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.896428571428572e-05,
+      "loss": 1.0091,
+      "step": 5716
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.8955357142857146e-05,
+      "loss": 1.0094,
+      "step": 5717
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 4.894642857142858e-05,
+      "loss": 1.0784,
+      "step": 5718
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 4.8937500000000004e-05,
+      "loss": 0.9711,
+      "step": 5719
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.892857142857143e-05,
+      "loss": 1.0806,
+      "step": 5720
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.891964285714286e-05,
+      "loss": 1.0647,
+      "step": 5721
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.891071428571429e-05,
+      "loss": 1.0205,
+      "step": 5722
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.890178571428571e-05,
+      "loss": 1.0657,
+      "step": 5723
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.8892857142857144e-05,
+      "loss": 1.0564,
+      "step": 5724
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 4.888392857142857e-05,
+      "loss": 1.0187,
+      "step": 5725
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 4.8875e-05,
+      "loss": 0.956,
+      "step": 5726
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.886607142857143e-05,
+      "loss": 1.0483,
+      "step": 5727
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.885714285714286e-05,
+      "loss": 1.0687,
+      "step": 5728
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.884821428571429e-05,
+      "loss": 1.0217,
+      "step": 5729
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.883928571428572e-05,
+      "loss": 1.0976,
+      "step": 5730
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.883035714285715e-05,
+      "loss": 1.0529,
+      "step": 5731
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 4.8821428571428575e-05,
+      "loss": 1.0156,
+      "step": 5732
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 4.88125e-05,
+      "loss": 1.1173,
+      "step": 5733
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 4.880357142857143e-05,
+      "loss": 1.0887,
+      "step": 5734
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 4.879464285714286e-05,
+      "loss": 1.072,
+      "step": 5735
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.878571428571429e-05,
+      "loss": 1.0321,
+      "step": 5736
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.8776785714285716e-05,
+      "loss": 0.9963,
+      "step": 5737
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 4.876785714285714e-05,
+      "loss": 1.005,
+      "step": 5738
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 4.8758928571428574e-05,
+      "loss": 1.0119,
+      "step": 5739
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 4.875e-05,
+      "loss": 1.0504,
+      "step": 5740
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 4.874107142857143e-05,
+      "loss": 1.1372,
+      "step": 5741
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 4.873214285714286e-05,
+      "loss": 1.0254,
+      "step": 5742
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.872321428571429e-05,
+      "loss": 1.0462,
+      "step": 5743
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.8714285714285714e-05,
+      "loss": 0.9225,
+      "step": 5744
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 4.870535714285715e-05,
+      "loss": 1.0483,
+      "step": 5745
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 4.869642857142858e-05,
+      "loss": 1.074,
+      "step": 5746
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.8687500000000004e-05,
+      "loss": 1.0153,
+      "step": 5747
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.867857142857143e-05,
+      "loss": 1.0901,
+      "step": 5748
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.866964285714286e-05,
+      "loss": 1.0206,
+      "step": 5749
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.866071428571429e-05,
+      "loss": 1.0724,
+      "step": 5750
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.865178571428572e-05,
+      "loss": 1.0367,
+      "step": 5751
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 4.8642857142857145e-05,
+      "loss": 1.0361,
+      "step": 5752
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 4.863392857142857e-05,
+      "loss": 1.0254,
+      "step": 5753
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.8625e-05,
+      "loss": 1.074,
+      "step": 5754
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.861607142857143e-05,
+      "loss": 1.0263,
+      "step": 5755
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 4.860714285714286e-05,
+      "loss": 0.993,
+      "step": 5756
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 4.8598214285714286e-05,
+      "loss": 0.9878,
+      "step": 5757
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 4.858928571428572e-05,
+      "loss": 1.0182,
+      "step": 5758
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 4.8580357142857144e-05,
+      "loss": 1.0017,
+      "step": 5759
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 4.8571428571428576e-05,
+      "loss": 1.017,
+      "step": 5760
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.85625e-05,
+      "loss": 1.0238,
+      "step": 5761
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.8553571428571434e-05,
+      "loss": 1.0451,
+      "step": 5762
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.854464285714286e-05,
+      "loss": 1.0195,
+      "step": 5763
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.853571428571429e-05,
+      "loss": 1.0113,
+      "step": 5764
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 4.852678571428572e-05,
+      "loss": 1.0229,
+      "step": 5765
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 4.851785714285714e-05,
+      "loss": 0.9953,
+      "step": 5766
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.8508928571428574e-05,
+      "loss": 1.0325,
+      "step": 5767
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.85e-05,
+      "loss": 1.0372,
+      "step": 5768
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.849107142857143e-05,
+      "loss": 0.9694,
+      "step": 5769
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.848214285714286e-05,
+      "loss": 1.0878,
+      "step": 5770
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.847321428571428e-05,
+      "loss": 1.0423,
+      "step": 5771
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 4.8464285714285715e-05,
+      "loss": 1.0236,
+      "step": 5772
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 4.845535714285714e-05,
+      "loss": 1.1239,
+      "step": 5773
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.844642857142857e-05,
+      "loss": 1.0003,
+      "step": 5774
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 1.0761,
+      "step": 5775
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.842857142857143e-05,
+      "loss": 1.0515,
+      "step": 5776
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.841964285714286e-05,
+      "loss": 0.9745,
+      "step": 5777
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.841071428571429e-05,
+      "loss": 1.014,
+      "step": 5778
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 4.840178571428572e-05,
+      "loss": 1.0188,
+      "step": 5779
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 4.8392857142857146e-05,
+      "loss": 1.0585,
+      "step": 5780
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 4.838392857142857e-05,
+      "loss": 1.017,
+      "step": 5781
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 4.8375000000000004e-05,
+      "loss": 0.966,
+      "step": 5782
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.836607142857143e-05,
+      "loss": 1.0597,
+      "step": 5783
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.835714285714286e-05,
+      "loss": 1.0088,
+      "step": 5784
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.834821428571429e-05,
+      "loss": 1.0596,
+      "step": 5785
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.833928571428571e-05,
+      "loss": 1.0679,
+      "step": 5786
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.8330357142857144e-05,
+      "loss": 1.0616,
+      "step": 5787
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 4.832142857142857e-05,
+      "loss": 1.0355,
+      "step": 5788
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 4.83125e-05,
+      "loss": 1.0258,
+      "step": 5789
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.8303571428571434e-05,
+      "loss": 1.0452,
+      "step": 5790
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.829464285714286e-05,
+      "loss": 1.0027,
+      "step": 5791
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 4.828571428571429e-05,
+      "loss": 1.0925,
+      "step": 5792
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 4.827678571428572e-05,
+      "loss": 1.0204,
+      "step": 5793
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.826785714285714e-05,
+      "loss": 0.9807,
+      "step": 5794
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.8258928571428575e-05,
+      "loss": 1.0201,
+      "step": 5795
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.825e-05,
+      "loss": 1.0801,
+      "step": 5796
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.824107142857143e-05,
+      "loss": 1.1007,
+      "step": 5797
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.823214285714286e-05,
+      "loss": 1.0251,
+      "step": 5798
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 4.8223214285714284e-05,
+      "loss": 1.0065,
+      "step": 5799
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.9962,
+      "step": 5800
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.820535714285714e-05,
+      "loss": 1.0684,
+      "step": 5801
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.8196428571428574e-05,
+      "loss": 1.0914,
+      "step": 5802
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.81875e-05,
+      "loss": 0.9947,
+      "step": 5803
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.817857142857143e-05,
+      "loss": 1.0568,
+      "step": 5804
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.8169642857142863e-05,
+      "loss": 1.014,
+      "step": 5805
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.816071428571429e-05,
+      "loss": 1.0556,
+      "step": 5806
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.815178571428572e-05,
+      "loss": 1.0108,
+      "step": 5807
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.8142857142857147e-05,
+      "loss": 0.9709,
+      "step": 5808
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.813392857142857e-05,
+      "loss": 1.0301,
+      "step": 5809
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 0.9607,
+      "step": 5810
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.811607142857143e-05,
+      "loss": 1.0871,
+      "step": 5811
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 4.810714285714286e-05,
+      "loss": 0.9151,
+      "step": 5812
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 4.809821428571429e-05,
+      "loss": 1.0136,
+      "step": 5813
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.808928571428571e-05,
+      "loss": 1.0335,
+      "step": 5814
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.8080357142857145e-05,
+      "loss": 1.0519,
+      "step": 5815
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.807142857142857e-05,
+      "loss": 1.0784,
+      "step": 5816
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.80625e-05,
+      "loss": 1.0122,
+      "step": 5817
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.805357142857143e-05,
+      "loss": 1.036,
+      "step": 5818
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 4.804464285714286e-05,
+      "loss": 0.9619,
+      "step": 5819
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 4.803571428571429e-05,
+      "loss": 0.9737,
+      "step": 5820
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.802678571428572e-05,
+      "loss": 1.0307,
+      "step": 5821
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.801785714285715e-05,
+      "loss": 0.9702,
+      "step": 5822
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.8008928571428576e-05,
+      "loss": 0.9976,
+      "step": 5823
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.9153,
+      "step": 5824
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.7991071428571433e-05,
+      "loss": 0.985,
+      "step": 5825
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.798214285714286e-05,
+      "loss": 0.9928,
+      "step": 5826
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.797321428571429e-05,
+      "loss": 0.966,
+      "step": 5827
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 4.7964285714285717e-05,
+      "loss": 0.9486,
+      "step": 5828
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 4.795535714285714e-05,
+      "loss": 1.0135,
+      "step": 5829
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 4.7946428571428574e-05,
+      "loss": 1.0082,
+      "step": 5830
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 4.79375e-05,
+      "loss": 1.0275,
+      "step": 5831
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 4.7928571428571425e-05,
+      "loss": 1.034,
+      "step": 5832
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 4.791964285714286e-05,
+      "loss": 1.0364,
+      "step": 5833
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 4.791071428571429e-05,
+      "loss": 0.9585,
+      "step": 5834
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 4.790178571428572e-05,
+      "loss": 1.0627,
+      "step": 5835
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 4.789285714285715e-05,
+      "loss": 0.99,
+      "step": 5836
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 4.788392857142857e-05,
+      "loss": 1.0216,
+      "step": 5837
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 4.7875000000000005e-05,
+      "loss": 1.0519,
+      "step": 5838
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 4.786607142857143e-05,
+      "loss": 1.0611,
+      "step": 5839
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 4.785714285714286e-05,
+      "loss": 1.1278,
+      "step": 5840
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 4.784821428571429e-05,
+      "loss": 1.0203,
+      "step": 5841
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 4.7839285714285714e-05,
+      "loss": 1.0278,
+      "step": 5842
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 4.7830357142857146e-05,
+      "loss": 1.04,
+      "step": 5843
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 4.782142857142857e-05,
+      "loss": 1.0488,
+      "step": 5844
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 4.7812500000000003e-05,
+      "loss": 0.9992,
+      "step": 5845
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 4.780357142857143e-05,
+      "loss": 1.0232,
+      "step": 5846
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 4.7794642857142854e-05,
+      "loss": 0.9508,
+      "step": 5847
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 4.7785714285714287e-05,
+      "loss": 1.0881,
+      "step": 5848
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 4.777678571428572e-05,
+      "loss": 0.991,
+      "step": 5849
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 4.7767857142857144e-05,
+      "loss": 1.0448,
+      "step": 5850
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 4.7758928571428576e-05,
+      "loss": 1.0887,
+      "step": 5851
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 4.775e-05,
+      "loss": 1.0349,
+      "step": 5852
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 4.7741071428571434e-05,
+      "loss": 0.9565,
+      "step": 5853
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 4.773214285714286e-05,
+      "loss": 1.0145,
+      "step": 5854
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 4.772321428571429e-05,
+      "loss": 0.9815,
+      "step": 5855
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 4.771428571428572e-05,
+      "loss": 0.9636,
+      "step": 5856
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 4.770535714285714e-05,
+      "loss": 0.9761,
+      "step": 5857
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 4.7696428571428575e-05,
+      "loss": 0.9345,
+      "step": 5858
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 4.76875e-05,
+      "loss": 0.9313,
+      "step": 5859
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 4.767857142857143e-05,
+      "loss": 0.9452,
+      "step": 5860
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 4.766964285714286e-05,
+      "loss": 1.0791,
+      "step": 5861
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 4.7660714285714284e-05,
+      "loss": 0.9739,
+      "step": 5862
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 4.7651785714285716e-05,
+      "loss": 1.0337,
+      "step": 5863
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 4.764285714285715e-05,
+      "loss": 1.0437,
+      "step": 5864
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 4.7633928571428573e-05,
+      "loss": 1.0482,
+      "step": 5865
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 4.7625000000000006e-05,
+      "loss": 1.0308,
+      "step": 5866
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 4.761607142857143e-05,
+      "loss": 1.0602,
+      "step": 5867
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 4.760714285714286e-05,
+      "loss": 1.01,
+      "step": 5868
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 4.759821428571429e-05,
+      "loss": 0.9667,
+      "step": 5869
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 4.7589285714285714e-05,
+      "loss": 0.9341,
+      "step": 5870
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 4.7580357142857146e-05,
+      "loss": 0.9849,
+      "step": 5871
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 4.757142857142857e-05,
+      "loss": 0.9385,
+      "step": 5872
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 4.7562500000000004e-05,
+      "loss": 1.062,
+      "step": 5873
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 4.755357142857143e-05,
+      "loss": 1.049,
+      "step": 5874
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 4.7544642857142855e-05,
+      "loss": 1.0816,
+      "step": 5875
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 4.753571428571429e-05,
+      "loss": 1.0838,
+      "step": 5876
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 4.752678571428571e-05,
+      "loss": 0.9878,
+      "step": 5877
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 4.7517857142857145e-05,
+      "loss": 1.0145,
+      "step": 5878
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 4.750892857142857e-05,
+      "loss": 1.0456,
+      "step": 5879
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 4.75e-05,
+      "loss": 1.003,
+      "step": 5880
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 4.7491071428571435e-05,
+      "loss": 1.0029,
+      "step": 5881
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 4.748214285714286e-05,
+      "loss": 1.0999,
+      "step": 5882
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 4.747321428571429e-05,
+      "loss": 1.0014,
+      "step": 5883
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 4.746428571428572e-05,
+      "loss": 1.0773,
+      "step": 5884
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 4.7455357142857143e-05,
+      "loss": 0.9853,
+      "step": 5885
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 4.7446428571428576e-05,
+      "loss": 0.9878,
+      "step": 5886
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 4.74375e-05,
+      "loss": 1.1099,
+      "step": 5887
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 4.742857142857143e-05,
+      "loss": 0.9345,
+      "step": 5888
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 4.741964285714286e-05,
+      "loss": 1.0926,
+      "step": 5889
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 4.7410714285714284e-05,
+      "loss": 1.0197,
+      "step": 5890
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 4.7401785714285716e-05,
+      "loss": 1.057,
+      "step": 5891
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 4.739285714285714e-05,
+      "loss": 1.0823,
+      "step": 5892
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 4.7383928571428574e-05,
+      "loss": 1.0015,
+      "step": 5893
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 4.7375e-05,
+      "loss": 1.0154,
+      "step": 5894
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 4.736607142857143e-05,
+      "loss": 0.9738,
+      "step": 5895
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 4.7357142857142864e-05,
+      "loss": 1.0628,
+      "step": 5896
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 4.734821428571429e-05,
+      "loss": 0.9779,
+      "step": 5897
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 4.7339285714285715e-05,
+      "loss": 0.9506,
+      "step": 5898
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 4.733035714285715e-05,
+      "loss": 1.0157,
+      "step": 5899
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 4.732142857142857e-05,
+      "loss": 1.0371,
+      "step": 5900
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 4.7312500000000005e-05,
+      "loss": 1.0417,
+      "step": 5901
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 4.730357142857143e-05,
+      "loss": 0.9927,
+      "step": 5902
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 4.7294642857142856e-05,
+      "loss": 0.9316,
+      "step": 5903
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 4.728571428571429e-05,
+      "loss": 1.0183,
+      "step": 5904
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 4.7276785714285713e-05,
+      "loss": 1.0183,
+      "step": 5905
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 4.7267857142857146e-05,
+      "loss": 1.05,
+      "step": 5906
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 4.725892857142857e-05,
+      "loss": 1.0801,
+      "step": 5907
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 1.0456,
+      "step": 5908
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 4.724107142857143e-05,
+      "loss": 1.0208,
+      "step": 5909
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 4.723214285714286e-05,
+      "loss": 0.9952,
+      "step": 5910
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 4.722321428571429e-05,
+      "loss": 0.9711,
+      "step": 5911
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 4.721428571428572e-05,
+      "loss": 1.0167,
+      "step": 5912
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 4.7205357142857144e-05,
+      "loss": 1.0601,
+      "step": 5913
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 4.7196428571428576e-05,
+      "loss": 1.165,
+      "step": 5914
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 4.71875e-05,
+      "loss": 0.9864,
+      "step": 5915
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 4.7178571428571434e-05,
+      "loss": 0.9995,
+      "step": 5916
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 4.716964285714286e-05,
+      "loss": 0.97,
+      "step": 5917
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 4.7160714285714285e-05,
+      "loss": 1.0723,
+      "step": 5918
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 4.715178571428572e-05,
+      "loss": 1.0114,
+      "step": 5919
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 1.0005,
+      "step": 5920
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 4.7133928571428575e-05,
+      "loss": 0.9936,
+      "step": 5921
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 4.7125e-05,
+      "loss": 1.0766,
+      "step": 5922
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 4.7116071428571426e-05,
+      "loss": 0.9941,
+      "step": 5923
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 4.710714285714286e-05,
+      "loss": 1.0262,
+      "step": 5924
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 4.709821428571429e-05,
+      "loss": 1.101,
+      "step": 5925
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 4.708928571428572e-05,
+      "loss": 1.0878,
+      "step": 5926
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 4.708035714285715e-05,
+      "loss": 1.0217,
+      "step": 5927
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 4.707142857142857e-05,
+      "loss": 0.9992,
+      "step": 5928
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 4.7062500000000006e-05,
+      "loss": 0.9851,
+      "step": 5929
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 4.705357142857143e-05,
+      "loss": 0.9508,
+      "step": 5930
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 4.704464285714286e-05,
+      "loss": 1.0706,
+      "step": 5931
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 4.703571428571429e-05,
+      "loss": 0.9765,
+      "step": 5932
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 4.7026785714285714e-05,
+      "loss": 1.0037,
+      "step": 5933
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 4.7017857142857146e-05,
+      "loss": 0.9799,
+      "step": 5934
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 4.700892857142857e-05,
+      "loss": 1.031,
+      "step": 5935
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 4.7e-05,
+      "loss": 0.9945,
+      "step": 5936
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 4.699107142857143e-05,
+      "loss": 1.0332,
+      "step": 5937
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 4.6982142857142855e-05,
+      "loss": 0.9397,
+      "step": 5938
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 4.697321428571429e-05,
+      "loss": 1.0227,
+      "step": 5939
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 4.696428571428572e-05,
+      "loss": 1.0559,
+      "step": 5940
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 4.6955357142857145e-05,
+      "loss": 1.0651,
+      "step": 5941
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 4.694642857142858e-05,
+      "loss": 0.9923,
+      "step": 5942
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 4.69375e-05,
+      "loss": 1.0493,
+      "step": 5943
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 4.6928571428571435e-05,
+      "loss": 1.0602,
+      "step": 5944
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 4.691964285714286e-05,
+      "loss": 1.0562,
+      "step": 5945
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 4.6910714285714286e-05,
+      "loss": 1.0625,
+      "step": 5946
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 4.690178571428572e-05,
+      "loss": 0.927,
+      "step": 5947
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 4.689285714285714e-05,
+      "loss": 1.0491,
+      "step": 5948
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 4.6883928571428576e-05,
+      "loss": 1.0653,
+      "step": 5949
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.9151,
+      "step": 5950
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 4.6866071428571426e-05,
+      "loss": 1.0731,
+      "step": 5951
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 4.685714285714286e-05,
+      "loss": 0.9177,
+      "step": 5952
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 4.6848214285714284e-05,
+      "loss": 1.032,
+      "step": 5953
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 4.6839285714285716e-05,
+      "loss": 1.0325,
+      "step": 5954
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 4.683035714285715e-05,
+      "loss": 1.0259,
+      "step": 5955
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 4.6821428571428574e-05,
+      "loss": 1.0579,
+      "step": 5956
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 4.6812500000000006e-05,
+      "loss": 1.0529,
+      "step": 5957
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 4.680357142857143e-05,
+      "loss": 0.9945,
+      "step": 5958
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 4.6794642857142864e-05,
+      "loss": 0.9893,
+      "step": 5959
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 4.678571428571429e-05,
+      "loss": 0.9843,
+      "step": 5960
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 4.6776785714285715e-05,
+      "loss": 1.076,
+      "step": 5961
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 4.676785714285715e-05,
+      "loss": 1.0363,
+      "step": 5962
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 4.675892857142857e-05,
+      "loss": 1.0608,
+      "step": 5963
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 1.0476,
+      "step": 5964
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 4.674107142857143e-05,
+      "loss": 1.023,
+      "step": 5965
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 4.6732142857142856e-05,
+      "loss": 1.0269,
+      "step": 5966
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 4.672321428571429e-05,
+      "loss": 1.0487,
+      "step": 5967
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 4.671428571428571e-05,
+      "loss": 0.9852,
+      "step": 5968
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 4.6705357142857146e-05,
+      "loss": 1.0785,
+      "step": 5969
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 4.669642857142857e-05,
+      "loss": 0.9885,
+      "step": 5970
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 4.66875e-05,
+      "loss": 1.0157,
+      "step": 5971
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 4.6678571428571435e-05,
+      "loss": 0.9618,
+      "step": 5972
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 4.666964285714286e-05,
+      "loss": 1.0124,
+      "step": 5973
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 4.6660714285714286e-05,
+      "loss": 1.0448,
+      "step": 5974
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 4.665178571428572e-05,
+      "loss": 1.0227,
+      "step": 5975
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 4.6642857142857144e-05,
+      "loss": 0.9525,
+      "step": 5976
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 4.6633928571428576e-05,
+      "loss": 1.0737,
+      "step": 5977
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 4.6625e-05,
+      "loss": 0.9904,
+      "step": 5978
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 4.661607142857143e-05,
+      "loss": 0.991,
+      "step": 5979
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 4.660714285714286e-05,
+      "loss": 1.0122,
+      "step": 5980
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 4.6598214285714285e-05,
+      "loss": 1.0292,
+      "step": 5981
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 4.658928571428572e-05,
+      "loss": 1.0385,
+      "step": 5982
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 4.658035714285714e-05,
+      "loss": 1.0011,
+      "step": 5983
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 4.6571428571428575e-05,
+      "loss": 1.0123,
+      "step": 5984
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 4.65625e-05,
+      "loss": 1.0566,
+      "step": 5985
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 4.655357142857143e-05,
+      "loss": 1.0228,
+      "step": 5986
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 4.6544642857142865e-05,
+      "loss": 0.9867,
+      "step": 5987
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 4.653571428571429e-05,
+      "loss": 1.0695,
+      "step": 5988
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 4.6526785714285716e-05,
+      "loss": 1.0587,
+      "step": 5989
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 4.651785714285715e-05,
+      "loss": 1.0437,
+      "step": 5990
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 4.650892857142857e-05,
+      "loss": 1.0364,
+      "step": 5991
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.0516,
+      "step": 5992
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 4.649107142857143e-05,
+      "loss": 0.9755,
+      "step": 5993
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 4.6482142857142856e-05,
+      "loss": 1.0496,
+      "step": 5994
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 4.647321428571429e-05,
+      "loss": 1.0206,
+      "step": 5995
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 4.6464285714285714e-05,
+      "loss": 1.0315,
+      "step": 5996
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 4.6455357142857146e-05,
+      "loss": 1.0652,
+      "step": 5997
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 4.644642857142857e-05,
+      "loss": 0.9603,
+      "step": 5998
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 4.64375e-05,
+      "loss": 0.9896,
+      "step": 5999
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 1.0746,
+      "step": 6000
+    },
+    {
+      "epoch": 26.79,
+      "eval_loss": 1.1184805631637573,
+      "eval_runtime": 45.0383,
+      "eval_samples_per_second": 290.73,
+      "eval_steps_per_second": 1.532,
+      "step": 6000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 4.641964285714286e-05,
+      "loss": 1.0188,
+      "step": 6001
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 4.6410714285714294e-05,
+      "loss": 0.9633,
+      "step": 6002
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 4.640178571428572e-05,
+      "loss": 1.0573,
+      "step": 6003
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 4.6392857142857145e-05,
+      "loss": 1.0697,
+      "step": 6004
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 4.638392857142858e-05,
+      "loss": 1.0349,
+      "step": 6005
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 4.6375e-05,
+      "loss": 1.0683,
+      "step": 6006
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 4.636607142857143e-05,
+      "loss": 1.0072,
+      "step": 6007
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 4.635714285714286e-05,
+      "loss": 1.0182,
+      "step": 6008
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 4.6348214285714286e-05,
+      "loss": 1.0448,
+      "step": 6009
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 4.633928571428572e-05,
+      "loss": 0.9847,
+      "step": 6010
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 4.633035714285714e-05,
+      "loss": 1.0056,
+      "step": 6011
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 4.632142857142857e-05,
+      "loss": 1.0193,
+      "step": 6012
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 4.63125e-05,
+      "loss": 1.0262,
+      "step": 6013
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 4.6303571428571426e-05,
+      "loss": 1.0873,
+      "step": 6014
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 4.629464285714286e-05,
+      "loss": 1.0114,
+      "step": 6015
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 4.628571428571429e-05,
+      "loss": 0.9716,
+      "step": 6016
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 4.6276785714285716e-05,
+      "loss": 1.0552,
+      "step": 6017
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 4.626785714285715e-05,
+      "loss": 1.0991,
+      "step": 6018
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 4.6258928571428574e-05,
+      "loss": 1.0443,
+      "step": 6019
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 1.054,
+      "step": 6020
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 4.624107142857143e-05,
+      "loss": 0.9315,
+      "step": 6021
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 4.623214285714286e-05,
+      "loss": 1.0076,
+      "step": 6022
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 4.622321428571429e-05,
+      "loss": 1.0155,
+      "step": 6023
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 4.6214285714285715e-05,
+      "loss": 1.0059,
+      "step": 6024
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 4.620535714285715e-05,
+      "loss": 0.9982,
+      "step": 6025
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 4.619642857142857e-05,
+      "loss": 1.051,
+      "step": 6026
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 4.61875e-05,
+      "loss": 1.1056,
+      "step": 6027
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 4.617857142857143e-05,
+      "loss": 1.0392,
+      "step": 6028
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 4.6169642857142855e-05,
+      "loss": 0.9554,
+      "step": 6029
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 4.616071428571429e-05,
+      "loss": 1.0048,
+      "step": 6030
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 4.615178571428572e-05,
+      "loss": 1.0381,
+      "step": 6031
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 4.6142857142857145e-05,
+      "loss": 1.0437,
+      "step": 6032
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 4.613392857142858e-05,
+      "loss": 1.02,
+      "step": 6033
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 4.6125e-05,
+      "loss": 1.0684,
+      "step": 6034
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 4.6116071428571435e-05,
+      "loss": 0.924,
+      "step": 6035
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 4.610714285714286e-05,
+      "loss": 1.0124,
+      "step": 6036
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 4.6098214285714286e-05,
+      "loss": 1.0956,
+      "step": 6037
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 4.608928571428572e-05,
+      "loss": 0.988,
+      "step": 6038
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 4.6080357142857144e-05,
+      "loss": 1.0106,
+      "step": 6039
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 4.607142857142857e-05,
+      "loss": 1.0644,
+      "step": 6040
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 4.60625e-05,
+      "loss": 0.9706,
+      "step": 6041
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 4.605357142857143e-05,
+      "loss": 0.9967,
+      "step": 6042
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 4.604464285714286e-05,
+      "loss": 1.0288,
+      "step": 6043
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 4.6035714285714285e-05,
+      "loss": 0.9694,
+      "step": 6044
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 4.602678571428572e-05,
+      "loss": 0.9585,
+      "step": 6045
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 4.601785714285715e-05,
+      "loss": 1.0556,
+      "step": 6046
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.6008928571428575e-05,
+      "loss": 0.9845,
+      "step": 6047
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.0647,
+      "step": 6048
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.599107142857143e-05,
+      "loss": 1.03,
+      "step": 6049
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 4.598214285714286e-05,
+      "loss": 1.0226,
+      "step": 6050
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 4.597321428571429e-05,
+      "loss": 0.9741,
+      "step": 6051
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 4.5964285714285715e-05,
+      "loss": 0.9975,
+      "step": 6052
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 4.595535714285715e-05,
+      "loss": 0.9812,
+      "step": 6053
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 4.594642857142857e-05,
+      "loss": 0.9788,
+      "step": 6054
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 4.59375e-05,
+      "loss": 1.063,
+      "step": 6055
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 4.592857142857143e-05,
+      "loss": 0.9565,
+      "step": 6056
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 4.5919642857142856e-05,
+      "loss": 1.0514,
+      "step": 6057
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 4.591071428571429e-05,
+      "loss": 0.9533,
+      "step": 6058
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 4.5901785714285714e-05,
+      "loss": 0.9879,
+      "step": 6059
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 4.5892857142857146e-05,
+      "loss": 1.0005,
+      "step": 6060
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 4.588392857142857e-05,
+      "loss": 1.0466,
+      "step": 6061
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 4.5875000000000004e-05,
+      "loss": 0.9979,
+      "step": 6062
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 4.5866071428571436e-05,
+      "loss": 0.9784,
+      "step": 6063
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 4.585714285714286e-05,
+      "loss": 1.0346,
+      "step": 6064
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 4.584821428571429e-05,
+      "loss": 1.0057,
+      "step": 6065
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 4.583928571428572e-05,
+      "loss": 0.9709,
+      "step": 6066
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 4.5830357142857145e-05,
+      "loss": 1.0062,
+      "step": 6067
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 4.582142857142858e-05,
+      "loss": 1.0073,
+      "step": 6068
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 4.58125e-05,
+      "loss": 1.0142,
+      "step": 6069
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 4.580357142857143e-05,
+      "loss": 1.0275,
+      "step": 6070
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 4.579464285714286e-05,
+      "loss": 1.0027,
+      "step": 6071
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 4.5785714285714285e-05,
+      "loss": 1.0584,
+      "step": 6072
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 4.577678571428571e-05,
+      "loss": 0.9465,
+      "step": 6073
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 4.576785714285714e-05,
+      "loss": 1.0136,
+      "step": 6074
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 4.5758928571428575e-05,
+      "loss": 0.9943,
+      "step": 6075
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 4.575e-05,
+      "loss": 1.0617,
+      "step": 6076
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 4.574107142857143e-05,
+      "loss": 0.9764,
+      "step": 6077
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 4.573214285714286e-05,
+      "loss": 1.0357,
+      "step": 6078
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 4.572321428571429e-05,
+      "loss": 0.9595,
+      "step": 6079
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 1.0658,
+      "step": 6080
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 4.570535714285715e-05,
+      "loss": 1.0119,
+      "step": 6081
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 4.5696428571428574e-05,
+      "loss": 1.066,
+      "step": 6082
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 4.56875e-05,
+      "loss": 1.0409,
+      "step": 6083
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 4.567857142857143e-05,
+      "loss": 1.054,
+      "step": 6084
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 4.566964285714286e-05,
+      "loss": 1.047,
+      "step": 6085
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 4.566071428571429e-05,
+      "loss": 1.0153,
+      "step": 6086
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 4.5651785714285715e-05,
+      "loss": 0.9671,
+      "step": 6087
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 4.564285714285714e-05,
+      "loss": 0.9697,
+      "step": 6088
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 4.563392857142857e-05,
+      "loss": 1.0337,
+      "step": 6089
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 4.5625e-05,
+      "loss": 0.9889,
+      "step": 6090
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 4.561607142857143e-05,
+      "loss": 1.0039,
+      "step": 6091
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 4.560714285714286e-05,
+      "loss": 0.9447,
+      "step": 6092
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 4.559821428571429e-05,
+      "loss": 1.0503,
+      "step": 6093
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 4.558928571428572e-05,
+      "loss": 0.9948,
+      "step": 6094
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 4.5580357142857145e-05,
+      "loss": 0.994,
+      "step": 6095
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 4.557142857142858e-05,
+      "loss": 1.0252,
+      "step": 6096
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 4.55625e-05,
+      "loss": 0.9534,
+      "step": 6097
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 4.555357142857143e-05,
+      "loss": 0.9885,
+      "step": 6098
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 4.554464285714286e-05,
+      "loss": 1.0108,
+      "step": 6099
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 4.5535714285714286e-05,
+      "loss": 0.9906,
+      "step": 6100
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 4.552678571428572e-05,
+      "loss": 1.0443,
+      "step": 6101
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 4.5517857142857144e-05,
+      "loss": 1.005,
+      "step": 6102
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 4.550892857142857e-05,
+      "loss": 0.9593,
+      "step": 6103
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 4.55e-05,
+      "loss": 1.0339,
+      "step": 6104
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 4.549107142857143e-05,
+      "loss": 1.0127,
+      "step": 6105
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 4.548214285714286e-05,
+      "loss": 1.0282,
+      "step": 6106
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 4.547321428571429e-05,
+      "loss": 1.0319,
+      "step": 6107
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 4.546428571428572e-05,
+      "loss": 1.0396,
+      "step": 6108
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 4.545535714285715e-05,
+      "loss": 1.0659,
+      "step": 6109
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 4.5446428571428574e-05,
+      "loss": 1.0713,
+      "step": 6110
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 4.54375e-05,
+      "loss": 0.9523,
+      "step": 6111
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 4.542857142857143e-05,
+      "loss": 0.9222,
+      "step": 6112
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 4.541964285714286e-05,
+      "loss": 1.0544,
+      "step": 6113
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 4.541071428571429e-05,
+      "loss": 0.996,
+      "step": 6114
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 4.5401785714285715e-05,
+      "loss": 1.0512,
+      "step": 6115
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 4.539285714285714e-05,
+      "loss": 0.9499,
+      "step": 6116
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 4.538392857142857e-05,
+      "loss": 0.9835,
+      "step": 6117
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 4.5375e-05,
+      "loss": 1.013,
+      "step": 6118
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 4.536607142857143e-05,
+      "loss": 1.0531,
+      "step": 6119
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 4.5357142857142856e-05,
+      "loss": 1.0258,
+      "step": 6120
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 4.534821428571429e-05,
+      "loss": 1.0037,
+      "step": 6121
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 4.533928571428572e-05,
+      "loss": 1.0145,
+      "step": 6122
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 4.5330357142857146e-05,
+      "loss": 1.0191,
+      "step": 6123
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 4.532142857142858e-05,
+      "loss": 1.0552,
+      "step": 6124
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 1.0552,
+      "step": 6125
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 4.530357142857143e-05,
+      "loss": 1.0283,
+      "step": 6126
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 4.529464285714286e-05,
+      "loss": 1.0169,
+      "step": 6127
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 4.528571428571429e-05,
+      "loss": 0.9822,
+      "step": 6128
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 4.527678571428572e-05,
+      "loss": 0.9861,
+      "step": 6129
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 4.5267857142857144e-05,
+      "loss": 1.0199,
+      "step": 6130
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 4.525892857142857e-05,
+      "loss": 1.0171,
+      "step": 6131
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 4.525e-05,
+      "loss": 0.9437,
+      "step": 6132
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 4.524107142857143e-05,
+      "loss": 0.9426,
+      "step": 6133
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 4.523214285714286e-05,
+      "loss": 0.99,
+      "step": 6134
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 4.5223214285714285e-05,
+      "loss": 1.0568,
+      "step": 6135
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 4.521428571428572e-05,
+      "loss": 1.0685,
+      "step": 6136
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 4.520535714285715e-05,
+      "loss": 1.0142,
+      "step": 6137
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 4.5196428571428575e-05,
+      "loss": 1.038,
+      "step": 6138
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 4.518750000000001e-05,
+      "loss": 1.0233,
+      "step": 6139
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 4.517857142857143e-05,
+      "loss": 1.0379,
+      "step": 6140
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 4.516964285714286e-05,
+      "loss": 1.001,
+      "step": 6141
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 4.516071428571429e-05,
+      "loss": 1.0056,
+      "step": 6142
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 4.5151785714285716e-05,
+      "loss": 0.9718,
+      "step": 6143
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 4.514285714285714e-05,
+      "loss": 1.0917,
+      "step": 6144
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 4.5133928571428574e-05,
+      "loss": 1.0585,
+      "step": 6145
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 4.5125e-05,
+      "loss": 1.0258,
+      "step": 6146
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 4.511607142857143e-05,
+      "loss": 0.9832,
+      "step": 6147
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 4.510714285714286e-05,
+      "loss": 1.0075,
+      "step": 6148
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 4.509821428571428e-05,
+      "loss": 1.0215,
+      "step": 6149
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 4.5089285714285714e-05,
+      "loss": 1.083,
+      "step": 6150
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 4.508035714285715e-05,
+      "loss": 0.9911,
+      "step": 6151
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 4.507142857142858e-05,
+      "loss": 1.0369,
+      "step": 6152
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 4.5062500000000004e-05,
+      "loss": 1.0886,
+      "step": 6153
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 4.505357142857143e-05,
+      "loss": 1.0184,
+      "step": 6154
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 4.504464285714286e-05,
+      "loss": 0.9497,
+      "step": 6155
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 4.503571428571429e-05,
+      "loss": 0.9767,
+      "step": 6156
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 4.502678571428572e-05,
+      "loss": 1.0866,
+      "step": 6157
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 4.5017857142857145e-05,
+      "loss": 0.9857,
+      "step": 6158
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 4.500892857142857e-05,
+      "loss": 1.0461,
+      "step": 6159
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 4.5e-05,
+      "loss": 1.0239,
+      "step": 6160
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 4.499107142857143e-05,
+      "loss": 1.0543,
+      "step": 6161
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 4.498214285714286e-05,
+      "loss": 0.991,
+      "step": 6162
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 4.4973214285714286e-05,
+      "loss": 1.1062,
+      "step": 6163
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 4.496428571428571e-05,
+      "loss": 0.9564,
+      "step": 6164
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 4.4955357142857144e-05,
+      "loss": 1.081,
+      "step": 6165
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 4.4946428571428576e-05,
+      "loss": 1.0022,
+      "step": 6166
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 4.49375e-05,
+      "loss": 1.0029,
+      "step": 6167
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 4.4928571428571434e-05,
+      "loss": 0.9653,
+      "step": 6168
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 4.491964285714286e-05,
+      "loss": 0.9923,
+      "step": 6169
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 4.491071428571429e-05,
+      "loss": 1.0056,
+      "step": 6170
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 4.490178571428572e-05,
+      "loss": 1.0358,
+      "step": 6171
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 4.489285714285715e-05,
+      "loss": 1.0455,
+      "step": 6172
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 4.4883928571428574e-05,
+      "loss": 0.9764,
+      "step": 6173
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 4.4875e-05,
+      "loss": 1.0276,
+      "step": 6174
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 4.486607142857143e-05,
+      "loss": 1.0631,
+      "step": 6175
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 4.485714285714286e-05,
+      "loss": 1.0815,
+      "step": 6176
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 4.484821428571429e-05,
+      "loss": 1.0526,
+      "step": 6177
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 4.4839285714285715e-05,
+      "loss": 0.9708,
+      "step": 6178
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 4.483035714285714e-05,
+      "loss": 1.0031,
+      "step": 6179
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 4.482142857142857e-05,
+      "loss": 1.032,
+      "step": 6180
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 4.4812500000000005e-05,
+      "loss": 0.9389,
+      "step": 6181
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 4.480357142857143e-05,
+      "loss": 0.9779,
+      "step": 6182
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 4.479464285714286e-05,
+      "loss": 1.0504,
+      "step": 6183
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 4.478571428571429e-05,
+      "loss": 1.0048,
+      "step": 6184
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 4.477678571428572e-05,
+      "loss": 0.9726,
+      "step": 6185
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 4.4767857142857146e-05,
+      "loss": 1.02,
+      "step": 6186
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 4.475892857142857e-05,
+      "loss": 1.1257,
+      "step": 6187
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 0.915,
+      "step": 6188
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 4.474107142857143e-05,
+      "loss": 1.0295,
+      "step": 6189
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 4.473214285714286e-05,
+      "loss": 1.0459,
+      "step": 6190
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 4.472321428571429e-05,
+      "loss": 1.0448,
+      "step": 6191
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 4.471428571428571e-05,
+      "loss": 0.9371,
+      "step": 6192
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 4.4705357142857144e-05,
+      "loss": 0.9886,
+      "step": 6193
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 4.469642857142857e-05,
+      "loss": 1.0555,
+      "step": 6194
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 4.46875e-05,
+      "loss": 0.9683,
+      "step": 6195
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 4.467857142857143e-05,
+      "loss": 0.9799,
+      "step": 6196
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 4.466964285714286e-05,
+      "loss": 1.0489,
+      "step": 6197
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 4.466071428571429e-05,
+      "loss": 0.9799,
+      "step": 6198
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 4.465178571428572e-05,
+      "loss": 1.0032,
+      "step": 6199
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 1.008,
+      "step": 6200
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 4.4633928571428575e-05,
+      "loss": 0.9653,
+      "step": 6201
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 4.4625e-05,
+      "loss": 1.0205,
+      "step": 6202
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 4.461607142857143e-05,
+      "loss": 0.9843,
+      "step": 6203
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 4.460714285714286e-05,
+      "loss": 1.0065,
+      "step": 6204
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 4.459821428571429e-05,
+      "loss": 1.1466,
+      "step": 6205
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 4.4589285714285716e-05,
+      "loss": 1.0384,
+      "step": 6206
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 4.458035714285714e-05,
+      "loss": 0.9625,
+      "step": 6207
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 4.4571428571428574e-05,
+      "loss": 0.993,
+      "step": 6208
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 4.45625e-05,
+      "loss": 1.0182,
+      "step": 6209
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 4.455357142857143e-05,
+      "loss": 1.0551,
+      "step": 6210
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 4.454464285714286e-05,
+      "loss": 1.042,
+      "step": 6211
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 4.453571428571429e-05,
+      "loss": 0.9417,
+      "step": 6212
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 4.452678571428572e-05,
+      "loss": 0.9851,
+      "step": 6213
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 4.4517857142857147e-05,
+      "loss": 0.9619,
+      "step": 6214
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 4.450892857142857e-05,
+      "loss": 1.0109,
+      "step": 6215
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.1254,
+      "step": 6216
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 4.449107142857143e-05,
+      "loss": 1.0046,
+      "step": 6217
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 4.448214285714286e-05,
+      "loss": 1.0207,
+      "step": 6218
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 4.447321428571429e-05,
+      "loss": 1.0316,
+      "step": 6219
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 4.446428571428571e-05,
+      "loss": 1.091,
+      "step": 6220
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 4.4455357142857145e-05,
+      "loss": 0.9373,
+      "step": 6221
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 4.444642857142857e-05,
+      "loss": 1.0183,
+      "step": 6222
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 4.44375e-05,
+      "loss": 1.0052,
+      "step": 6223
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 4.442857142857143e-05,
+      "loss": 0.9942,
+      "step": 6224
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 4.4419642857142854e-05,
+      "loss": 1.0626,
+      "step": 6225
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 4.4410714285714286e-05,
+      "loss": 1.0211,
+      "step": 6226
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 4.440178571428572e-05,
+      "loss": 0.9942,
+      "step": 6227
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 4.439285714285715e-05,
+      "loss": 1.0441,
+      "step": 6228
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 4.4383928571428576e-05,
+      "loss": 0.9899,
+      "step": 6229
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 4.4375e-05,
+      "loss": 1.0623,
+      "step": 6230
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 4.4366071428571433e-05,
+      "loss": 0.9746,
+      "step": 6231
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 4.435714285714286e-05,
+      "loss": 0.9784,
+      "step": 6232
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 4.434821428571429e-05,
+      "loss": 0.9641,
+      "step": 6233
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 4.4339285714285717e-05,
+      "loss": 0.9804,
+      "step": 6234
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 4.433035714285714e-05,
+      "loss": 0.9607,
+      "step": 6235
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 4.4321428571428574e-05,
+      "loss": 1.074,
+      "step": 6236
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 4.43125e-05,
+      "loss": 1.0696,
+      "step": 6237
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 4.430357142857143e-05,
+      "loss": 1.0226,
+      "step": 6238
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 4.429464285714286e-05,
+      "loss": 0.9844,
+      "step": 6239
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 4.428571428571428e-05,
+      "loss": 1.0414,
+      "step": 6240
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 4.4276785714285715e-05,
+      "loss": 1.0331,
+      "step": 6241
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 4.426785714285715e-05,
+      "loss": 0.9785,
+      "step": 6242
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 4.425892857142858e-05,
+      "loss": 1.0341,
+      "step": 6243
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 0.9943,
+      "step": 6244
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 4.424107142857143e-05,
+      "loss": 0.9302,
+      "step": 6245
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 4.423214285714286e-05,
+      "loss": 0.9913,
+      "step": 6246
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 4.422321428571429e-05,
+      "loss": 1.04,
+      "step": 6247
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 4.4214285714285714e-05,
+      "loss": 1.0195,
+      "step": 6248
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 4.4205357142857146e-05,
+      "loss": 1.0205,
+      "step": 6249
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 4.419642857142857e-05,
+      "loss": 1.0036,
+      "step": 6250
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 4.4187500000000003e-05,
+      "loss": 0.9954,
+      "step": 6251
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 4.417857142857143e-05,
+      "loss": 1.0381,
+      "step": 6252
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 4.4169642857142854e-05,
+      "loss": 0.977,
+      "step": 6253
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 4.4160714285714287e-05,
+      "loss": 0.9651,
+      "step": 6254
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 4.415178571428571e-05,
+      "loss": 0.9899,
+      "step": 6255
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 4.4142857142857144e-05,
+      "loss": 1.0857,
+      "step": 6256
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 4.4133928571428576e-05,
+      "loss": 0.9805,
+      "step": 6257
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 4.4125e-05,
+      "loss": 0.9697,
+      "step": 6258
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 4.4116071428571434e-05,
+      "loss": 1.0165,
+      "step": 6259
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 4.410714285714286e-05,
+      "loss": 1.0152,
+      "step": 6260
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 4.409821428571429e-05,
+      "loss": 1.0948,
+      "step": 6261
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 4.408928571428572e-05,
+      "loss": 0.9804,
+      "step": 6262
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 4.408035714285714e-05,
+      "loss": 1.0019,
+      "step": 6263
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 4.4071428571428575e-05,
+      "loss": 0.9833,
+      "step": 6264
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 4.40625e-05,
+      "loss": 1.0095,
+      "step": 6265
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 4.405357142857143e-05,
+      "loss": 1.0683,
+      "step": 6266
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 4.404464285714286e-05,
+      "loss": 1.0749,
+      "step": 6267
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 4.4035714285714284e-05,
+      "loss": 1.0107,
+      "step": 6268
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 4.4026785714285716e-05,
+      "loss": 0.9856,
+      "step": 6269
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 4.401785714285714e-05,
+      "loss": 1.0198,
+      "step": 6270
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.4008928571428573e-05,
+      "loss": 0.9809,
+      "step": 6271
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.9789,
+      "step": 6272
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.399107142857143e-05,
+      "loss": 1.044,
+      "step": 6273
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.398214285714286e-05,
+      "loss": 1.0385,
+      "step": 6274
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.397321428571429e-05,
+      "loss": 1.0468,
+      "step": 6275
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.396428571428572e-05,
+      "loss": 0.9559,
+      "step": 6276
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.3955357142857146e-05,
+      "loss": 1.0755,
+      "step": 6277
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.394642857142857e-05,
+      "loss": 0.9988,
+      "step": 6278
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.3937500000000004e-05,
+      "loss": 0.9871,
+      "step": 6279
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.392857142857143e-05,
+      "loss": 1.0816,
+      "step": 6280
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.391964285714286e-05,
+      "loss": 1.0263,
+      "step": 6281
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.391071428571429e-05,
+      "loss": 1.033,
+      "step": 6282
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.390178571428571e-05,
+      "loss": 1.0986,
+      "step": 6283
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.3892857142857145e-05,
+      "loss": 1.0583,
+      "step": 6284
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.388392857142857e-05,
+      "loss": 1.0928,
+      "step": 6285
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.3875e-05,
+      "loss": 1.012,
+      "step": 6286
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.386607142857143e-05,
+      "loss": 1.0831,
+      "step": 6287
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.385714285714286e-05,
+      "loss": 1.0554,
+      "step": 6288
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.384821428571429e-05,
+      "loss": 1.0056,
+      "step": 6289
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.383928571428572e-05,
+      "loss": 0.9176,
+      "step": 6290
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.3830357142857143e-05,
+      "loss": 0.9399,
+      "step": 6291
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.3821428571428576e-05,
+      "loss": 1.0108,
+      "step": 6292
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.38125e-05,
+      "loss": 1.0283,
+      "step": 6293
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.380357142857143e-05,
+      "loss": 0.9553,
+      "step": 6294
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.379464285714286e-05,
+      "loss": 0.9909,
+      "step": 6295
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.3785714285714284e-05,
+      "loss": 1.0185,
+      "step": 6296
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.3776785714285716e-05,
+      "loss": 1.0249,
+      "step": 6297
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.376785714285714e-05,
+      "loss": 1.0252,
+      "step": 6298
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.3758928571428574e-05,
+      "loss": 0.9117,
+      "step": 6299
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.375e-05,
+      "loss": 1.0315,
+      "step": 6300
+    },
+    {
+      "epoch": 28.12,
+      "eval_loss": 1.1168270111083984,
+      "eval_runtime": 45.1181,
+      "eval_samples_per_second": 290.216,
+      "eval_steps_per_second": 1.529,
+      "step": 6300
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.374107142857143e-05,
+      "loss": 1.023,
+      "step": 6301
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.373214285714286e-05,
+      "loss": 1.0114,
+      "step": 6302
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.372321428571429e-05,
+      "loss": 1.042,
+      "step": 6303
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.371428571428572e-05,
+      "loss": 1.0129,
+      "step": 6304
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.370535714285715e-05,
+      "loss": 1.0603,
+      "step": 6305
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.369642857142857e-05,
+      "loss": 1.0111,
+      "step": 6306
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.3687500000000005e-05,
+      "loss": 1.0305,
+      "step": 6307
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.367857142857143e-05,
+      "loss": 1.0516,
+      "step": 6308
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.366964285714286e-05,
+      "loss": 0.9264,
+      "step": 6309
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.366071428571429e-05,
+      "loss": 0.998,
+      "step": 6310
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.3651785714285713e-05,
+      "loss": 0.9827,
+      "step": 6311
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.3642857142857146e-05,
+      "loss": 0.9179,
+      "step": 6312
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.363392857142857e-05,
+      "loss": 1.0284,
+      "step": 6313
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.3625e-05,
+      "loss": 1.0777,
+      "step": 6314
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.361607142857143e-05,
+      "loss": 1.0635,
+      "step": 6315
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.3607142857142854e-05,
+      "loss": 1.0432,
+      "step": 6316
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.3598214285714286e-05,
+      "loss": 1.0422,
+      "step": 6317
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.358928571428572e-05,
+      "loss": 0.9926,
+      "step": 6318
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.3580357142857144e-05,
+      "loss": 0.9553,
+      "step": 6319
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.3571428571428576e-05,
+      "loss": 1.0297,
+      "step": 6320
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.35625e-05,
+      "loss": 0.9911,
+      "step": 6321
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.3553571428571434e-05,
+      "loss": 0.9937,
+      "step": 6322
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.354464285714286e-05,
+      "loss": 1.0335,
+      "step": 6323
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.3535714285714285e-05,
+      "loss": 1.0031,
+      "step": 6324
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.352678571428572e-05,
+      "loss": 0.9722,
+      "step": 6325
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.351785714285714e-05,
+      "loss": 1.0481,
+      "step": 6326
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.3508928571428575e-05,
+      "loss": 1.0201,
+      "step": 6327
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.35e-05,
+      "loss": 0.9476,
+      "step": 6328
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.3491071428571426e-05,
+      "loss": 0.9279,
+      "step": 6329
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.348214285714286e-05,
+      "loss": 1.0543,
+      "step": 6330
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.347321428571428e-05,
+      "loss": 0.9723,
+      "step": 6331
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3464285714285716e-05,
+      "loss": 0.9861,
+      "step": 6332
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.345535714285715e-05,
+      "loss": 1.002,
+      "step": 6333
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.344642857142857e-05,
+      "loss": 1.0728,
+      "step": 6334
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.3437500000000006e-05,
+      "loss": 0.9847,
+      "step": 6335
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.342857142857143e-05,
+      "loss": 0.9995,
+      "step": 6336
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.341964285714286e-05,
+      "loss": 0.998,
+      "step": 6337
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.341071428571429e-05,
+      "loss": 0.9597,
+      "step": 6338
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.3401785714285714e-05,
+      "loss": 0.977,
+      "step": 6339
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.3392857142857146e-05,
+      "loss": 1.0112,
+      "step": 6340
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.338392857142857e-05,
+      "loss": 1.0379,
+      "step": 6341
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.3375000000000004e-05,
+      "loss": 0.9952,
+      "step": 6342
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.336607142857143e-05,
+      "loss": 1.1227,
+      "step": 6343
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.3357142857142855e-05,
+      "loss": 1.0,
+      "step": 6344
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.334821428571429e-05,
+      "loss": 1.0155,
+      "step": 6345
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.333928571428571e-05,
+      "loss": 0.8964,
+      "step": 6346
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.3330357142857145e-05,
+      "loss": 0.9914,
+      "step": 6347
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.332142857142858e-05,
+      "loss": 1.0069,
+      "step": 6348
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.33125e-05,
+      "loss": 1.0511,
+      "step": 6349
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.3303571428571435e-05,
+      "loss": 1.0847,
+      "step": 6350
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.329464285714286e-05,
+      "loss": 1.0265,
+      "step": 6351
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.328571428571429e-05,
+      "loss": 0.9304,
+      "step": 6352
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.327678571428572e-05,
+      "loss": 0.9548,
+      "step": 6353
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.326785714285714e-05,
+      "loss": 0.9992,
+      "step": 6354
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.3258928571428576e-05,
+      "loss": 0.9146,
+      "step": 6355
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.325e-05,
+      "loss": 0.9621,
+      "step": 6356
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.3241071428571426e-05,
+      "loss": 1.0177,
+      "step": 6357
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.323214285714286e-05,
+      "loss": 1.081,
+      "step": 6358
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.3223214285714284e-05,
+      "loss": 0.9316,
+      "step": 6359
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.3214285714285716e-05,
+      "loss": 1.0286,
+      "step": 6360
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.320535714285714e-05,
+      "loss": 1.0397,
+      "step": 6361
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.3196428571428574e-05,
+      "loss": 1.0179,
+      "step": 6362
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 4.3187500000000006e-05,
+      "loss": 0.9308,
+      "step": 6363
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 4.317857142857143e-05,
+      "loss": 1.0458,
+      "step": 6364
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 4.3169642857142864e-05,
+      "loss": 0.988,
+      "step": 6365
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 4.316071428571429e-05,
+      "loss": 0.981,
+      "step": 6366
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 4.3151785714285715e-05,
+      "loss": 1.0727,
+      "step": 6367
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 4.314285714285715e-05,
+      "loss": 1.0025,
+      "step": 6368
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 4.313392857142857e-05,
+      "loss": 1.0074,
+      "step": 6369
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 4.3125000000000005e-05,
+      "loss": 1.0005,
+      "step": 6370
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 4.311607142857143e-05,
+      "loss": 1.0307,
+      "step": 6371
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 4.3107142857142856e-05,
+      "loss": 1.0105,
+      "step": 6372
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 4.309821428571429e-05,
+      "loss": 0.9356,
+      "step": 6373
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 4.308928571428571e-05,
+      "loss": 1.013,
+      "step": 6374
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 4.3080357142857145e-05,
+      "loss": 1.0044,
+      "step": 6375
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 4.307142857142857e-05,
+      "loss": 0.956,
+      "step": 6376
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 4.30625e-05,
+      "loss": 0.9653,
+      "step": 6377
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 4.305357142857143e-05,
+      "loss": 0.9817,
+      "step": 6378
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 4.304464285714286e-05,
+      "loss": 0.9807,
+      "step": 6379
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 4.303571428571429e-05,
+      "loss": 0.9788,
+      "step": 6380
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 4.302678571428572e-05,
+      "loss": 1.0132,
+      "step": 6381
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 4.3017857142857144e-05,
+      "loss": 1.0423,
+      "step": 6382
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 4.3008928571428576e-05,
+      "loss": 1.0317,
+      "step": 6383
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 4.3e-05,
+      "loss": 1.0011,
+      "step": 6384
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 4.2991071428571434e-05,
+      "loss": 0.9473,
+      "step": 6385
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 4.298214285714286e-05,
+      "loss": 1.0037,
+      "step": 6386
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 4.2973214285714285e-05,
+      "loss": 1.0472,
+      "step": 6387
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 4.296428571428572e-05,
+      "loss": 0.9674,
+      "step": 6388
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 4.295535714285714e-05,
+      "loss": 1.0604,
+      "step": 6389
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 4.294642857142857e-05,
+      "loss": 1.1098,
+      "step": 6390
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 4.29375e-05,
+      "loss": 1.0367,
+      "step": 6391
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 4.292857142857143e-05,
+      "loss": 1.092,
+      "step": 6392
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 4.291964285714286e-05,
+      "loss": 1.0256,
+      "step": 6393
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 4.291071428571429e-05,
+      "loss": 1.0405,
+      "step": 6394
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 4.2901785714285715e-05,
+      "loss": 0.9454,
+      "step": 6395
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 4.289285714285715e-05,
+      "loss": 0.9675,
+      "step": 6396
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 4.288392857142857e-05,
+      "loss": 0.9733,
+      "step": 6397
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 4.2875000000000005e-05,
+      "loss": 1.0036,
+      "step": 6398
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 4.286607142857143e-05,
+      "loss": 1.066,
+      "step": 6399
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 1.0131,
+      "step": 6400
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 4.284821428571429e-05,
+      "loss": 1.0175,
+      "step": 6401
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 4.2839285714285714e-05,
+      "loss": 0.9924,
+      "step": 6402
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 4.2830357142857146e-05,
+      "loss": 1.0543,
+      "step": 6403
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 4.282142857142857e-05,
+      "loss": 0.9693,
+      "step": 6404
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 4.28125e-05,
+      "loss": 1.03,
+      "step": 6405
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 4.280357142857143e-05,
+      "loss": 1.1262,
+      "step": 6406
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 4.2794642857142855e-05,
+      "loss": 0.9656,
+      "step": 6407
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 4.278571428571429e-05,
+      "loss": 0.9901,
+      "step": 6408
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 4.277678571428572e-05,
+      "loss": 0.9921,
+      "step": 6409
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 4.2767857142857145e-05,
+      "loss": 1.0726,
+      "step": 6410
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 4.275892857142858e-05,
+      "loss": 0.9937,
+      "step": 6411
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 4.275e-05,
+      "loss": 0.9893,
+      "step": 6412
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 4.2741071428571435e-05,
+      "loss": 0.9806,
+      "step": 6413
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 4.273214285714286e-05,
+      "loss": 0.963,
+      "step": 6414
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 4.2723214285714285e-05,
+      "loss": 0.9756,
+      "step": 6415
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 4.271428571428572e-05,
+      "loss": 1.0045,
+      "step": 6416
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 4.270535714285714e-05,
+      "loss": 0.9731,
+      "step": 6417
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 4.2696428571428575e-05,
+      "loss": 1.0598,
+      "step": 6418
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 4.26875e-05,
+      "loss": 1.0317,
+      "step": 6419
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 4.2678571428571426e-05,
+      "loss": 0.9755,
+      "step": 6420
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 4.266964285714286e-05,
+      "loss": 1.0172,
+      "step": 6421
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 4.2660714285714284e-05,
+      "loss": 0.9362,
+      "step": 6422
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 4.2651785714285716e-05,
+      "loss": 1.0332,
+      "step": 6423
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 4.264285714285715e-05,
+      "loss": 0.9613,
+      "step": 6424
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 4.2633928571428574e-05,
+      "loss": 1.0207,
+      "step": 6425
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 4.2625000000000006e-05,
+      "loss": 1.0196,
+      "step": 6426
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 4.261607142857143e-05,
+      "loss": 1.0282,
+      "step": 6427
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 4.260714285714286e-05,
+      "loss": 0.9852,
+      "step": 6428
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 4.259821428571429e-05,
+      "loss": 1.0258,
+      "step": 6429
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 4.2589285714285715e-05,
+      "loss": 1.0032,
+      "step": 6430
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 4.258035714285715e-05,
+      "loss": 1.0039,
+      "step": 6431
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 4.257142857142857e-05,
+      "loss": 0.9879,
+      "step": 6432
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 4.25625e-05,
+      "loss": 1.0326,
+      "step": 6433
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 4.255357142857143e-05,
+      "loss": 1.0152,
+      "step": 6434
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 4.2544642857142855e-05,
+      "loss": 1.0042,
+      "step": 6435
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 4.253571428571429e-05,
+      "loss": 1.004,
+      "step": 6436
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 4.252678571428571e-05,
+      "loss": 1.0325,
+      "step": 6437
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 4.2517857142857145e-05,
+      "loss": 1.0524,
+      "step": 6438
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 4.250892857142858e-05,
+      "loss": 1.026,
+      "step": 6439
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 4.25e-05,
+      "loss": 1.0202,
+      "step": 6440
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 4.2491071428571435e-05,
+      "loss": 0.9876,
+      "step": 6441
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 4.248214285714286e-05,
+      "loss": 0.9687,
+      "step": 6442
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 4.2473214285714286e-05,
+      "loss": 0.9787,
+      "step": 6443
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 4.246428571428572e-05,
+      "loss": 0.9739,
+      "step": 6444
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 4.2455357142857144e-05,
+      "loss": 0.9537,
+      "step": 6445
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 4.2446428571428576e-05,
+      "loss": 0.9294,
+      "step": 6446
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 4.24375e-05,
+      "loss": 1.049,
+      "step": 6447
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 4.242857142857143e-05,
+      "loss": 0.9879,
+      "step": 6448
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 4.241964285714286e-05,
+      "loss": 0.9467,
+      "step": 6449
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 4.2410714285714285e-05,
+      "loss": 1.0516,
+      "step": 6450
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 4.240178571428572e-05,
+      "loss": 0.9766,
+      "step": 6451
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 4.239285714285714e-05,
+      "loss": 0.9871,
+      "step": 6452
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 4.2383928571428575e-05,
+      "loss": 1.0112,
+      "step": 6453
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 4.237500000000001e-05,
+      "loss": 1.0017,
+      "step": 6454
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 4.236607142857143e-05,
+      "loss": 0.9634,
+      "step": 6455
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 4.2357142857142864e-05,
+      "loss": 1.0702,
+      "step": 6456
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 4.234821428571429e-05,
+      "loss": 0.967,
+      "step": 6457
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 4.2339285714285715e-05,
+      "loss": 0.9706,
+      "step": 6458
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 4.233035714285715e-05,
+      "loss": 1.0042,
+      "step": 6459
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 4.232142857142857e-05,
+      "loss": 0.986,
+      "step": 6460
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 4.23125e-05,
+      "loss": 1.0579,
+      "step": 6461
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 4.230357142857143e-05,
+      "loss": 0.9455,
+      "step": 6462
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 4.2294642857142856e-05,
+      "loss": 0.9403,
+      "step": 6463
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 4.228571428571429e-05,
+      "loss": 1.0223,
+      "step": 6464
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 4.2276785714285714e-05,
+      "loss": 0.9138,
+      "step": 6465
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 4.226785714285714e-05,
+      "loss": 0.9652,
+      "step": 6466
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 4.225892857142857e-05,
+      "loss": 0.9437,
+      "step": 6467
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 0.9521,
+      "step": 6468
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 4.2241071428571436e-05,
+      "loss": 0.9412,
+      "step": 6469
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 4.223214285714286e-05,
+      "loss": 1.0286,
+      "step": 6470
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 4.222321428571429e-05,
+      "loss": 0.9627,
+      "step": 6471
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 4.221428571428572e-05,
+      "loss": 1.0156,
+      "step": 6472
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 4.2205357142857145e-05,
+      "loss": 1.0073,
+      "step": 6473
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 4.219642857142858e-05,
+      "loss": 1.0101,
+      "step": 6474
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.9959,
+      "step": 6475
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 4.217857142857143e-05,
+      "loss": 0.9314,
+      "step": 6476
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 4.216964285714286e-05,
+      "loss": 1.1159,
+      "step": 6477
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 4.2160714285714285e-05,
+      "loss": 0.9711,
+      "step": 6478
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 4.215178571428572e-05,
+      "loss": 0.9954,
+      "step": 6479
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 4.214285714285714e-05,
+      "loss": 0.9584,
+      "step": 6480
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 4.213392857142857e-05,
+      "loss": 0.9534,
+      "step": 6481
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 4.2125e-05,
+      "loss": 0.9934,
+      "step": 6482
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 4.211607142857143e-05,
+      "loss": 1.0314,
+      "step": 6483
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 4.210714285714286e-05,
+      "loss": 0.9986,
+      "step": 6484
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 4.209821428571429e-05,
+      "loss": 1.0112,
+      "step": 6485
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 4.2089285714285716e-05,
+      "loss": 1.0313,
+      "step": 6486
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 4.208035714285715e-05,
+      "loss": 0.9958,
+      "step": 6487
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 4.2071428571428574e-05,
+      "loss": 1.0041,
+      "step": 6488
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 4.2062500000000006e-05,
+      "loss": 0.9196,
+      "step": 6489
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 4.205357142857143e-05,
+      "loss": 0.9731,
+      "step": 6490
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 4.204464285714286e-05,
+      "loss": 1.0888,
+      "step": 6491
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 4.203571428571429e-05,
+      "loss": 1.0154,
+      "step": 6492
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 4.2026785714285715e-05,
+      "loss": 0.996,
+      "step": 6493
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 4.201785714285714e-05,
+      "loss": 1.0168,
+      "step": 6494
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 4.200892857142857e-05,
+      "loss": 1.0263,
+      "step": 6495
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 4.2e-05,
+      "loss": 1.0976,
+      "step": 6496
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 4.199107142857143e-05,
+      "loss": 0.956,
+      "step": 6497
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 4.198214285714286e-05,
+      "loss": 0.9959,
+      "step": 6498
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 4.197321428571429e-05,
+      "loss": 0.9721,
+      "step": 6499
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 4.196428571428572e-05,
+      "loss": 0.9941,
+      "step": 6500
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 4.1955357142857145e-05,
+      "loss": 1.0685,
+      "step": 6501
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 4.194642857142858e-05,
+      "loss": 1.0195,
+      "step": 6502
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 4.19375e-05,
+      "loss": 1.0036,
+      "step": 6503
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 4.192857142857143e-05,
+      "loss": 0.9568,
+      "step": 6504
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 4.191964285714286e-05,
+      "loss": 1.0136,
+      "step": 6505
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 4.1910714285714286e-05,
+      "loss": 1.0533,
+      "step": 6506
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 4.190178571428572e-05,
+      "loss": 0.9835,
+      "step": 6507
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 4.1892857142857144e-05,
+      "loss": 1.0226,
+      "step": 6508
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 4.188392857142857e-05,
+      "loss": 1.0096,
+      "step": 6509
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 4.1875e-05,
+      "loss": 1.0055,
+      "step": 6510
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 4.186607142857143e-05,
+      "loss": 0.989,
+      "step": 6511
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 4.185714285714286e-05,
+      "loss": 0.9527,
+      "step": 6512
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 4.1848214285714285e-05,
+      "loss": 1.0472,
+      "step": 6513
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 4.183928571428572e-05,
+      "loss": 1.0421,
+      "step": 6514
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 4.183035714285715e-05,
+      "loss": 0.9692,
+      "step": 6515
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 4.1821428571428574e-05,
+      "loss": 1.0394,
+      "step": 6516
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 4.181250000000001e-05,
+      "loss": 0.9589,
+      "step": 6517
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 4.180357142857143e-05,
+      "loss": 0.9534,
+      "step": 6518
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 4.179464285714286e-05,
+      "loss": 0.9717,
+      "step": 6519
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 4.178571428571429e-05,
+      "loss": 1.0469,
+      "step": 6520
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 4.1776785714285715e-05,
+      "loss": 1.0126,
+      "step": 6521
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 4.176785714285715e-05,
+      "loss": 1.0134,
+      "step": 6522
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 4.175892857142857e-05,
+      "loss": 0.9669,
+      "step": 6523
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 4.175e-05,
+      "loss": 0.9856,
+      "step": 6524
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 4.174107142857143e-05,
+      "loss": 1.0969,
+      "step": 6525
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 4.1732142857142856e-05,
+      "loss": 1.0048,
+      "step": 6526
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 4.172321428571429e-05,
+      "loss": 0.9454,
+      "step": 6527
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 4.1714285714285714e-05,
+      "loss": 1.021,
+      "step": 6528
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 4.1705357142857146e-05,
+      "loss": 0.9268,
+      "step": 6529
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 4.169642857142858e-05,
+      "loss": 0.9809,
+      "step": 6530
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 4.1687500000000004e-05,
+      "loss": 0.9855,
+      "step": 6531
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 4.167857142857143e-05,
+      "loss": 0.9822,
+      "step": 6532
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 4.166964285714286e-05,
+      "loss": 0.9185,
+      "step": 6533
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 4.166071428571429e-05,
+      "loss": 0.9587,
+      "step": 6534
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 4.165178571428572e-05,
+      "loss": 1.0579,
+      "step": 6535
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 4.1642857142857144e-05,
+      "loss": 0.9948,
+      "step": 6536
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 4.163392857142857e-05,
+      "loss": 1.0486,
+      "step": 6537
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 4.1625e-05,
+      "loss": 1.0113,
+      "step": 6538
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 4.161607142857143e-05,
+      "loss": 1.0347,
+      "step": 6539
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 4.160714285714286e-05,
+      "loss": 1.0129,
+      "step": 6540
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 4.1598214285714285e-05,
+      "loss": 0.9416,
+      "step": 6541
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 4.158928571428571e-05,
+      "loss": 1.0408,
+      "step": 6542
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 4.158035714285714e-05,
+      "loss": 0.9913,
+      "step": 6543
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 4.1571428571428575e-05,
+      "loss": 0.98,
+      "step": 6544
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 4.156250000000001e-05,
+      "loss": 1.0396,
+      "step": 6545
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 4.155357142857143e-05,
+      "loss": 0.9479,
+      "step": 6546
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 4.154464285714286e-05,
+      "loss": 1.0017,
+      "step": 6547
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 4.153571428571429e-05,
+      "loss": 1.0197,
+      "step": 6548
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 4.1526785714285716e-05,
+      "loss": 0.9584,
+      "step": 6549
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 4.151785714285715e-05,
+      "loss": 1.0162,
+      "step": 6550
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 4.1508928571428574e-05,
+      "loss": 0.9921,
+      "step": 6551
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 4.15e-05,
+      "loss": 0.9828,
+      "step": 6552
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 4.149107142857143e-05,
+      "loss": 0.9788,
+      "step": 6553
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 4.148214285714286e-05,
+      "loss": 1.0227,
+      "step": 6554
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 4.147321428571429e-05,
+      "loss": 1.0206,
+      "step": 6555
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 4.1464285714285714e-05,
+      "loss": 1.0709,
+      "step": 6556
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 4.145535714285714e-05,
+      "loss": 1.0352,
+      "step": 6557
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 4.144642857142857e-05,
+      "loss": 0.945,
+      "step": 6558
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 4.1437500000000004e-05,
+      "loss": 0.9799,
+      "step": 6559
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 1.0551,
+      "step": 6560
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 4.141964285714286e-05,
+      "loss": 0.9795,
+      "step": 6561
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 4.141071428571429e-05,
+      "loss": 0.9578,
+      "step": 6562
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 4.140178571428572e-05,
+      "loss": 1.004,
+      "step": 6563
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 4.1392857142857145e-05,
+      "loss": 0.9996,
+      "step": 6564
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 4.138392857142857e-05,
+      "loss": 1.0125,
+      "step": 6565
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 4.1375e-05,
+      "loss": 1.02,
+      "step": 6566
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 4.136607142857143e-05,
+      "loss": 0.9758,
+      "step": 6567
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 4.135714285714286e-05,
+      "loss": 1.1091,
+      "step": 6568
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 4.1348214285714286e-05,
+      "loss": 1.0994,
+      "step": 6569
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 4.133928571428571e-05,
+      "loss": 0.9541,
+      "step": 6570
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 4.1330357142857144e-05,
+      "loss": 0.9831,
+      "step": 6571
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 4.132142857142857e-05,
+      "loss": 0.9999,
+      "step": 6572
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 4.13125e-05,
+      "loss": 0.9924,
+      "step": 6573
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 4.1303571428571434e-05,
+      "loss": 1.0753,
+      "step": 6574
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 4.129464285714286e-05,
+      "loss": 1.0105,
+      "step": 6575
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 4.128571428571429e-05,
+      "loss": 0.9665,
+      "step": 6576
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 4.127678571428572e-05,
+      "loss": 1.0379,
+      "step": 6577
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 4.126785714285715e-05,
+      "loss": 1.0236,
+      "step": 6578
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 4.1258928571428574e-05,
+      "loss": 0.9794,
+      "step": 6579
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 4.125e-05,
+      "loss": 0.9949,
+      "step": 6580
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 4.124107142857143e-05,
+      "loss": 1.0335,
+      "step": 6581
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 4.123214285714286e-05,
+      "loss": 1.0156,
+      "step": 6582
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 4.122321428571429e-05,
+      "loss": 1.0096,
+      "step": 6583
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 4.1214285714285715e-05,
+      "loss": 0.9882,
+      "step": 6584
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 4.120535714285714e-05,
+      "loss": 0.9632,
+      "step": 6585
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 4.119642857142857e-05,
+      "loss": 0.9855,
+      "step": 6586
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 4.11875e-05,
+      "loss": 0.971,
+      "step": 6587
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 4.117857142857143e-05,
+      "loss": 1.0016,
+      "step": 6588
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 4.116964285714286e-05,
+      "loss": 0.9409,
+      "step": 6589
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 4.116071428571429e-05,
+      "loss": 0.9828,
+      "step": 6590
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 4.115178571428572e-05,
+      "loss": 1.0014,
+      "step": 6591
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 4.1142857142857146e-05,
+      "loss": 0.978,
+      "step": 6592
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 4.113392857142858e-05,
+      "loss": 0.9761,
+      "step": 6593
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 4.1125000000000004e-05,
+      "loss": 0.9648,
+      "step": 6594
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 4.111607142857143e-05,
+      "loss": 1.0333,
+      "step": 6595
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 4.110714285714286e-05,
+      "loss": 0.9496,
+      "step": 6596
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 4.109821428571429e-05,
+      "loss": 1.0139,
+      "step": 6597
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 4.108928571428571e-05,
+      "loss": 0.952,
+      "step": 6598
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 4.1080357142857144e-05,
+      "loss": 1.0154,
+      "step": 6599
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 1.0237,
+      "step": 6600
+    },
+    {
+      "epoch": 29.46,
+      "eval_loss": 1.117234230041504,
+      "eval_runtime": 44.6169,
+      "eval_samples_per_second": 293.476,
+      "eval_steps_per_second": 1.546,
+      "step": 6600
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 4.10625e-05,
+      "loss": 0.9977,
+      "step": 6601
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 4.105357142857143e-05,
+      "loss": 1.0303,
+      "step": 6602
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 4.104464285714286e-05,
+      "loss": 0.9381,
+      "step": 6603
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 4.1035714285714285e-05,
+      "loss": 1.0536,
+      "step": 6604
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 4.102678571428572e-05,
+      "loss": 0.9705,
+      "step": 6605
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 4.101785714285715e-05,
+      "loss": 1.0458,
+      "step": 6606
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 4.1008928571428575e-05,
+      "loss": 1.0113,
+      "step": 6607
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 4.1e-05,
+      "loss": 0.9413,
+      "step": 6608
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 4.099107142857143e-05,
+      "loss": 0.98,
+      "step": 6609
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 4.098214285714286e-05,
+      "loss": 0.915,
+      "step": 6610
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 4.097321428571429e-05,
+      "loss": 1.0196,
+      "step": 6611
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 4.0964285714285716e-05,
+      "loss": 1.0336,
+      "step": 6612
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 4.095535714285714e-05,
+      "loss": 0.9605,
+      "step": 6613
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 4.0946428571428574e-05,
+      "loss": 0.9877,
+      "step": 6614
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 4.09375e-05,
+      "loss": 0.9923,
+      "step": 6615
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 4.092857142857143e-05,
+      "loss": 0.9746,
+      "step": 6616
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 4.091964285714286e-05,
+      "loss": 1.0632,
+      "step": 6617
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 4.091071428571429e-05,
+      "loss": 0.9748,
+      "step": 6618
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 4.0901785714285714e-05,
+      "loss": 1.0116,
+      "step": 6619
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 4.0892857142857147e-05,
+      "loss": 1.032,
+      "step": 6620
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 4.088392857142858e-05,
+      "loss": 0.9967,
+      "step": 6621
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 4.0875000000000004e-05,
+      "loss": 1.0119,
+      "step": 6622
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 4.086607142857143e-05,
+      "loss": 0.9696,
+      "step": 6623
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 4.085714285714286e-05,
+      "loss": 0.9512,
+      "step": 6624
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 4.084821428571429e-05,
+      "loss": 1.0289,
+      "step": 6625
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 4.083928571428572e-05,
+      "loss": 1.0497,
+      "step": 6626
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 4.0830357142857145e-05,
+      "loss": 1.0037,
+      "step": 6627
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 4.082142857142857e-05,
+      "loss": 0.9973,
+      "step": 6628
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 4.08125e-05,
+      "loss": 0.9777,
+      "step": 6629
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 4.080357142857143e-05,
+      "loss": 1.082,
+      "step": 6630
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 4.079464285714286e-05,
+      "loss": 1.0493,
+      "step": 6631
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 4.0785714285714286e-05,
+      "loss": 1.0799,
+      "step": 6632
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 4.077678571428571e-05,
+      "loss": 0.9764,
+      "step": 6633
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 4.0767857142857143e-05,
+      "loss": 1.0187,
+      "step": 6634
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 4.0758928571428576e-05,
+      "loss": 1.0248,
+      "step": 6635
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 4.075e-05,
+      "loss": 0.9877,
+      "step": 6636
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 4.0741071428571433e-05,
+      "loss": 0.9812,
+      "step": 6637
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 4.073214285714286e-05,
+      "loss": 0.9827,
+      "step": 6638
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 4.072321428571429e-05,
+      "loss": 1.0302,
+      "step": 6639
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 4.0714285714285717e-05,
+      "loss": 0.9247,
+      "step": 6640
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 4.070535714285714e-05,
+      "loss": 1.0386,
+      "step": 6641
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 4.0696428571428574e-05,
+      "loss": 1.0119,
+      "step": 6642
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 4.06875e-05,
+      "loss": 1.0425,
+      "step": 6643
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 4.067857142857143e-05,
+      "loss": 1.0146,
+      "step": 6644
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 4.066964285714286e-05,
+      "loss": 1.002,
+      "step": 6645
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 4.066071428571428e-05,
+      "loss": 0.9974,
+      "step": 6646
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 4.0651785714285715e-05,
+      "loss": 0.9108,
+      "step": 6647
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 4.064285714285714e-05,
+      "loss": 0.9877,
+      "step": 6648
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 4.063392857142857e-05,
+      "loss": 0.9617,
+      "step": 6649
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.9254,
+      "step": 6650
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 4.061607142857143e-05,
+      "loss": 0.9929,
+      "step": 6651
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 4.060714285714286e-05,
+      "loss": 0.9987,
+      "step": 6652
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 4.059821428571429e-05,
+      "loss": 1.0963,
+      "step": 6653
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 4.058928571428572e-05,
+      "loss": 0.9904,
+      "step": 6654
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 4.0580357142857146e-05,
+      "loss": 0.9445,
+      "step": 6655
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 4.057142857142857e-05,
+      "loss": 0.9644,
+      "step": 6656
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 4.0562500000000003e-05,
+      "loss": 1.0297,
+      "step": 6657
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 4.055357142857143e-05,
+      "loss": 1.0102,
+      "step": 6658
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 4.054464285714286e-05,
+      "loss": 1.007,
+      "step": 6659
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 4.0535714285714287e-05,
+      "loss": 0.9651,
+      "step": 6660
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 4.052678571428571e-05,
+      "loss": 1.0222,
+      "step": 6661
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 4.0517857142857144e-05,
+      "loss": 0.9681,
+      "step": 6662
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 4.050892857142857e-05,
+      "loss": 0.9931,
+      "step": 6663
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 4.05e-05,
+      "loss": 0.9535,
+      "step": 6664
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 4.0491071428571434e-05,
+      "loss": 0.991,
+      "step": 6665
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 4.048214285714286e-05,
+      "loss": 0.9775,
+      "step": 6666
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 4.047321428571429e-05,
+      "loss": 1.0109,
+      "step": 6667
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 4.046428571428572e-05,
+      "loss": 0.9701,
+      "step": 6668
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 4.045535714285714e-05,
+      "loss": 0.998,
+      "step": 6669
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 4.0446428571428575e-05,
+      "loss": 1.0499,
+      "step": 6670
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 4.04375e-05,
+      "loss": 1.0144,
+      "step": 6671
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 4.042857142857143e-05,
+      "loss": 1.0532,
+      "step": 6672
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 4.041964285714286e-05,
+      "loss": 0.9945,
+      "step": 6673
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 4.0410714285714283e-05,
+      "loss": 1.0129,
+      "step": 6674
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.0401785714285716e-05,
+      "loss": 1.0027,
+      "step": 6675
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.039285714285714e-05,
+      "loss": 1.0376,
+      "step": 6676
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.038392857142857e-05,
+      "loss": 0.9979,
+      "step": 6677
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.0375e-05,
+      "loss": 1.0326,
+      "step": 6678
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.036607142857143e-05,
+      "loss": 0.9718,
+      "step": 6679
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.035714285714286e-05,
+      "loss": 1.0238,
+      "step": 6680
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.034821428571429e-05,
+      "loss": 0.9949,
+      "step": 6681
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.033928571428572e-05,
+      "loss": 0.9378,
+      "step": 6682
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.0330357142857146e-05,
+      "loss": 0.9952,
+      "step": 6683
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.032142857142857e-05,
+      "loss": 1.0144,
+      "step": 6684
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.0312500000000004e-05,
+      "loss": 1.0622,
+      "step": 6685
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 4.030357142857143e-05,
+      "loss": 0.9691,
+      "step": 6686
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 4.029464285714286e-05,
+      "loss": 1.007,
+      "step": 6687
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 4.028571428571429e-05,
+      "loss": 1.0292,
+      "step": 6688
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 4.027678571428571e-05,
+      "loss": 1.0273,
+      "step": 6689
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 4.0267857142857145e-05,
+      "loss": 1.0155,
+      "step": 6690
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 4.025892857142857e-05,
+      "loss": 1.02,
+      "step": 6691
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 4.025e-05,
+      "loss": 0.9994,
+      "step": 6692
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 4.024107142857143e-05,
+      "loss": 1.0022,
+      "step": 6693
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 4.023214285714286e-05,
+      "loss": 1.0167,
+      "step": 6694
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 4.0223214285714286e-05,
+      "loss": 0.9424,
+      "step": 6695
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 4.021428571428572e-05,
+      "loss": 1.0125,
+      "step": 6696
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 4.020535714285715e-05,
+      "loss": 0.9307,
+      "step": 6697
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 4.0196428571428576e-05,
+      "loss": 0.9767,
+      "step": 6698
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 4.01875e-05,
+      "loss": 1.119,
+      "step": 6699
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 4.017857142857143e-05,
+      "loss": 1.0225,
+      "step": 6700
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 4.016964285714286e-05,
+      "loss": 1.0305,
+      "step": 6701
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 4.016071428571429e-05,
+      "loss": 0.9861,
+      "step": 6702
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 4.0151785714285716e-05,
+      "loss": 1.1166,
+      "step": 6703
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 4.014285714285714e-05,
+      "loss": 0.9909,
+      "step": 6704
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 4.0133928571428574e-05,
+      "loss": 0.9866,
+      "step": 6705
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 4.0125e-05,
+      "loss": 0.9938,
+      "step": 6706
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 4.0116071428571425e-05,
+      "loss": 0.9749,
+      "step": 6707
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 4.010714285714286e-05,
+      "loss": 1.0025,
+      "step": 6708
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 4.009821428571429e-05,
+      "loss": 0.9387,
+      "step": 6709
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 4.0089285714285715e-05,
+      "loss": 0.9319,
+      "step": 6710
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 4.008035714285715e-05,
+      "loss": 1.016,
+      "step": 6711
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 4.007142857142857e-05,
+      "loss": 1.0363,
+      "step": 6712
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 4.0062500000000005e-05,
+      "loss": 1.001,
+      "step": 6713
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 4.005357142857143e-05,
+      "loss": 1.0312,
+      "step": 6714
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.004464285714286e-05,
+      "loss": 0.9828,
+      "step": 6715
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.003571428571429e-05,
+      "loss": 0.9064,
+      "step": 6716
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 4.002678571428571e-05,
+      "loss": 1.0201,
+      "step": 6717
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 4.0017857142857146e-05,
+      "loss": 0.9675,
+      "step": 6718
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 4.000892857142857e-05,
+      "loss": 1.0368,
+      "step": 6719
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 4e-05,
+      "loss": 0.9896,
+      "step": 6720
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3.999107142857143e-05,
+      "loss": 0.9505,
+      "step": 6721
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3.9982142857142854e-05,
+      "loss": 0.9766,
+      "step": 6722
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3.9973214285714286e-05,
+      "loss": 0.9997,
+      "step": 6723
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 3.996428571428571e-05,
+      "loss": 1.0011,
+      "step": 6724
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 3.9955357142857144e-05,
+      "loss": 0.9899,
+      "step": 6725
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 3.9946428571428576e-05,
+      "loss": 0.9999,
+      "step": 6726
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 3.99375e-05,
+      "loss": 0.9966,
+      "step": 6727
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 3.9928571428571434e-05,
+      "loss": 0.9687,
+      "step": 6728
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 3.991964285714286e-05,
+      "loss": 0.9916,
+      "step": 6729
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 3.991071428571429e-05,
+      "loss": 1.0886,
+      "step": 6730
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 3.990178571428572e-05,
+      "loss": 0.9526,
+      "step": 6731
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 3.989285714285714e-05,
+      "loss": 0.9606,
+      "step": 6732
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 3.9883928571428575e-05,
+      "loss": 0.9955,
+      "step": 6733
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 3.9875e-05,
+      "loss": 1.0679,
+      "step": 6734
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 3.986607142857143e-05,
+      "loss": 0.9786,
+      "step": 6735
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 3.985714285714286e-05,
+      "loss": 1.004,
+      "step": 6736
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 3.984821428571428e-05,
+      "loss": 0.9971,
+      "step": 6737
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 3.9839285714285716e-05,
+      "loss": 0.9918,
+      "step": 6738
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 3.983035714285714e-05,
+      "loss": 0.9268,
+      "step": 6739
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 3.982142857142857e-05,
+      "loss": 1.0211,
+      "step": 6740
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 3.9812500000000005e-05,
+      "loss": 0.9706,
+      "step": 6741
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 3.980357142857143e-05,
+      "loss": 1.0134,
+      "step": 6742
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 3.979464285714286e-05,
+      "loss": 1.0134,
+      "step": 6743
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 3.978571428571429e-05,
+      "loss": 0.9732,
+      "step": 6744
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 3.9776785714285714e-05,
+      "loss": 0.977,
+      "step": 6745
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 3.9767857142857146e-05,
+      "loss": 0.9463,
+      "step": 6746
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 3.975892857142857e-05,
+      "loss": 0.9903,
+      "step": 6747
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 0.9484,
+      "step": 6748
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 3.974107142857143e-05,
+      "loss": 1.0285,
+      "step": 6749
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 3.9732142857142855e-05,
+      "loss": 0.9848,
+      "step": 6750
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 3.972321428571429e-05,
+      "loss": 0.9553,
+      "step": 6751
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 3.971428571428571e-05,
+      "loss": 1.0228,
+      "step": 6752
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 3.9705357142857145e-05,
+      "loss": 1.0161,
+      "step": 6753
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 3.969642857142857e-05,
+      "loss": 0.895,
+      "step": 6754
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 3.96875e-05,
+      "loss": 0.9708,
+      "step": 6755
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 3.9678571428571435e-05,
+      "loss": 0.9695,
+      "step": 6756
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 3.966964285714286e-05,
+      "loss": 1.0096,
+      "step": 6757
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 3.966071428571429e-05,
+      "loss": 0.9644,
+      "step": 6758
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 3.965178571428572e-05,
+      "loss": 0.9644,
+      "step": 6759
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 3.964285714285714e-05,
+      "loss": 1.0254,
+      "step": 6760
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 3.9633928571428575e-05,
+      "loss": 1.0165,
+      "step": 6761
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 3.9625e-05,
+      "loss": 0.9833,
+      "step": 6762
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 3.961607142857143e-05,
+      "loss": 1.0333,
+      "step": 6763
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 3.960714285714286e-05,
+      "loss": 0.994,
+      "step": 6764
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 3.9598214285714284e-05,
+      "loss": 1.0703,
+      "step": 6765
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 3.9589285714285716e-05,
+      "loss": 0.9643,
+      "step": 6766
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 3.958035714285714e-05,
+      "loss": 1.026,
+      "step": 6767
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 3.9571428571428574e-05,
+      "loss": 0.9827,
+      "step": 6768
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 3.95625e-05,
+      "loss": 0.9591,
+      "step": 6769
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 3.955357142857143e-05,
+      "loss": 1.0246,
+      "step": 6770
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 3.9544642857142864e-05,
+      "loss": 0.9041,
+      "step": 6771
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 3.953571428571429e-05,
+      "loss": 1.0574,
+      "step": 6772
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 3.9526785714285715e-05,
+      "loss": 1.0866,
+      "step": 6773
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 3.951785714285715e-05,
+      "loss": 1.0067,
+      "step": 6774
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 3.950892857142857e-05,
+      "loss": 1.0335,
+      "step": 6775
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.0323,
+      "step": 6776
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 3.949107142857143e-05,
+      "loss": 1.063,
+      "step": 6777
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 3.9482142857142856e-05,
+      "loss": 0.9378,
+      "step": 6778
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 3.947321428571429e-05,
+      "loss": 0.9357,
+      "step": 6779
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 3.946428571428571e-05,
+      "loss": 0.9446,
+      "step": 6780
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 3.9455357142857145e-05,
+      "loss": 0.9872,
+      "step": 6781
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 3.944642857142857e-05,
+      "loss": 0.9476,
+      "step": 6782
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 3.9437499999999996e-05,
+      "loss": 1.0328,
+      "step": 6783
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 3.942857142857143e-05,
+      "loss": 0.9701,
+      "step": 6784
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 3.941964285714286e-05,
+      "loss": 0.9792,
+      "step": 6785
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 3.941071428571429e-05,
+      "loss": 0.9995,
+      "step": 6786
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 3.940178571428572e-05,
+      "loss": 1.0534,
+      "step": 6787
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 3.9392857142857144e-05,
+      "loss": 0.9841,
+      "step": 6788
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 3.9383928571428576e-05,
+      "loss": 1.0166,
+      "step": 6789
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 3.9375e-05,
+      "loss": 0.9591,
+      "step": 6790
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 3.9366071428571434e-05,
+      "loss": 1.0337,
+      "step": 6791
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 3.935714285714286e-05,
+      "loss": 0.9676,
+      "step": 6792
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 3.9348214285714285e-05,
+      "loss": 0.9571,
+      "step": 6793
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 3.933928571428572e-05,
+      "loss": 0.9589,
+      "step": 6794
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 3.933035714285714e-05,
+      "loss": 1.0369,
+      "step": 6795
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 3.9321428571428575e-05,
+      "loss": 1.0137,
+      "step": 6796
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 3.93125e-05,
+      "loss": 0.9679,
+      "step": 6797
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 3.9303571428571426e-05,
+      "loss": 1.0038,
+      "step": 6798
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 3.929464285714286e-05,
+      "loss": 0.9513,
+      "step": 6799
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 1.0412,
+      "step": 6800
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 3.9276785714285715e-05,
+      "loss": 0.9448,
+      "step": 6801
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 3.926785714285715e-05,
+      "loss": 0.9471,
+      "step": 6802
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 3.925892857142857e-05,
+      "loss": 0.9471,
+      "step": 6803
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 0.9855,
+      "step": 6804
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 3.924107142857143e-05,
+      "loss": 0.9987,
+      "step": 6805
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 3.923214285714286e-05,
+      "loss": 1.0597,
+      "step": 6806
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 3.922321428571429e-05,
+      "loss": 0.9951,
+      "step": 6807
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 3.9214285714285714e-05,
+      "loss": 0.9883,
+      "step": 6808
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 3.9205357142857146e-05,
+      "loss": 0.9748,
+      "step": 6809
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 3.919642857142857e-05,
+      "loss": 0.9714,
+      "step": 6810
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 3.91875e-05,
+      "loss": 1.0107,
+      "step": 6811
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 3.917857142857143e-05,
+      "loss": 0.9941,
+      "step": 6812
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 3.9169642857142855e-05,
+      "loss": 0.9972,
+      "step": 6813
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 3.916071428571429e-05,
+      "loss": 0.9928,
+      "step": 6814
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 3.915178571428572e-05,
+      "loss": 1.0463,
+      "step": 6815
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 3.9142857142857145e-05,
+      "loss": 0.991,
+      "step": 6816
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 3.913392857142858e-05,
+      "loss": 0.9983,
+      "step": 6817
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 3.9125e-05,
+      "loss": 0.9786,
+      "step": 6818
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 3.9116071428571435e-05,
+      "loss": 0.9649,
+      "step": 6819
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 3.910714285714286e-05,
+      "loss": 0.9917,
+      "step": 6820
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 3.9098214285714285e-05,
+      "loss": 0.9461,
+      "step": 6821
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 3.908928571428572e-05,
+      "loss": 0.9628,
+      "step": 6822
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 3.908035714285714e-05,
+      "loss": 0.9422,
+      "step": 6823
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 3.9071428571428575e-05,
+      "loss": 1.0198,
+      "step": 6824
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 3.90625e-05,
+      "loss": 1.0215,
+      "step": 6825
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 3.9053571428571426e-05,
+      "loss": 1.0681,
+      "step": 6826
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 3.904464285714286e-05,
+      "loss": 0.9749,
+      "step": 6827
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 3.9035714285714284e-05,
+      "loss": 1.0081,
+      "step": 6828
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 3.9026785714285716e-05,
+      "loss": 1.0011,
+      "step": 6829
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 3.901785714285714e-05,
+      "loss": 0.9147,
+      "step": 6830
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 3.9008928571428574e-05,
+      "loss": 0.9782,
+      "step": 6831
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.0197,
+      "step": 6832
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 3.899107142857143e-05,
+      "loss": 0.9603,
+      "step": 6833
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 3.8982142857142864e-05,
+      "loss": 1.0051,
+      "step": 6834
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 3.897321428571429e-05,
+      "loss": 1.0266,
+      "step": 6835
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 3.8964285714285715e-05,
+      "loss": 1.0686,
+      "step": 6836
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 3.895535714285715e-05,
+      "loss": 0.968,
+      "step": 6837
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 3.894642857142857e-05,
+      "loss": 0.9552,
+      "step": 6838
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 3.8937500000000005e-05,
+      "loss": 0.9788,
+      "step": 6839
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 3.892857142857143e-05,
+      "loss": 1.0118,
+      "step": 6840
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 3.8919642857142855e-05,
+      "loss": 0.9956,
+      "step": 6841
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 3.891071428571429e-05,
+      "loss": 1.0088,
+      "step": 6842
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 3.890178571428571e-05,
+      "loss": 1.0543,
+      "step": 6843
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 3.8892857142857145e-05,
+      "loss": 0.9733,
+      "step": 6844
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 3.888392857142857e-05,
+      "loss": 1.019,
+      "step": 6845
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 3.8875e-05,
+      "loss": 0.9936,
+      "step": 6846
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 3.8866071428571435e-05,
+      "loss": 0.991,
+      "step": 6847
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 3.885714285714286e-05,
+      "loss": 0.963,
+      "step": 6848
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 3.8848214285714286e-05,
+      "loss": 0.9845,
+      "step": 6849
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 3.883928571428572e-05,
+      "loss": 0.9697,
+      "step": 6850
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 3.8830357142857144e-05,
+      "loss": 0.9341,
+      "step": 6851
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 3.8821428571428576e-05,
+      "loss": 0.9897,
+      "step": 6852
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 3.88125e-05,
+      "loss": 0.9671,
+      "step": 6853
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 3.880357142857143e-05,
+      "loss": 1.0551,
+      "step": 6854
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 3.879464285714286e-05,
+      "loss": 1.0736,
+      "step": 6855
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 3.8785714285714285e-05,
+      "loss": 0.9134,
+      "step": 6856
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 3.877678571428572e-05,
+      "loss": 1.0333,
+      "step": 6857
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 3.876785714285714e-05,
+      "loss": 1.0203,
+      "step": 6858
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 3.875892857142857e-05,
+      "loss": 0.9748,
+      "step": 6859
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 3.875e-05,
+      "loss": 0.9874,
+      "step": 6860
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 3.874107142857143e-05,
+      "loss": 1.0861,
+      "step": 6861
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 3.8732142857142864e-05,
+      "loss": 0.9801,
+      "step": 6862
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 3.872321428571429e-05,
+      "loss": 1.0319,
+      "step": 6863
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 3.8714285714285715e-05,
+      "loss": 0.9695,
+      "step": 6864
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 3.870535714285715e-05,
+      "loss": 0.9902,
+      "step": 6865
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 3.869642857142857e-05,
+      "loss": 0.9534,
+      "step": 6866
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 3.8687500000000005e-05,
+      "loss": 0.9704,
+      "step": 6867
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 3.867857142857143e-05,
+      "loss": 0.9965,
+      "step": 6868
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 3.8669642857142856e-05,
+      "loss": 0.9238,
+      "step": 6869
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 3.866071428571429e-05,
+      "loss": 0.9526,
+      "step": 6870
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 3.8651785714285714e-05,
+      "loss": 0.987,
+      "step": 6871
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 3.8642857142857146e-05,
+      "loss": 1.0288,
+      "step": 6872
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 3.863392857142857e-05,
+      "loss": 1.0217,
+      "step": 6873
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 3.8625e-05,
+      "loss": 1.024,
+      "step": 6874
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 3.861607142857143e-05,
+      "loss": 1.0396,
+      "step": 6875
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 3.860714285714286e-05,
+      "loss": 0.958,
+      "step": 6876
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 3.8598214285714294e-05,
+      "loss": 0.9247,
+      "step": 6877
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 3.858928571428572e-05,
+      "loss": 1.0182,
+      "step": 6878
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 3.8580357142857145e-05,
+      "loss": 0.9663,
+      "step": 6879
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 1.0234,
+      "step": 6880
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 3.85625e-05,
+      "loss": 0.9949,
+      "step": 6881
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 3.855357142857143e-05,
+      "loss": 1.0077,
+      "step": 6882
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 3.854464285714286e-05,
+      "loss": 1.0112,
+      "step": 6883
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 3.8535714285714285e-05,
+      "loss": 0.9831,
+      "step": 6884
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 3.852678571428572e-05,
+      "loss": 0.9474,
+      "step": 6885
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 3.851785714285714e-05,
+      "loss": 0.9918,
+      "step": 6886
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 3.850892857142857e-05,
+      "loss": 0.8942,
+      "step": 6887
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 3.85e-05,
+      "loss": 1.0418,
+      "step": 6888
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 3.8491071428571426e-05,
+      "loss": 0.9964,
+      "step": 6889
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 3.848214285714286e-05,
+      "loss": 0.9845,
+      "step": 6890
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 3.847321428571429e-05,
+      "loss": 1.0097,
+      "step": 6891
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 3.8464285714285716e-05,
+      "loss": 0.9528,
+      "step": 6892
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 3.845535714285715e-05,
+      "loss": 1.046,
+      "step": 6893
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 3.8446428571428574e-05,
+      "loss": 1.0052,
+      "step": 6894
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 3.8437500000000006e-05,
+      "loss": 1.042,
+      "step": 6895
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 3.842857142857143e-05,
+      "loss": 0.9635,
+      "step": 6896
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 3.841964285714286e-05,
+      "loss": 1.0106,
+      "step": 6897
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 3.841071428571429e-05,
+      "loss": 1.0133,
+      "step": 6898
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 3.8401785714285715e-05,
+      "loss": 0.9835,
+      "step": 6899
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 3.839285714285715e-05,
+      "loss": 0.9515,
+      "step": 6900
+    },
+    {
+      "epoch": 30.8,
+      "eval_loss": 1.1136188507080078,
+      "eval_runtime": 44.7908,
+      "eval_samples_per_second": 292.337,
+      "eval_steps_per_second": 1.54,
+      "step": 6900
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 3.838392857142857e-05,
+      "loss": 1.0019,
+      "step": 6901
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 3.8375e-05,
+      "loss": 1.0159,
+      "step": 6902
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 3.836607142857143e-05,
+      "loss": 0.99,
+      "step": 6903
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 3.8357142857142855e-05,
+      "loss": 0.8969,
+      "step": 6904
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 3.834821428571429e-05,
+      "loss": 0.975,
+      "step": 6905
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 3.833928571428572e-05,
+      "loss": 0.9576,
+      "step": 6906
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 3.8330357142857145e-05,
+      "loss": 1.0439,
+      "step": 6907
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 3.832142857142858e-05,
+      "loss": 0.9347,
+      "step": 6908
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 3.83125e-05,
+      "loss": 0.9483,
+      "step": 6909
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 3.8303571428571435e-05,
+      "loss": 0.9881,
+      "step": 6910
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 3.829464285714286e-05,
+      "loss": 0.9942,
+      "step": 6911
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 3.8285714285714286e-05,
+      "loss": 1.0454,
+      "step": 6912
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 3.827678571428572e-05,
+      "loss": 1.0458,
+      "step": 6913
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 3.8267857142857144e-05,
+      "loss": 1.0398,
+      "step": 6914
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 3.825892857142857e-05,
+      "loss": 1.0282,
+      "step": 6915
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 3.825e-05,
+      "loss": 1.0448,
+      "step": 6916
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 3.824107142857143e-05,
+      "loss": 1.0074,
+      "step": 6917
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 3.823214285714286e-05,
+      "loss": 1.0098,
+      "step": 6918
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 3.8223214285714285e-05,
+      "loss": 1.0393,
+      "step": 6919
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 3.821428571428572e-05,
+      "loss": 1.02,
+      "step": 6920
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 3.820535714285714e-05,
+      "loss": 1.0276,
+      "step": 6921
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 3.8196428571428574e-05,
+      "loss": 0.9205,
+      "step": 6922
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 3.818750000000001e-05,
+      "loss": 1.0016,
+      "step": 6923
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 3.817857142857143e-05,
+      "loss": 0.9704,
+      "step": 6924
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 3.816964285714286e-05,
+      "loss": 0.9707,
+      "step": 6925
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 3.816071428571429e-05,
+      "loss": 1.0236,
+      "step": 6926
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 3.8151785714285715e-05,
+      "loss": 1.0274,
+      "step": 6927
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 3.814285714285715e-05,
+      "loss": 0.9953,
+      "step": 6928
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 3.813392857142857e-05,
+      "loss": 0.9815,
+      "step": 6929
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 3.8125e-05,
+      "loss": 0.9454,
+      "step": 6930
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 3.811607142857143e-05,
+      "loss": 1.0099,
+      "step": 6931
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 3.8107142857142856e-05,
+      "loss": 1.0449,
+      "step": 6932
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 3.809821428571429e-05,
+      "loss": 0.9744,
+      "step": 6933
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 3.8089285714285714e-05,
+      "loss": 1.0518,
+      "step": 6934
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 3.8080357142857146e-05,
+      "loss": 1.0033,
+      "step": 6935
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 3.807142857142857e-05,
+      "loss": 1.0117,
+      "step": 6936
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 3.8062500000000004e-05,
+      "loss": 1.0176,
+      "step": 6937
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 3.8053571428571436e-05,
+      "loss": 1.0022,
+      "step": 6938
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 3.804464285714286e-05,
+      "loss": 1.0849,
+      "step": 6939
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 3.803571428571429e-05,
+      "loss": 1.0294,
+      "step": 6940
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 3.802678571428572e-05,
+      "loss": 1.0083,
+      "step": 6941
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 3.8017857142857144e-05,
+      "loss": 0.9229,
+      "step": 6942
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.800892857142858e-05,
+      "loss": 0.9848,
+      "step": 6943
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.8e-05,
+      "loss": 1.1378,
+      "step": 6944
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.799107142857143e-05,
+      "loss": 1.0319,
+      "step": 6945
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3.798214285714286e-05,
+      "loss": 0.9646,
+      "step": 6946
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3.7973214285714285e-05,
+      "loss": 1.0451,
+      "step": 6947
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 3.796428571428571e-05,
+      "loss": 0.9313,
+      "step": 6948
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 3.795535714285714e-05,
+      "loss": 0.9884,
+      "step": 6949
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 3.794642857142857e-05,
+      "loss": 0.9537,
+      "step": 6950
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 3.79375e-05,
+      "loss": 1.0151,
+      "step": 6951
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 3.792857142857143e-05,
+      "loss": 0.9753,
+      "step": 6952
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 3.791964285714286e-05,
+      "loss": 0.9783,
+      "step": 6953
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 3.791071428571429e-05,
+      "loss": 0.9526,
+      "step": 6954
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 3.7901785714285716e-05,
+      "loss": 0.9629,
+      "step": 6955
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 3.789285714285715e-05,
+      "loss": 0.9438,
+      "step": 6956
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 3.7883928571428574e-05,
+      "loss": 1.0104,
+      "step": 6957
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 3.7875e-05,
+      "loss": 0.9974,
+      "step": 6958
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 3.786607142857143e-05,
+      "loss": 1.0285,
+      "step": 6959
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 1.0817,
+      "step": 6960
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 3.784821428571429e-05,
+      "loss": 1.0032,
+      "step": 6961
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 3.7839285714285714e-05,
+      "loss": 0.9744,
+      "step": 6962
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 3.783035714285714e-05,
+      "loss": 1.0004,
+      "step": 6963
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 3.782142857142857e-05,
+      "loss": 0.9777,
+      "step": 6964
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 3.78125e-05,
+      "loss": 0.9658,
+      "step": 6965
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 3.780357142857143e-05,
+      "loss": 0.9322,
+      "step": 6966
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 3.779464285714286e-05,
+      "loss": 1.0001,
+      "step": 6967
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 3.778571428571429e-05,
+      "loss": 0.958,
+      "step": 6968
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 3.777678571428572e-05,
+      "loss": 0.9607,
+      "step": 6969
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 3.7767857142857145e-05,
+      "loss": 1.0026,
+      "step": 6970
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 3.775892857142858e-05,
+      "loss": 0.9689,
+      "step": 6971
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 3.775e-05,
+      "loss": 0.937,
+      "step": 6972
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 3.774107142857143e-05,
+      "loss": 0.9831,
+      "step": 6973
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 3.773214285714286e-05,
+      "loss": 1.0707,
+      "step": 6974
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 3.7723214285714286e-05,
+      "loss": 0.9713,
+      "step": 6975
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 3.771428571428572e-05,
+      "loss": 0.9409,
+      "step": 6976
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 3.7705357142857144e-05,
+      "loss": 1.022,
+      "step": 6977
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 3.769642857142857e-05,
+      "loss": 0.9256,
+      "step": 6978
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 3.76875e-05,
+      "loss": 0.918,
+      "step": 6979
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 3.767857142857143e-05,
+      "loss": 0.873,
+      "step": 6980
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 3.766964285714286e-05,
+      "loss": 1.0269,
+      "step": 6981
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 3.766071428571429e-05,
+      "loss": 0.9928,
+      "step": 6982
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 3.765178571428572e-05,
+      "loss": 0.9936,
+      "step": 6983
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 3.764285714285715e-05,
+      "loss": 0.9623,
+      "step": 6984
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 3.7633928571428574e-05,
+      "loss": 0.9606,
+      "step": 6985
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 3.7625e-05,
+      "loss": 1.0266,
+      "step": 6986
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 3.761607142857143e-05,
+      "loss": 1.0106,
+      "step": 6987
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 3.760714285714286e-05,
+      "loss": 0.9654,
+      "step": 6988
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 3.759821428571429e-05,
+      "loss": 1.0352,
+      "step": 6989
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 3.7589285714285715e-05,
+      "loss": 1.0936,
+      "step": 6990
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 3.758035714285714e-05,
+      "loss": 0.9431,
+      "step": 6991
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 3.757142857142857e-05,
+      "loss": 1.0008,
+      "step": 6992
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 3.75625e-05,
+      "loss": 1.0462,
+      "step": 6993
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 3.755357142857143e-05,
+      "loss": 1.0769,
+      "step": 6994
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 3.7544642857142856e-05,
+      "loss": 0.9975,
+      "step": 6995
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 3.753571428571429e-05,
+      "loss": 0.9539,
+      "step": 6996
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 3.752678571428572e-05,
+      "loss": 1.0048,
+      "step": 6997
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 3.7517857142857146e-05,
+      "loss": 1.0566,
+      "step": 6998
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 3.750892857142858e-05,
+      "loss": 1.0132,
+      "step": 6999
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.0581,
+      "step": 7000
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 3.749107142857143e-05,
+      "loss": 0.9944,
+      "step": 7001
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 3.748214285714286e-05,
+      "loss": 0.9828,
+      "step": 7002
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 3.747321428571429e-05,
+      "loss": 1.026,
+      "step": 7003
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 3.746428571428572e-05,
+      "loss": 0.9478,
+      "step": 7004
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 3.7455357142857144e-05,
+      "loss": 1.0009,
+      "step": 7005
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 3.744642857142857e-05,
+      "loss": 0.9748,
+      "step": 7006
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 3.74375e-05,
+      "loss": 0.9238,
+      "step": 7007
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 3.742857142857143e-05,
+      "loss": 0.9626,
+      "step": 7008
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 3.741964285714286e-05,
+      "loss": 0.9255,
+      "step": 7009
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 3.7410714285714285e-05,
+      "loss": 0.9471,
+      "step": 7010
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 3.740178571428572e-05,
+      "loss": 0.9929,
+      "step": 7011
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 3.739285714285714e-05,
+      "loss": 0.9725,
+      "step": 7012
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 3.7383928571428575e-05,
+      "loss": 0.9548,
+      "step": 7013
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 3.737500000000001e-05,
+      "loss": 0.9852,
+      "step": 7014
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 3.736607142857143e-05,
+      "loss": 0.9508,
+      "step": 7015
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 3.735714285714286e-05,
+      "loss": 0.9516,
+      "step": 7016
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 3.734821428571429e-05,
+      "loss": 0.9238,
+      "step": 7017
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 3.7339285714285716e-05,
+      "loss": 0.9433,
+      "step": 7018
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 3.733035714285714e-05,
+      "loss": 0.955,
+      "step": 7019
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 3.7321428571428573e-05,
+      "loss": 0.9801,
+      "step": 7020
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 3.73125e-05,
+      "loss": 0.9777,
+      "step": 7021
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 3.730357142857143e-05,
+      "loss": 0.9318,
+      "step": 7022
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 3.7294642857142857e-05,
+      "loss": 0.9333,
+      "step": 7023
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 3.728571428571428e-05,
+      "loss": 0.9813,
+      "step": 7024
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 3.7276785714285714e-05,
+      "loss": 1.0124,
+      "step": 7025
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 3.7267857142857147e-05,
+      "loss": 1.108,
+      "step": 7026
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 3.725892857142857e-05,
+      "loss": 0.9675,
+      "step": 7027
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 1.004,
+      "step": 7028
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 3.724107142857143e-05,
+      "loss": 0.986,
+      "step": 7029
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 3.723214285714286e-05,
+      "loss": 0.9574,
+      "step": 7030
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 3.722321428571429e-05,
+      "loss": 1.0033,
+      "step": 7031
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 3.721428571428572e-05,
+      "loss": 0.9925,
+      "step": 7032
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 3.7205357142857145e-05,
+      "loss": 1.0451,
+      "step": 7033
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 3.719642857142857e-05,
+      "loss": 0.9639,
+      "step": 7034
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 3.71875e-05,
+      "loss": 1.0334,
+      "step": 7035
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 3.717857142857143e-05,
+      "loss": 1.0447,
+      "step": 7036
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 3.716964285714286e-05,
+      "loss": 0.9406,
+      "step": 7037
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 3.7160714285714286e-05,
+      "loss": 0.9721,
+      "step": 7038
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 3.715178571428571e-05,
+      "loss": 0.9622,
+      "step": 7039
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 3.7142857142857143e-05,
+      "loss": 0.9738,
+      "step": 7040
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 3.713392857142857e-05,
+      "loss": 1.0169,
+      "step": 7041
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 3.7125e-05,
+      "loss": 0.9555,
+      "step": 7042
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 3.711607142857143e-05,
+      "loss": 1.0075,
+      "step": 7043
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 3.710714285714286e-05,
+      "loss": 0.9519,
+      "step": 7044
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 3.709821428571429e-05,
+      "loss": 0.9776,
+      "step": 7045
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 3.7089285714285717e-05,
+      "loss": 1.0084,
+      "step": 7046
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 3.708035714285715e-05,
+      "loss": 0.9227,
+      "step": 7047
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 3.7071428571428574e-05,
+      "loss": 0.9417,
+      "step": 7048
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 3.70625e-05,
+      "loss": 1.0252,
+      "step": 7049
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 3.705357142857143e-05,
+      "loss": 0.9913,
+      "step": 7050
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 3.704464285714286e-05,
+      "loss": 1.067,
+      "step": 7051
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 3.703571428571429e-05,
+      "loss": 0.9487,
+      "step": 7052
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 3.7026785714285715e-05,
+      "loss": 1.0171,
+      "step": 7053
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 3.701785714285714e-05,
+      "loss": 1.0504,
+      "step": 7054
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 3.700892857142857e-05,
+      "loss": 1.077,
+      "step": 7055
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 3.7e-05,
+      "loss": 1.0048,
+      "step": 7056
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 3.699107142857143e-05,
+      "loss": 0.9608,
+      "step": 7057
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 3.698214285714286e-05,
+      "loss": 1.0666,
+      "step": 7058
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 3.697321428571429e-05,
+      "loss": 0.9747,
+      "step": 7059
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 3.696428571428572e-05,
+      "loss": 0.9805,
+      "step": 7060
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 3.6955357142857146e-05,
+      "loss": 0.9932,
+      "step": 7061
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 3.694642857142857e-05,
+      "loss": 0.9636,
+      "step": 7062
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 3.69375e-05,
+      "loss": 0.9246,
+      "step": 7063
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 3.692857142857143e-05,
+      "loss": 1.0069,
+      "step": 7064
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 3.691964285714286e-05,
+      "loss": 1.0166,
+      "step": 7065
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 3.6910714285714286e-05,
+      "loss": 1.0423,
+      "step": 7066
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 3.690178571428571e-05,
+      "loss": 0.9831,
+      "step": 7067
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 3.6892857142857144e-05,
+      "loss": 1.0339,
+      "step": 7068
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 3.688392857142857e-05,
+      "loss": 1.0239,
+      "step": 7069
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 3.6875e-05,
+      "loss": 1.0335,
+      "step": 7070
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 3.686607142857143e-05,
+      "loss": 0.9962,
+      "step": 7071
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 3.685714285714286e-05,
+      "loss": 0.9825,
+      "step": 7072
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 3.684821428571429e-05,
+      "loss": 0.9762,
+      "step": 7073
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 3.683928571428572e-05,
+      "loss": 0.9977,
+      "step": 7074
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 3.683035714285715e-05,
+      "loss": 0.9623,
+      "step": 7075
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 3.6821428571428575e-05,
+      "loss": 0.9922,
+      "step": 7076
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 3.68125e-05,
+      "loss": 1.0113,
+      "step": 7077
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 3.680357142857143e-05,
+      "loss": 1.007,
+      "step": 7078
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 3.679464285714286e-05,
+      "loss": 0.9653,
+      "step": 7079
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 3.678571428571429e-05,
+      "loss": 1.0058,
+      "step": 7080
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 3.6776785714285716e-05,
+      "loss": 0.9455,
+      "step": 7081
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 3.676785714285714e-05,
+      "loss": 1.0106,
+      "step": 7082
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 3.675892857142857e-05,
+      "loss": 0.9955,
+      "step": 7083
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 3.675e-05,
+      "loss": 0.989,
+      "step": 7084
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 3.674107142857143e-05,
+      "loss": 1.0379,
+      "step": 7085
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 3.6732142857142856e-05,
+      "loss": 1.0279,
+      "step": 7086
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 3.672321428571429e-05,
+      "loss": 0.9877,
+      "step": 7087
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 3.671428571428572e-05,
+      "loss": 1.0515,
+      "step": 7088
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 3.6705357142857146e-05,
+      "loss": 0.9137,
+      "step": 7089
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 3.669642857142857e-05,
+      "loss": 0.9787,
+      "step": 7090
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 3.6687500000000004e-05,
+      "loss": 0.9582,
+      "step": 7091
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 3.667857142857143e-05,
+      "loss": 1.0288,
+      "step": 7092
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 3.666964285714286e-05,
+      "loss": 0.9778,
+      "step": 7093
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 3.666071428571429e-05,
+      "loss": 0.9445,
+      "step": 7094
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 3.665178571428571e-05,
+      "loss": 0.9896,
+      "step": 7095
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 3.6642857142857145e-05,
+      "loss": 0.9339,
+      "step": 7096
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 3.663392857142857e-05,
+      "loss": 0.9324,
+      "step": 7097
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 3.6625e-05,
+      "loss": 0.9525,
+      "step": 7098
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 3.661607142857143e-05,
+      "loss": 1.0089,
+      "step": 7099
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 3.6607142857142853e-05,
+      "loss": 0.9756,
+      "step": 7100
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 3.6598214285714286e-05,
+      "loss": 1.0134,
+      "step": 7101
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 3.658928571428572e-05,
+      "loss": 1.0203,
+      "step": 7102
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 3.658035714285715e-05,
+      "loss": 0.9761,
+      "step": 7103
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 3.6571428571428576e-05,
+      "loss": 1.011,
+      "step": 7104
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 3.65625e-05,
+      "loss": 0.9576,
+      "step": 7105
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 3.655357142857143e-05,
+      "loss": 0.9861,
+      "step": 7106
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 3.654464285714286e-05,
+      "loss": 1.011,
+      "step": 7107
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 3.653571428571429e-05,
+      "loss": 1.0256,
+      "step": 7108
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 3.6526785714285716e-05,
+      "loss": 0.9917,
+      "step": 7109
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 3.651785714285714e-05,
+      "loss": 1.002,
+      "step": 7110
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 3.6508928571428574e-05,
+      "loss": 1.0107,
+      "step": 7111
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 3.65e-05,
+      "loss": 1.0463,
+      "step": 7112
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 3.649107142857143e-05,
+      "loss": 1.0991,
+      "step": 7113
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 3.648214285714286e-05,
+      "loss": 0.9391,
+      "step": 7114
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 3.647321428571428e-05,
+      "loss": 0.9988,
+      "step": 7115
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 3.6464285714285715e-05,
+      "loss": 0.9417,
+      "step": 7116
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 3.645535714285715e-05,
+      "loss": 0.8857,
+      "step": 7117
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 3.644642857142857e-05,
+      "loss": 0.9819,
+      "step": 7118
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 3.6437500000000005e-05,
+      "loss": 1.0886,
+      "step": 7119
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 0.9712,
+      "step": 7120
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 3.641964285714286e-05,
+      "loss": 0.9837,
+      "step": 7121
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 3.641071428571429e-05,
+      "loss": 1.0009,
+      "step": 7122
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 3.640178571428571e-05,
+      "loss": 0.9511,
+      "step": 7123
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 3.6392857142857146e-05,
+      "loss": 1.0389,
+      "step": 7124
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 3.638392857142857e-05,
+      "loss": 0.9446,
+      "step": 7125
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 3.6375e-05,
+      "loss": 0.9284,
+      "step": 7126
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 3.636607142857143e-05,
+      "loss": 0.9411,
+      "step": 7127
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 3.6357142857142854e-05,
+      "loss": 1.004,
+      "step": 7128
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3.6348214285714286e-05,
+      "loss": 1.0575,
+      "step": 7129
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3.633928571428571e-05,
+      "loss": 0.9676,
+      "step": 7130
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3.6330357142857144e-05,
+      "loss": 1.0135,
+      "step": 7131
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 3.6321428571428576e-05,
+      "loss": 0.9323,
+      "step": 7132
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 3.63125e-05,
+      "loss": 0.9638,
+      "step": 7133
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 3.6303571428571434e-05,
+      "loss": 1.024,
+      "step": 7134
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 3.629464285714286e-05,
+      "loss": 0.988,
+      "step": 7135
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 3.628571428571429e-05,
+      "loss": 0.9982,
+      "step": 7136
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 3.627678571428572e-05,
+      "loss": 1.0372,
+      "step": 7137
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 3.626785714285714e-05,
+      "loss": 0.9681,
+      "step": 7138
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 3.6258928571428575e-05,
+      "loss": 0.9846,
+      "step": 7139
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 3.625e-05,
+      "loss": 0.9958,
+      "step": 7140
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 3.624107142857143e-05,
+      "loss": 1.0262,
+      "step": 7141
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 3.623214285714286e-05,
+      "loss": 1.0534,
+      "step": 7142
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3.622321428571428e-05,
+      "loss": 1.0207,
+      "step": 7143
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3.6214285714285716e-05,
+      "loss": 0.9706,
+      "step": 7144
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 3.620535714285714e-05,
+      "loss": 0.9628,
+      "step": 7145
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 3.619642857142857e-05,
+      "loss": 0.9987,
+      "step": 7146
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 3.61875e-05,
+      "loss": 0.9257,
+      "step": 7147
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 3.617857142857143e-05,
+      "loss": 1.0913,
+      "step": 7148
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 3.616964285714286e-05,
+      "loss": 1.0298,
+      "step": 7149
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 3.616071428571429e-05,
+      "loss": 1.0189,
+      "step": 7150
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 3.615178571428572e-05,
+      "loss": 1.0162,
+      "step": 7151
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 3.6142857142857146e-05,
+      "loss": 0.9607,
+      "step": 7152
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 3.613392857142857e-05,
+      "loss": 0.9575,
+      "step": 7153
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 3.6125000000000004e-05,
+      "loss": 0.9301,
+      "step": 7154
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 3.611607142857143e-05,
+      "loss": 1.0676,
+      "step": 7155
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 3.610714285714286e-05,
+      "loss": 0.9991,
+      "step": 7156
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 3.609821428571429e-05,
+      "loss": 0.9882,
+      "step": 7157
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 3.608928571428571e-05,
+      "loss": 1.0288,
+      "step": 7158
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 3.6080357142857145e-05,
+      "loss": 0.9301,
+      "step": 7159
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 3.607142857142857e-05,
+      "loss": 0.9484,
+      "step": 7160
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 3.60625e-05,
+      "loss": 1.0184,
+      "step": 7161
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 3.605357142857143e-05,
+      "loss": 1.0546,
+      "step": 7162
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 3.604464285714286e-05,
+      "loss": 0.9801,
+      "step": 7163
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 3.603571428571429e-05,
+      "loss": 0.9028,
+      "step": 7164
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 3.602678571428572e-05,
+      "loss": 1.0339,
+      "step": 7165
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 3.601785714285714e-05,
+      "loss": 0.9776,
+      "step": 7166
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.6008928571428575e-05,
+      "loss": 1.0163,
+      "step": 7167
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.0095,
+      "step": 7168
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.599107142857143e-05,
+      "loss": 0.999,
+      "step": 7169
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.598214285714286e-05,
+      "loss": 1.0367,
+      "step": 7170
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.5973214285714284e-05,
+      "loss": 0.9145,
+      "step": 7171
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 3.5964285714285716e-05,
+      "loss": 0.9756,
+      "step": 7172
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 3.595535714285714e-05,
+      "loss": 0.9445,
+      "step": 7173
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 3.5946428571428574e-05,
+      "loss": 0.9321,
+      "step": 7174
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.9573,
+      "step": 7175
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 3.5928571428571425e-05,
+      "loss": 0.9573,
+      "step": 7176
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 3.591964285714286e-05,
+      "loss": 1.0436,
+      "step": 7177
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 3.591071428571429e-05,
+      "loss": 0.9387,
+      "step": 7178
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 3.590178571428572e-05,
+      "loss": 0.9323,
+      "step": 7179
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 3.589285714285715e-05,
+      "loss": 0.9771,
+      "step": 7180
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 3.588392857142857e-05,
+      "loss": 1.0046,
+      "step": 7181
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 3.5875000000000005e-05,
+      "loss": 0.927,
+      "step": 7182
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 3.586607142857143e-05,
+      "loss": 0.9907,
+      "step": 7183
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 3.585714285714286e-05,
+      "loss": 1.0428,
+      "step": 7184
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 3.584821428571429e-05,
+      "loss": 1.0321,
+      "step": 7185
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 3.583928571428571e-05,
+      "loss": 1.0414,
+      "step": 7186
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 3.5830357142857145e-05,
+      "loss": 0.9831,
+      "step": 7187
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 3.582142857142857e-05,
+      "loss": 1.0098,
+      "step": 7188
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 3.58125e-05,
+      "loss": 0.9861,
+      "step": 7189
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 3.580357142857143e-05,
+      "loss": 0.9848,
+      "step": 7190
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 3.5794642857142854e-05,
+      "loss": 0.9718,
+      "step": 7191
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3.5785714285714286e-05,
+      "loss": 0.9213,
+      "step": 7192
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3.577678571428572e-05,
+      "loss": 1.0209,
+      "step": 7193
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 3.5767857142857144e-05,
+      "loss": 0.9313,
+      "step": 7194
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 3.5758928571428576e-05,
+      "loss": 1.0132,
+      "step": 7195
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 3.575e-05,
+      "loss": 1.0241,
+      "step": 7196
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 3.5741071428571434e-05,
+      "loss": 0.9598,
+      "step": 7197
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 3.573214285714286e-05,
+      "loss": 0.9742,
+      "step": 7198
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 3.5723214285714285e-05,
+      "loss": 1.0022,
+      "step": 7199
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.9109,
+      "step": 7200
+    },
+    {
+      "epoch": 32.14,
+      "eval_loss": 1.1131030321121216,
+      "eval_runtime": 44.5775,
+      "eval_samples_per_second": 293.735,
+      "eval_steps_per_second": 1.548,
+      "step": 7200
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 3.570535714285714e-05,
+      "loss": 1.0203,
+      "step": 7201
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 3.5696428571428575e-05,
+      "loss": 0.95,
+      "step": 7202
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 3.56875e-05,
+      "loss": 1.0407,
+      "step": 7203
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 3.5678571428571426e-05,
+      "loss": 0.9947,
+      "step": 7204
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3.566964285714286e-05,
+      "loss": 0.9779,
+      "step": 7205
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3.566071428571428e-05,
+      "loss": 0.9662,
+      "step": 7206
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3.5651785714285715e-05,
+      "loss": 0.9958,
+      "step": 7207
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 3.564285714285715e-05,
+      "loss": 1.0043,
+      "step": 7208
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 3.563392857142857e-05,
+      "loss": 0.9768,
+      "step": 7209
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 3.5625000000000005e-05,
+      "loss": 0.9229,
+      "step": 7210
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 3.561607142857143e-05,
+      "loss": 1.0073,
+      "step": 7211
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 3.560714285714286e-05,
+      "loss": 1.0272,
+      "step": 7212
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 3.559821428571429e-05,
+      "loss": 0.9785,
+      "step": 7213
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 3.5589285714285714e-05,
+      "loss": 0.9828,
+      "step": 7214
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 3.5580357142857146e-05,
+      "loss": 0.9878,
+      "step": 7215
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 3.557142857142857e-05,
+      "loss": 1.0233,
+      "step": 7216
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 3.5562500000000004e-05,
+      "loss": 0.9563,
+      "step": 7217
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 3.555357142857143e-05,
+      "loss": 0.9173,
+      "step": 7218
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 3.5544642857142855e-05,
+      "loss": 1.0237,
+      "step": 7219
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 3.553571428571429e-05,
+      "loss": 1.0127,
+      "step": 7220
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 3.552678571428571e-05,
+      "loss": 0.9425,
+      "step": 7221
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 3.5517857142857145e-05,
+      "loss": 1.0613,
+      "step": 7222
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3.550892857142858e-05,
+      "loss": 0.9491,
+      "step": 7223
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3.55e-05,
+      "loss": 0.9621,
+      "step": 7224
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3.5491071428571435e-05,
+      "loss": 1.0129,
+      "step": 7225
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 3.548214285714286e-05,
+      "loss": 0.9776,
+      "step": 7226
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 3.547321428571429e-05,
+      "loss": 0.9923,
+      "step": 7227
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 3.546428571428572e-05,
+      "loss": 0.9821,
+      "step": 7228
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 3.545535714285714e-05,
+      "loss": 0.9566,
+      "step": 7229
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 3.5446428571428575e-05,
+      "loss": 0.9683,
+      "step": 7230
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 3.54375e-05,
+      "loss": 0.9899,
+      "step": 7231
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 3.5428571428571426e-05,
+      "loss": 0.9797,
+      "step": 7232
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 3.541964285714286e-05,
+      "loss": 0.9834,
+      "step": 7233
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 3.5410714285714284e-05,
+      "loss": 0.9872,
+      "step": 7234
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 3.5401785714285716e-05,
+      "loss": 0.9472,
+      "step": 7235
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 3.539285714285714e-05,
+      "loss": 0.9977,
+      "step": 7236
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 3.5383928571428574e-05,
+      "loss": 1.0089,
+      "step": 7237
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 3.5375e-05,
+      "loss": 1.02,
+      "step": 7238
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 3.536607142857143e-05,
+      "loss": 0.9684,
+      "step": 7239
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 3.5357142857142864e-05,
+      "loss": 0.9542,
+      "step": 7240
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3.534821428571429e-05,
+      "loss": 0.9872,
+      "step": 7241
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3.5339285714285715e-05,
+      "loss": 0.9196,
+      "step": 7242
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3.533035714285715e-05,
+      "loss": 0.91,
+      "step": 7243
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 3.532142857142857e-05,
+      "loss": 0.9557,
+      "step": 7244
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 3.5312500000000005e-05,
+      "loss": 1.0249,
+      "step": 7245
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 3.530357142857143e-05,
+      "loss": 1.0114,
+      "step": 7246
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 3.5294642857142855e-05,
+      "loss": 0.9988,
+      "step": 7247
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 3.528571428571429e-05,
+      "loss": 0.9967,
+      "step": 7248
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 3.527678571428571e-05,
+      "loss": 0.9827,
+      "step": 7249
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 3.5267857142857145e-05,
+      "loss": 1.023,
+      "step": 7250
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 3.525892857142857e-05,
+      "loss": 0.9589,
+      "step": 7251
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 3.525e-05,
+      "loss": 1.0369,
+      "step": 7252
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 3.524107142857143e-05,
+      "loss": 0.9641,
+      "step": 7253
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 3.523214285714286e-05,
+      "loss": 0.9799,
+      "step": 7254
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 3.522321428571429e-05,
+      "loss": 0.9493,
+      "step": 7255
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 3.521428571428572e-05,
+      "loss": 1.0219,
+      "step": 7256
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 3.5205357142857144e-05,
+      "loss": 1.0341,
+      "step": 7257
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 3.5196428571428576e-05,
+      "loss": 1.0035,
+      "step": 7258
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 3.51875e-05,
+      "loss": 0.8885,
+      "step": 7259
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 3.5178571428571434e-05,
+      "loss": 0.9522,
+      "step": 7260
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 3.516964285714286e-05,
+      "loss": 0.9895,
+      "step": 7261
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 3.5160714285714285e-05,
+      "loss": 1.0154,
+      "step": 7262
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 3.515178571428572e-05,
+      "loss": 0.9674,
+      "step": 7263
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 3.514285714285714e-05,
+      "loss": 0.9407,
+      "step": 7264
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 3.513392857142857e-05,
+      "loss": 1.0291,
+      "step": 7265
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 3.5125e-05,
+      "loss": 0.9821,
+      "step": 7266
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 3.5116071428571425e-05,
+      "loss": 0.9133,
+      "step": 7267
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 3.510714285714286e-05,
+      "loss": 1.057,
+      "step": 7268
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 3.509821428571429e-05,
+      "loss": 0.9872,
+      "step": 7269
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 3.5089285714285715e-05,
+      "loss": 0.9953,
+      "step": 7270
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 3.508035714285715e-05,
+      "loss": 1.0533,
+      "step": 7271
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 3.507142857142857e-05,
+      "loss": 0.975,
+      "step": 7272
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 3.5062500000000005e-05,
+      "loss": 0.9775,
+      "step": 7273
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 3.505357142857143e-05,
+      "loss": 0.9829,
+      "step": 7274
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 3.5044642857142856e-05,
+      "loss": 1.0154,
+      "step": 7275
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 3.503571428571429e-05,
+      "loss": 1.0024,
+      "step": 7276
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 3.5026785714285714e-05,
+      "loss": 0.9503,
+      "step": 7277
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 3.5017857142857146e-05,
+      "loss": 1.024,
+      "step": 7278
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3.500892857142857e-05,
+      "loss": 0.9839,
+      "step": 7279
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3.5e-05,
+      "loss": 1.0059,
+      "step": 7280
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3.499107142857143e-05,
+      "loss": 1.0517,
+      "step": 7281
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 3.4982142857142855e-05,
+      "loss": 0.9868,
+      "step": 7282
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 3.497321428571429e-05,
+      "loss": 0.9978,
+      "step": 7283
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 3.496428571428572e-05,
+      "loss": 1.0186,
+      "step": 7284
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 3.4955357142857145e-05,
+      "loss": 0.9744,
+      "step": 7285
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 3.494642857142858e-05,
+      "loss": 0.9937,
+      "step": 7286
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 3.49375e-05,
+      "loss": 1.006,
+      "step": 7287
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 3.4928571428571434e-05,
+      "loss": 0.9522,
+      "step": 7288
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 3.491964285714286e-05,
+      "loss": 0.982,
+      "step": 7289
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 3.4910714285714285e-05,
+      "loss": 0.9831,
+      "step": 7290
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 3.490178571428572e-05,
+      "loss": 0.9429,
+      "step": 7291
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 3.489285714285714e-05,
+      "loss": 1.0104,
+      "step": 7292
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 3.4883928571428575e-05,
+      "loss": 0.9388,
+      "step": 7293
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 3.4875e-05,
+      "loss": 0.9724,
+      "step": 7294
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 3.4866071428571426e-05,
+      "loss": 1.0191,
+      "step": 7295
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 3.485714285714286e-05,
+      "loss": 0.9753,
+      "step": 7296
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 3.4848214285714284e-05,
+      "loss": 0.9949,
+      "step": 7297
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 3.4839285714285716e-05,
+      "loss": 0.9419,
+      "step": 7298
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 3.483035714285715e-05,
+      "loss": 0.951,
+      "step": 7299
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 3.4821428571428574e-05,
+      "loss": 0.9895,
+      "step": 7300
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 3.4812500000000006e-05,
+      "loss": 0.9717,
+      "step": 7301
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 3.480357142857143e-05,
+      "loss": 0.9582,
+      "step": 7302
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 3.479464285714286e-05,
+      "loss": 0.9862,
+      "step": 7303
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 3.478571428571429e-05,
+      "loss": 1.0013,
+      "step": 7304
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 3.4776785714285715e-05,
+      "loss": 0.9606,
+      "step": 7305
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 3.476785714285715e-05,
+      "loss": 1.0051,
+      "step": 7306
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 3.475892857142857e-05,
+      "loss": 0.9067,
+      "step": 7307
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 3.475e-05,
+      "loss": 1.0582,
+      "step": 7308
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 3.474107142857143e-05,
+      "loss": 0.9445,
+      "step": 7309
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 3.4732142857142855e-05,
+      "loss": 1.0077,
+      "step": 7310
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 3.472321428571429e-05,
+      "loss": 1.0217,
+      "step": 7311
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 3.471428571428571e-05,
+      "loss": 1.0626,
+      "step": 7312
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 3.4705357142857145e-05,
+      "loss": 1.0685,
+      "step": 7313
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 3.469642857142858e-05,
+      "loss": 0.9595,
+      "step": 7314
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 3.46875e-05,
+      "loss": 1.0401,
+      "step": 7315
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 3.4678571428571435e-05,
+      "loss": 0.9461,
+      "step": 7316
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3.466964285714286e-05,
+      "loss": 0.974,
+      "step": 7317
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3.4660714285714286e-05,
+      "loss": 0.9696,
+      "step": 7318
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3.465178571428572e-05,
+      "loss": 0.9904,
+      "step": 7319
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 3.4642857142857144e-05,
+      "loss": 0.9556,
+      "step": 7320
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 3.4633928571428576e-05,
+      "loss": 0.9966,
+      "step": 7321
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 3.4625e-05,
+      "loss": 0.9947,
+      "step": 7322
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 3.461607142857143e-05,
+      "loss": 1.0085,
+      "step": 7323
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 3.460714285714286e-05,
+      "loss": 0.989,
+      "step": 7324
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 3.4598214285714284e-05,
+      "loss": 1.0194,
+      "step": 7325
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 3.458928571428572e-05,
+      "loss": 0.9356,
+      "step": 7326
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 3.458035714285714e-05,
+      "loss": 0.9928,
+      "step": 7327
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 3.4571428571428574e-05,
+      "loss": 1.0093,
+      "step": 7328
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 3.45625e-05,
+      "loss": 1.0385,
+      "step": 7329
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 3.455357142857143e-05,
+      "loss": 0.9922,
+      "step": 7330
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 3.4544642857142864e-05,
+      "loss": 0.9334,
+      "step": 7331
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 3.453571428571429e-05,
+      "loss": 1.043,
+      "step": 7332
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 3.4526785714285715e-05,
+      "loss": 1.0452,
+      "step": 7333
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 3.451785714285715e-05,
+      "loss": 0.9988,
+      "step": 7334
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3.450892857142857e-05,
+      "loss": 0.9787,
+      "step": 7335
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3.45e-05,
+      "loss": 0.9327,
+      "step": 7336
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3.449107142857143e-05,
+      "loss": 0.9786,
+      "step": 7337
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 3.4482142857142856e-05,
+      "loss": 0.9837,
+      "step": 7338
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 3.447321428571429e-05,
+      "loss": 0.9921,
+      "step": 7339
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 3.4464285714285714e-05,
+      "loss": 1.0502,
+      "step": 7340
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 3.445535714285714e-05,
+      "loss": 0.9923,
+      "step": 7341
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 3.444642857142857e-05,
+      "loss": 1.0024,
+      "step": 7342
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 3.4437500000000004e-05,
+      "loss": 0.9689,
+      "step": 7343
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 3.442857142857143e-05,
+      "loss": 0.9811,
+      "step": 7344
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 3.441964285714286e-05,
+      "loss": 1.0034,
+      "step": 7345
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 3.441071428571429e-05,
+      "loss": 0.9355,
+      "step": 7346
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 3.440178571428572e-05,
+      "loss": 0.9522,
+      "step": 7347
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 3.4392857142857144e-05,
+      "loss": 0.9569,
+      "step": 7348
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 3.438392857142858e-05,
+      "loss": 0.9756,
+      "step": 7349
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.9954,
+      "step": 7350
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 3.436607142857143e-05,
+      "loss": 0.9805,
+      "step": 7351
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 3.435714285714286e-05,
+      "loss": 0.9365,
+      "step": 7352
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3.4348214285714285e-05,
+      "loss": 1.0222,
+      "step": 7353
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3.433928571428572e-05,
+      "loss": 0.934,
+      "step": 7354
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3.433035714285714e-05,
+      "loss": 0.9867,
+      "step": 7355
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 3.432142857142857e-05,
+      "loss": 1.0387,
+      "step": 7356
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 3.43125e-05,
+      "loss": 1.0375,
+      "step": 7357
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 3.4303571428571426e-05,
+      "loss": 0.9077,
+      "step": 7358
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 3.429464285714286e-05,
+      "loss": 1.0321,
+      "step": 7359
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 3.428571428571429e-05,
+      "loss": 1.0041,
+      "step": 7360
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 3.4276785714285716e-05,
+      "loss": 0.9758,
+      "step": 7361
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 3.426785714285715e-05,
+      "loss": 1.0035,
+      "step": 7362
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 3.4258928571428574e-05,
+      "loss": 0.9235,
+      "step": 7363
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 1.0419,
+      "step": 7364
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 3.424107142857143e-05,
+      "loss": 0.9615,
+      "step": 7365
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 3.423214285714286e-05,
+      "loss": 0.9348,
+      "step": 7366
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 3.422321428571429e-05,
+      "loss": 0.9586,
+      "step": 7367
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 3.4214285714285714e-05,
+      "loss": 0.9352,
+      "step": 7368
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 3.420535714285714e-05,
+      "loss": 1.0023,
+      "step": 7369
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 3.419642857142857e-05,
+      "loss": 0.9349,
+      "step": 7370
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 3.41875e-05,
+      "loss": 0.9918,
+      "step": 7371
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 3.417857142857143e-05,
+      "loss": 0.9495,
+      "step": 7372
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 3.4169642857142855e-05,
+      "loss": 0.9777,
+      "step": 7373
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 3.416071428571429e-05,
+      "loss": 1.0384,
+      "step": 7374
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 3.415178571428572e-05,
+      "loss": 0.9914,
+      "step": 7375
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 3.4142857142857145e-05,
+      "loss": 0.9234,
+      "step": 7376
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 3.413392857142858e-05,
+      "loss": 0.9772,
+      "step": 7377
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 3.4125e-05,
+      "loss": 0.9822,
+      "step": 7378
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 3.411607142857143e-05,
+      "loss": 0.9819,
+      "step": 7379
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 3.410714285714286e-05,
+      "loss": 1.0126,
+      "step": 7380
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 3.4098214285714286e-05,
+      "loss": 0.9631,
+      "step": 7381
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 3.408928571428572e-05,
+      "loss": 0.971,
+      "step": 7382
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 3.4080357142857144e-05,
+      "loss": 1.0404,
+      "step": 7383
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 3.407142857142857e-05,
+      "loss": 1.0172,
+      "step": 7384
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 3.40625e-05,
+      "loss": 0.8958,
+      "step": 7385
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 3.405357142857143e-05,
+      "loss": 0.9597,
+      "step": 7386
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 3.404464285714286e-05,
+      "loss": 0.9485,
+      "step": 7387
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 3.4035714285714284e-05,
+      "loss": 1.0219,
+      "step": 7388
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 3.4026785714285717e-05,
+      "loss": 0.9893,
+      "step": 7389
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 3.401785714285715e-05,
+      "loss": 1.0486,
+      "step": 7390
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.4008928571428574e-05,
+      "loss": 0.9441,
+      "step": 7391
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.1231,
+      "step": 7392
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.399107142857143e-05,
+      "loss": 0.9753,
+      "step": 7393
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3.398214285714286e-05,
+      "loss": 1.0565,
+      "step": 7394
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3.397321428571429e-05,
+      "loss": 0.9224,
+      "step": 7395
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 3.3964285714285715e-05,
+      "loss": 1.0262,
+      "step": 7396
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 3.395535714285715e-05,
+      "loss": 0.953,
+      "step": 7397
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 3.394642857142857e-05,
+      "loss": 0.9087,
+      "step": 7398
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 3.39375e-05,
+      "loss": 0.9344,
+      "step": 7399
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.9263,
+      "step": 7400
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 3.3919642857142856e-05,
+      "loss": 0.9383,
+      "step": 7401
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 3.391071428571429e-05,
+      "loss": 1.0147,
+      "step": 7402
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 3.3901785714285714e-05,
+      "loss": 0.9518,
+      "step": 7403
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 3.3892857142857146e-05,
+      "loss": 0.9625,
+      "step": 7404
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 3.388392857142858e-05,
+      "loss": 0.9812,
+      "step": 7405
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 3.3875000000000003e-05,
+      "loss": 1.0221,
+      "step": 7406
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 3.386607142857143e-05,
+      "loss": 1.0059,
+      "step": 7407
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 3.385714285714286e-05,
+      "loss": 0.9574,
+      "step": 7408
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 3.3848214285714287e-05,
+      "loss": 1.0432,
+      "step": 7409
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 3.383928571428572e-05,
+      "loss": 0.957,
+      "step": 7410
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 3.3830357142857144e-05,
+      "loss": 1.0171,
+      "step": 7411
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 3.382142857142857e-05,
+      "loss": 0.9427,
+      "step": 7412
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 3.38125e-05,
+      "loss": 0.9268,
+      "step": 7413
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 3.380357142857143e-05,
+      "loss": 1.0118,
+      "step": 7414
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 3.379464285714286e-05,
+      "loss": 0.9487,
+      "step": 7415
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 3.3785714285714285e-05,
+      "loss": 1.0306,
+      "step": 7416
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 3.377678571428571e-05,
+      "loss": 1.0371,
+      "step": 7417
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 3.376785714285714e-05,
+      "loss": 0.9451,
+      "step": 7418
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 3.3758928571428575e-05,
+      "loss": 0.9436,
+      "step": 7419
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 0.9721,
+      "step": 7420
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 3.374107142857143e-05,
+      "loss": 0.8664,
+      "step": 7421
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 3.373214285714286e-05,
+      "loss": 0.9706,
+      "step": 7422
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 3.372321428571429e-05,
+      "loss": 1.0273,
+      "step": 7423
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 3.3714285714285716e-05,
+      "loss": 1.0422,
+      "step": 7424
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 3.370535714285715e-05,
+      "loss": 0.9897,
+      "step": 7425
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 3.3696428571428573e-05,
+      "loss": 0.9307,
+      "step": 7426
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 3.36875e-05,
+      "loss": 0.9024,
+      "step": 7427
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 3.367857142857143e-05,
+      "loss": 0.9275,
+      "step": 7428
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 3.3669642857142857e-05,
+      "loss": 1.0219,
+      "step": 7429
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 3.366071428571429e-05,
+      "loss": 0.9069,
+      "step": 7430
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 3.3651785714285714e-05,
+      "loss": 0.9649,
+      "step": 7431
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 3.364285714285714e-05,
+      "loss": 1.0007,
+      "step": 7432
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 3.363392857142857e-05,
+      "loss": 1.0499,
+      "step": 7433
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 3.3625000000000004e-05,
+      "loss": 0.9779,
+      "step": 7434
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 3.361607142857143e-05,
+      "loss": 1.0069,
+      "step": 7435
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 3.360714285714286e-05,
+      "loss": 0.9904,
+      "step": 7436
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 3.359821428571429e-05,
+      "loss": 0.9275,
+      "step": 7437
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 3.358928571428572e-05,
+      "loss": 1.0246,
+      "step": 7438
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 3.3580357142857145e-05,
+      "loss": 0.9578,
+      "step": 7439
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 3.357142857142857e-05,
+      "loss": 0.9503,
+      "step": 7440
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 3.35625e-05,
+      "loss": 0.9872,
+      "step": 7441
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 3.355357142857143e-05,
+      "loss": 1.0235,
+      "step": 7442
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 3.354464285714286e-05,
+      "loss": 0.9469,
+      "step": 7443
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 3.3535714285714286e-05,
+      "loss": 0.998,
+      "step": 7444
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 3.352678571428571e-05,
+      "loss": 0.9157,
+      "step": 7445
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 3.3517857142857143e-05,
+      "loss": 0.9521,
+      "step": 7446
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 3.350892857142857e-05,
+      "loss": 0.9097,
+      "step": 7447
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 3.35e-05,
+      "loss": 1.0138,
+      "step": 7448
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 3.349107142857143e-05,
+      "loss": 1.0304,
+      "step": 7449
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 3.348214285714286e-05,
+      "loss": 1.0055,
+      "step": 7450
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 3.347321428571429e-05,
+      "loss": 1.0196,
+      "step": 7451
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 3.3464285714285716e-05,
+      "loss": 0.9797,
+      "step": 7452
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 3.345535714285715e-05,
+      "loss": 0.959,
+      "step": 7453
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 3.3446428571428574e-05,
+      "loss": 0.9645,
+      "step": 7454
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 3.34375e-05,
+      "loss": 0.922,
+      "step": 7455
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 3.342857142857143e-05,
+      "loss": 1.0349,
+      "step": 7456
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 3.341964285714286e-05,
+      "loss": 1.0581,
+      "step": 7457
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 3.341071428571429e-05,
+      "loss": 1.0159,
+      "step": 7458
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 3.3401785714285715e-05,
+      "loss": 1.023,
+      "step": 7459
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 3.339285714285714e-05,
+      "loss": 0.9861,
+      "step": 7460
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 3.338392857142857e-05,
+      "loss": 0.983,
+      "step": 7461
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 3.3375e-05,
+      "loss": 1.0037,
+      "step": 7462
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 3.336607142857143e-05,
+      "loss": 0.9757,
+      "step": 7463
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 3.3357142857142856e-05,
+      "loss": 0.9435,
+      "step": 7464
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 3.334821428571429e-05,
+      "loss": 0.9851,
+      "step": 7465
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 3.333928571428572e-05,
+      "loss": 0.9613,
+      "step": 7466
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 3.3330357142857146e-05,
+      "loss": 0.8844,
+      "step": 7467
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 3.332142857142858e-05,
+      "loss": 0.9511,
+      "step": 7468
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 3.33125e-05,
+      "loss": 0.9314,
+      "step": 7469
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 3.330357142857143e-05,
+      "loss": 0.9282,
+      "step": 7470
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 3.329464285714286e-05,
+      "loss": 0.9171,
+      "step": 7471
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 3.3285714285714286e-05,
+      "loss": 0.9726,
+      "step": 7472
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 3.327678571428571e-05,
+      "loss": 0.9922,
+      "step": 7473
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 3.3267857142857144e-05,
+      "loss": 0.8873,
+      "step": 7474
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 3.325892857142857e-05,
+      "loss": 0.975,
+      "step": 7475
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 3.325e-05,
+      "loss": 0.9402,
+      "step": 7476
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 3.324107142857143e-05,
+      "loss": 0.9762,
+      "step": 7477
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 3.323214285714286e-05,
+      "loss": 0.9544,
+      "step": 7478
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 3.3223214285714285e-05,
+      "loss": 1.0032,
+      "step": 7479
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 3.321428571428572e-05,
+      "loss": 0.9631,
+      "step": 7480
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 3.320535714285715e-05,
+      "loss": 0.9423,
+      "step": 7481
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 3.3196428571428575e-05,
+      "loss": 0.9509,
+      "step": 7482
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 3.31875e-05,
+      "loss": 1.0154,
+      "step": 7483
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 3.317857142857143e-05,
+      "loss": 0.9549,
+      "step": 7484
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 3.316964285714286e-05,
+      "loss": 0.9875,
+      "step": 7485
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 3.316071428571429e-05,
+      "loss": 0.9659,
+      "step": 7486
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 3.3151785714285716e-05,
+      "loss": 1.0221,
+      "step": 7487
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 3.314285714285714e-05,
+      "loss": 0.9879,
+      "step": 7488
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 3.313392857142857e-05,
+      "loss": 0.9979,
+      "step": 7489
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 3.3125e-05,
+      "loss": 0.9518,
+      "step": 7490
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 3.311607142857143e-05,
+      "loss": 0.9373,
+      "step": 7491
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 3.3107142857142856e-05,
+      "loss": 0.9689,
+      "step": 7492
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 3.309821428571428e-05,
+      "loss": 0.984,
+      "step": 7493
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 3.3089285714285714e-05,
+      "loss": 0.9803,
+      "step": 7494
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 3.3080357142857146e-05,
+      "loss": 0.9278,
+      "step": 7495
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 3.307142857142858e-05,
+      "loss": 0.9463,
+      "step": 7496
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 3.3062500000000004e-05,
+      "loss": 0.9666,
+      "step": 7497
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 3.305357142857143e-05,
+      "loss": 0.9559,
+      "step": 7498
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 3.304464285714286e-05,
+      "loss": 1.0242,
+      "step": 7499
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 3.303571428571429e-05,
+      "loss": 0.933,
+      "step": 7500
+    },
+    {
+      "epoch": 33.48,
+      "eval_loss": 1.1135326623916626,
+      "eval_runtime": 44.4876,
+      "eval_samples_per_second": 294.329,
+      "eval_steps_per_second": 1.551,
+      "step": 7500
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 3.302678571428572e-05,
+      "loss": 1.0287,
+      "step": 7501
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 3.3017857142857145e-05,
+      "loss": 0.9809,
+      "step": 7502
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 3.300892857142857e-05,
+      "loss": 1.0373,
+      "step": 7503
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 3.3e-05,
+      "loss": 0.9934,
+      "step": 7504
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 3.299107142857143e-05,
+      "loss": 0.9893,
+      "step": 7505
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 3.298214285714286e-05,
+      "loss": 0.9606,
+      "step": 7506
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 3.2973214285714286e-05,
+      "loss": 0.9746,
+      "step": 7507
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 3.296428571428571e-05,
+      "loss": 0.9652,
+      "step": 7508
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 3.295535714285714e-05,
+      "loss": 1.0064,
+      "step": 7509
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 3.2946428571428576e-05,
+      "loss": 0.9702,
+      "step": 7510
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 3.29375e-05,
+      "loss": 0.9982,
+      "step": 7511
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 3.292857142857143e-05,
+      "loss": 1.0327,
+      "step": 7512
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 3.291964285714286e-05,
+      "loss": 0.9504,
+      "step": 7513
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 3.291071428571429e-05,
+      "loss": 0.9104,
+      "step": 7514
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 3.2901785714285716e-05,
+      "loss": 0.9391,
+      "step": 7515
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 3.289285714285714e-05,
+      "loss": 1.0642,
+      "step": 7516
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 3.2883928571428574e-05,
+      "loss": 0.9264,
+      "step": 7517
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 3.2875e-05,
+      "loss": 0.964,
+      "step": 7518
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 3.286607142857143e-05,
+      "loss": 1.046,
+      "step": 7519
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.9754,
+      "step": 7520
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 3.284821428571428e-05,
+      "loss": 0.9209,
+      "step": 7521
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 3.2839285714285715e-05,
+      "loss": 0.9825,
+      "step": 7522
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 3.283035714285714e-05,
+      "loss": 1.059,
+      "step": 7523
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 3.282142857142857e-05,
+      "loss": 0.9813,
+      "step": 7524
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.9888,
+      "step": 7525
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 3.280357142857143e-05,
+      "loss": 0.9937,
+      "step": 7526
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 3.279464285714286e-05,
+      "loss": 0.9986,
+      "step": 7527
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 3.278571428571429e-05,
+      "loss": 1.0141,
+      "step": 7528
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 3.277678571428572e-05,
+      "loss": 1.026,
+      "step": 7529
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 3.2767857142857146e-05,
+      "loss": 0.9825,
+      "step": 7530
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 3.275892857142857e-05,
+      "loss": 0.9651,
+      "step": 7531
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 3.275e-05,
+      "loss": 0.9073,
+      "step": 7532
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 3.274107142857143e-05,
+      "loss": 0.9525,
+      "step": 7533
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 3.273214285714286e-05,
+      "loss": 0.9989,
+      "step": 7534
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 3.2723214285714286e-05,
+      "loss": 1.0252,
+      "step": 7535
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 3.271428571428571e-05,
+      "loss": 1.0119,
+      "step": 7536
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 3.2705357142857144e-05,
+      "loss": 0.9813,
+      "step": 7537
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 3.269642857142857e-05,
+      "loss": 1.0466,
+      "step": 7538
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 3.26875e-05,
+      "loss": 0.9183,
+      "step": 7539
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 3.2678571428571434e-05,
+      "loss": 0.9764,
+      "step": 7540
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 3.266964285714286e-05,
+      "loss": 1.0086,
+      "step": 7541
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 3.266071428571429e-05,
+      "loss": 0.9841,
+      "step": 7542
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 3.265178571428572e-05,
+      "loss": 0.9732,
+      "step": 7543
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 3.264285714285714e-05,
+      "loss": 1.0116,
+      "step": 7544
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 3.2633928571428575e-05,
+      "loss": 0.9742,
+      "step": 7545
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 3.2625e-05,
+      "loss": 0.9425,
+      "step": 7546
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 3.261607142857143e-05,
+      "loss": 0.9079,
+      "step": 7547
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 3.260714285714286e-05,
+      "loss": 0.904,
+      "step": 7548
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 3.259821428571428e-05,
+      "loss": 0.9152,
+      "step": 7549
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 3.2589285714285716e-05,
+      "loss": 0.9903,
+      "step": 7550
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 3.258035714285714e-05,
+      "loss": 0.9901,
+      "step": 7551
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 3.257142857142857e-05,
+      "loss": 0.9455,
+      "step": 7552
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 3.25625e-05,
+      "loss": 0.972,
+      "step": 7553
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 3.255357142857143e-05,
+      "loss": 0.9144,
+      "step": 7554
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 3.2544642857142856e-05,
+      "loss": 0.8712,
+      "step": 7555
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 3.253571428571429e-05,
+      "loss": 0.9595,
+      "step": 7556
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 3.252678571428572e-05,
+      "loss": 1.0237,
+      "step": 7557
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 3.2517857142857146e-05,
+      "loss": 0.9843,
+      "step": 7558
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 3.250892857142857e-05,
+      "loss": 1.0603,
+      "step": 7559
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.0647,
+      "step": 7560
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 3.249107142857143e-05,
+      "loss": 0.9896,
+      "step": 7561
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 3.248214285714286e-05,
+      "loss": 0.9247,
+      "step": 7562
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 3.247321428571429e-05,
+      "loss": 0.9521,
+      "step": 7563
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 3.246428571428571e-05,
+      "loss": 0.9476,
+      "step": 7564
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 3.2455357142857145e-05,
+      "loss": 1.0187,
+      "step": 7565
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 3.244642857142857e-05,
+      "loss": 1.0679,
+      "step": 7566
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 3.24375e-05,
+      "loss": 1.0452,
+      "step": 7567
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 3.242857142857143e-05,
+      "loss": 1.0018,
+      "step": 7568
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 3.241964285714286e-05,
+      "loss": 1.0025,
+      "step": 7569
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 3.2410714285714286e-05,
+      "loss": 0.9238,
+      "step": 7570
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 3.240178571428572e-05,
+      "loss": 1.0036,
+      "step": 7571
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 3.239285714285715e-05,
+      "loss": 0.945,
+      "step": 7572
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 3.2383928571428575e-05,
+      "loss": 0.9239,
+      "step": 7573
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 3.2375e-05,
+      "loss": 0.9405,
+      "step": 7574
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 3.236607142857143e-05,
+      "loss": 1.0068,
+      "step": 7575
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 3.235714285714286e-05,
+      "loss": 0.9943,
+      "step": 7576
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 3.234821428571429e-05,
+      "loss": 1.0094,
+      "step": 7577
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 3.2339285714285716e-05,
+      "loss": 0.975,
+      "step": 7578
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 3.233035714285714e-05,
+      "loss": 0.9625,
+      "step": 7579
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 3.2321428571428574e-05,
+      "loss": 0.9223,
+      "step": 7580
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 3.23125e-05,
+      "loss": 0.9956,
+      "step": 7581
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 3.2303571428571425e-05,
+      "loss": 1.0337,
+      "step": 7582
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 3.229464285714286e-05,
+      "loss": 1.002,
+      "step": 7583
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 3.228571428571428e-05,
+      "loss": 1.0332,
+      "step": 7584
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 3.2276785714285715e-05,
+      "loss": 0.9899,
+      "step": 7585
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 3.226785714285715e-05,
+      "loss": 1.0332,
+      "step": 7586
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 3.225892857142857e-05,
+      "loss": 1.022,
+      "step": 7587
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 1.0039,
+      "step": 7588
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 3.224107142857143e-05,
+      "loss": 0.9975,
+      "step": 7589
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 3.223214285714286e-05,
+      "loss": 0.9502,
+      "step": 7590
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 3.222321428571429e-05,
+      "loss": 1.064,
+      "step": 7591
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 3.221428571428571e-05,
+      "loss": 0.9939,
+      "step": 7592
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 3.2205357142857145e-05,
+      "loss": 0.9963,
+      "step": 7593
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 3.219642857142857e-05,
+      "loss": 0.995,
+      "step": 7594
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 3.21875e-05,
+      "loss": 1.0295,
+      "step": 7595
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 3.217857142857143e-05,
+      "loss": 1.0215,
+      "step": 7596
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 3.2169642857142854e-05,
+      "loss": 0.936,
+      "step": 7597
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 3.2160714285714286e-05,
+      "loss": 0.9941,
+      "step": 7598
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 3.215178571428571e-05,
+      "loss": 1.0161,
+      "step": 7599
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.9528,
+      "step": 7600
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 3.2133928571428576e-05,
+      "loss": 1.0333,
+      "step": 7601
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 3.2125e-05,
+      "loss": 0.9749,
+      "step": 7602
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 3.2116071428571434e-05,
+      "loss": 1.0009,
+      "step": 7603
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 3.210714285714286e-05,
+      "loss": 0.9533,
+      "step": 7604
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 3.209821428571429e-05,
+      "loss": 0.9687,
+      "step": 7605
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 3.208928571428572e-05,
+      "loss": 0.9443,
+      "step": 7606
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 3.208035714285714e-05,
+      "loss": 0.966,
+      "step": 7607
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 3.2071428571428575e-05,
+      "loss": 0.9607,
+      "step": 7608
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 3.20625e-05,
+      "loss": 0.9757,
+      "step": 7609
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 3.205357142857143e-05,
+      "loss": 0.9763,
+      "step": 7610
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 3.204464285714286e-05,
+      "loss": 0.9742,
+      "step": 7611
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 3.203571428571428e-05,
+      "loss": 1.0146,
+      "step": 7612
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 3.2026785714285715e-05,
+      "loss": 1.0386,
+      "step": 7613
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 3.201785714285714e-05,
+      "loss": 1.0376,
+      "step": 7614
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.200892857142857e-05,
+      "loss": 1.0031,
+      "step": 7615
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.875,
+      "step": 7616
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.199107142857143e-05,
+      "loss": 1.0026,
+      "step": 7617
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.198214285714286e-05,
+      "loss": 1.0034,
+      "step": 7618
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.197321428571429e-05,
+      "loss": 0.9738,
+      "step": 7619
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.1964285714285714e-05,
+      "loss": 0.8953,
+      "step": 7620
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.1955357142857146e-05,
+      "loss": 0.984,
+      "step": 7621
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.194642857142857e-05,
+      "loss": 1.0094,
+      "step": 7622
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.1937500000000004e-05,
+      "loss": 0.9646,
+      "step": 7623
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.192857142857143e-05,
+      "loss": 0.9672,
+      "step": 7624
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.1919642857142855e-05,
+      "loss": 0.9237,
+      "step": 7625
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.191071428571429e-05,
+      "loss": 0.9705,
+      "step": 7626
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.190178571428571e-05,
+      "loss": 0.9412,
+      "step": 7627
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.1892857142857145e-05,
+      "loss": 0.9793,
+      "step": 7628
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.188392857142857e-05,
+      "loss": 1.0357,
+      "step": 7629
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.1875e-05,
+      "loss": 0.94,
+      "step": 7630
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.1866071428571435e-05,
+      "loss": 1.0327,
+      "step": 7631
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.185714285714286e-05,
+      "loss": 0.9266,
+      "step": 7632
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.184821428571429e-05,
+      "loss": 0.9638,
+      "step": 7633
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.183928571428572e-05,
+      "loss": 0.9504,
+      "step": 7634
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.183035714285714e-05,
+      "loss": 0.9192,
+      "step": 7635
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.1821428571428575e-05,
+      "loss": 1.0087,
+      "step": 7636
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.18125e-05,
+      "loss": 1.0049,
+      "step": 7637
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.180357142857143e-05,
+      "loss": 0.9829,
+      "step": 7638
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.179464285714286e-05,
+      "loss": 1.0213,
+      "step": 7639
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.1785714285714284e-05,
+      "loss": 0.9423,
+      "step": 7640
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.1776785714285716e-05,
+      "loss": 0.972,
+      "step": 7641
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.176785714285714e-05,
+      "loss": 0.9397,
+      "step": 7642
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.1758928571428574e-05,
+      "loss": 0.9446,
+      "step": 7643
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.175e-05,
+      "loss": 1.0121,
+      "step": 7644
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.174107142857143e-05,
+      "loss": 0.8949,
+      "step": 7645
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.173214285714286e-05,
+      "loss": 0.9411,
+      "step": 7646
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.172321428571429e-05,
+      "loss": 0.9862,
+      "step": 7647
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.1714285714285715e-05,
+      "loss": 0.9772,
+      "step": 7648
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.170535714285715e-05,
+      "loss": 0.9641,
+      "step": 7649
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.169642857142857e-05,
+      "loss": 0.9812,
+      "step": 7650
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.1687500000000005e-05,
+      "loss": 1.0426,
+      "step": 7651
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.167857142857143e-05,
+      "loss": 0.9533,
+      "step": 7652
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.1669642857142855e-05,
+      "loss": 0.9684,
+      "step": 7653
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.166071428571429e-05,
+      "loss": 0.914,
+      "step": 7654
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.165178571428571e-05,
+      "loss": 1.0175,
+      "step": 7655
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.1642857142857145e-05,
+      "loss": 0.9485,
+      "step": 7656
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.163392857142857e-05,
+      "loss": 0.9609,
+      "step": 7657
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.1624999999999996e-05,
+      "loss": 0.9182,
+      "step": 7658
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.161607142857143e-05,
+      "loss": 1.0299,
+      "step": 7659
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.160714285714286e-05,
+      "loss": 1.0193,
+      "step": 7660
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.1598214285714286e-05,
+      "loss": 1.0268,
+      "step": 7661
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.158928571428572e-05,
+      "loss": 0.956,
+      "step": 7662
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.1580357142857144e-05,
+      "loss": 1.0471,
+      "step": 7663
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.1571428571428576e-05,
+      "loss": 1.0075,
+      "step": 7664
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.15625e-05,
+      "loss": 0.9725,
+      "step": 7665
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.1553571428571434e-05,
+      "loss": 1.0385,
+      "step": 7666
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.154464285714286e-05,
+      "loss": 0.9597,
+      "step": 7667
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.1535714285714285e-05,
+      "loss": 0.9657,
+      "step": 7668
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.152678571428572e-05,
+      "loss": 0.9946,
+      "step": 7669
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.151785714285714e-05,
+      "loss": 0.9937,
+      "step": 7670
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.1508928571428575e-05,
+      "loss": 0.943,
+      "step": 7671
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.15e-05,
+      "loss": 0.9775,
+      "step": 7672
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.1491071428571425e-05,
+      "loss": 0.9354,
+      "step": 7673
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.148214285714286e-05,
+      "loss": 1.0117,
+      "step": 7674
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.147321428571428e-05,
+      "loss": 1.0384,
+      "step": 7675
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.1464285714285715e-05,
+      "loss": 0.9932,
+      "step": 7676
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.145535714285715e-05,
+      "loss": 1.0113,
+      "step": 7677
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.144642857142857e-05,
+      "loss": 0.9943,
+      "step": 7678
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.1437500000000005e-05,
+      "loss": 1.035,
+      "step": 7679
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.9417,
+      "step": 7680
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.141964285714286e-05,
+      "loss": 1.0201,
+      "step": 7681
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.141071428571429e-05,
+      "loss": 0.9842,
+      "step": 7682
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.1401785714285714e-05,
+      "loss": 0.9587,
+      "step": 7683
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.1392857142857146e-05,
+      "loss": 0.9668,
+      "step": 7684
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.138392857142857e-05,
+      "loss": 0.9628,
+      "step": 7685
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.1375e-05,
+      "loss": 0.9524,
+      "step": 7686
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.136607142857143e-05,
+      "loss": 0.937,
+      "step": 7687
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.1357142857142855e-05,
+      "loss": 0.9992,
+      "step": 7688
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.134821428571429e-05,
+      "loss": 1.0279,
+      "step": 7689
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.133928571428571e-05,
+      "loss": 0.9404,
+      "step": 7690
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.1330357142857144e-05,
+      "loss": 0.9399,
+      "step": 7691
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.132142857142858e-05,
+      "loss": 0.9955,
+      "step": 7692
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.13125e-05,
+      "loss": 0.9335,
+      "step": 7693
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.1303571428571434e-05,
+      "loss": 0.935,
+      "step": 7694
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.129464285714286e-05,
+      "loss": 0.9244,
+      "step": 7695
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.1285714285714285e-05,
+      "loss": 0.9677,
+      "step": 7696
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.127678571428572e-05,
+      "loss": 0.9598,
+      "step": 7697
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.126785714285714e-05,
+      "loss": 0.9816,
+      "step": 7698
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.1258928571428575e-05,
+      "loss": 0.9867,
+      "step": 7699
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.125e-05,
+      "loss": 0.9609,
+      "step": 7700
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.1241071428571426e-05,
+      "loss": 0.9946,
+      "step": 7701
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.123214285714286e-05,
+      "loss": 0.9901,
+      "step": 7702
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.1223214285714284e-05,
+      "loss": 0.9503,
+      "step": 7703
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.1214285714285716e-05,
+      "loss": 0.9527,
+      "step": 7704
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.120535714285714e-05,
+      "loss": 1.0343,
+      "step": 7705
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.1196428571428574e-05,
+      "loss": 0.9651,
+      "step": 7706
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.1187500000000006e-05,
+      "loss": 0.9203,
+      "step": 7707
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.117857142857143e-05,
+      "loss": 0.9872,
+      "step": 7708
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.1169642857142864e-05,
+      "loss": 0.949,
+      "step": 7709
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.116071428571429e-05,
+      "loss": 0.9425,
+      "step": 7710
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.1151785714285714e-05,
+      "loss": 1.0099,
+      "step": 7711
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.114285714285715e-05,
+      "loss": 0.9996,
+      "step": 7712
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.113392857142857e-05,
+      "loss": 1.0086,
+      "step": 7713
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.1125000000000004e-05,
+      "loss": 0.9741,
+      "step": 7714
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.111607142857143e-05,
+      "loss": 0.9942,
+      "step": 7715
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.1107142857142855e-05,
+      "loss": 0.965,
+      "step": 7716
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.109821428571429e-05,
+      "loss": 0.9826,
+      "step": 7717
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.108928571428571e-05,
+      "loss": 1.0194,
+      "step": 7718
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.108035714285714e-05,
+      "loss": 1.0347,
+      "step": 7719
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.107142857142857e-05,
+      "loss": 0.997,
+      "step": 7720
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.10625e-05,
+      "loss": 0.956,
+      "step": 7721
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.1053571428571435e-05,
+      "loss": 1.003,
+      "step": 7722
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.104464285714286e-05,
+      "loss": 0.9695,
+      "step": 7723
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.1035714285714286e-05,
+      "loss": 1.02,
+      "step": 7724
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.102678571428572e-05,
+      "loss": 0.9927,
+      "step": 7725
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.1017857142857144e-05,
+      "loss": 0.9075,
+      "step": 7726
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.1008928571428576e-05,
+      "loss": 0.9653,
+      "step": 7727
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.1e-05,
+      "loss": 1.0349,
+      "step": 7728
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.099107142857143e-05,
+      "loss": 0.9812,
+      "step": 7729
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.098214285714286e-05,
+      "loss": 0.9894,
+      "step": 7730
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.0973214285714284e-05,
+      "loss": 0.9479,
+      "step": 7731
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.096428571428572e-05,
+      "loss": 1.0082,
+      "step": 7732
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.095535714285714e-05,
+      "loss": 0.9848,
+      "step": 7733
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.094642857142857e-05,
+      "loss": 1.0025,
+      "step": 7734
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.09375e-05,
+      "loss": 1.0427,
+      "step": 7735
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.092857142857143e-05,
+      "loss": 0.8958,
+      "step": 7736
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.0919642857142864e-05,
+      "loss": 0.9355,
+      "step": 7737
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.091071428571429e-05,
+      "loss": 0.9016,
+      "step": 7738
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.0901785714285715e-05,
+      "loss": 0.9931,
+      "step": 7739
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.089285714285715e-05,
+      "loss": 0.9232,
+      "step": 7740
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.088392857142857e-05,
+      "loss": 1.024,
+      "step": 7741
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.0875000000000005e-05,
+      "loss": 0.9719,
+      "step": 7742
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.086607142857143e-05,
+      "loss": 0.9666,
+      "step": 7743
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.0857142857142856e-05,
+      "loss": 0.9915,
+      "step": 7744
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.084821428571429e-05,
+      "loss": 1.0215,
+      "step": 7745
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.0839285714285714e-05,
+      "loss": 0.9902,
+      "step": 7746
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.0830357142857146e-05,
+      "loss": 0.9657,
+      "step": 7747
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.082142857142857e-05,
+      "loss": 1.0007,
+      "step": 7748
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.08125e-05,
+      "loss": 0.9801,
+      "step": 7749
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.080357142857143e-05,
+      "loss": 1.0018,
+      "step": 7750
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.079464285714286e-05,
+      "loss": 0.9671,
+      "step": 7751
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.078571428571429e-05,
+      "loss": 0.9293,
+      "step": 7752
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.077678571428572e-05,
+      "loss": 0.9271,
+      "step": 7753
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.0767857142857144e-05,
+      "loss": 0.9512,
+      "step": 7754
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.0758928571428577e-05,
+      "loss": 0.9634,
+      "step": 7755
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.075e-05,
+      "loss": 0.9692,
+      "step": 7756
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.074107142857143e-05,
+      "loss": 1.0299,
+      "step": 7757
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.073214285714286e-05,
+      "loss": 1.0096,
+      "step": 7758
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.0723214285714285e-05,
+      "loss": 0.9565,
+      "step": 7759
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.071428571428572e-05,
+      "loss": 1.016,
+      "step": 7760
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.070535714285714e-05,
+      "loss": 0.9368,
+      "step": 7761
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.069642857142857e-05,
+      "loss": 0.95,
+      "step": 7762
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.06875e-05,
+      "loss": 0.9102,
+      "step": 7763
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.0678571428571426e-05,
+      "loss": 0.9453,
+      "step": 7764
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.066964285714286e-05,
+      "loss": 0.947,
+      "step": 7765
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.066071428571429e-05,
+      "loss": 1.0706,
+      "step": 7766
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.0651785714285716e-05,
+      "loss": 0.9807,
+      "step": 7767
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.064285714285715e-05,
+      "loss": 0.9746,
+      "step": 7768
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.0633928571428574e-05,
+      "loss": 0.9864,
+      "step": 7769
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.0625000000000006e-05,
+      "loss": 0.9628,
+      "step": 7770
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.061607142857143e-05,
+      "loss": 0.9685,
+      "step": 7771
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.060714285714286e-05,
+      "loss": 0.992,
+      "step": 7772
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.059821428571429e-05,
+      "loss": 0.9382,
+      "step": 7773
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.0589285714285714e-05,
+      "loss": 0.9846,
+      "step": 7774
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.0580357142857147e-05,
+      "loss": 0.9222,
+      "step": 7775
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.057142857142857e-05,
+      "loss": 0.9404,
+      "step": 7776
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.05625e-05,
+      "loss": 0.9792,
+      "step": 7777
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.055357142857143e-05,
+      "loss": 0.9523,
+      "step": 7778
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.0544642857142855e-05,
+      "loss": 0.9991,
+      "step": 7779
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.053571428571429e-05,
+      "loss": 0.967,
+      "step": 7780
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.052678571428571e-05,
+      "loss": 1.0051,
+      "step": 7781
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.0517857142857145e-05,
+      "loss": 0.8841,
+      "step": 7782
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.0508928571428574e-05,
+      "loss": 0.9656,
+      "step": 7783
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.05e-05,
+      "loss": 1.0433,
+      "step": 7784
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.049107142857143e-05,
+      "loss": 0.9752,
+      "step": 7785
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.048214285714286e-05,
+      "loss": 0.9812,
+      "step": 7786
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.0473214285714286e-05,
+      "loss": 0.9926,
+      "step": 7787
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.0464285714285718e-05,
+      "loss": 0.9554,
+      "step": 7788
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.0455357142857144e-05,
+      "loss": 1.0103,
+      "step": 7789
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.044642857142857e-05,
+      "loss": 0.9639,
+      "step": 7790
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.04375e-05,
+      "loss": 1.0041,
+      "step": 7791
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.042857142857143e-05,
+      "loss": 0.9726,
+      "step": 7792
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.041964285714286e-05,
+      "loss": 0.9838,
+      "step": 7793
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.0410714285714288e-05,
+      "loss": 1.0438,
+      "step": 7794
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.0401785714285713e-05,
+      "loss": 0.9695,
+      "step": 7795
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.0392857142857145e-05,
+      "loss": 1.0132,
+      "step": 7796
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.038392857142857e-05,
+      "loss": 1.0327,
+      "step": 7797
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.0375000000000003e-05,
+      "loss": 0.9635,
+      "step": 7798
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.036607142857143e-05,
+      "loss": 0.939,
+      "step": 7799
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.9672,
+      "step": 7800
+    },
+    {
+      "epoch": 34.82,
+      "eval_loss": 1.1125041246414185,
+      "eval_runtime": 44.5135,
+      "eval_samples_per_second": 294.158,
+      "eval_steps_per_second": 1.55,
+      "step": 7800
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.034821428571429e-05,
+      "loss": 0.9554,
+      "step": 7801
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.0339285714285715e-05,
+      "loss": 0.9583,
+      "step": 7802
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.0330357142857147e-05,
+      "loss": 0.9695,
+      "step": 7803
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.0321428571428573e-05,
+      "loss": 0.9042,
+      "step": 7804
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.0312499999999998e-05,
+      "loss": 0.9505,
+      "step": 7805
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.030357142857143e-05,
+      "loss": 1.0298,
+      "step": 7806
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.0294642857142856e-05,
+      "loss": 0.9454,
+      "step": 7807
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.0285714285714288e-05,
+      "loss": 1.0363,
+      "step": 7808
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.0276785714285717e-05,
+      "loss": 0.9165,
+      "step": 7809
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.0267857142857142e-05,
+      "loss": 1.0477,
+      "step": 7810
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.0258928571428575e-05,
+      "loss": 0.9913,
+      "step": 7811
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.025e-05,
+      "loss": 0.957,
+      "step": 7812
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.0241071428571432e-05,
+      "loss": 1.0002,
+      "step": 7813
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.0232142857142858e-05,
+      "loss": 0.9679,
+      "step": 7814
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.0223214285714287e-05,
+      "loss": 0.91,
+      "step": 7815
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.021428571428572e-05,
+      "loss": 1.0024,
+      "step": 7816
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.0205357142857144e-05,
+      "loss": 0.9354,
+      "step": 7817
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.0196428571428576e-05,
+      "loss": 0.8938,
+      "step": 7818
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.0187500000000002e-05,
+      "loss": 1.0057,
+      "step": 7819
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.0178571428571427e-05,
+      "loss": 0.9655,
+      "step": 7820
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.016964285714286e-05,
+      "loss": 1.0613,
+      "step": 7821
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.0160714285714285e-05,
+      "loss": 0.9896,
+      "step": 7822
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.0151785714285714e-05,
+      "loss": 0.9066,
+      "step": 7823
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.0142857142857146e-05,
+      "loss": 1.0312,
+      "step": 7824
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.013392857142857e-05,
+      "loss": 0.9655,
+      "step": 7825
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.0125000000000004e-05,
+      "loss": 0.9706,
+      "step": 7826
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.011607142857143e-05,
+      "loss": 0.9742,
+      "step": 7827
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.0107142857142855e-05,
+      "loss": 1.0459,
+      "step": 7828
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.0098214285714287e-05,
+      "loss": 0.9445,
+      "step": 7829
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.0089285714285716e-05,
+      "loss": 0.947,
+      "step": 7830
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.0080357142857145e-05,
+      "loss": 1.0067,
+      "step": 7831
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.0071428571428573e-05,
+      "loss": 0.9461,
+      "step": 7832
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.00625e-05,
+      "loss": 0.9863,
+      "step": 7833
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.005357142857143e-05,
+      "loss": 0.8928,
+      "step": 7834
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.0044642857142857e-05,
+      "loss": 0.9257,
+      "step": 7835
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.003571428571429e-05,
+      "loss": 0.9278,
+      "step": 7836
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.0026785714285714e-05,
+      "loss": 0.9362,
+      "step": 7837
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.0017857142857143e-05,
+      "loss": 0.9283,
+      "step": 7838
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.0008928571428575e-05,
+      "loss": 0.9882,
+      "step": 7839
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3e-05,
+      "loss": 1.0151,
+      "step": 7840
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 2.9991071428571433e-05,
+      "loss": 1.0076,
+      "step": 7841
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 2.998214285714286e-05,
+      "loss": 0.9547,
+      "step": 7842
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 2.9973214285714284e-05,
+      "loss": 0.9488,
+      "step": 7843
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 2.9964285714285716e-05,
+      "loss": 0.9328,
+      "step": 7844
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 2.9955357142857145e-05,
+      "loss": 0.9482,
+      "step": 7845
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 2.9946428571428574e-05,
+      "loss": 1.0114,
+      "step": 7846
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 2.9937500000000003e-05,
+      "loss": 0.8944,
+      "step": 7847
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 2.9928571428571428e-05,
+      "loss": 0.8788,
+      "step": 7848
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 2.991964285714286e-05,
+      "loss": 1.0399,
+      "step": 7849
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 2.9910714285714286e-05,
+      "loss": 0.9583,
+      "step": 7850
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 2.9901785714285718e-05,
+      "loss": 0.9688,
+      "step": 7851
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 2.9892857142857143e-05,
+      "loss": 1.0409,
+      "step": 7852
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 2.9883928571428572e-05,
+      "loss": 0.974,
+      "step": 7853
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 2.9875000000000004e-05,
+      "loss": 0.9961,
+      "step": 7854
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 2.986607142857143e-05,
+      "loss": 0.9479,
+      "step": 7855
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 2.9857142857142862e-05,
+      "loss": 1.0159,
+      "step": 7856
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 2.9848214285714288e-05,
+      "loss": 0.9886,
+      "step": 7857
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 2.9839285714285713e-05,
+      "loss": 0.9495,
+      "step": 7858
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 2.9830357142857145e-05,
+      "loss": 0.9905,
+      "step": 7859
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 2.982142857142857e-05,
+      "loss": 0.9594,
+      "step": 7860
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 2.98125e-05,
+      "loss": 1.0078,
+      "step": 7861
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 2.9803571428571432e-05,
+      "loss": 0.9291,
+      "step": 7862
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 2.9794642857142857e-05,
+      "loss": 0.9554,
+      "step": 7863
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 2.978571428571429e-05,
+      "loss": 0.9491,
+      "step": 7864
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 2.9776785714285715e-05,
+      "loss": 0.9344,
+      "step": 7865
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 2.976785714285714e-05,
+      "loss": 0.911,
+      "step": 7866
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 2.9758928571428573e-05,
+      "loss": 0.9884,
+      "step": 7867
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 2.975e-05,
+      "loss": 0.865,
+      "step": 7868
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 2.9741071428571434e-05,
+      "loss": 0.9488,
+      "step": 7869
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 2.973214285714286e-05,
+      "loss": 1.0366,
+      "step": 7870
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 2.9723214285714285e-05,
+      "loss": 0.8939,
+      "step": 7871
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 2.9714285714285717e-05,
+      "loss": 0.9012,
+      "step": 7872
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 2.9705357142857142e-05,
+      "loss": 0.9629,
+      "step": 7873
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 2.9696428571428574e-05,
+      "loss": 0.9708,
+      "step": 7874
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 2.96875e-05,
+      "loss": 1.0526,
+      "step": 7875
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 2.967857142857143e-05,
+      "loss": 0.9402,
+      "step": 7876
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 2.966964285714286e-05,
+      "loss": 0.9279,
+      "step": 7877
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 2.9660714285714286e-05,
+      "loss": 0.9396,
+      "step": 7878
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 2.965178571428572e-05,
+      "loss": 0.9261,
+      "step": 7879
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 2.9642857142857144e-05,
+      "loss": 1.0254,
+      "step": 7880
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 2.963392857142857e-05,
+      "loss": 1.052,
+      "step": 7881
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 2.9625000000000002e-05,
+      "loss": 1.0002,
+      "step": 7882
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 2.961607142857143e-05,
+      "loss": 0.9976,
+      "step": 7883
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 2.960714285714286e-05,
+      "loss": 0.9731,
+      "step": 7884
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 2.959821428571429e-05,
+      "loss": 0.9092,
+      "step": 7885
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 2.9589285714285714e-05,
+      "loss": 0.984,
+      "step": 7886
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 2.9580357142857146e-05,
+      "loss": 0.9804,
+      "step": 7887
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 2.957142857142857e-05,
+      "loss": 1.0069,
+      "step": 7888
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 2.9562500000000004e-05,
+      "loss": 1.006,
+      "step": 7889
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 2.955357142857143e-05,
+      "loss": 0.9629,
+      "step": 7890
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 2.9544642857142858e-05,
+      "loss": 0.9543,
+      "step": 7891
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 2.953571428571429e-05,
+      "loss": 1.0015,
+      "step": 7892
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 2.9526785714285716e-05,
+      "loss": 0.9829,
+      "step": 7893
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 2.951785714285714e-05,
+      "loss": 1.0018,
+      "step": 7894
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 2.9508928571428573e-05,
+      "loss": 0.9286,
+      "step": 7895
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 2.95e-05,
+      "loss": 1.0424,
+      "step": 7896
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 2.949107142857143e-05,
+      "loss": 0.9555,
+      "step": 7897
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 2.948214285714286e-05,
+      "loss": 0.9338,
+      "step": 7898
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 2.9473214285714285e-05,
+      "loss": 1.0159,
+      "step": 7899
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 2.9464285714285718e-05,
+      "loss": 0.9923,
+      "step": 7900
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 2.9455357142857143e-05,
+      "loss": 1.0409,
+      "step": 7901
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 2.9446428571428575e-05,
+      "loss": 0.9846,
+      "step": 7902
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 2.94375e-05,
+      "loss": 0.9767,
+      "step": 7903
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 2.9428571428571426e-05,
+      "loss": 1.0078,
+      "step": 7904
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 2.941964285714286e-05,
+      "loss": 1.011,
+      "step": 7905
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 2.9410714285714287e-05,
+      "loss": 0.9861,
+      "step": 7906
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 2.940178571428572e-05,
+      "loss": 0.9884,
+      "step": 7907
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 2.9392857142857145e-05,
+      "loss": 0.9329,
+      "step": 7908
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 2.938392857142857e-05,
+      "loss": 0.959,
+      "step": 7909
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 2.9375000000000003e-05,
+      "loss": 0.9995,
+      "step": 7910
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 2.9366071428571428e-05,
+      "loss": 1.0067,
+      "step": 7911
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 2.935714285714286e-05,
+      "loss": 0.945,
+      "step": 7912
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 2.9348214285714286e-05,
+      "loss": 0.8528,
+      "step": 7913
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 2.9339285714285714e-05,
+      "loss": 0.9913,
+      "step": 7914
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 2.9330357142857147e-05,
+      "loss": 0.9416,
+      "step": 7915
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 2.9321428571428572e-05,
+      "loss": 0.9937,
+      "step": 7916
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 2.9312500000000004e-05,
+      "loss": 1.0036,
+      "step": 7917
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 2.930357142857143e-05,
+      "loss": 1.0573,
+      "step": 7918
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 2.9294642857142855e-05,
+      "loss": 1.0266,
+      "step": 7919
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 2.9285714285714288e-05,
+      "loss": 0.9337,
+      "step": 7920
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 2.9276785714285716e-05,
+      "loss": 0.8933,
+      "step": 7921
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 2.9267857142857145e-05,
+      "loss": 0.8887,
+      "step": 7922
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 2.9258928571428574e-05,
+      "loss": 0.9208,
+      "step": 7923
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 2.925e-05,
+      "loss": 0.9459,
+      "step": 7924
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 2.9241071428571432e-05,
+      "loss": 0.9356,
+      "step": 7925
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 2.9232142857142857e-05,
+      "loss": 0.9723,
+      "step": 7926
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 2.922321428571429e-05,
+      "loss": 0.9786,
+      "step": 7927
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 2.9214285714285715e-05,
+      "loss": 0.9121,
+      "step": 7928
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 2.9205357142857144e-05,
+      "loss": 0.9839,
+      "step": 7929
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 2.9196428571428576e-05,
+      "loss": 0.929,
+      "step": 7930
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 2.91875e-05,
+      "loss": 0.952,
+      "step": 7931
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 2.9178571428571427e-05,
+      "loss": 0.9565,
+      "step": 7932
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 2.916964285714286e-05,
+      "loss": 0.9938,
+      "step": 7933
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 2.9160714285714284e-05,
+      "loss": 0.9664,
+      "step": 7934
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 2.9151785714285717e-05,
+      "loss": 0.9956,
+      "step": 7935
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 2.9142857142857146e-05,
+      "loss": 1.0151,
+      "step": 7936
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 2.913392857142857e-05,
+      "loss": 1.0182,
+      "step": 7937
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 2.9125000000000003e-05,
+      "loss": 1.0154,
+      "step": 7938
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 2.911607142857143e-05,
+      "loss": 0.9463,
+      "step": 7939
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 2.910714285714286e-05,
+      "loss": 0.9824,
+      "step": 7940
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 2.9098214285714286e-05,
+      "loss": 1.0106,
+      "step": 7941
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 2.9089285714285712e-05,
+      "loss": 0.9894,
+      "step": 7942
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 2.9080357142857144e-05,
+      "loss": 0.9473,
+      "step": 7943
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 2.9071428571428573e-05,
+      "loss": 1.0177,
+      "step": 7944
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 2.9062500000000005e-05,
+      "loss": 0.9746,
+      "step": 7945
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 2.905357142857143e-05,
+      "loss": 0.9865,
+      "step": 7946
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 2.9044642857142856e-05,
+      "loss": 0.9801,
+      "step": 7947
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 2.9035714285714288e-05,
+      "loss": 0.9323,
+      "step": 7948
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 2.9026785714285714e-05,
+      "loss": 0.9383,
+      "step": 7949
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 2.9017857142857146e-05,
+      "loss": 0.9459,
+      "step": 7950
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 2.900892857142857e-05,
+      "loss": 0.9944,
+      "step": 7951
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 2.9e-05,
+      "loss": 0.9977,
+      "step": 7952
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 2.8991071428571432e-05,
+      "loss": 0.9661,
+      "step": 7953
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 2.8982142857142858e-05,
+      "loss": 0.9762,
+      "step": 7954
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 2.897321428571429e-05,
+      "loss": 0.9554,
+      "step": 7955
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 2.8964285714285716e-05,
+      "loss": 0.9864,
+      "step": 7956
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 2.895535714285714e-05,
+      "loss": 1.0475,
+      "step": 7957
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 2.8946428571428573e-05,
+      "loss": 1.0507,
+      "step": 7958
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 2.8937500000000002e-05,
+      "loss": 0.9792,
+      "step": 7959
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 2.8928571428571434e-05,
+      "loss": 0.8541,
+      "step": 7960
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 2.891964285714286e-05,
+      "loss": 0.9417,
+      "step": 7961
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 2.8910714285714285e-05,
+      "loss": 0.9809,
+      "step": 7962
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 2.8901785714285717e-05,
+      "loss": 0.9528,
+      "step": 7963
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 2.8892857142857143e-05,
+      "loss": 1.0301,
+      "step": 7964
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 2.888392857142857e-05,
+      "loss": 0.9573,
+      "step": 7965
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 2.8875e-05,
+      "loss": 0.9803,
+      "step": 7966
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 2.886607142857143e-05,
+      "loss": 1.0553,
+      "step": 7967
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 2.885714285714286e-05,
+      "loss": 1.0092,
+      "step": 7968
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 2.8848214285714287e-05,
+      "loss": 1.009,
+      "step": 7969
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 2.8839285714285712e-05,
+      "loss": 0.9087,
+      "step": 7970
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 2.8830357142857145e-05,
+      "loss": 1.0495,
+      "step": 7971
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 2.882142857142857e-05,
+      "loss": 0.9069,
+      "step": 7972
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 2.8812500000000002e-05,
+      "loss": 0.9825,
+      "step": 7973
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 2.880357142857143e-05,
+      "loss": 0.9651,
+      "step": 7974
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 2.8794642857142857e-05,
+      "loss": 0.9924,
+      "step": 7975
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 2.878571428571429e-05,
+      "loss": 0.933,
+      "step": 7976
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 2.8776785714285714e-05,
+      "loss": 0.9753,
+      "step": 7977
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 2.8767857142857147e-05,
+      "loss": 1.0143,
+      "step": 7978
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 2.8758928571428572e-05,
+      "loss": 0.9571,
+      "step": 7979
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 0.9372,
+      "step": 7980
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 2.874107142857143e-05,
+      "loss": 0.9438,
+      "step": 7981
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 2.873214285714286e-05,
+      "loss": 0.9394,
+      "step": 7982
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 2.872321428571429e-05,
+      "loss": 0.9206,
+      "step": 7983
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 2.8714285714285716e-05,
+      "loss": 0.9918,
+      "step": 7984
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 2.870535714285714e-05,
+      "loss": 0.9611,
+      "step": 7985
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 2.8696428571428574e-05,
+      "loss": 0.9612,
+      "step": 7986
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 2.86875e-05,
+      "loss": 0.9916,
+      "step": 7987
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 2.867857142857143e-05,
+      "loss": 0.9445,
+      "step": 7988
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 2.866964285714286e-05,
+      "loss": 0.9901,
+      "step": 7989
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 2.8660714285714286e-05,
+      "loss": 0.9521,
+      "step": 7990
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 2.8651785714285718e-05,
+      "loss": 0.9435,
+      "step": 7991
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 2.8642857142857144e-05,
+      "loss": 0.9937,
+      "step": 7992
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 2.8633928571428576e-05,
+      "loss": 0.9475,
+      "step": 7993
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 2.8625e-05,
+      "loss": 0.9565,
+      "step": 7994
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 2.8616071428571427e-05,
+      "loss": 0.9317,
+      "step": 7995
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 2.860714285714286e-05,
+      "loss": 0.9761,
+      "step": 7996
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 2.8598214285714288e-05,
+      "loss": 1.0223,
+      "step": 7997
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 2.8589285714285713e-05,
+      "loss": 0.9609,
+      "step": 7998
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 2.8580357142857145e-05,
+      "loss": 0.9318,
+      "step": 7999
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.8781,
+      "step": 8000
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 2.8562500000000003e-05,
+      "loss": 1.0224,
+      "step": 8001
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 2.855357142857143e-05,
+      "loss": 0.9461,
+      "step": 8002
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 2.8544642857142857e-05,
+      "loss": 1.0374,
+      "step": 8003
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 2.8535714285714286e-05,
+      "loss": 0.9598,
+      "step": 8004
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 2.8526785714285715e-05,
+      "loss": 0.9142,
+      "step": 8005
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 2.8517857142857147e-05,
+      "loss": 0.9291,
+      "step": 8006
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 2.8508928571428573e-05,
+      "loss": 1.0003,
+      "step": 8007
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.9875,
+      "step": 8008
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 2.849107142857143e-05,
+      "loss": 0.9635,
+      "step": 8009
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 2.8482142857142856e-05,
+      "loss": 0.9356,
+      "step": 8010
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 2.8473214285714288e-05,
+      "loss": 0.9602,
+      "step": 8011
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 2.8464285714285717e-05,
+      "loss": 0.932,
+      "step": 8012
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 2.8455357142857142e-05,
+      "loss": 1.0133,
+      "step": 8013
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 2.8446428571428575e-05,
+      "loss": 0.948,
+      "step": 8014
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 2.84375e-05,
+      "loss": 1.0119,
+      "step": 8015
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 2.8428571428571432e-05,
+      "loss": 0.8828,
+      "step": 8016
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 2.8419642857142858e-05,
+      "loss": 1.0142,
+      "step": 8017
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 2.8410714285714287e-05,
+      "loss": 0.9482,
+      "step": 8018
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 2.8401785714285715e-05,
+      "loss": 1.0666,
+      "step": 8019
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 2.8392857142857144e-05,
+      "loss": 1.0589,
+      "step": 8020
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 2.8383928571428576e-05,
+      "loss": 0.9865,
+      "step": 8021
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 2.8375000000000002e-05,
+      "loss": 0.9576,
+      "step": 8022
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 2.8366071428571427e-05,
+      "loss": 0.9111,
+      "step": 8023
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 2.835714285714286e-05,
+      "loss": 0.9998,
+      "step": 8024
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 2.8348214285714285e-05,
+      "loss": 1.0134,
+      "step": 8025
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 2.8339285714285717e-05,
+      "loss": 0.9737,
+      "step": 8026
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 2.8330357142857146e-05,
+      "loss": 0.8384,
+      "step": 8027
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 2.832142857142857e-05,
+      "loss": 0.9363,
+      "step": 8028
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 2.8312500000000004e-05,
+      "loss": 0.9696,
+      "step": 8029
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 2.830357142857143e-05,
+      "loss": 0.9647,
+      "step": 8030
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 2.829464285714286e-05,
+      "loss": 1.0053,
+      "step": 8031
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 2.8285714285714287e-05,
+      "loss": 1.0237,
+      "step": 8032
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 2.8276785714285712e-05,
+      "loss": 0.9538,
+      "step": 8033
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 2.8267857142857145e-05,
+      "loss": 0.9189,
+      "step": 8034
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 2.8258928571428573e-05,
+      "loss": 0.9048,
+      "step": 8035
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 2.825e-05,
+      "loss": 1.0095,
+      "step": 8036
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 2.824107142857143e-05,
+      "loss": 0.8616,
+      "step": 8037
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 2.8232142857142857e-05,
+      "loss": 0.9642,
+      "step": 8038
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 2.822321428571429e-05,
+      "loss": 0.9895,
+      "step": 8039
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 2.8214285714285714e-05,
+      "loss": 0.9309,
+      "step": 8040
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 2.8205357142857143e-05,
+      "loss": 0.968,
+      "step": 8041
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 2.8196428571428575e-05,
+      "loss": 1.0083,
+      "step": 8042
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 2.81875e-05,
+      "loss": 0.9667,
+      "step": 8043
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 2.8178571428571433e-05,
+      "loss": 0.987,
+      "step": 8044
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 2.816964285714286e-05,
+      "loss": 0.9709,
+      "step": 8045
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 2.8160714285714284e-05,
+      "loss": 0.9542,
+      "step": 8046
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 2.8151785714285716e-05,
+      "loss": 0.96,
+      "step": 8047
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 2.814285714285714e-05,
+      "loss": 1.0241,
+      "step": 8048
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 2.8133928571428574e-05,
+      "loss": 0.9226,
+      "step": 8049
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.9802,
+      "step": 8050
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 2.8116071428571428e-05,
+      "loss": 1.0091,
+      "step": 8051
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 2.810714285714286e-05,
+      "loss": 0.9719,
+      "step": 8052
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 2.8098214285714286e-05,
+      "loss": 0.934,
+      "step": 8053
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 2.8089285714285718e-05,
+      "loss": 0.9266,
+      "step": 8054
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 2.8080357142857143e-05,
+      "loss": 0.9598,
+      "step": 8055
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 2.8071428571428572e-05,
+      "loss": 0.9395,
+      "step": 8056
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 2.80625e-05,
+      "loss": 1.0781,
+      "step": 8057
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 2.805357142857143e-05,
+      "loss": 0.9424,
+      "step": 8058
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 2.8044642857142862e-05,
+      "loss": 1.0626,
+      "step": 8059
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 2.8035714285714288e-05,
+      "loss": 0.9427,
+      "step": 8060
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 2.8026785714285713e-05,
+      "loss": 1.0234,
+      "step": 8061
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 2.8017857142857145e-05,
+      "loss": 1.0194,
+      "step": 8062
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 2.800892857142857e-05,
+      "loss": 0.9694,
+      "step": 8063
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.9808,
+      "step": 8064
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 2.7991071428571432e-05,
+      "loss": 0.9362,
+      "step": 8065
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 2.7982142857142857e-05,
+      "loss": 0.9325,
+      "step": 8066
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 2.797321428571429e-05,
+      "loss": 1.0079,
+      "step": 8067
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 2.7964285714285715e-05,
+      "loss": 0.9071,
+      "step": 8068
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 2.795535714285714e-05,
+      "loss": 0.9379,
+      "step": 8069
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 2.7946428571428573e-05,
+      "loss": 0.9867,
+      "step": 8070
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 2.79375e-05,
+      "loss": 1.0266,
+      "step": 8071
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 2.792857142857143e-05,
+      "loss": 0.9723,
+      "step": 8072
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 2.791964285714286e-05,
+      "loss": 1.0163,
+      "step": 8073
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 2.7910714285714285e-05,
+      "loss": 0.9563,
+      "step": 8074
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 2.7901785714285717e-05,
+      "loss": 0.9735,
+      "step": 8075
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 2.7892857142857142e-05,
+      "loss": 0.9889,
+      "step": 8076
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 2.7883928571428574e-05,
+      "loss": 0.9039,
+      "step": 8077
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 2.7875e-05,
+      "loss": 0.8872,
+      "step": 8078
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 2.786607142857143e-05,
+      "loss": 0.9677,
+      "step": 8079
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 2.785714285714286e-05,
+      "loss": 1.0103,
+      "step": 8080
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 2.7848214285714286e-05,
+      "loss": 1.008,
+      "step": 8081
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 2.783928571428572e-05,
+      "loss": 0.9786,
+      "step": 8082
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 2.7830357142857144e-05,
+      "loss": 0.8598,
+      "step": 8083
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 2.782142857142857e-05,
+      "loss": 0.9056,
+      "step": 8084
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 2.7812500000000002e-05,
+      "loss": 0.9994,
+      "step": 8085
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 2.7803571428571427e-05,
+      "loss": 0.9165,
+      "step": 8086
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 2.779464285714286e-05,
+      "loss": 0.9735,
+      "step": 8087
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 2.778571428571429e-05,
+      "loss": 0.9958,
+      "step": 8088
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 2.7776785714285714e-05,
+      "loss": 0.9436,
+      "step": 8089
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 2.7767857142857146e-05,
+      "loss": 0.9269,
+      "step": 8090
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 2.775892857142857e-05,
+      "loss": 1.0035,
+      "step": 8091
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 1.0225,
+      "step": 8092
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 2.774107142857143e-05,
+      "loss": 0.9388,
+      "step": 8093
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 2.7732142857142858e-05,
+      "loss": 0.9308,
+      "step": 8094
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 2.7723214285714287e-05,
+      "loss": 0.8994,
+      "step": 8095
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 2.7714285714285716e-05,
+      "loss": 0.9904,
+      "step": 8096
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 2.7705357142857148e-05,
+      "loss": 0.9019,
+      "step": 8097
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 2.7696428571428573e-05,
+      "loss": 1.0006,
+      "step": 8098
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 2.76875e-05,
+      "loss": 0.9724,
+      "step": 8099
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 2.767857142857143e-05,
+      "loss": 0.9244,
+      "step": 8100
+    },
+    {
+      "epoch": 36.16,
+      "eval_loss": 1.111864447593689,
+      "eval_runtime": 44.6818,
+      "eval_samples_per_second": 293.05,
+      "eval_steps_per_second": 1.544,
+      "step": 8100
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 2.7669642857142856e-05,
+      "loss": 0.9964,
+      "step": 8101
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 2.766071428571429e-05,
+      "loss": 0.9506,
+      "step": 8102
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 2.7651785714285718e-05,
+      "loss": 1.0436,
+      "step": 8103
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 2.7642857142857143e-05,
+      "loss": 1.0104,
+      "step": 8104
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 2.7633928571428575e-05,
+      "loss": 0.9374,
+      "step": 8105
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 2.7625e-05,
+      "loss": 1.0111,
+      "step": 8106
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 2.7616071428571426e-05,
+      "loss": 0.9598,
+      "step": 8107
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 2.760714285714286e-05,
+      "loss": 0.9333,
+      "step": 8108
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 2.7598214285714287e-05,
+      "loss": 0.998,
+      "step": 8109
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 2.7589285714285716e-05,
+      "loss": 1.0153,
+      "step": 8110
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 2.7580357142857145e-05,
+      "loss": 0.9303,
+      "step": 8111
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 2.757142857142857e-05,
+      "loss": 1.055,
+      "step": 8112
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 2.7562500000000002e-05,
+      "loss": 0.9741,
+      "step": 8113
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 2.7553571428571428e-05,
+      "loss": 0.9263,
+      "step": 8114
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 2.754464285714286e-05,
+      "loss": 0.9584,
+      "step": 8115
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 2.7535714285714286e-05,
+      "loss": 0.9325,
+      "step": 8116
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 2.7526785714285714e-05,
+      "loss": 0.9817,
+      "step": 8117
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 2.7517857142857147e-05,
+      "loss": 0.9325,
+      "step": 8118
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 2.7508928571428572e-05,
+      "loss": 1.0278,
+      "step": 8119
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.0503,
+      "step": 8120
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 2.749107142857143e-05,
+      "loss": 1.0386,
+      "step": 8121
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 2.7482142857142855e-05,
+      "loss": 0.9964,
+      "step": 8122
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 2.7473214285714287e-05,
+      "loss": 0.9551,
+      "step": 8123
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 2.7464285714285713e-05,
+      "loss": 0.9462,
+      "step": 8124
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 2.7455357142857145e-05,
+      "loss": 0.9982,
+      "step": 8125
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 2.7446428571428574e-05,
+      "loss": 0.9908,
+      "step": 8126
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 2.74375e-05,
+      "loss": 1.0147,
+      "step": 8127
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 2.742857142857143e-05,
+      "loss": 1.0208,
+      "step": 8128
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 2.7419642857142857e-05,
+      "loss": 1.0169,
+      "step": 8129
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 2.741071428571429e-05,
+      "loss": 0.9542,
+      "step": 8130
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 2.7401785714285715e-05,
+      "loss": 0.9837,
+      "step": 8131
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 2.7392857142857144e-05,
+      "loss": 0.9594,
+      "step": 8132
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 2.7383928571428576e-05,
+      "loss": 0.9134,
+      "step": 8133
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 2.7375e-05,
+      "loss": 0.9649,
+      "step": 8134
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 2.7366071428571434e-05,
+      "loss": 0.9376,
+      "step": 8135
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 2.735714285714286e-05,
+      "loss": 0.9192,
+      "step": 8136
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 2.7348214285714284e-05,
+      "loss": 1.0621,
+      "step": 8137
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 2.7339285714285717e-05,
+      "loss": 1.0191,
+      "step": 8138
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 2.7330357142857142e-05,
+      "loss": 0.9982,
+      "step": 8139
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 2.732142857142857e-05,
+      "loss": 0.9833,
+      "step": 8140
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 2.7312500000000003e-05,
+      "loss": 0.9847,
+      "step": 8141
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 2.730357142857143e-05,
+      "loss": 0.9807,
+      "step": 8142
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 2.729464285714286e-05,
+      "loss": 0.9385,
+      "step": 8143
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 2.7285714285714286e-05,
+      "loss": 0.975,
+      "step": 8144
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 2.7276785714285712e-05,
+      "loss": 0.9449,
+      "step": 8145
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 2.7267857142857144e-05,
+      "loss": 0.994,
+      "step": 8146
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 2.7258928571428573e-05,
+      "loss": 0.9583,
+      "step": 8147
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 2.725e-05,
+      "loss": 1.0296,
+      "step": 8148
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 2.724107142857143e-05,
+      "loss": 0.9853,
+      "step": 8149
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 2.7232142857142856e-05,
+      "loss": 0.9833,
+      "step": 8150
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 2.7223214285714288e-05,
+      "loss": 0.9788,
+      "step": 8151
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 2.7214285714285714e-05,
+      "loss": 0.9304,
+      "step": 8152
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 2.7205357142857146e-05,
+      "loss": 0.9756,
+      "step": 8153
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 2.719642857142857e-05,
+      "loss": 0.8935,
+      "step": 8154
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 2.71875e-05,
+      "loss": 0.9798,
+      "step": 8155
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 2.7178571428571432e-05,
+      "loss": 0.8677,
+      "step": 8156
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 2.7169642857142858e-05,
+      "loss": 1.0644,
+      "step": 8157
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 2.716071428571429e-05,
+      "loss": 0.9551,
+      "step": 8158
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 2.7151785714285716e-05,
+      "loss": 0.9607,
+      "step": 8159
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 2.714285714285714e-05,
+      "loss": 0.9576,
+      "step": 8160
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 2.7133928571428573e-05,
+      "loss": 0.9709,
+      "step": 8161
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 2.7125000000000002e-05,
+      "loss": 0.9353,
+      "step": 8162
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 2.711607142857143e-05,
+      "loss": 0.9843,
+      "step": 8163
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 2.710714285714286e-05,
+      "loss": 0.9922,
+      "step": 8164
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 2.7098214285714285e-05,
+      "loss": 0.9224,
+      "step": 8165
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 2.7089285714285717e-05,
+      "loss": 0.982,
+      "step": 8166
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 2.7080357142857143e-05,
+      "loss": 0.9007,
+      "step": 8167
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 2.7071428571428575e-05,
+      "loss": 0.952,
+      "step": 8168
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 2.70625e-05,
+      "loss": 0.9303,
+      "step": 8169
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 2.705357142857143e-05,
+      "loss": 0.992,
+      "step": 8170
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 2.704464285714286e-05,
+      "loss": 0.9429,
+      "step": 8171
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 2.7035714285714287e-05,
+      "loss": 0.9995,
+      "step": 8172
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 2.7026785714285712e-05,
+      "loss": 0.9845,
+      "step": 8173
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 2.7017857142857145e-05,
+      "loss": 0.9194,
+      "step": 8174
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 2.700892857142857e-05,
+      "loss": 0.9763,
+      "step": 8175
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.9568,
+      "step": 8176
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 2.6991071428571428e-05,
+      "loss": 0.9281,
+      "step": 8177
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 2.6982142857142857e-05,
+      "loss": 1.0697,
+      "step": 8178
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 2.697321428571429e-05,
+      "loss": 1.0412,
+      "step": 8179
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 2.6964285714285714e-05,
+      "loss": 0.8719,
+      "step": 8180
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 2.6955357142857147e-05,
+      "loss": 0.965,
+      "step": 8181
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 2.6946428571428572e-05,
+      "loss": 0.9649,
+      "step": 8182
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 2.6937499999999997e-05,
+      "loss": 0.9387,
+      "step": 8183
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 2.692857142857143e-05,
+      "loss": 0.967,
+      "step": 8184
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 2.691964285714286e-05,
+      "loss": 0.9811,
+      "step": 8185
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 2.691071428571429e-05,
+      "loss": 0.9415,
+      "step": 8186
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 2.6901785714285716e-05,
+      "loss": 0.982,
+      "step": 8187
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 2.689285714285714e-05,
+      "loss": 0.9494,
+      "step": 8188
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 2.6883928571428574e-05,
+      "loss": 0.975,
+      "step": 8189
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 2.6875e-05,
+      "loss": 1.0267,
+      "step": 8190
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 2.686607142857143e-05,
+      "loss": 0.9966,
+      "step": 8191
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 2.6857142857142857e-05,
+      "loss": 0.9707,
+      "step": 8192
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 2.6848214285714286e-05,
+      "loss": 0.9991,
+      "step": 8193
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 2.6839285714285718e-05,
+      "loss": 0.9545,
+      "step": 8194
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 2.6830357142857144e-05,
+      "loss": 0.9121,
+      "step": 8195
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 2.6821428571428576e-05,
+      "loss": 0.986,
+      "step": 8196
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 2.68125e-05,
+      "loss": 0.9439,
+      "step": 8197
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 2.6803571428571427e-05,
+      "loss": 0.9501,
+      "step": 8198
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 2.679464285714286e-05,
+      "loss": 0.991,
+      "step": 8199
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.9725,
+      "step": 8200
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 2.6776785714285717e-05,
+      "loss": 1.0031,
+      "step": 8201
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 2.6767857142857145e-05,
+      "loss": 0.9992,
+      "step": 8202
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 2.675892857142857e-05,
+      "loss": 0.9693,
+      "step": 8203
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 0.9381,
+      "step": 8204
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 2.674107142857143e-05,
+      "loss": 0.9987,
+      "step": 8205
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 2.673214285714286e-05,
+      "loss": 0.9208,
+      "step": 8206
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 2.6723214285714286e-05,
+      "loss": 0.9635,
+      "step": 8207
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 2.6714285714285715e-05,
+      "loss": 0.9489,
+      "step": 8208
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 2.6705357142857147e-05,
+      "loss": 1.0272,
+      "step": 8209
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 2.6696428571428573e-05,
+      "loss": 0.9308,
+      "step": 8210
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 2.6687499999999998e-05,
+      "loss": 0.9154,
+      "step": 8211
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 2.667857142857143e-05,
+      "loss": 0.9115,
+      "step": 8212
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 2.6669642857142856e-05,
+      "loss": 0.9691,
+      "step": 8213
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 2.6660714285714288e-05,
+      "loss": 0.9505,
+      "step": 8214
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 2.6651785714285717e-05,
+      "loss": 0.9075,
+      "step": 8215
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 2.6642857142857142e-05,
+      "loss": 0.8754,
+      "step": 8216
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 2.6633928571428575e-05,
+      "loss": 0.8855,
+      "step": 8217
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 2.6625e-05,
+      "loss": 1.0005,
+      "step": 8218
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 2.6616071428571432e-05,
+      "loss": 0.9991,
+      "step": 8219
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 2.6607142857142858e-05,
+      "loss": 0.9561,
+      "step": 8220
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 2.6598214285714283e-05,
+      "loss": 0.9777,
+      "step": 8221
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 2.6589285714285715e-05,
+      "loss": 0.9122,
+      "step": 8222
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 2.6580357142857144e-05,
+      "loss": 0.9896,
+      "step": 8223
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 2.6571428571428576e-05,
+      "loss": 0.994,
+      "step": 8224
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 0.9504,
+      "step": 8225
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 2.6553571428571427e-05,
+      "loss": 0.9687,
+      "step": 8226
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 2.654464285714286e-05,
+      "loss": 1.021,
+      "step": 8227
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 2.6535714285714285e-05,
+      "loss": 0.938,
+      "step": 8228
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 2.6526785714285717e-05,
+      "loss": 1.0334,
+      "step": 8229
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 2.6517857142857143e-05,
+      "loss": 0.9541,
+      "step": 8230
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 2.650892857142857e-05,
+      "loss": 0.9375,
+      "step": 8231
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.9472,
+      "step": 8232
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 2.649107142857143e-05,
+      "loss": 0.9499,
+      "step": 8233
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 2.648214285714286e-05,
+      "loss": 0.9578,
+      "step": 8234
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 2.6473214285714287e-05,
+      "loss": 0.9217,
+      "step": 8235
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 2.6464285714285712e-05,
+      "loss": 0.9859,
+      "step": 8236
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 2.6455357142857145e-05,
+      "loss": 0.9496,
+      "step": 8237
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 2.6446428571428573e-05,
+      "loss": 0.9957,
+      "step": 8238
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 2.6437500000000002e-05,
+      "loss": 0.988,
+      "step": 8239
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 2.642857142857143e-05,
+      "loss": 0.9819,
+      "step": 8240
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 2.6419642857142857e-05,
+      "loss": 1.0064,
+      "step": 8241
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 2.641071428571429e-05,
+      "loss": 0.9897,
+      "step": 8242
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 2.6401785714285714e-05,
+      "loss": 0.9013,
+      "step": 8243
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 2.6392857142857143e-05,
+      "loss": 0.9567,
+      "step": 8244
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 2.6383928571428572e-05,
+      "loss": 0.9381,
+      "step": 8245
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 2.6375e-05,
+      "loss": 0.9354,
+      "step": 8246
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 2.6366071428571433e-05,
+      "loss": 0.9539,
+      "step": 8247
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 2.635714285714286e-05,
+      "loss": 0.9481,
+      "step": 8248
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 2.6348214285714284e-05,
+      "loss": 0.9448,
+      "step": 8249
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 2.6339285714285716e-05,
+      "loss": 1.0276,
+      "step": 8250
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 2.633035714285714e-05,
+      "loss": 0.9996,
+      "step": 8251
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 2.6321428571428574e-05,
+      "loss": 0.9433,
+      "step": 8252
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 2.6312500000000003e-05,
+      "loss": 0.9574,
+      "step": 8253
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 2.6303571428571428e-05,
+      "loss": 0.9259,
+      "step": 8254
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 2.629464285714286e-05,
+      "loss": 0.9734,
+      "step": 8255
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 2.6285714285714286e-05,
+      "loss": 0.9702,
+      "step": 8256
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 2.6276785714285718e-05,
+      "loss": 0.9756,
+      "step": 8257
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 2.6267857142857143e-05,
+      "loss": 1.0041,
+      "step": 8258
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 2.625892857142857e-05,
+      "loss": 0.9052,
+      "step": 8259
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 2.625e-05,
+      "loss": 1.022,
+      "step": 8260
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 2.624107142857143e-05,
+      "loss": 0.9944,
+      "step": 8261
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 2.6232142857142862e-05,
+      "loss": 0.8941,
+      "step": 8262
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 2.6223214285714288e-05,
+      "loss": 0.9751,
+      "step": 8263
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 2.6214285714285713e-05,
+      "loss": 1.0193,
+      "step": 8264
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 2.6205357142857145e-05,
+      "loss": 0.9491,
+      "step": 8265
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 2.619642857142857e-05,
+      "loss": 0.9248,
+      "step": 8266
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 2.6187500000000003e-05,
+      "loss": 0.9935,
+      "step": 8267
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 2.617857142857143e-05,
+      "loss": 0.9733,
+      "step": 8268
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 2.6169642857142857e-05,
+      "loss": 0.9783,
+      "step": 8269
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 2.616071428571429e-05,
+      "loss": 0.994,
+      "step": 8270
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 2.6151785714285715e-05,
+      "loss": 0.8516,
+      "step": 8271
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 2.6142857142857147e-05,
+      "loss": 1.0298,
+      "step": 8272
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 2.6133928571428573e-05,
+      "loss": 0.9686,
+      "step": 8273
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 2.6124999999999998e-05,
+      "loss": 1.0065,
+      "step": 8274
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 2.611607142857143e-05,
+      "loss": 0.979,
+      "step": 8275
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 2.610714285714286e-05,
+      "loss": 0.9983,
+      "step": 8276
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 2.609821428571429e-05,
+      "loss": 0.9316,
+      "step": 8277
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 2.6089285714285717e-05,
+      "loss": 0.9357,
+      "step": 8278
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 2.6080357142857142e-05,
+      "loss": 0.9094,
+      "step": 8279
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 2.6071428571428574e-05,
+      "loss": 0.961,
+      "step": 8280
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 2.60625e-05,
+      "loss": 0.9536,
+      "step": 8281
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 2.605357142857143e-05,
+      "loss": 0.9097,
+      "step": 8282
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 2.6044642857142858e-05,
+      "loss": 1.0141,
+      "step": 8283
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 2.6035714285714286e-05,
+      "loss": 0.9931,
+      "step": 8284
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 2.602678571428572e-05,
+      "loss": 0.8934,
+      "step": 8285
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 2.6017857142857144e-05,
+      "loss": 0.9846,
+      "step": 8286
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 2.600892857142857e-05,
+      "loss": 1.0166,
+      "step": 8287
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.997,
+      "step": 8288
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 2.5991071428571427e-05,
+      "loss": 0.9552,
+      "step": 8289
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 2.598214285714286e-05,
+      "loss": 0.937,
+      "step": 8290
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 2.5973214285714288e-05,
+      "loss": 0.9467,
+      "step": 8291
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 2.5964285714285714e-05,
+      "loss": 0.9268,
+      "step": 8292
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 2.5955357142857146e-05,
+      "loss": 0.9421,
+      "step": 8293
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 2.594642857142857e-05,
+      "loss": 0.9592,
+      "step": 8294
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 2.5937500000000004e-05,
+      "loss": 0.9536,
+      "step": 8295
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 2.592857142857143e-05,
+      "loss": 0.9554,
+      "step": 8296
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 2.5919642857142855e-05,
+      "loss": 0.9602,
+      "step": 8297
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 2.5910714285714287e-05,
+      "loss": 0.962,
+      "step": 8298
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 2.5901785714285716e-05,
+      "loss": 0.9431,
+      "step": 8299
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 2.5892857142857148e-05,
+      "loss": 0.9828,
+      "step": 8300
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 2.5883928571428573e-05,
+      "loss": 0.9142,
+      "step": 8301
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 2.5875e-05,
+      "loss": 0.9588,
+      "step": 8302
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 2.586607142857143e-05,
+      "loss": 0.9192,
+      "step": 8303
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 2.5857142857142856e-05,
+      "loss": 0.9254,
+      "step": 8304
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 2.584821428571429e-05,
+      "loss": 0.9362,
+      "step": 8305
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 2.5839285714285717e-05,
+      "loss": 0.9214,
+      "step": 8306
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 2.5830357142857143e-05,
+      "loss": 0.9614,
+      "step": 8307
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 2.5821428571428575e-05,
+      "loss": 1.0142,
+      "step": 8308
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 2.58125e-05,
+      "loss": 0.9733,
+      "step": 8309
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 2.5803571428571433e-05,
+      "loss": 0.9459,
+      "step": 8310
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 2.5794642857142858e-05,
+      "loss": 0.9301,
+      "step": 8311
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 2.5785714285714284e-05,
+      "loss": 0.9912,
+      "step": 8312
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 2.5776785714285716e-05,
+      "loss": 1.0015,
+      "step": 8313
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 2.5767857142857145e-05,
+      "loss": 0.9821,
+      "step": 8314
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 2.575892857142857e-05,
+      "loss": 1.0045,
+      "step": 8315
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 0.9654,
+      "step": 8316
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 2.5741071428571428e-05,
+      "loss": 1.0107,
+      "step": 8317
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 2.573214285714286e-05,
+      "loss": 0.988,
+      "step": 8318
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 2.5723214285714286e-05,
+      "loss": 0.8914,
+      "step": 8319
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.8863,
+      "step": 8320
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 2.5705357142857143e-05,
+      "loss": 0.9588,
+      "step": 8321
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 2.5696428571428572e-05,
+      "loss": 0.8992,
+      "step": 8322
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 2.5687500000000004e-05,
+      "loss": 0.9411,
+      "step": 8323
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 2.567857142857143e-05,
+      "loss": 1.0064,
+      "step": 8324
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 2.5669642857142855e-05,
+      "loss": 0.9862,
+      "step": 8325
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 2.5660714285714287e-05,
+      "loss": 0.9344,
+      "step": 8326
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 2.5651785714285713e-05,
+      "loss": 1.0552,
+      "step": 8327
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 2.5642857142857145e-05,
+      "loss": 0.9242,
+      "step": 8328
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 2.5633928571428574e-05,
+      "loss": 1.0021,
+      "step": 8329
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 2.5625e-05,
+      "loss": 0.963,
+      "step": 8330
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 2.561607142857143e-05,
+      "loss": 0.9734,
+      "step": 8331
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 2.5607142857142857e-05,
+      "loss": 1.0332,
+      "step": 8332
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 2.559821428571429e-05,
+      "loss": 0.9274,
+      "step": 8333
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 2.5589285714285715e-05,
+      "loss": 0.9248,
+      "step": 8334
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 2.5580357142857144e-05,
+      "loss": 0.9719,
+      "step": 8335
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 2.5571428571428572e-05,
+      "loss": 1.0028,
+      "step": 8336
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 2.55625e-05,
+      "loss": 0.9189,
+      "step": 8337
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 2.5553571428571434e-05,
+      "loss": 0.9738,
+      "step": 8338
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 2.554464285714286e-05,
+      "loss": 0.9135,
+      "step": 8339
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 2.5535714285714284e-05,
+      "loss": 0.929,
+      "step": 8340
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 2.5526785714285717e-05,
+      "loss": 0.9526,
+      "step": 8341
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 2.5517857142857142e-05,
+      "loss": 1.0276,
+      "step": 8342
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 2.5508928571428574e-05,
+      "loss": 0.9294,
+      "step": 8343
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.9995,
+      "step": 8344
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 2.549107142857143e-05,
+      "loss": 0.9626,
+      "step": 8345
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 2.548214285714286e-05,
+      "loss": 1.0808,
+      "step": 8346
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 2.5473214285714286e-05,
+      "loss": 0.9489,
+      "step": 8347
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 2.5464285714285712e-05,
+      "loss": 1.0389,
+      "step": 8348
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 2.5455357142857144e-05,
+      "loss": 0.9473,
+      "step": 8349
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 2.544642857142857e-05,
+      "loss": 1.0083,
+      "step": 8350
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 2.54375e-05,
+      "loss": 0.9236,
+      "step": 8351
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 2.542857142857143e-05,
+      "loss": 1.0097,
+      "step": 8352
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 2.5419642857142856e-05,
+      "loss": 1.0716,
+      "step": 8353
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 2.5410714285714288e-05,
+      "loss": 0.9378,
+      "step": 8354
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 2.5401785714285714e-05,
+      "loss": 1.0082,
+      "step": 8355
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 2.5392857142857146e-05,
+      "loss": 0.9011,
+      "step": 8356
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 2.538392857142857e-05,
+      "loss": 0.9676,
+      "step": 8357
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 2.5375e-05,
+      "loss": 0.962,
+      "step": 8358
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 2.5366071428571432e-05,
+      "loss": 0.9636,
+      "step": 8359
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 2.5357142857142858e-05,
+      "loss": 0.9539,
+      "step": 8360
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 2.534821428571429e-05,
+      "loss": 0.9143,
+      "step": 8361
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 2.5339285714285716e-05,
+      "loss": 0.992,
+      "step": 8362
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 2.533035714285714e-05,
+      "loss": 1.0548,
+      "step": 8363
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 2.5321428571428573e-05,
+      "loss": 0.9296,
+      "step": 8364
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 2.53125e-05,
+      "loss": 0.935,
+      "step": 8365
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 2.530357142857143e-05,
+      "loss": 0.9316,
+      "step": 8366
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 2.529464285714286e-05,
+      "loss": 0.9213,
+      "step": 8367
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 2.5285714285714285e-05,
+      "loss": 1.0483,
+      "step": 8368
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 2.5276785714285717e-05,
+      "loss": 0.9508,
+      "step": 8369
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 2.5267857142857143e-05,
+      "loss": 0.9638,
+      "step": 8370
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 2.5258928571428575e-05,
+      "loss": 0.875,
+      "step": 8371
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 2.525e-05,
+      "loss": 0.9647,
+      "step": 8372
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 2.524107142857143e-05,
+      "loss": 0.9472,
+      "step": 8373
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 2.5232142857142858e-05,
+      "loss": 0.9996,
+      "step": 8374
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 2.5223214285714287e-05,
+      "loss": 0.9293,
+      "step": 8375
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 2.521428571428572e-05,
+      "loss": 0.9821,
+      "step": 8376
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 2.5205357142857145e-05,
+      "loss": 0.9411,
+      "step": 8377
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 2.519642857142857e-05,
+      "loss": 0.9667,
+      "step": 8378
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 2.5187500000000002e-05,
+      "loss": 0.9444,
+      "step": 8379
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 2.5178571428571428e-05,
+      "loss": 0.9253,
+      "step": 8380
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 2.516964285714286e-05,
+      "loss": 0.9904,
+      "step": 8381
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 2.516071428571429e-05,
+      "loss": 0.9342,
+      "step": 8382
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 2.5151785714285714e-05,
+      "loss": 1.0334,
+      "step": 8383
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 2.5142857142857147e-05,
+      "loss": 0.9658,
+      "step": 8384
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 2.5133928571428572e-05,
+      "loss": 1.0035,
+      "step": 8385
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 2.5124999999999997e-05,
+      "loss": 0.9827,
+      "step": 8386
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 2.511607142857143e-05,
+      "loss": 0.9861,
+      "step": 8387
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 2.510714285714286e-05,
+      "loss": 0.9722,
+      "step": 8388
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 2.5098214285714287e-05,
+      "loss": 0.9136,
+      "step": 8389
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 2.5089285714285716e-05,
+      "loss": 0.8973,
+      "step": 8390
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 2.508035714285714e-05,
+      "loss": 0.8769,
+      "step": 8391
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 2.5071428571428574e-05,
+      "loss": 1.0324,
+      "step": 8392
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 2.50625e-05,
+      "loss": 0.9373,
+      "step": 8393
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 2.505357142857143e-05,
+      "loss": 0.9871,
+      "step": 8394
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 2.5044642857142857e-05,
+      "loss": 0.947,
+      "step": 8395
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 2.5035714285714286e-05,
+      "loss": 0.9548,
+      "step": 8396
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 2.5026785714285718e-05,
+      "loss": 0.9696,
+      "step": 8397
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 2.5017857142857144e-05,
+      "loss": 0.8809,
+      "step": 8398
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 2.5008928571428576e-05,
+      "loss": 0.9599,
+      "step": 8399
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9546,
+      "step": 8400
+    },
+    {
+      "epoch": 37.5,
+      "eval_loss": 1.113389492034912,
+      "eval_runtime": 44.4504,
+      "eval_samples_per_second": 294.576,
+      "eval_steps_per_second": 1.552,
+      "step": 8400
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 2.499107142857143e-05,
+      "loss": 0.8675,
+      "step": 8401
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 2.498214285714286e-05,
+      "loss": 0.9921,
+      "step": 8402
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 2.4973214285714284e-05,
+      "loss": 0.972,
+      "step": 8403
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 2.4964285714285717e-05,
+      "loss": 0.9206,
+      "step": 8404
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 2.4955357142857145e-05,
+      "loss": 1.0239,
+      "step": 8405
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 2.4946428571428574e-05,
+      "loss": 0.9651,
+      "step": 8406
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 2.4937500000000003e-05,
+      "loss": 0.9995,
+      "step": 8407
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 2.492857142857143e-05,
+      "loss": 0.9433,
+      "step": 8408
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 2.4919642857142857e-05,
+      "loss": 0.9713,
+      "step": 8409
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 2.4910714285714286e-05,
+      "loss": 0.9267,
+      "step": 8410
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 2.4901785714285715e-05,
+      "loss": 0.9966,
+      "step": 8411
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 2.4892857142857144e-05,
+      "loss": 0.9488,
+      "step": 8412
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 2.4883928571428573e-05,
+      "loss": 0.8988,
+      "step": 8413
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 2.4875e-05,
+      "loss": 0.947,
+      "step": 8414
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 2.486607142857143e-05,
+      "loss": 0.9527,
+      "step": 8415
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 2.485714285714286e-05,
+      "loss": 0.9685,
+      "step": 8416
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 2.4848214285714285e-05,
+      "loss": 0.9114,
+      "step": 8417
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 2.4839285714285714e-05,
+      "loss": 0.9816,
+      "step": 8418
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 2.4830357142857146e-05,
+      "loss": 0.9816,
+      "step": 8419
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 2.4821428571428575e-05,
+      "loss": 1.0463,
+      "step": 8420
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 2.4812500000000003e-05,
+      "loss": 0.9662,
+      "step": 8421
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 2.480357142857143e-05,
+      "loss": 0.9839,
+      "step": 8422
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 2.4794642857142858e-05,
+      "loss": 0.9305,
+      "step": 8423
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 2.4785714285714287e-05,
+      "loss": 0.9487,
+      "step": 8424
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 2.4776785714285715e-05,
+      "loss": 0.936,
+      "step": 8425
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 2.4767857142857144e-05,
+      "loss": 0.9383,
+      "step": 8426
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 2.4758928571428573e-05,
+      "loss": 0.979,
+      "step": 8427
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 0.9805,
+      "step": 8428
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 2.474107142857143e-05,
+      "loss": 1.0166,
+      "step": 8429
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 2.473214285714286e-05,
+      "loss": 0.9242,
+      "step": 8430
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 2.4723214285714285e-05,
+      "loss": 0.9398,
+      "step": 8431
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 2.4714285714285714e-05,
+      "loss": 0.97,
+      "step": 8432
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 2.4705357142857143e-05,
+      "loss": 0.9226,
+      "step": 8433
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 2.469642857142857e-05,
+      "loss": 0.8932,
+      "step": 8434
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 2.4687500000000004e-05,
+      "loss": 0.9916,
+      "step": 8435
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 2.467857142857143e-05,
+      "loss": 0.9872,
+      "step": 8436
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 2.4669642857142858e-05,
+      "loss": 0.9299,
+      "step": 8437
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 2.4660714285714287e-05,
+      "loss": 1.0551,
+      "step": 8438
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 2.4651785714285716e-05,
+      "loss": 1.0623,
+      "step": 8439
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 2.4642857142857145e-05,
+      "loss": 1.0207,
+      "step": 8440
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 2.463392857142857e-05,
+      "loss": 0.9714,
+      "step": 8441
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 2.4625000000000002e-05,
+      "loss": 0.9895,
+      "step": 8442
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 2.461607142857143e-05,
+      "loss": 0.9949,
+      "step": 8443
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 2.460714285714286e-05,
+      "loss": 0.9877,
+      "step": 8444
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 2.4598214285714285e-05,
+      "loss": 0.9694,
+      "step": 8445
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 2.4589285714285714e-05,
+      "loss": 1.045,
+      "step": 8446
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 2.4580357142857143e-05,
+      "loss": 0.9372,
+      "step": 8447
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 2.4571428571428572e-05,
+      "loss": 1.0089,
+      "step": 8448
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 2.45625e-05,
+      "loss": 0.9803,
+      "step": 8449
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 2.455357142857143e-05,
+      "loss": 0.9285,
+      "step": 8450
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 2.454464285714286e-05,
+      "loss": 1.0595,
+      "step": 8451
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 2.4535714285714287e-05,
+      "loss": 0.9719,
+      "step": 8452
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 2.4526785714285716e-05,
+      "loss": 1.0186,
+      "step": 8453
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 2.4517857142857145e-05,
+      "loss": 0.978,
+      "step": 8454
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 2.450892857142857e-05,
+      "loss": 0.9199,
+      "step": 8455
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 2.45e-05,
+      "loss": 0.9304,
+      "step": 8456
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 2.449107142857143e-05,
+      "loss": 0.9485,
+      "step": 8457
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 2.448214285714286e-05,
+      "loss": 1.0108,
+      "step": 8458
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 2.447321428571429e-05,
+      "loss": 0.9847,
+      "step": 8459
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 2.4464285714285715e-05,
+      "loss": 0.993,
+      "step": 8460
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 2.4455357142857143e-05,
+      "loss": 0.954,
+      "step": 8461
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 2.4446428571428572e-05,
+      "loss": 0.9486,
+      "step": 8462
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 2.44375e-05,
+      "loss": 0.9834,
+      "step": 8463
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.442857142857143e-05,
+      "loss": 1.005,
+      "step": 8464
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.441964285714286e-05,
+      "loss": 0.9697,
+      "step": 8465
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.4410714285714288e-05,
+      "loss": 0.9593,
+      "step": 8466
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 2.4401785714285716e-05,
+      "loss": 0.9346,
+      "step": 8467
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 2.4392857142857145e-05,
+      "loss": 0.8787,
+      "step": 8468
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 2.438392857142857e-05,
+      "loss": 0.9942,
+      "step": 8469
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 2.4375e-05,
+      "loss": 0.9025,
+      "step": 8470
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 2.436607142857143e-05,
+      "loss": 0.9479,
+      "step": 8471
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 2.4357142857142857e-05,
+      "loss": 0.9696,
+      "step": 8472
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 2.434821428571429e-05,
+      "loss": 0.8909,
+      "step": 8473
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 2.4339285714285715e-05,
+      "loss": 0.9261,
+      "step": 8474
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 2.4330357142857144e-05,
+      "loss": 1.017,
+      "step": 8475
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 2.4321428571428573e-05,
+      "loss": 0.9355,
+      "step": 8476
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 2.43125e-05,
+      "loss": 0.9677,
+      "step": 8477
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 2.430357142857143e-05,
+      "loss": 0.9026,
+      "step": 8478
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 2.429464285714286e-05,
+      "loss": 0.977,
+      "step": 8479
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 0.965,
+      "step": 8480
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 2.4276785714285717e-05,
+      "loss": 0.9816,
+      "step": 8481
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 2.4267857142857146e-05,
+      "loss": 0.9803,
+      "step": 8482
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 2.425892857142857e-05,
+      "loss": 0.9746,
+      "step": 8483
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 2.425e-05,
+      "loss": 0.9995,
+      "step": 8484
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 2.424107142857143e-05,
+      "loss": 1.0,
+      "step": 8485
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 2.4232142857142858e-05,
+      "loss": 0.9764,
+      "step": 8486
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 2.4223214285714286e-05,
+      "loss": 0.9992,
+      "step": 8487
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 2.4214285714285715e-05,
+      "loss": 0.8871,
+      "step": 8488
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 2.4205357142857144e-05,
+      "loss": 1.0373,
+      "step": 8489
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 2.4196428571428573e-05,
+      "loss": 0.9558,
+      "step": 8490
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 2.4187500000000002e-05,
+      "loss": 1.0746,
+      "step": 8491
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 2.417857142857143e-05,
+      "loss": 0.9168,
+      "step": 8492
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 2.4169642857142856e-05,
+      "loss": 0.9768,
+      "step": 8493
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 2.4160714285714285e-05,
+      "loss": 0.9697,
+      "step": 8494
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 2.4151785714285717e-05,
+      "loss": 0.9904,
+      "step": 8495
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 2.4142857142857146e-05,
+      "loss": 0.96,
+      "step": 8496
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 2.413392857142857e-05,
+      "loss": 0.965,
+      "step": 8497
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 2.4125e-05,
+      "loss": 0.9894,
+      "step": 8498
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 2.411607142857143e-05,
+      "loss": 0.9921,
+      "step": 8499
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 2.4107142857142858e-05,
+      "loss": 0.9303,
+      "step": 8500
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 2.4098214285714287e-05,
+      "loss": 0.9824,
+      "step": 8501
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 2.4089285714285716e-05,
+      "loss": 0.9813,
+      "step": 8502
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 2.4080357142857144e-05,
+      "loss": 0.915,
+      "step": 8503
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 2.4071428571428573e-05,
+      "loss": 1.0256,
+      "step": 8504
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 2.4062500000000002e-05,
+      "loss": 0.9998,
+      "step": 8505
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 2.405357142857143e-05,
+      "loss": 0.9591,
+      "step": 8506
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 2.4044642857142856e-05,
+      "loss": 1.0041,
+      "step": 8507
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 2.4035714285714285e-05,
+      "loss": 1.0395,
+      "step": 8508
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 2.4026785714285714e-05,
+      "loss": 0.9547,
+      "step": 8509
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 2.4017857142857146e-05,
+      "loss": 0.9393,
+      "step": 8510
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 2.4008928571428575e-05,
+      "loss": 0.9527,
+      "step": 8511
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.0017,
+      "step": 8512
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 2.399107142857143e-05,
+      "loss": 0.8918,
+      "step": 8513
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 2.3982142857142858e-05,
+      "loss": 0.9995,
+      "step": 8514
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 2.3973214285714287e-05,
+      "loss": 0.9155,
+      "step": 8515
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 2.3964285714285713e-05,
+      "loss": 0.9196,
+      "step": 8516
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 2.3955357142857145e-05,
+      "loss": 0.982,
+      "step": 8517
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 2.3946428571428574e-05,
+      "loss": 0.9321,
+      "step": 8518
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 2.3937500000000002e-05,
+      "loss": 0.9951,
+      "step": 8519
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 2.392857142857143e-05,
+      "loss": 0.9677,
+      "step": 8520
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 2.3919642857142857e-05,
+      "loss": 1.0172,
+      "step": 8521
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 2.3910714285714286e-05,
+      "loss": 1.0407,
+      "step": 8522
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 2.3901785714285714e-05,
+      "loss": 0.9218,
+      "step": 8523
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 2.3892857142857143e-05,
+      "loss": 1.0661,
+      "step": 8524
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 2.3883928571428572e-05,
+      "loss": 0.9472,
+      "step": 8525
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 2.3875e-05,
+      "loss": 1.0005,
+      "step": 8526
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 2.386607142857143e-05,
+      "loss": 0.8777,
+      "step": 8527
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 2.385714285714286e-05,
+      "loss": 0.9701,
+      "step": 8528
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 2.3848214285714287e-05,
+      "loss": 0.9869,
+      "step": 8529
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 2.3839285714285716e-05,
+      "loss": 0.9511,
+      "step": 8530
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 2.3830357142857142e-05,
+      "loss": 0.9713,
+      "step": 8531
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 2.3821428571428574e-05,
+      "loss": 0.9701,
+      "step": 8532
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 2.3812500000000003e-05,
+      "loss": 0.9957,
+      "step": 8533
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 2.380357142857143e-05,
+      "loss": 0.9214,
+      "step": 8534
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 2.3794642857142857e-05,
+      "loss": 0.9816,
+      "step": 8535
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 2.3785714285714286e-05,
+      "loss": 0.9023,
+      "step": 8536
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 2.3776785714285715e-05,
+      "loss": 1.003,
+      "step": 8537
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 2.3767857142857144e-05,
+      "loss": 0.9157,
+      "step": 8538
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 2.3758928571428572e-05,
+      "loss": 0.9817,
+      "step": 8539
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 2.375e-05,
+      "loss": 1.0229,
+      "step": 8540
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 2.374107142857143e-05,
+      "loss": 0.9489,
+      "step": 8541
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 2.373214285714286e-05,
+      "loss": 0.9408,
+      "step": 8542
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 2.3723214285714288e-05,
+      "loss": 0.8975,
+      "step": 8543
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 2.3714285714285717e-05,
+      "loss": 0.9565,
+      "step": 8544
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 2.3705357142857142e-05,
+      "loss": 0.9538,
+      "step": 8545
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 2.369642857142857e-05,
+      "loss": 0.9448,
+      "step": 8546
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 2.36875e-05,
+      "loss": 0.9681,
+      "step": 8547
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 2.3678571428571432e-05,
+      "loss": 0.9788,
+      "step": 8548
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 2.3669642857142857e-05,
+      "loss": 0.9871,
+      "step": 8549
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 2.3660714285714286e-05,
+      "loss": 0.9826,
+      "step": 8550
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 2.3651785714285715e-05,
+      "loss": 0.9224,
+      "step": 8551
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 2.3642857142857144e-05,
+      "loss": 0.8751,
+      "step": 8552
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 2.3633928571428573e-05,
+      "loss": 0.9461,
+      "step": 8553
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 2.3624999999999998e-05,
+      "loss": 1.0164,
+      "step": 8554
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 2.361607142857143e-05,
+      "loss": 0.9393,
+      "step": 8555
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 2.360714285714286e-05,
+      "loss": 0.9495,
+      "step": 8556
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 2.3598214285714288e-05,
+      "loss": 1.0089,
+      "step": 8557
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 2.3589285714285717e-05,
+      "loss": 1.0048,
+      "step": 8558
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 2.3580357142857142e-05,
+      "loss": 0.9672,
+      "step": 8559
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 1.0029,
+      "step": 8560
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 2.35625e-05,
+      "loss": 0.9952,
+      "step": 8561
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 2.355357142857143e-05,
+      "loss": 0.9475,
+      "step": 8562
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 2.354464285714286e-05,
+      "loss": 0.9968,
+      "step": 8563
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 2.3535714285714287e-05,
+      "loss": 0.9344,
+      "step": 8564
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 2.3526785714285715e-05,
+      "loss": 0.9988,
+      "step": 8565
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 2.3517857142857144e-05,
+      "loss": 0.9498,
+      "step": 8566
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 2.3508928571428573e-05,
+      "loss": 0.9362,
+      "step": 8567
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 2.35e-05,
+      "loss": 0.9,
+      "step": 8568
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 2.3491071428571427e-05,
+      "loss": 0.9802,
+      "step": 8569
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 2.348214285714286e-05,
+      "loss": 0.9932,
+      "step": 8570
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 2.347321428571429e-05,
+      "loss": 0.9978,
+      "step": 8571
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 2.3464285714285717e-05,
+      "loss": 1.011,
+      "step": 8572
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 2.3455357142857143e-05,
+      "loss": 0.9874,
+      "step": 8573
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 2.344642857142857e-05,
+      "loss": 0.9937,
+      "step": 8574
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 2.34375e-05,
+      "loss": 1.0166,
+      "step": 8575
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 2.342857142857143e-05,
+      "loss": 0.9676,
+      "step": 8576
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 2.3419642857142858e-05,
+      "loss": 0.9173,
+      "step": 8577
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 2.3410714285714287e-05,
+      "loss": 0.9681,
+      "step": 8578
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 2.3401785714285716e-05,
+      "loss": 0.9441,
+      "step": 8579
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 2.3392857142857145e-05,
+      "loss": 0.9814,
+      "step": 8580
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 2.3383928571428574e-05,
+      "loss": 0.9695,
+      "step": 8581
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 2.3375000000000002e-05,
+      "loss": 0.9566,
+      "step": 8582
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 2.3366071428571428e-05,
+      "loss": 0.8889,
+      "step": 8583
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 2.3357142857142857e-05,
+      "loss": 1.0237,
+      "step": 8584
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 2.3348214285714285e-05,
+      "loss": 0.928,
+      "step": 8585
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 2.3339285714285718e-05,
+      "loss": 0.9996,
+      "step": 8586
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 2.3330357142857143e-05,
+      "loss": 0.9377,
+      "step": 8587
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 2.3321428571428572e-05,
+      "loss": 0.9009,
+      "step": 8588
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 2.33125e-05,
+      "loss": 1.0505,
+      "step": 8589
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 2.330357142857143e-05,
+      "loss": 0.9132,
+      "step": 8590
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 2.329464285714286e-05,
+      "loss": 0.9667,
+      "step": 8591
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 2.3285714285714287e-05,
+      "loss": 0.9867,
+      "step": 8592
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 2.3276785714285716e-05,
+      "loss": 1.0061,
+      "step": 8593
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 2.3267857142857145e-05,
+      "loss": 1.0219,
+      "step": 8594
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 2.3258928571428574e-05,
+      "loss": 0.9816,
+      "step": 8595
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 0.9001,
+      "step": 8596
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 2.3241071428571428e-05,
+      "loss": 0.9327,
+      "step": 8597
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 2.3232142857142857e-05,
+      "loss": 0.9764,
+      "step": 8598
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 2.3223214285714286e-05,
+      "loss": 0.9253,
+      "step": 8599
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.9567,
+      "step": 8600
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 2.3205357142857147e-05,
+      "loss": 1.0065,
+      "step": 8601
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 2.3196428571428572e-05,
+      "loss": 1.0148,
+      "step": 8602
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 2.31875e-05,
+      "loss": 0.943,
+      "step": 8603
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 2.317857142857143e-05,
+      "loss": 0.9746,
+      "step": 8604
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 2.316964285714286e-05,
+      "loss": 0.9298,
+      "step": 8605
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 2.3160714285714284e-05,
+      "loss": 0.973,
+      "step": 8606
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 2.3151785714285713e-05,
+      "loss": 1.0017,
+      "step": 8607
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 2.3142857142857145e-05,
+      "loss": 0.9077,
+      "step": 8608
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 2.3133928571428574e-05,
+      "loss": 0.9,
+      "step": 8609
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 2.3125000000000003e-05,
+      "loss": 0.927,
+      "step": 8610
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 2.311607142857143e-05,
+      "loss": 0.8826,
+      "step": 8611
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 2.3107142857142857e-05,
+      "loss": 0.9313,
+      "step": 8612
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 2.3098214285714286e-05,
+      "loss": 0.9866,
+      "step": 8613
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 2.3089285714285715e-05,
+      "loss": 0.9742,
+      "step": 8614
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 2.3080357142857144e-05,
+      "loss": 1.0279,
+      "step": 8615
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 2.3071428571428573e-05,
+      "loss": 0.9965,
+      "step": 8616
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 2.30625e-05,
+      "loss": 0.9354,
+      "step": 8617
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 2.305357142857143e-05,
+      "loss": 0.9833,
+      "step": 8618
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 2.304464285714286e-05,
+      "loss": 1.0043,
+      "step": 8619
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 2.3035714285714285e-05,
+      "loss": 0.9127,
+      "step": 8620
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 2.3026785714285714e-05,
+      "loss": 0.8872,
+      "step": 8621
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 2.3017857142857142e-05,
+      "loss": 0.9874,
+      "step": 8622
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 2.3008928571428575e-05,
+      "loss": 0.9787,
+      "step": 8623
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.0171,
+      "step": 8624
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 2.299107142857143e-05,
+      "loss": 0.9161,
+      "step": 8625
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 2.2982142857142858e-05,
+      "loss": 0.9564,
+      "step": 8626
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 2.2973214285714287e-05,
+      "loss": 0.9327,
+      "step": 8627
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 2.2964285714285715e-05,
+      "loss": 0.9286,
+      "step": 8628
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 2.2955357142857144e-05,
+      "loss": 0.9696,
+      "step": 8629
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 2.2946428571428573e-05,
+      "loss": 0.9765,
+      "step": 8630
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 2.2937500000000002e-05,
+      "loss": 0.968,
+      "step": 8631
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 2.292857142857143e-05,
+      "loss": 0.9404,
+      "step": 8632
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 2.291964285714286e-05,
+      "loss": 0.9787,
+      "step": 8633
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 2.291071428571429e-05,
+      "loss": 0.9773,
+      "step": 8634
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 2.2901785714285714e-05,
+      "loss": 0.9375,
+      "step": 8635
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 2.2892857142857143e-05,
+      "loss": 0.9565,
+      "step": 8636
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 2.288392857142857e-05,
+      "loss": 0.912,
+      "step": 8637
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 2.2875e-05,
+      "loss": 1.0047,
+      "step": 8638
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 2.286607142857143e-05,
+      "loss": 0.9505,
+      "step": 8639
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.9338,
+      "step": 8640
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 2.2848214285714287e-05,
+      "loss": 0.8671,
+      "step": 8641
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 2.2839285714285716e-05,
+      "loss": 0.9612,
+      "step": 8642
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 2.2830357142857145e-05,
+      "loss": 0.9553,
+      "step": 8643
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 2.282142857142857e-05,
+      "loss": 0.9241,
+      "step": 8644
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 2.28125e-05,
+      "loss": 0.9186,
+      "step": 8645
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 2.280357142857143e-05,
+      "loss": 1.0005,
+      "step": 8646
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 2.279464285714286e-05,
+      "loss": 0.9707,
+      "step": 8647
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 2.278571428571429e-05,
+      "loss": 0.9382,
+      "step": 8648
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 2.2776785714285714e-05,
+      "loss": 0.9195,
+      "step": 8649
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 2.2767857142857143e-05,
+      "loss": 1.0253,
+      "step": 8650
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 2.2758928571428572e-05,
+      "loss": 1.0197,
+      "step": 8651
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 2.275e-05,
+      "loss": 0.9524,
+      "step": 8652
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 2.274107142857143e-05,
+      "loss": 0.9771,
+      "step": 8653
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 2.273214285714286e-05,
+      "loss": 0.9449,
+      "step": 8654
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 2.2723214285714287e-05,
+      "loss": 0.9273,
+      "step": 8655
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 2.2714285714285716e-05,
+      "loss": 0.95,
+      "step": 8656
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 2.2705357142857145e-05,
+      "loss": 0.9202,
+      "step": 8657
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 2.269642857142857e-05,
+      "loss": 1.0053,
+      "step": 8658
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 2.26875e-05,
+      "loss": 0.928,
+      "step": 8659
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 2.2678571428571428e-05,
+      "loss": 0.9995,
+      "step": 8660
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 2.266964285714286e-05,
+      "loss": 0.989,
+      "step": 8661
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 2.266071428571429e-05,
+      "loss": 0.9918,
+      "step": 8662
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 2.2651785714285715e-05,
+      "loss": 0.9435,
+      "step": 8663
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 2.2642857142857143e-05,
+      "loss": 0.9463,
+      "step": 8664
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 2.2633928571428572e-05,
+      "loss": 0.9242,
+      "step": 8665
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 2.2625e-05,
+      "loss": 0.9969,
+      "step": 8666
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 2.261607142857143e-05,
+      "loss": 0.97,
+      "step": 8667
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 2.260714285714286e-05,
+      "loss": 0.9662,
+      "step": 8668
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 2.2598214285714288e-05,
+      "loss": 1.0625,
+      "step": 8669
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 2.2589285714285716e-05,
+      "loss": 0.9161,
+      "step": 8670
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 2.2580357142857145e-05,
+      "loss": 0.995,
+      "step": 8671
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 2.257142857142857e-05,
+      "loss": 0.9766,
+      "step": 8672
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 2.25625e-05,
+      "loss": 0.9276,
+      "step": 8673
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 2.255357142857143e-05,
+      "loss": 0.9614,
+      "step": 8674
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 2.2544642857142857e-05,
+      "loss": 0.8915,
+      "step": 8675
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 2.253571428571429e-05,
+      "loss": 0.9846,
+      "step": 8676
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 2.2526785714285715e-05,
+      "loss": 0.9856,
+      "step": 8677
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 2.2517857142857144e-05,
+      "loss": 0.9701,
+      "step": 8678
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 2.2508928571428573e-05,
+      "loss": 0.9294,
+      "step": 8679
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 2.25e-05,
+      "loss": 0.9027,
+      "step": 8680
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 2.249107142857143e-05,
+      "loss": 0.9678,
+      "step": 8681
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 2.2482142857142856e-05,
+      "loss": 0.9813,
+      "step": 8682
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 2.2473214285714288e-05,
+      "loss": 0.9699,
+      "step": 8683
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 2.2464285714285717e-05,
+      "loss": 0.9338,
+      "step": 8684
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 2.2455357142857146e-05,
+      "loss": 0.997,
+      "step": 8685
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 2.2446428571428574e-05,
+      "loss": 0.9974,
+      "step": 8686
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 2.24375e-05,
+      "loss": 0.9237,
+      "step": 8687
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 2.242857142857143e-05,
+      "loss": 0.9767,
+      "step": 8688
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 2.2419642857142858e-05,
+      "loss": 0.9878,
+      "step": 8689
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 2.2410714285714286e-05,
+      "loss": 0.9805,
+      "step": 8690
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 2.2401785714285715e-05,
+      "loss": 0.9698,
+      "step": 8691
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 2.2392857142857144e-05,
+      "loss": 0.9317,
+      "step": 8692
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 2.2383928571428573e-05,
+      "loss": 1.0201,
+      "step": 8693
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 2.2375000000000002e-05,
+      "loss": 0.946,
+      "step": 8694
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 2.236607142857143e-05,
+      "loss": 0.9051,
+      "step": 8695
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 2.2357142857142856e-05,
+      "loss": 0.9129,
+      "step": 8696
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 2.2348214285714285e-05,
+      "loss": 0.9872,
+      "step": 8697
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 2.2339285714285714e-05,
+      "loss": 0.9409,
+      "step": 8698
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 2.2330357142857146e-05,
+      "loss": 0.9768,
+      "step": 8699
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 2.2321428571428575e-05,
+      "loss": 0.9618,
+      "step": 8700
+    },
+    {
+      "epoch": 38.84,
+      "eval_loss": 1.1104276180267334,
+      "eval_runtime": 44.5159,
+      "eval_samples_per_second": 294.142,
+      "eval_steps_per_second": 1.55,
+      "step": 8700
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 2.23125e-05,
+      "loss": 0.923,
+      "step": 8701
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 2.230357142857143e-05,
+      "loss": 0.986,
+      "step": 8702
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 2.2294642857142858e-05,
+      "loss": 0.93,
+      "step": 8703
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 2.2285714285714287e-05,
+      "loss": 0.9273,
+      "step": 8704
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 2.2276785714285716e-05,
+      "loss": 0.9821,
+      "step": 8705
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 2.2267857142857144e-05,
+      "loss": 0.9787,
+      "step": 8706
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 2.2258928571428573e-05,
+      "loss": 0.9465,
+      "step": 8707
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 0.9397,
+      "step": 8708
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 2.224107142857143e-05,
+      "loss": 0.9547,
+      "step": 8709
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 2.2232142857142856e-05,
+      "loss": 0.9758,
+      "step": 8710
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 2.2223214285714285e-05,
+      "loss": 0.9659,
+      "step": 8711
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 2.2214285714285714e-05,
+      "loss": 0.9271,
+      "step": 8712
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 2.2205357142857143e-05,
+      "loss": 0.9979,
+      "step": 8713
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 2.2196428571428575e-05,
+      "loss": 0.9383,
+      "step": 8714
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 2.21875e-05,
+      "loss": 0.979,
+      "step": 8715
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 2.217857142857143e-05,
+      "loss": 0.9369,
+      "step": 8716
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 2.2169642857142858e-05,
+      "loss": 0.9601,
+      "step": 8717
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 2.2160714285714287e-05,
+      "loss": 0.9033,
+      "step": 8718
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 2.2151785714285716e-05,
+      "loss": 0.9486,
+      "step": 8719
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 2.214285714285714e-05,
+      "loss": 0.9461,
+      "step": 8720
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 2.2133928571428574e-05,
+      "loss": 0.9501,
+      "step": 8721
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 2.2125000000000002e-05,
+      "loss": 1.0106,
+      "step": 8722
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 2.211607142857143e-05,
+      "loss": 0.9725,
+      "step": 8723
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 2.2107142857142857e-05,
+      "loss": 0.9457,
+      "step": 8724
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 2.2098214285714286e-05,
+      "loss": 0.9811,
+      "step": 8725
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 2.2089285714285714e-05,
+      "loss": 1.05,
+      "step": 8726
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 2.2080357142857143e-05,
+      "loss": 0.9108,
+      "step": 8727
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 2.2071428571428572e-05,
+      "loss": 0.9436,
+      "step": 8728
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 2.20625e-05,
+      "loss": 0.974,
+      "step": 8729
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 2.205357142857143e-05,
+      "loss": 0.9999,
+      "step": 8730
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 2.204464285714286e-05,
+      "loss": 0.9768,
+      "step": 8731
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 2.2035714285714287e-05,
+      "loss": 0.9247,
+      "step": 8732
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 2.2026785714285716e-05,
+      "loss": 0.949,
+      "step": 8733
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 2.2017857142857142e-05,
+      "loss": 0.9752,
+      "step": 8734
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 2.200892857142857e-05,
+      "loss": 0.9505,
+      "step": 8735
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.8794,
+      "step": 8736
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 2.199107142857143e-05,
+      "loss": 0.9696,
+      "step": 8737
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 2.198214285714286e-05,
+      "loss": 1.0153,
+      "step": 8738
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 2.1973214285714286e-05,
+      "loss": 1.0333,
+      "step": 8739
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 2.1964285714285715e-05,
+      "loss": 0.9239,
+      "step": 8740
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 2.1955357142857144e-05,
+      "loss": 0.9313,
+      "step": 8741
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 2.1946428571428572e-05,
+      "loss": 0.9415,
+      "step": 8742
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 2.19375e-05,
+      "loss": 0.9892,
+      "step": 8743
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 2.192857142857143e-05,
+      "loss": 0.8925,
+      "step": 8744
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 2.191964285714286e-05,
+      "loss": 0.9757,
+      "step": 8745
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 2.1910714285714288e-05,
+      "loss": 0.9436,
+      "step": 8746
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 2.1901785714285717e-05,
+      "loss": 0.9125,
+      "step": 8747
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 2.1892857142857142e-05,
+      "loss": 0.9693,
+      "step": 8748
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 2.188392857142857e-05,
+      "loss": 0.98,
+      "step": 8749
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.9556,
+      "step": 8750
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 2.186607142857143e-05,
+      "loss": 0.8972,
+      "step": 8751
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 2.185714285714286e-05,
+      "loss": 1.0106,
+      "step": 8752
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 2.1848214285714286e-05,
+      "loss": 0.9302,
+      "step": 8753
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 2.1839285714285715e-05,
+      "loss": 0.9793,
+      "step": 8754
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 2.1830357142857144e-05,
+      "loss": 0.9573,
+      "step": 8755
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 2.1821428571428573e-05,
+      "loss": 0.9812,
+      "step": 8756
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 2.18125e-05,
+      "loss": 0.8696,
+      "step": 8757
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 2.1803571428571427e-05,
+      "loss": 1.0192,
+      "step": 8758
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 2.179464285714286e-05,
+      "loss": 0.9621,
+      "step": 8759
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 2.1785714285714288e-05,
+      "loss": 1.0396,
+      "step": 8760
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 2.1776785714285717e-05,
+      "loss": 0.9687,
+      "step": 8761
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 2.1767857142857142e-05,
+      "loss": 0.9495,
+      "step": 8762
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 2.175892857142857e-05,
+      "loss": 0.9598,
+      "step": 8763
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 2.175e-05,
+      "loss": 0.9284,
+      "step": 8764
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 2.174107142857143e-05,
+      "loss": 0.9873,
+      "step": 8765
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 2.1732142857142858e-05,
+      "loss": 0.989,
+      "step": 8766
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 2.1723214285714287e-05,
+      "loss": 0.9901,
+      "step": 8767
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 2.1714285714285715e-05,
+      "loss": 0.962,
+      "step": 8768
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 2.1705357142857144e-05,
+      "loss": 0.8231,
+      "step": 8769
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 2.1696428571428573e-05,
+      "loss": 0.97,
+      "step": 8770
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 2.1687500000000002e-05,
+      "loss": 0.9893,
+      "step": 8771
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 2.1678571428571427e-05,
+      "loss": 0.8868,
+      "step": 8772
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 2.1669642857142856e-05,
+      "loss": 1.0428,
+      "step": 8773
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 2.166071428571429e-05,
+      "loss": 1.0124,
+      "step": 8774
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 2.1651785714285717e-05,
+      "loss": 0.9798,
+      "step": 8775
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 2.1642857142857146e-05,
+      "loss": 0.9067,
+      "step": 8776
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 2.163392857142857e-05,
+      "loss": 0.9296,
+      "step": 8777
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 2.1625e-05,
+      "loss": 0.9031,
+      "step": 8778
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 2.161607142857143e-05,
+      "loss": 0.954,
+      "step": 8779
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 2.1607142857142858e-05,
+      "loss": 0.9649,
+      "step": 8780
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 2.1598214285714287e-05,
+      "loss": 1.0408,
+      "step": 8781
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 2.1589285714285716e-05,
+      "loss": 0.9116,
+      "step": 8782
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 2.1580357142857145e-05,
+      "loss": 0.9199,
+      "step": 8783
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 2.1571428571428574e-05,
+      "loss": 0.9418,
+      "step": 8784
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 2.1562500000000002e-05,
+      "loss": 0.9616,
+      "step": 8785
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 2.1553571428571428e-05,
+      "loss": 0.9254,
+      "step": 8786
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 2.1544642857142857e-05,
+      "loss": 0.9722,
+      "step": 8787
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 2.1535714285714285e-05,
+      "loss": 1.0537,
+      "step": 8788
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 2.1526785714285714e-05,
+      "loss": 0.9355,
+      "step": 8789
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 2.1517857142857147e-05,
+      "loss": 0.9182,
+      "step": 8790
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 2.1508928571428572e-05,
+      "loss": 0.9617,
+      "step": 8791
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 2.15e-05,
+      "loss": 0.9232,
+      "step": 8792
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 2.149107142857143e-05,
+      "loss": 0.9422,
+      "step": 8793
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 2.148214285714286e-05,
+      "loss": 0.9136,
+      "step": 8794
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 2.1473214285714284e-05,
+      "loss": 0.9335,
+      "step": 8795
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 2.1464285714285716e-05,
+      "loss": 0.9298,
+      "step": 8796
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 2.1455357142857145e-05,
+      "loss": 1.0158,
+      "step": 8797
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 2.1446428571428574e-05,
+      "loss": 0.8799,
+      "step": 8798
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 2.1437500000000003e-05,
+      "loss": 1.0315,
+      "step": 8799
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.9627,
+      "step": 8800
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 2.1419642857142857e-05,
+      "loss": 0.8837,
+      "step": 8801
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 2.1410714285714286e-05,
+      "loss": 0.9574,
+      "step": 8802
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 2.1401785714285715e-05,
+      "loss": 0.9624,
+      "step": 8803
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 2.1392857142857143e-05,
+      "loss": 0.919,
+      "step": 8804
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 2.1383928571428572e-05,
+      "loss": 0.9569,
+      "step": 8805
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 2.1375e-05,
+      "loss": 1.0085,
+      "step": 8806
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 2.136607142857143e-05,
+      "loss": 0.953,
+      "step": 8807
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 2.135714285714286e-05,
+      "loss": 0.9749,
+      "step": 8808
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 2.1348214285714288e-05,
+      "loss": 0.9111,
+      "step": 8809
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 2.1339285714285713e-05,
+      "loss": 0.9768,
+      "step": 8810
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 2.1330357142857142e-05,
+      "loss": 0.8728,
+      "step": 8811
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 2.1321428571428574e-05,
+      "loss": 1.0076,
+      "step": 8812
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 2.1312500000000003e-05,
+      "loss": 0.9742,
+      "step": 8813
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 2.130357142857143e-05,
+      "loss": 0.913,
+      "step": 8814
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 2.1294642857142857e-05,
+      "loss": 0.9422,
+      "step": 8815
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 2.1285714285714286e-05,
+      "loss": 0.9334,
+      "step": 8816
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 2.1276785714285715e-05,
+      "loss": 0.952,
+      "step": 8817
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 2.1267857142857144e-05,
+      "loss": 0.9547,
+      "step": 8818
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 2.1258928571428573e-05,
+      "loss": 1.0192,
+      "step": 8819
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 2.125e-05,
+      "loss": 1.0146,
+      "step": 8820
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 2.124107142857143e-05,
+      "loss": 0.9607,
+      "step": 8821
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 2.123214285714286e-05,
+      "loss": 0.986,
+      "step": 8822
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 2.1223214285714288e-05,
+      "loss": 0.9091,
+      "step": 8823
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 2.1214285714285713e-05,
+      "loss": 1.0016,
+      "step": 8824
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 2.1205357142857142e-05,
+      "loss": 1.0003,
+      "step": 8825
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 2.119642857142857e-05,
+      "loss": 0.9857,
+      "step": 8826
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 2.1187500000000003e-05,
+      "loss": 0.9236,
+      "step": 8827
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 2.1178571428571432e-05,
+      "loss": 0.9453,
+      "step": 8828
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 2.1169642857142858e-05,
+      "loss": 0.9889,
+      "step": 8829
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 2.1160714285714287e-05,
+      "loss": 0.9389,
+      "step": 8830
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 2.1151785714285715e-05,
+      "loss": 0.972,
+      "step": 8831
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 2.1142857142857144e-05,
+      "loss": 0.9356,
+      "step": 8832
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 2.113392857142857e-05,
+      "loss": 0.9734,
+      "step": 8833
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 2.1125000000000002e-05,
+      "loss": 0.9256,
+      "step": 8834
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 2.111607142857143e-05,
+      "loss": 1.0003,
+      "step": 8835
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 2.110714285714286e-05,
+      "loss": 0.9604,
+      "step": 8836
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 2.109821428571429e-05,
+      "loss": 0.9856,
+      "step": 8837
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 2.1089285714285714e-05,
+      "loss": 0.9959,
+      "step": 8838
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 2.1080357142857143e-05,
+      "loss": 0.9651,
+      "step": 8839
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 2.107142857142857e-05,
+      "loss": 0.9569,
+      "step": 8840
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 2.10625e-05,
+      "loss": 0.8481,
+      "step": 8841
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 2.105357142857143e-05,
+      "loss": 0.964,
+      "step": 8842
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 2.1044642857142858e-05,
+      "loss": 1.0099,
+      "step": 8843
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 2.1035714285714287e-05,
+      "loss": 0.9281,
+      "step": 8844
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 2.1026785714285716e-05,
+      "loss": 0.9679,
+      "step": 8845
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 2.1017857142857145e-05,
+      "loss": 0.9851,
+      "step": 8846
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 2.100892857142857e-05,
+      "loss": 0.9932,
+      "step": 8847
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 2.1e-05,
+      "loss": 0.9596,
+      "step": 8848
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 2.099107142857143e-05,
+      "loss": 0.9306,
+      "step": 8849
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 2.098214285714286e-05,
+      "loss": 0.9994,
+      "step": 8850
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 2.097321428571429e-05,
+      "loss": 0.8918,
+      "step": 8851
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 2.0964285714285714e-05,
+      "loss": 0.964,
+      "step": 8852
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 2.0955357142857143e-05,
+      "loss": 0.9574,
+      "step": 8853
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 2.0946428571428572e-05,
+      "loss": 1.0064,
+      "step": 8854
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 2.09375e-05,
+      "loss": 0.9498,
+      "step": 8855
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 2.092857142857143e-05,
+      "loss": 0.983,
+      "step": 8856
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 2.091964285714286e-05,
+      "loss": 0.9407,
+      "step": 8857
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 2.0910714285714287e-05,
+      "loss": 0.8775,
+      "step": 8858
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 2.0901785714285716e-05,
+      "loss": 0.9655,
+      "step": 8859
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 2.0892857142857145e-05,
+      "loss": 0.9898,
+      "step": 8860
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 2.0883928571428574e-05,
+      "loss": 1.0126,
+      "step": 8861
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 2.0875e-05,
+      "loss": 0.9749,
+      "step": 8862
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 2.0866071428571428e-05,
+      "loss": 0.9623,
+      "step": 8863
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 2.0857142857142857e-05,
+      "loss": 0.9396,
+      "step": 8864
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 2.084821428571429e-05,
+      "loss": 0.9261,
+      "step": 8865
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 2.0839285714285715e-05,
+      "loss": 0.9971,
+      "step": 8866
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 2.0830357142857143e-05,
+      "loss": 0.9812,
+      "step": 8867
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 2.0821428571428572e-05,
+      "loss": 0.9172,
+      "step": 8868
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 2.08125e-05,
+      "loss": 0.9046,
+      "step": 8869
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 2.080357142857143e-05,
+      "loss": 0.9723,
+      "step": 8870
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 2.0794642857142855e-05,
+      "loss": 0.9234,
+      "step": 8871
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 2.0785714285714288e-05,
+      "loss": 0.9782,
+      "step": 8872
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 2.0776785714285716e-05,
+      "loss": 0.9655,
+      "step": 8873
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 2.0767857142857145e-05,
+      "loss": 0.956,
+      "step": 8874
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 2.0758928571428574e-05,
+      "loss": 0.929,
+      "step": 8875
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 2.075e-05,
+      "loss": 0.9691,
+      "step": 8876
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 2.074107142857143e-05,
+      "loss": 0.9325,
+      "step": 8877
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 2.0732142857142857e-05,
+      "loss": 0.9984,
+      "step": 8878
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 2.0723214285714286e-05,
+      "loss": 0.8993,
+      "step": 8879
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 2.0714285714285718e-05,
+      "loss": 0.9254,
+      "step": 8880
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 2.0705357142857144e-05,
+      "loss": 0.9859,
+      "step": 8881
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 2.0696428571428573e-05,
+      "loss": 0.9179,
+      "step": 8882
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 2.06875e-05,
+      "loss": 1.0214,
+      "step": 8883
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 2.067857142857143e-05,
+      "loss": 0.9288,
+      "step": 8884
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 2.0669642857142856e-05,
+      "loss": 0.8933,
+      "step": 8885
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 2.0660714285714285e-05,
+      "loss": 0.9289,
+      "step": 8886
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 2.0651785714285717e-05,
+      "loss": 0.9101,
+      "step": 8887
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 2.0642857142857146e-05,
+      "loss": 0.989,
+      "step": 8888
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 2.0633928571428574e-05,
+      "loss": 0.9425,
+      "step": 8889
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 2.0625e-05,
+      "loss": 0.9623,
+      "step": 8890
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 2.061607142857143e-05,
+      "loss": 0.9327,
+      "step": 8891
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 2.0607142857142858e-05,
+      "loss": 0.9269,
+      "step": 8892
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 2.0598214285714286e-05,
+      "loss": 0.9308,
+      "step": 8893
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 2.0589285714285715e-05,
+      "loss": 0.9457,
+      "step": 8894
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 2.0580357142857144e-05,
+      "loss": 1.0101,
+      "step": 8895
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 2.0571428571428573e-05,
+      "loss": 0.9608,
+      "step": 8896
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 2.0562500000000002e-05,
+      "loss": 0.9775,
+      "step": 8897
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 2.055357142857143e-05,
+      "loss": 0.9869,
+      "step": 8898
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 2.0544642857142856e-05,
+      "loss": 0.9794,
+      "step": 8899
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 2.0535714285714285e-05,
+      "loss": 1.0098,
+      "step": 8900
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 2.0526785714285714e-05,
+      "loss": 0.9938,
+      "step": 8901
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 2.0517857142857143e-05,
+      "loss": 0.9609,
+      "step": 8902
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 2.0508928571428575e-05,
+      "loss": 1.0089,
+      "step": 8903
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 2.05e-05,
+      "loss": 1.0822,
+      "step": 8904
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 2.049107142857143e-05,
+      "loss": 1.0192,
+      "step": 8905
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 2.0482142857142858e-05,
+      "loss": 0.9915,
+      "step": 8906
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 2.0473214285714287e-05,
+      "loss": 0.8953,
+      "step": 8907
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 2.0464285714285716e-05,
+      "loss": 1.0001,
+      "step": 8908
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 2.0455357142857144e-05,
+      "loss": 0.9233,
+      "step": 8909
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 2.0446428571428573e-05,
+      "loss": 0.9647,
+      "step": 8910
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 2.0437500000000002e-05,
+      "loss": 0.9466,
+      "step": 8911
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 2.042857142857143e-05,
+      "loss": 0.9268,
+      "step": 8912
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 2.041964285714286e-05,
+      "loss": 0.9737,
+      "step": 8913
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 2.0410714285714285e-05,
+      "loss": 0.9439,
+      "step": 8914
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 2.0401785714285714e-05,
+      "loss": 1.0333,
+      "step": 8915
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 2.0392857142857143e-05,
+      "loss": 0.9772,
+      "step": 8916
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 2.0383928571428572e-05,
+      "loss": 0.9831,
+      "step": 8917
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 2.0375e-05,
+      "loss": 0.974,
+      "step": 8918
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 2.036607142857143e-05,
+      "loss": 0.9795,
+      "step": 8919
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 2.0357142857142858e-05,
+      "loss": 0.98,
+      "step": 8920
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 2.0348214285714287e-05,
+      "loss": 0.9229,
+      "step": 8921
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 2.0339285714285716e-05,
+      "loss": 0.9159,
+      "step": 8922
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 2.033035714285714e-05,
+      "loss": 0.9196,
+      "step": 8923
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 2.032142857142857e-05,
+      "loss": 0.9601,
+      "step": 8924
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.9249,
+      "step": 8925
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 2.030357142857143e-05,
+      "loss": 0.9516,
+      "step": 8926
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 2.029464285714286e-05,
+      "loss": 0.9062,
+      "step": 8927
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 2.0285714285714286e-05,
+      "loss": 0.9737,
+      "step": 8928
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 2.0276785714285714e-05,
+      "loss": 0.8768,
+      "step": 8929
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 2.0267857142857143e-05,
+      "loss": 0.8801,
+      "step": 8930
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 2.0258928571428572e-05,
+      "loss": 0.9615,
+      "step": 8931
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 2.025e-05,
+      "loss": 0.9457,
+      "step": 8932
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 2.024107142857143e-05,
+      "loss": 0.9294,
+      "step": 8933
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 2.023214285714286e-05,
+      "loss": 1.0005,
+      "step": 8934
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 2.0223214285714287e-05,
+      "loss": 0.8933,
+      "step": 8935
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 2.0214285714285716e-05,
+      "loss": 0.9832,
+      "step": 8936
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 2.0205357142857142e-05,
+      "loss": 0.9757,
+      "step": 8937
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 2.019642857142857e-05,
+      "loss": 0.9169,
+      "step": 8938
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 2.01875e-05,
+      "loss": 0.9431,
+      "step": 8939
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 2.017857142857143e-05,
+      "loss": 0.9737,
+      "step": 8940
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 2.016964285714286e-05,
+      "loss": 0.93,
+      "step": 8941
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 2.0160714285714286e-05,
+      "loss": 0.9687,
+      "step": 8942
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 2.0151785714285715e-05,
+      "loss": 0.9449,
+      "step": 8943
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 2.0142857142857144e-05,
+      "loss": 0.9864,
+      "step": 8944
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 2.0133928571428572e-05,
+      "loss": 1.067,
+      "step": 8945
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 2.0125e-05,
+      "loss": 0.9927,
+      "step": 8946
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 2.011607142857143e-05,
+      "loss": 0.8845,
+      "step": 8947
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 2.010714285714286e-05,
+      "loss": 0.9364,
+      "step": 8948
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 2.0098214285714288e-05,
+      "loss": 0.9462,
+      "step": 8949
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 2.0089285714285717e-05,
+      "loss": 0.9655,
+      "step": 8950
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 2.0080357142857145e-05,
+      "loss": 0.9415,
+      "step": 8951
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 2.007142857142857e-05,
+      "loss": 0.9621,
+      "step": 8952
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 2.00625e-05,
+      "loss": 0.909,
+      "step": 8953
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 2.005357142857143e-05,
+      "loss": 0.952,
+      "step": 8954
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 2.0044642857142857e-05,
+      "loss": 0.9492,
+      "step": 8955
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 2.0035714285714286e-05,
+      "loss": 0.9746,
+      "step": 8956
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 2.0026785714285715e-05,
+      "loss": 0.9761,
+      "step": 8957
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 2.0017857142857144e-05,
+      "loss": 0.9124,
+      "step": 8958
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 2.0008928571428573e-05,
+      "loss": 0.9423,
+      "step": 8959
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 2e-05,
+      "loss": 1.1006,
+      "step": 8960
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 1.9991071428571427e-05,
+      "loss": 0.9184,
+      "step": 8961
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 1.9982142857142856e-05,
+      "loss": 0.907,
+      "step": 8962
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 1.9973214285714288e-05,
+      "loss": 0.94,
+      "step": 8963
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 1.9964285714285717e-05,
+      "loss": 0.8658,
+      "step": 8964
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 1.9955357142857146e-05,
+      "loss": 0.9677,
+      "step": 8965
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 1.994642857142857e-05,
+      "loss": 0.9334,
+      "step": 8966
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 1.99375e-05,
+      "loss": 0.9199,
+      "step": 8967
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 1.992857142857143e-05,
+      "loss": 0.9037,
+      "step": 8968
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 1.9919642857142858e-05,
+      "loss": 0.9815,
+      "step": 8969
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 1.9910714285714287e-05,
+      "loss": 0.9602,
+      "step": 8970
+    },
+    {
+      "epoch": 40.05,
+      "learning_rate": 1.9901785714285715e-05,
+      "loss": 0.9347,
+      "step": 8971
+    },
+    {
+      "epoch": 40.05,
+      "learning_rate": 1.9892857142857144e-05,
+      "loss": 0.9097,
+      "step": 8972
+    },
+    {
+      "epoch": 40.06,
+      "learning_rate": 1.9883928571428573e-05,
+      "loss": 0.9269,
+      "step": 8973
+    },
+    {
+      "epoch": 40.06,
+      "learning_rate": 1.9875000000000002e-05,
+      "loss": 0.9704,
+      "step": 8974
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 1.9866071428571427e-05,
+      "loss": 0.9287,
+      "step": 8975
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 1.9857142857142856e-05,
+      "loss": 0.9774,
+      "step": 8976
+    },
+    {
+      "epoch": 40.08,
+      "learning_rate": 1.9848214285714285e-05,
+      "loss": 0.8872,
+      "step": 8977
+    },
+    {
+      "epoch": 40.08,
+      "learning_rate": 1.9839285714285717e-05,
+      "loss": 0.9537,
+      "step": 8978
+    },
+    {
+      "epoch": 40.08,
+      "learning_rate": 1.9830357142857146e-05,
+      "loss": 0.95,
+      "step": 8979
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 1.982142857142857e-05,
+      "loss": 0.9339,
+      "step": 8980
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 1.98125e-05,
+      "loss": 0.9547,
+      "step": 8981
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 1.980357142857143e-05,
+      "loss": 0.9827,
+      "step": 8982
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 1.9794642857142858e-05,
+      "loss": 0.911,
+      "step": 8983
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 1.9785714285714287e-05,
+      "loss": 0.9903,
+      "step": 8984
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 1.9776785714285716e-05,
+      "loss": 1.0269,
+      "step": 8985
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 1.9767857142857145e-05,
+      "loss": 0.9658,
+      "step": 8986
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 1.9758928571428573e-05,
+      "loss": 0.9255,
+      "step": 8987
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 0.9159,
+      "step": 8988
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 1.9741071428571428e-05,
+      "loss": 0.8914,
+      "step": 8989
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 1.9732142857142857e-05,
+      "loss": 0.9585,
+      "step": 8990
+    },
+    {
+      "epoch": 40.14,
+      "learning_rate": 1.9723214285714285e-05,
+      "loss": 0.9716,
+      "step": 8991
+    },
+    {
+      "epoch": 40.14,
+      "learning_rate": 1.9714285714285714e-05,
+      "loss": 0.9833,
+      "step": 8992
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 1.9705357142857147e-05,
+      "loss": 0.933,
+      "step": 8993
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 1.9696428571428572e-05,
+      "loss": 0.9751,
+      "step": 8994
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 1.96875e-05,
+      "loss": 0.9417,
+      "step": 8995
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 1.967857142857143e-05,
+      "loss": 0.8866,
+      "step": 8996
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 1.966964285714286e-05,
+      "loss": 1.0274,
+      "step": 8997
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 1.9660714285714287e-05,
+      "loss": 0.9666,
+      "step": 8998
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 1.9651785714285713e-05,
+      "loss": 0.8663,
+      "step": 8999
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.9838,
+      "step": 9000
+    },
+    {
+      "epoch": 40.18,
+      "eval_loss": 1.1106640100479126,
+      "eval_runtime": 44.4782,
+      "eval_samples_per_second": 294.391,
+      "eval_steps_per_second": 1.551,
+      "step": 9000
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 1.9633928571428574e-05,
+      "loss": 0.9594,
+      "step": 9001
+    },
+    {
+      "epoch": 40.19,
+      "learning_rate": 1.9625000000000003e-05,
+      "loss": 0.9627,
+      "step": 9002
+    },
+    {
+      "epoch": 40.19,
+      "learning_rate": 1.961607142857143e-05,
+      "loss": 0.91,
+      "step": 9003
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 1.9607142857142857e-05,
+      "loss": 0.9362,
+      "step": 9004
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 1.9598214285714286e-05,
+      "loss": 0.9477,
+      "step": 9005
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 1.9589285714285715e-05,
+      "loss": 1.0349,
+      "step": 9006
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 1.9580357142857143e-05,
+      "loss": 0.974,
+      "step": 9007
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 0.9829,
+      "step": 9008
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 1.95625e-05,
+      "loss": 0.9958,
+      "step": 9009
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 1.955357142857143e-05,
+      "loss": 0.9246,
+      "step": 9010
+    },
+    {
+      "epoch": 40.23,
+      "learning_rate": 1.954464285714286e-05,
+      "loss": 0.9284,
+      "step": 9011
+    },
+    {
+      "epoch": 40.23,
+      "learning_rate": 1.9535714285714288e-05,
+      "loss": 0.9888,
+      "step": 9012
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 1.9526785714285713e-05,
+      "loss": 0.9802,
+      "step": 9013
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 1.9517857142857142e-05,
+      "loss": 0.9223,
+      "step": 9014
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 1.950892857142857e-05,
+      "loss": 0.9369,
+      "step": 9015
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.9906,
+      "step": 9016
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 1.9491071428571432e-05,
+      "loss": 0.9171,
+      "step": 9017
+    },
+    {
+      "epoch": 40.26,
+      "learning_rate": 1.9482142857142857e-05,
+      "loss": 0.98,
+      "step": 9018
+    },
+    {
+      "epoch": 40.26,
+      "learning_rate": 1.9473214285714286e-05,
+      "loss": 0.9615,
+      "step": 9019
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 1.9464285714285715e-05,
+      "loss": 0.8812,
+      "step": 9020
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 1.9455357142857144e-05,
+      "loss": 0.9417,
+      "step": 9021
+    },
+    {
+      "epoch": 40.28,
+      "learning_rate": 1.9446428571428573e-05,
+      "loss": 0.9343,
+      "step": 9022
+    },
+    {
+      "epoch": 40.28,
+      "learning_rate": 1.94375e-05,
+      "loss": 0.9808,
+      "step": 9023
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.9645,
+      "step": 9024
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 1.941964285714286e-05,
+      "loss": 0.9434,
+      "step": 9025
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 1.9410714285714288e-05,
+      "loss": 0.884,
+      "step": 9026
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 1.9401785714285713e-05,
+      "loss": 0.9512,
+      "step": 9027
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 1.9392857142857142e-05,
+      "loss": 0.9673,
+      "step": 9028
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 1.938392857142857e-05,
+      "loss": 0.9195,
+      "step": 9029
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 1.9375e-05,
+      "loss": 0.9335,
+      "step": 9030
+    },
+    {
+      "epoch": 40.32,
+      "learning_rate": 1.9366071428571432e-05,
+      "loss": 0.9397,
+      "step": 9031
+    },
+    {
+      "epoch": 40.32,
+      "learning_rate": 1.9357142857142858e-05,
+      "loss": 0.908,
+      "step": 9032
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 1.9348214285714287e-05,
+      "loss": 0.9952,
+      "step": 9033
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 1.9339285714285715e-05,
+      "loss": 0.9008,
+      "step": 9034
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 1.9330357142857144e-05,
+      "loss": 0.9586,
+      "step": 9035
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 1.9321428571428573e-05,
+      "loss": 0.9965,
+      "step": 9036
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 1.93125e-05,
+      "loss": 0.9226,
+      "step": 9037
+    },
+    {
+      "epoch": 40.35,
+      "learning_rate": 1.930357142857143e-05,
+      "loss": 0.9307,
+      "step": 9038
+    },
+    {
+      "epoch": 40.35,
+      "learning_rate": 1.929464285714286e-05,
+      "loss": 0.9639,
+      "step": 9039
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.9057,
+      "step": 9040
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 1.9276785714285714e-05,
+      "loss": 0.9483,
+      "step": 9041
+    },
+    {
+      "epoch": 40.37,
+      "learning_rate": 1.9267857142857143e-05,
+      "loss": 0.9514,
+      "step": 9042
+    },
+    {
+      "epoch": 40.37,
+      "learning_rate": 1.925892857142857e-05,
+      "loss": 0.9369,
+      "step": 9043
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 1.925e-05,
+      "loss": 0.9543,
+      "step": 9044
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 1.924107142857143e-05,
+      "loss": 1.0037,
+      "step": 9045
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 1.9232142857142858e-05,
+      "loss": 1.0081,
+      "step": 9046
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 1.9223214285714287e-05,
+      "loss": 0.9208,
+      "step": 9047
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 1.9214285714285716e-05,
+      "loss": 0.9966,
+      "step": 9048
+    },
+    {
+      "epoch": 40.4,
+      "learning_rate": 1.9205357142857145e-05,
+      "loss": 1.0077,
+      "step": 9049
+    },
+    {
+      "epoch": 40.4,
+      "learning_rate": 1.9196428571428573e-05,
+      "loss": 0.9336,
+      "step": 9050
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 1.91875e-05,
+      "loss": 0.994,
+      "step": 9051
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 1.9178571428571428e-05,
+      "loss": 0.9312,
+      "step": 9052
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 1.916964285714286e-05,
+      "loss": 0.9207,
+      "step": 9053
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 1.916071428571429e-05,
+      "loss": 0.9306,
+      "step": 9054
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 1.9151785714285718e-05,
+      "loss": 0.9643,
+      "step": 9055
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 1.9142857142857143e-05,
+      "loss": 1.0169,
+      "step": 9056
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 1.9133928571428572e-05,
+      "loss": 0.9404,
+      "step": 9057
+    },
+    {
+      "epoch": 40.44,
+      "learning_rate": 1.9125e-05,
+      "loss": 0.9724,
+      "step": 9058
+    },
+    {
+      "epoch": 40.44,
+      "learning_rate": 1.911607142857143e-05,
+      "loss": 0.8653,
+      "step": 9059
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 1.910714285714286e-05,
+      "loss": 0.9057,
+      "step": 9060
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 1.9098214285714287e-05,
+      "loss": 0.9378,
+      "step": 9061
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 1.9089285714285716e-05,
+      "loss": 0.9404,
+      "step": 9062
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 1.9080357142857145e-05,
+      "loss": 0.9564,
+      "step": 9063
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 1.9071428571428574e-05,
+      "loss": 0.8801,
+      "step": 9064
+    },
+    {
+      "epoch": 40.47,
+      "learning_rate": 1.90625e-05,
+      "loss": 0.9723,
+      "step": 9065
+    },
+    {
+      "epoch": 40.47,
+      "learning_rate": 1.9053571428571428e-05,
+      "loss": 0.9396,
+      "step": 9066
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 1.9044642857142857e-05,
+      "loss": 0.9601,
+      "step": 9067
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 1.9035714285714286e-05,
+      "loss": 0.9558,
+      "step": 9068
+    },
+    {
+      "epoch": 40.49,
+      "learning_rate": 1.9026785714285718e-05,
+      "loss": 0.9191,
+      "step": 9069
+    },
+    {
+      "epoch": 40.49,
+      "learning_rate": 1.9017857142857143e-05,
+      "loss": 0.99,
+      "step": 9070
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 1.9008928571428572e-05,
+      "loss": 0.9517,
+      "step": 9071
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 1.9e-05,
+      "loss": 0.947,
+      "step": 9072
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 1.899107142857143e-05,
+      "loss": 0.9614,
+      "step": 9073
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 1.8982142857142855e-05,
+      "loss": 0.9734,
+      "step": 9074
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 1.8973214285714284e-05,
+      "loss": 1.0597,
+      "step": 9075
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 1.8964285714285716e-05,
+      "loss": 1.0062,
+      "step": 9076
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 1.8955357142857145e-05,
+      "loss": 0.9403,
+      "step": 9077
+    },
+    {
+      "epoch": 40.53,
+      "learning_rate": 1.8946428571428574e-05,
+      "loss": 0.8847,
+      "step": 9078
+    },
+    {
+      "epoch": 40.53,
+      "learning_rate": 1.89375e-05,
+      "loss": 0.9107,
+      "step": 9079
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.9822,
+      "step": 9080
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 1.8919642857142857e-05,
+      "loss": 0.9195,
+      "step": 9081
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 1.8910714285714286e-05,
+      "loss": 0.9888,
+      "step": 9082
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 1.8901785714285715e-05,
+      "loss": 1.0078,
+      "step": 9083
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 1.8892857142857144e-05,
+      "loss": 0.9317,
+      "step": 9084
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 1.8883928571428573e-05,
+      "loss": 0.9404,
+      "step": 9085
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 1.8875e-05,
+      "loss": 0.9389,
+      "step": 9086
+    },
+    {
+      "epoch": 40.57,
+      "learning_rate": 1.886607142857143e-05,
+      "loss": 1.0393,
+      "step": 9087
+    },
+    {
+      "epoch": 40.57,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 1.0623,
+      "step": 9088
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 1.8848214285714285e-05,
+      "loss": 0.8845,
+      "step": 9089
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 1.8839285714285713e-05,
+      "loss": 0.9802,
+      "step": 9090
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 1.8830357142857146e-05,
+      "loss": 0.9085,
+      "step": 9091
+    },
+    {
+      "epoch": 40.59,
+      "learning_rate": 1.8821428571428574e-05,
+      "loss": 0.9826,
+      "step": 9092
+    },
+    {
+      "epoch": 40.59,
+      "learning_rate": 1.88125e-05,
+      "loss": 0.9565,
+      "step": 9093
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 1.880357142857143e-05,
+      "loss": 1.0635,
+      "step": 9094
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 1.8794642857142858e-05,
+      "loss": 0.9852,
+      "step": 9095
+    },
+    {
+      "epoch": 40.61,
+      "learning_rate": 1.8785714285714286e-05,
+      "loss": 0.979,
+      "step": 9096
+    },
+    {
+      "epoch": 40.61,
+      "learning_rate": 1.8776785714285715e-05,
+      "loss": 0.9635,
+      "step": 9097
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 1.8767857142857144e-05,
+      "loss": 0.9473,
+      "step": 9098
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 1.8758928571428573e-05,
+      "loss": 0.9137,
+      "step": 9099
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.8863,
+      "step": 9100
+    },
+    {
+      "epoch": 40.63,
+      "learning_rate": 1.874107142857143e-05,
+      "loss": 0.972,
+      "step": 9101
+    },
+    {
+      "epoch": 40.63,
+      "learning_rate": 1.873214285714286e-05,
+      "loss": 0.878,
+      "step": 9102
+    },
+    {
+      "epoch": 40.64,
+      "learning_rate": 1.8723214285714285e-05,
+      "loss": 0.9294,
+      "step": 9103
+    },
+    {
+      "epoch": 40.64,
+      "learning_rate": 1.8714285714285714e-05,
+      "loss": 0.9504,
+      "step": 9104
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 1.8705357142857143e-05,
+      "loss": 1.0225,
+      "step": 9105
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 1.869642857142857e-05,
+      "loss": 0.9726,
+      "step": 9106
+    },
+    {
+      "epoch": 40.66,
+      "learning_rate": 1.8687500000000004e-05,
+      "loss": 0.9633,
+      "step": 9107
+    },
+    {
+      "epoch": 40.66,
+      "learning_rate": 1.867857142857143e-05,
+      "loss": 0.928,
+      "step": 9108
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 1.8669642857142858e-05,
+      "loss": 0.9487,
+      "step": 9109
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 1.8660714285714287e-05,
+      "loss": 0.918,
+      "step": 9110
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 1.8651785714285716e-05,
+      "loss": 0.9651,
+      "step": 9111
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 1.864285714285714e-05,
+      "loss": 1.07,
+      "step": 9112
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 1.8633928571428573e-05,
+      "loss": 0.9803,
+      "step": 9113
+    },
+    {
+      "epoch": 40.69,
+      "learning_rate": 1.8625000000000002e-05,
+      "loss": 0.9185,
+      "step": 9114
+    },
+    {
+      "epoch": 40.69,
+      "learning_rate": 1.861607142857143e-05,
+      "loss": 0.9201,
+      "step": 9115
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 1.860714285714286e-05,
+      "loss": 0.9032,
+      "step": 9116
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 1.8598214285714285e-05,
+      "loss": 0.9387,
+      "step": 9117
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 1.8589285714285714e-05,
+      "loss": 0.9544,
+      "step": 9118
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 1.8580357142857143e-05,
+      "loss": 0.9586,
+      "step": 9119
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.9452,
+      "step": 9120
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 1.85625e-05,
+      "loss": 0.9823,
+      "step": 9121
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 1.855357142857143e-05,
+      "loss": 0.9872,
+      "step": 9122
+    },
+    {
+      "epoch": 40.73,
+      "learning_rate": 1.8544642857142858e-05,
+      "loss": 0.9157,
+      "step": 9123
+    },
+    {
+      "epoch": 40.73,
+      "learning_rate": 1.8535714285714287e-05,
+      "loss": 0.9636,
+      "step": 9124
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 1.8526785714285716e-05,
+      "loss": 0.9034,
+      "step": 9125
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 1.8517857142857145e-05,
+      "loss": 0.918,
+      "step": 9126
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 1.850892857142857e-05,
+      "loss": 0.9782,
+      "step": 9127
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 1.85e-05,
+      "loss": 1.001,
+      "step": 9128
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 1.849107142857143e-05,
+      "loss": 0.9431,
+      "step": 9129
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 1.848214285714286e-05,
+      "loss": 0.9669,
+      "step": 9130
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 1.8473214285714286e-05,
+      "loss": 1.0056,
+      "step": 9131
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 1.8464285714285714e-05,
+      "loss": 0.9298,
+      "step": 9132
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 1.8455357142857143e-05,
+      "loss": 0.9748,
+      "step": 9133
+    },
+    {
+      "epoch": 40.78,
+      "learning_rate": 1.8446428571428572e-05,
+      "loss": 0.9965,
+      "step": 9134
+    },
+    {
+      "epoch": 40.78,
+      "learning_rate": 1.84375e-05,
+      "loss": 0.9905,
+      "step": 9135
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.9815,
+      "step": 9136
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 1.841964285714286e-05,
+      "loss": 0.9355,
+      "step": 9137
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 1.8410714285714287e-05,
+      "loss": 0.9021,
+      "step": 9138
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 1.8401785714285716e-05,
+      "loss": 0.9952,
+      "step": 9139
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 1.8392857142857145e-05,
+      "loss": 0.9504,
+      "step": 9140
+    },
+    {
+      "epoch": 40.81,
+      "learning_rate": 1.838392857142857e-05,
+      "loss": 1.0171,
+      "step": 9141
+    },
+    {
+      "epoch": 40.81,
+      "learning_rate": 1.8375e-05,
+      "loss": 0.8542,
+      "step": 9142
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 1.8366071428571428e-05,
+      "loss": 0.9448,
+      "step": 9143
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 1.835714285714286e-05,
+      "loss": 0.9607,
+      "step": 9144
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 1.8348214285714286e-05,
+      "loss": 1.0065,
+      "step": 9145
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 1.8339285714285715e-05,
+      "loss": 0.8849,
+      "step": 9146
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 1.8330357142857144e-05,
+      "loss": 0.9015,
+      "step": 9147
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 1.8321428571428572e-05,
+      "loss": 0.9767,
+      "step": 9148
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 1.83125e-05,
+      "loss": 0.9905,
+      "step": 9149
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 1.8303571428571427e-05,
+      "loss": 1.0306,
+      "step": 9150
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 1.829464285714286e-05,
+      "loss": 0.9617,
+      "step": 9151
+    },
+    {
+      "epoch": 40.86,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.9731,
+      "step": 9152
+    },
+    {
+      "epoch": 40.86,
+      "learning_rate": 1.8276785714285717e-05,
+      "loss": 0.9806,
+      "step": 9153
+    },
+    {
+      "epoch": 40.87,
+      "learning_rate": 1.8267857142857145e-05,
+      "loss": 0.9698,
+      "step": 9154
+    },
+    {
+      "epoch": 40.87,
+      "learning_rate": 1.825892857142857e-05,
+      "loss": 0.994,
+      "step": 9155
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 1.825e-05,
+      "loss": 0.952,
+      "step": 9156
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 1.824107142857143e-05,
+      "loss": 0.95,
+      "step": 9157
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 1.8232142857142857e-05,
+      "loss": 0.9804,
+      "step": 9158
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 1.8223214285714286e-05,
+      "loss": 0.9529,
+      "step": 9159
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.9048,
+      "step": 9160
+    },
+    {
+      "epoch": 40.9,
+      "learning_rate": 1.8205357142857144e-05,
+      "loss": 0.9532,
+      "step": 9161
+    },
+    {
+      "epoch": 40.9,
+      "learning_rate": 1.8196428571428573e-05,
+      "loss": 0.9467,
+      "step": 9162
+    },
+    {
+      "epoch": 40.91,
+      "learning_rate": 1.81875e-05,
+      "loss": 0.965,
+      "step": 9163
+    },
+    {
+      "epoch": 40.91,
+      "learning_rate": 1.8178571428571427e-05,
+      "loss": 0.9007,
+      "step": 9164
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 1.8169642857142856e-05,
+      "loss": 0.9694,
+      "step": 9165
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 1.8160714285714288e-05,
+      "loss": 0.9855,
+      "step": 9166
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 1.8151785714285717e-05,
+      "loss": 1.029,
+      "step": 9167
+    },
+    {
+      "epoch": 40.93,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.9753,
+      "step": 9168
+    },
+    {
+      "epoch": 40.93,
+      "learning_rate": 1.813392857142857e-05,
+      "loss": 0.9507,
+      "step": 9169
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 1.8125e-05,
+      "loss": 1.0219,
+      "step": 9170
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 1.811607142857143e-05,
+      "loss": 0.9773,
+      "step": 9171
+    },
+    {
+      "epoch": 40.95,
+      "learning_rate": 1.8107142857142858e-05,
+      "loss": 0.9414,
+      "step": 9172
+    },
+    {
+      "epoch": 40.95,
+      "learning_rate": 1.8098214285714287e-05,
+      "loss": 1.0073,
+      "step": 9173
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 1.8089285714285715e-05,
+      "loss": 0.9236,
+      "step": 9174
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 1.8080357142857144e-05,
+      "loss": 1.0444,
+      "step": 9175
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 1.8071428571428573e-05,
+      "loss": 1.0224,
+      "step": 9176
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 1.8062500000000002e-05,
+      "loss": 0.9401,
+      "step": 9177
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 1.805357142857143e-05,
+      "loss": 0.9348,
+      "step": 9178
+    },
+    {
+      "epoch": 40.98,
+      "learning_rate": 1.8044642857142856e-05,
+      "loss": 0.9489,
+      "step": 9179
+    },
+    {
+      "epoch": 40.98,
+      "learning_rate": 1.8035714285714285e-05,
+      "loss": 1.0133,
+      "step": 9180
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 1.8026785714285714e-05,
+      "loss": 0.9211,
+      "step": 9181
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 1.8017857142857146e-05,
+      "loss": 0.9182,
+      "step": 9182
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 1.800892857142857e-05,
+      "loss": 0.9298,
+      "step": 9183
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.7567,
+      "step": 9184
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 1.799107142857143e-05,
+      "loss": 0.9659,
+      "step": 9185
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 1.7982142857142858e-05,
+      "loss": 0.9155,
+      "step": 9186
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 1.7973214285714287e-05,
+      "loss": 0.9179,
+      "step": 9187
+    },
+    {
+      "epoch": 41.02,
+      "learning_rate": 1.7964285714285712e-05,
+      "loss": 0.9284,
+      "step": 9188
+    },
+    {
+      "epoch": 41.02,
+      "learning_rate": 1.7955357142857145e-05,
+      "loss": 0.9779,
+      "step": 9189
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 1.7946428571428573e-05,
+      "loss": 0.9115,
+      "step": 9190
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 1.7937500000000002e-05,
+      "loss": 0.9501,
+      "step": 9191
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 1.792857142857143e-05,
+      "loss": 0.9278,
+      "step": 9192
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 1.7919642857142857e-05,
+      "loss": 0.9745,
+      "step": 9193
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 1.7910714285714285e-05,
+      "loss": 0.9887,
+      "step": 9194
+    },
+    {
+      "epoch": 41.05,
+      "learning_rate": 1.7901785714285714e-05,
+      "loss": 0.9897,
+      "step": 9195
+    },
+    {
+      "epoch": 41.05,
+      "learning_rate": 1.7892857142857143e-05,
+      "loss": 0.9706,
+      "step": 9196
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 1.7883928571428572e-05,
+      "loss": 0.9483,
+      "step": 9197
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 1.7875e-05,
+      "loss": 0.9445,
+      "step": 9198
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 1.786607142857143e-05,
+      "loss": 0.9206,
+      "step": 9199
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 1.006,
+      "step": 9200
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 1.7848214285714287e-05,
+      "loss": 0.9674,
+      "step": 9201
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 1.7839285714285713e-05,
+      "loss": 1.0128,
+      "step": 9202
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 1.783035714285714e-05,
+      "loss": 0.992,
+      "step": 9203
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 1.7821428571428574e-05,
+      "loss": 0.9818,
+      "step": 9204
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 1.7812500000000003e-05,
+      "loss": 0.9353,
+      "step": 9205
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 1.780357142857143e-05,
+      "loss": 0.9615,
+      "step": 9206
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 1.7794642857142857e-05,
+      "loss": 0.9245,
+      "step": 9207
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 1.7785714285714286e-05,
+      "loss": 0.9055,
+      "step": 9208
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 1.7776785714285715e-05,
+      "loss": 0.95,
+      "step": 9209
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 1.7767857142857143e-05,
+      "loss": 1.0078,
+      "step": 9210
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 1.7758928571428572e-05,
+      "loss": 0.9558,
+      "step": 9211
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 1.775e-05,
+      "loss": 0.8877,
+      "step": 9212
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 1.774107142857143e-05,
+      "loss": 0.935,
+      "step": 9213
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 1.773214285714286e-05,
+      "loss": 0.9915,
+      "step": 9214
+    },
+    {
+      "epoch": 41.14,
+      "learning_rate": 1.7723214285714288e-05,
+      "loss": 0.9528,
+      "step": 9215
+    },
+    {
+      "epoch": 41.14,
+      "learning_rate": 1.7714285714285713e-05,
+      "loss": 0.9359,
+      "step": 9216
+    },
+    {
+      "epoch": 41.15,
+      "learning_rate": 1.7705357142857142e-05,
+      "loss": 0.9764,
+      "step": 9217
+    },
+    {
+      "epoch": 41.15,
+      "learning_rate": 1.769642857142857e-05,
+      "loss": 0.9797,
+      "step": 9218
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 1.76875e-05,
+      "loss": 0.9133,
+      "step": 9219
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 1.7678571428571432e-05,
+      "loss": 0.977,
+      "step": 9220
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 1.7669642857142857e-05,
+      "loss": 0.9711,
+      "step": 9221
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 1.7660714285714286e-05,
+      "loss": 0.9698,
+      "step": 9222
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 1.7651785714285715e-05,
+      "loss": 0.911,
+      "step": 9223
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 1.7642857142857144e-05,
+      "loss": 0.9282,
+      "step": 9224
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 1.7633928571428573e-05,
+      "loss": 0.9594,
+      "step": 9225
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 1.7625e-05,
+      "loss": 0.9266,
+      "step": 9226
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 1.761607142857143e-05,
+      "loss": 0.8737,
+      "step": 9227
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 1.760714285714286e-05,
+      "loss": 0.9999,
+      "step": 9228
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 1.7598214285714288e-05,
+      "loss": 0.9717,
+      "step": 9229
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 1.7589285714285717e-05,
+      "loss": 0.9717,
+      "step": 9230
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 1.7580357142857142e-05,
+      "loss": 0.9606,
+      "step": 9231
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.967,
+      "step": 9232
+    },
+    {
+      "epoch": 41.22,
+      "learning_rate": 1.75625e-05,
+      "loss": 0.957,
+      "step": 9233
+    },
+    {
+      "epoch": 41.22,
+      "learning_rate": 1.755357142857143e-05,
+      "loss": 0.9558,
+      "step": 9234
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 1.7544642857142858e-05,
+      "loss": 0.925,
+      "step": 9235
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 1.7535714285714287e-05,
+      "loss": 0.9725,
+      "step": 9236
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 1.7526785714285715e-05,
+      "loss": 0.9849,
+      "step": 9237
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 1.7517857142857144e-05,
+      "loss": 0.9831,
+      "step": 9238
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 1.7508928571428573e-05,
+      "loss": 0.9221,
+      "step": 9239
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 1.75e-05,
+      "loss": 0.9022,
+      "step": 9240
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 1.7491071428571427e-05,
+      "loss": 0.9048,
+      "step": 9241
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 1.748214285714286e-05,
+      "loss": 1.0014,
+      "step": 9242
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 1.747321428571429e-05,
+      "loss": 0.9426,
+      "step": 9243
+    },
+    {
+      "epoch": 41.27,
+      "learning_rate": 1.7464285714285717e-05,
+      "loss": 1.0311,
+      "step": 9244
+    },
+    {
+      "epoch": 41.27,
+      "learning_rate": 1.7455357142857143e-05,
+      "loss": 0.95,
+      "step": 9245
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 1.744642857142857e-05,
+      "loss": 0.9237,
+      "step": 9246
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 1.74375e-05,
+      "loss": 0.9524,
+      "step": 9247
+    },
+    {
+      "epoch": 41.29,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.9927,
+      "step": 9248
+    },
+    {
+      "epoch": 41.29,
+      "learning_rate": 1.7419642857142858e-05,
+      "loss": 0.9668,
+      "step": 9249
+    },
+    {
+      "epoch": 41.29,
+      "learning_rate": 1.7410714285714287e-05,
+      "loss": 0.9613,
+      "step": 9250
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 1.7401785714285716e-05,
+      "loss": 0.9592,
+      "step": 9251
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 1.7392857142857145e-05,
+      "loss": 0.9987,
+      "step": 9252
+    },
+    {
+      "epoch": 41.31,
+      "learning_rate": 1.7383928571428573e-05,
+      "loss": 1.0229,
+      "step": 9253
+    },
+    {
+      "epoch": 41.31,
+      "learning_rate": 1.7375e-05,
+      "loss": 0.9508,
+      "step": 9254
+    },
+    {
+      "epoch": 41.32,
+      "learning_rate": 1.7366071428571428e-05,
+      "loss": 0.9644,
+      "step": 9255
+    },
+    {
+      "epoch": 41.32,
+      "learning_rate": 1.7357142857142856e-05,
+      "loss": 0.9343,
+      "step": 9256
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 1.734821428571429e-05,
+      "loss": 1.0119,
+      "step": 9257
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 1.7339285714285718e-05,
+      "loss": 0.9294,
+      "step": 9258
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 1.7330357142857143e-05,
+      "loss": 0.9264,
+      "step": 9259
+    },
+    {
+      "epoch": 41.34,
+      "learning_rate": 1.7321428571428572e-05,
+      "loss": 0.9403,
+      "step": 9260
+    },
+    {
+      "epoch": 41.34,
+      "learning_rate": 1.73125e-05,
+      "loss": 0.9455,
+      "step": 9261
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 1.730357142857143e-05,
+      "loss": 1.0097,
+      "step": 9262
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 1.729464285714286e-05,
+      "loss": 1.0346,
+      "step": 9263
+    },
+    {
+      "epoch": 41.36,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.9156,
+      "step": 9264
+    },
+    {
+      "epoch": 41.36,
+      "learning_rate": 1.7276785714285716e-05,
+      "loss": 0.9442,
+      "step": 9265
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 1.7267857142857145e-05,
+      "loss": 0.9465,
+      "step": 9266
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 1.7258928571428574e-05,
+      "loss": 0.9234,
+      "step": 9267
+    },
+    {
+      "epoch": 41.38,
+      "learning_rate": 1.725e-05,
+      "loss": 0.9736,
+      "step": 9268
+    },
+    {
+      "epoch": 41.38,
+      "learning_rate": 1.7241071428571428e-05,
+      "loss": 0.9392,
+      "step": 9269
+    },
+    {
+      "epoch": 41.38,
+      "learning_rate": 1.7232142857142857e-05,
+      "loss": 0.9603,
+      "step": 9270
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 1.7223214285714286e-05,
+      "loss": 0.9317,
+      "step": 9271
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 1.7214285714285715e-05,
+      "loss": 0.9796,
+      "step": 9272
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 1.7205357142857143e-05,
+      "loss": 0.9547,
+      "step": 9273
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 1.7196428571428572e-05,
+      "loss": 0.9519,
+      "step": 9274
+    },
+    {
+      "epoch": 41.41,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.9732,
+      "step": 9275
+    },
+    {
+      "epoch": 41.41,
+      "learning_rate": 1.717857142857143e-05,
+      "loss": 0.9685,
+      "step": 9276
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 1.716964285714286e-05,
+      "loss": 0.9315,
+      "step": 9277
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 1.7160714285714284e-05,
+      "loss": 0.9525,
+      "step": 9278
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 1.7151785714285713e-05,
+      "loss": 0.9625,
+      "step": 9279
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 1.7142857142857145e-05,
+      "loss": 0.9283,
+      "step": 9280
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 1.7133928571428574e-05,
+      "loss": 0.9485,
+      "step": 9281
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 1.7125000000000003e-05,
+      "loss": 1.0634,
+      "step": 9282
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 1.711607142857143e-05,
+      "loss": 0.9528,
+      "step": 9283
+    },
+    {
+      "epoch": 41.45,
+      "learning_rate": 1.7107142857142857e-05,
+      "loss": 0.8826,
+      "step": 9284
+    },
+    {
+      "epoch": 41.45,
+      "learning_rate": 1.7098214285714286e-05,
+      "loss": 0.9581,
+      "step": 9285
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 1.7089285714285715e-05,
+      "loss": 0.9476,
+      "step": 9286
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 1.7080357142857144e-05,
+      "loss": 0.9432,
+      "step": 9287
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 1.7071428571428573e-05,
+      "loss": 0.9241,
+      "step": 9288
+    },
+    {
+      "epoch": 41.47,
+      "learning_rate": 1.70625e-05,
+      "loss": 0.9385,
+      "step": 9289
+    },
+    {
+      "epoch": 41.47,
+      "learning_rate": 1.705357142857143e-05,
+      "loss": 0.9029,
+      "step": 9290
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 1.704464285714286e-05,
+      "loss": 0.9143,
+      "step": 9291
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 1.7035714285714285e-05,
+      "loss": 0.9106,
+      "step": 9292
+    },
+    {
+      "epoch": 41.49,
+      "learning_rate": 1.7026785714285713e-05,
+      "loss": 0.9636,
+      "step": 9293
+    },
+    {
+      "epoch": 41.49,
+      "learning_rate": 1.7017857142857142e-05,
+      "loss": 0.9779,
+      "step": 9294
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 1.7008928571428574e-05,
+      "loss": 0.9257,
+      "step": 9295
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.03,
+      "step": 9296
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 1.699107142857143e-05,
+      "loss": 0.8347,
+      "step": 9297
+    },
+    {
+      "epoch": 41.51,
+      "learning_rate": 1.6982142857142858e-05,
+      "loss": 0.9388,
+      "step": 9298
+    },
+    {
+      "epoch": 41.51,
+      "learning_rate": 1.6973214285714286e-05,
+      "loss": 0.9672,
+      "step": 9299
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 1.6964285714285715e-05,
+      "loss": 0.9803,
+      "step": 9300
+    },
+    {
+      "epoch": 41.52,
+      "eval_loss": 1.1102186441421509,
+      "eval_runtime": 44.3449,
+      "eval_samples_per_second": 295.276,
+      "eval_steps_per_second": 1.556,
+      "step": 9300
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 1.6955357142857144e-05,
+      "loss": 0.9383,
+      "step": 9301
+    },
+    {
+      "epoch": 41.53,
+      "learning_rate": 1.6946428571428573e-05,
+      "loss": 0.934,
+      "step": 9302
+    },
+    {
+      "epoch": 41.53,
+      "learning_rate": 1.6937500000000002e-05,
+      "loss": 0.9435,
+      "step": 9303
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 1.692857142857143e-05,
+      "loss": 0.9819,
+      "step": 9304
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 1.691964285714286e-05,
+      "loss": 0.9396,
+      "step": 9305
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 1.6910714285714285e-05,
+      "loss": 1.0131,
+      "step": 9306
+    },
+    {
+      "epoch": 41.55,
+      "learning_rate": 1.6901785714285714e-05,
+      "loss": 0.9314,
+      "step": 9307
+    },
+    {
+      "epoch": 41.55,
+      "learning_rate": 1.6892857142857143e-05,
+      "loss": 0.9456,
+      "step": 9308
+    },
+    {
+      "epoch": 41.56,
+      "learning_rate": 1.688392857142857e-05,
+      "loss": 0.9483,
+      "step": 9309
+    },
+    {
+      "epoch": 41.56,
+      "learning_rate": 1.6875000000000004e-05,
+      "loss": 0.9303,
+      "step": 9310
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 1.686607142857143e-05,
+      "loss": 0.9053,
+      "step": 9311
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.9481,
+      "step": 9312
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 1.6848214285714287e-05,
+      "loss": 0.9536,
+      "step": 9313
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 1.6839285714285716e-05,
+      "loss": 0.9245,
+      "step": 9314
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 1.6830357142857144e-05,
+      "loss": 0.9393,
+      "step": 9315
+    },
+    {
+      "epoch": 41.59,
+      "learning_rate": 1.682142857142857e-05,
+      "loss": 0.9981,
+      "step": 9316
+    },
+    {
+      "epoch": 41.59,
+      "learning_rate": 1.6812500000000002e-05,
+      "loss": 0.8765,
+      "step": 9317
+    },
+    {
+      "epoch": 41.6,
+      "learning_rate": 1.680357142857143e-05,
+      "loss": 0.9898,
+      "step": 9318
+    },
+    {
+      "epoch": 41.6,
+      "learning_rate": 1.679464285714286e-05,
+      "loss": 0.9668,
+      "step": 9319
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 1.6785714285714285e-05,
+      "loss": 0.9481,
+      "step": 9320
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 1.6776785714285714e-05,
+      "loss": 0.9284,
+      "step": 9321
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 1.6767857142857143e-05,
+      "loss": 0.9786,
+      "step": 9322
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 1.6758928571428572e-05,
+      "loss": 0.9558,
+      "step": 9323
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 1.675e-05,
+      "loss": 1.0257,
+      "step": 9324
+    },
+    {
+      "epoch": 41.63,
+      "learning_rate": 1.674107142857143e-05,
+      "loss": 0.9582,
+      "step": 9325
+    },
+    {
+      "epoch": 41.63,
+      "learning_rate": 1.6732142857142858e-05,
+      "loss": 0.9997,
+      "step": 9326
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 1.6723214285714287e-05,
+      "loss": 1.0135,
+      "step": 9327
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.9401,
+      "step": 9328
+    },
+    {
+      "epoch": 41.65,
+      "learning_rate": 1.6705357142857145e-05,
+      "loss": 0.9571,
+      "step": 9329
+    },
+    {
+      "epoch": 41.65,
+      "learning_rate": 1.669642857142857e-05,
+      "loss": 1.0256,
+      "step": 9330
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 1.66875e-05,
+      "loss": 0.9905,
+      "step": 9331
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 1.6678571428571428e-05,
+      "loss": 0.9739,
+      "step": 9332
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 1.666964285714286e-05,
+      "loss": 0.9528,
+      "step": 9333
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 1.666071428571429e-05,
+      "loss": 0.9413,
+      "step": 9334
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 1.6651785714285714e-05,
+      "loss": 0.9407,
+      "step": 9335
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 1.6642857142857143e-05,
+      "loss": 1.0252,
+      "step": 9336
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 1.6633928571428572e-05,
+      "loss": 0.9544,
+      "step": 9337
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 1.6625e-05,
+      "loss": 0.941,
+      "step": 9338
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 1.661607142857143e-05,
+      "loss": 0.9927,
+      "step": 9339
+    },
+    {
+      "epoch": 41.7,
+      "learning_rate": 1.660714285714286e-05,
+      "loss": 0.9213,
+      "step": 9340
+    },
+    {
+      "epoch": 41.7,
+      "learning_rate": 1.6598214285714287e-05,
+      "loss": 0.8633,
+      "step": 9341
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 1.6589285714285716e-05,
+      "loss": 0.9535,
+      "step": 9342
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 1.6580357142857145e-05,
+      "loss": 0.987,
+      "step": 9343
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 1.657142857142857e-05,
+      "loss": 0.984,
+      "step": 9344
+    },
+    {
+      "epoch": 41.72,
+      "learning_rate": 1.65625e-05,
+      "loss": 0.9958,
+      "step": 9345
+    },
+    {
+      "epoch": 41.72,
+      "learning_rate": 1.6553571428571428e-05,
+      "loss": 0.99,
+      "step": 9346
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 1.6544642857142857e-05,
+      "loss": 0.9502,
+      "step": 9347
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 1.653571428571429e-05,
+      "loss": 0.9107,
+      "step": 9348
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 1.6526785714285715e-05,
+      "loss": 0.934,
+      "step": 9349
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 1.6517857142857144e-05,
+      "loss": 0.9038,
+      "step": 9350
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 1.6508928571428572e-05,
+      "loss": 0.9329,
+      "step": 9351
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 1.65e-05,
+      "loss": 0.9377,
+      "step": 9352
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 1.649107142857143e-05,
+      "loss": 0.9356,
+      "step": 9353
+    },
+    {
+      "epoch": 41.76,
+      "learning_rate": 1.6482142857142856e-05,
+      "loss": 0.9888,
+      "step": 9354
+    },
+    {
+      "epoch": 41.76,
+      "learning_rate": 1.6473214285714288e-05,
+      "loss": 0.9074,
+      "step": 9355
+    },
+    {
+      "epoch": 41.77,
+      "learning_rate": 1.6464285714285717e-05,
+      "loss": 0.9985,
+      "step": 9356
+    },
+    {
+      "epoch": 41.77,
+      "learning_rate": 1.6455357142857145e-05,
+      "loss": 0.884,
+      "step": 9357
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 1.644642857142857e-05,
+      "loss": 0.9456,
+      "step": 9358
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 1.64375e-05,
+      "loss": 0.9188,
+      "step": 9359
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.9536,
+      "step": 9360
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 1.6419642857142857e-05,
+      "loss": 0.9865,
+      "step": 9361
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 1.6410714285714286e-05,
+      "loss": 0.9356,
+      "step": 9362
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 1.6401785714285715e-05,
+      "loss": 0.939,
+      "step": 9363
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 1.6392857142857144e-05,
+      "loss": 0.9507,
+      "step": 9364
+    },
+    {
+      "epoch": 41.81,
+      "learning_rate": 1.6383928571428573e-05,
+      "loss": 0.9304,
+      "step": 9365
+    },
+    {
+      "epoch": 41.81,
+      "learning_rate": 1.6375e-05,
+      "loss": 0.9983,
+      "step": 9366
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 1.636607142857143e-05,
+      "loss": 0.9333,
+      "step": 9367
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 1.6357142857142856e-05,
+      "loss": 0.9979,
+      "step": 9368
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 1.6348214285714285e-05,
+      "loss": 1.0113,
+      "step": 9369
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 1.6339285714285717e-05,
+      "loss": 0.9719,
+      "step": 9370
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 1.6330357142857146e-05,
+      "loss": 0.9662,
+      "step": 9371
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 1.632142857142857e-05,
+      "loss": 0.9805,
+      "step": 9372
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 1.63125e-05,
+      "loss": 0.9003,
+      "step": 9373
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 1.630357142857143e-05,
+      "loss": 0.9905,
+      "step": 9374
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 1.6294642857142858e-05,
+      "loss": 0.9208,
+      "step": 9375
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.9499,
+      "step": 9376
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 1.6276785714285715e-05,
+      "loss": 0.9895,
+      "step": 9377
+    },
+    {
+      "epoch": 41.87,
+      "learning_rate": 1.6267857142857144e-05,
+      "loss": 0.918,
+      "step": 9378
+    },
+    {
+      "epoch": 41.87,
+      "learning_rate": 1.6258928571428573e-05,
+      "loss": 0.9083,
+      "step": 9379
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.9877,
+      "step": 9380
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 1.624107142857143e-05,
+      "loss": 0.9781,
+      "step": 9381
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 1.6232142857142856e-05,
+      "loss": 0.9097,
+      "step": 9382
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 1.6223214285714285e-05,
+      "loss": 0.8833,
+      "step": 9383
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 1.6214285714285714e-05,
+      "loss": 0.9587,
+      "step": 9384
+    },
+    {
+      "epoch": 41.9,
+      "learning_rate": 1.6205357142857143e-05,
+      "loss": 0.9388,
+      "step": 9385
+    },
+    {
+      "epoch": 41.9,
+      "learning_rate": 1.6196428571428575e-05,
+      "loss": 0.9749,
+      "step": 9386
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 1.61875e-05,
+      "loss": 0.9096,
+      "step": 9387
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 1.617857142857143e-05,
+      "loss": 0.9236,
+      "step": 9388
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 1.6169642857142858e-05,
+      "loss": 1.0268,
+      "step": 9389
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 1.6160714285714287e-05,
+      "loss": 0.9779,
+      "step": 9390
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 1.6151785714285712e-05,
+      "loss": 0.9377,
+      "step": 9391
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 1.614285714285714e-05,
+      "loss": 0.982,
+      "step": 9392
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 1.6133928571428573e-05,
+      "loss": 0.9684,
+      "step": 9393
+    },
+    {
+      "epoch": 41.94,
+      "learning_rate": 1.6125000000000002e-05,
+      "loss": 1.0099,
+      "step": 9394
+    },
+    {
+      "epoch": 41.94,
+      "learning_rate": 1.611607142857143e-05,
+      "loss": 0.8831,
+      "step": 9395
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 1.6107142857142857e-05,
+      "loss": 0.939,
+      "step": 9396
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 1.6098214285714285e-05,
+      "loss": 0.9277,
+      "step": 9397
+    },
+    {
+      "epoch": 41.96,
+      "learning_rate": 1.6089285714285714e-05,
+      "loss": 0.9557,
+      "step": 9398
+    },
+    {
+      "epoch": 41.96,
+      "learning_rate": 1.6080357142857143e-05,
+      "loss": 1.0009,
+      "step": 9399
+    },
+    {
+      "epoch": 41.96,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.9441,
+      "step": 9400
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 1.60625e-05,
+      "loss": 0.9881,
+      "step": 9401
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 1.605357142857143e-05,
+      "loss": 0.9265,
+      "step": 9402
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 1.604464285714286e-05,
+      "loss": 0.9539,
+      "step": 9403
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 1.6035714285714287e-05,
+      "loss": 0.9329,
+      "step": 9404
+    },
+    {
+      "epoch": 41.99,
+      "learning_rate": 1.6026785714285716e-05,
+      "loss": 0.8789,
+      "step": 9405
+    },
+    {
+      "epoch": 41.99,
+      "learning_rate": 1.601785714285714e-05,
+      "loss": 0.9685,
+      "step": 9406
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 1.600892857142857e-05,
+      "loss": 0.9021,
+      "step": 9407
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9588,
+      "step": 9408
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 1.599107142857143e-05,
+      "loss": 0.8883,
+      "step": 9409
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 1.5982142857142857e-05,
+      "loss": 0.9412,
+      "step": 9410
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 1.5973214285714286e-05,
+      "loss": 0.9092,
+      "step": 9411
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 1.5964285714285715e-05,
+      "loss": 0.9341,
+      "step": 9412
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 1.5955357142857143e-05,
+      "loss": 0.9238,
+      "step": 9413
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 1.5946428571428572e-05,
+      "loss": 0.9391,
+      "step": 9414
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 1.59375e-05,
+      "loss": 0.914,
+      "step": 9415
+    },
+    {
+      "epoch": 42.04,
+      "learning_rate": 1.592857142857143e-05,
+      "loss": 0.9277,
+      "step": 9416
+    },
+    {
+      "epoch": 42.04,
+      "learning_rate": 1.591964285714286e-05,
+      "loss": 0.9441,
+      "step": 9417
+    },
+    {
+      "epoch": 42.04,
+      "learning_rate": 1.5910714285714288e-05,
+      "loss": 0.9388,
+      "step": 9418
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 1.5901785714285716e-05,
+      "loss": 0.9699,
+      "step": 9419
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 1.5892857142857142e-05,
+      "loss": 0.9108,
+      "step": 9420
+    },
+    {
+      "epoch": 42.06,
+      "learning_rate": 1.588392857142857e-05,
+      "loss": 0.8762,
+      "step": 9421
+    },
+    {
+      "epoch": 42.06,
+      "learning_rate": 1.5875e-05,
+      "loss": 0.9583,
+      "step": 9422
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 1.586607142857143e-05,
+      "loss": 0.959,
+      "step": 9423
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.9418,
+      "step": 9424
+    },
+    {
+      "epoch": 42.08,
+      "learning_rate": 1.5848214285714286e-05,
+      "loss": 0.9239,
+      "step": 9425
+    },
+    {
+      "epoch": 42.08,
+      "learning_rate": 1.5839285714285715e-05,
+      "loss": 0.9579,
+      "step": 9426
+    },
+    {
+      "epoch": 42.08,
+      "learning_rate": 1.5830357142857144e-05,
+      "loss": 0.942,
+      "step": 9427
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 1.5821428571428573e-05,
+      "loss": 1.0827,
+      "step": 9428
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 1.5812499999999998e-05,
+      "loss": 0.9324,
+      "step": 9429
+    },
+    {
+      "epoch": 42.1,
+      "learning_rate": 1.580357142857143e-05,
+      "loss": 0.9504,
+      "step": 9430
+    },
+    {
+      "epoch": 42.1,
+      "learning_rate": 1.579464285714286e-05,
+      "loss": 1.0278,
+      "step": 9431
+    },
+    {
+      "epoch": 42.11,
+      "learning_rate": 1.5785714285714288e-05,
+      "loss": 0.9718,
+      "step": 9432
+    },
+    {
+      "epoch": 42.11,
+      "learning_rate": 1.5776785714285717e-05,
+      "loss": 1.0233,
+      "step": 9433
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 1.5767857142857142e-05,
+      "loss": 0.955,
+      "step": 9434
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 1.575892857142857e-05,
+      "loss": 0.9703,
+      "step": 9435
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 1.575e-05,
+      "loss": 0.9759,
+      "step": 9436
+    },
+    {
+      "epoch": 42.13,
+      "learning_rate": 1.574107142857143e-05,
+      "loss": 0.9887,
+      "step": 9437
+    },
+    {
+      "epoch": 42.13,
+      "learning_rate": 1.5732142857142858e-05,
+      "loss": 0.9804,
+      "step": 9438
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 1.5723214285714286e-05,
+      "loss": 1.0117,
+      "step": 9439
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.9353,
+      "step": 9440
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 1.5705357142857144e-05,
+      "loss": 0.9175,
+      "step": 9441
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 1.5696428571428573e-05,
+      "loss": 0.9271,
+      "step": 9442
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 1.56875e-05,
+      "loss": 0.9196,
+      "step": 9443
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 1.5678571428571427e-05,
+      "loss": 0.9607,
+      "step": 9444
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 1.5669642857142856e-05,
+      "loss": 0.9571,
+      "step": 9445
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 1.566071428571429e-05,
+      "loss": 0.9524,
+      "step": 9446
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 1.5651785714285717e-05,
+      "loss": 0.9223,
+      "step": 9447
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 1.5642857142857143e-05,
+      "loss": 0.9945,
+      "step": 9448
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 1.563392857142857e-05,
+      "loss": 0.9069,
+      "step": 9449
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.9713,
+      "step": 9450
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 1.561607142857143e-05,
+      "loss": 0.9634,
+      "step": 9451
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 1.5607142857142858e-05,
+      "loss": 1.0079,
+      "step": 9452
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 1.5598214285714287e-05,
+      "loss": 0.9314,
+      "step": 9453
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 1.5589285714285716e-05,
+      "loss": 0.9331,
+      "step": 9454
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 1.5580357142857145e-05,
+      "loss": 0.884,
+      "step": 9455
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.9803,
+      "step": 9456
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 1.5562500000000002e-05,
+      "loss": 0.9037,
+      "step": 9457
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 1.5553571428571428e-05,
+      "loss": 1.01,
+      "step": 9458
+    },
+    {
+      "epoch": 42.23,
+      "learning_rate": 1.5544642857142856e-05,
+      "loss": 0.9797,
+      "step": 9459
+    },
+    {
+      "epoch": 42.23,
+      "learning_rate": 1.5535714285714285e-05,
+      "loss": 0.9416,
+      "step": 9460
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 1.5526785714285718e-05,
+      "loss": 0.9299,
+      "step": 9461
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 1.5517857142857143e-05,
+      "loss": 0.9199,
+      "step": 9462
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 1.5508928571428572e-05,
+      "loss": 0.9057,
+      "step": 9463
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 1.55e-05,
+      "loss": 1.0054,
+      "step": 9464
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 1.549107142857143e-05,
+      "loss": 0.948,
+      "step": 9465
+    },
+    {
+      "epoch": 42.26,
+      "learning_rate": 1.548214285714286e-05,
+      "loss": 1.0133,
+      "step": 9466
+    },
+    {
+      "epoch": 42.26,
+      "learning_rate": 1.5473214285714284e-05,
+      "loss": 0.9421,
+      "step": 9467
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 1.5464285714285716e-05,
+      "loss": 0.958,
+      "step": 9468
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 1.5455357142857145e-05,
+      "loss": 0.9612,
+      "step": 9469
+    },
+    {
+      "epoch": 42.28,
+      "learning_rate": 1.5446428571428574e-05,
+      "loss": 0.954,
+      "step": 9470
+    },
+    {
+      "epoch": 42.28,
+      "learning_rate": 1.5437500000000003e-05,
+      "loss": 0.9127,
+      "step": 9471
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 1.5428571428571428e-05,
+      "loss": 0.8651,
+      "step": 9472
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 1.5419642857142857e-05,
+      "loss": 0.9532,
+      "step": 9473
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 1.5410714285714286e-05,
+      "loss": 0.9466,
+      "step": 9474
+    },
+    {
+      "epoch": 42.3,
+      "learning_rate": 1.5401785714285715e-05,
+      "loss": 1.004,
+      "step": 9475
+    },
+    {
+      "epoch": 42.3,
+      "learning_rate": 1.5392857142857143e-05,
+      "loss": 0.9424,
+      "step": 9476
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 1.5383928571428572e-05,
+      "loss": 0.9556,
+      "step": 9477
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 1.5375e-05,
+      "loss": 0.9827,
+      "step": 9478
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 1.536607142857143e-05,
+      "loss": 1.0029,
+      "step": 9479
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.9995,
+      "step": 9480
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 1.5348214285714284e-05,
+      "loss": 0.8987,
+      "step": 9481
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 1.5339285714285713e-05,
+      "loss": 0.9031,
+      "step": 9482
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 1.5330357142857145e-05,
+      "loss": 0.9206,
+      "step": 9483
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 1.5321428571428574e-05,
+      "loss": 0.957,
+      "step": 9484
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 1.5312500000000003e-05,
+      "loss": 1.0057,
+      "step": 9485
+    },
+    {
+      "epoch": 42.35,
+      "learning_rate": 1.530357142857143e-05,
+      "loss": 0.9809,
+      "step": 9486
+    },
+    {
+      "epoch": 42.35,
+      "learning_rate": 1.5294642857142857e-05,
+      "loss": 0.964,
+      "step": 9487
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 1.0124,
+      "step": 9488
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 1.5276785714285715e-05,
+      "loss": 0.9664,
+      "step": 9489
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 1.5267857142857144e-05,
+      "loss": 0.9758,
+      "step": 9490
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 1.5258928571428573e-05,
+      "loss": 0.9548,
+      "step": 9491
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 1.525e-05,
+      "loss": 0.9311,
+      "step": 9492
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 1.524107142857143e-05,
+      "loss": 0.9152,
+      "step": 9493
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 1.5232142857142859e-05,
+      "loss": 0.9638,
+      "step": 9494
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 1.5223214285714285e-05,
+      "loss": 0.9925,
+      "step": 9495
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 1.5214285714285715e-05,
+      "loss": 0.94,
+      "step": 9496
+    },
+    {
+      "epoch": 42.4,
+      "learning_rate": 1.5205357142857144e-05,
+      "loss": 1.0298,
+      "step": 9497
+    },
+    {
+      "epoch": 42.4,
+      "learning_rate": 1.5196428571428573e-05,
+      "loss": 0.976,
+      "step": 9498
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 1.5187500000000002e-05,
+      "loss": 1.0145,
+      "step": 9499
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 1.5178571428571429e-05,
+      "loss": 0.9667,
+      "step": 9500
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 1.5169642857142858e-05,
+      "loss": 0.9328,
+      "step": 9501
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 1.5160714285714286e-05,
+      "loss": 0.9622,
+      "step": 9502
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 1.5151785714285715e-05,
+      "loss": 0.9994,
+      "step": 9503
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.9087,
+      "step": 9504
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 1.5133928571428571e-05,
+      "loss": 0.9016,
+      "step": 9505
+    },
+    {
+      "epoch": 42.44,
+      "learning_rate": 1.5125e-05,
+      "loss": 0.9014,
+      "step": 9506
+    },
+    {
+      "epoch": 42.44,
+      "learning_rate": 1.5116071428571429e-05,
+      "loss": 0.9764,
+      "step": 9507
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 1.510714285714286e-05,
+      "loss": 0.902,
+      "step": 9508
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 1.5098214285714288e-05,
+      "loss": 0.9939,
+      "step": 9509
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 1.5089285714285714e-05,
+      "loss": 0.9891,
+      "step": 9510
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 1.5080357142857143e-05,
+      "loss": 0.9677,
+      "step": 9511
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 1.5071428571428573e-05,
+      "loss": 0.8727,
+      "step": 9512
+    },
+    {
+      "epoch": 42.47,
+      "learning_rate": 1.5062500000000002e-05,
+      "loss": 0.9411,
+      "step": 9513
+    },
+    {
+      "epoch": 42.47,
+      "learning_rate": 1.5053571428571427e-05,
+      "loss": 0.9618,
+      "step": 9514
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 1.5044642857142858e-05,
+      "loss": 0.9304,
+      "step": 9515
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 1.5035714285714287e-05,
+      "loss": 0.9366,
+      "step": 9516
+    },
+    {
+      "epoch": 42.49,
+      "learning_rate": 1.5026785714285716e-05,
+      "loss": 0.9718,
+      "step": 9517
+    },
+    {
+      "epoch": 42.49,
+      "learning_rate": 1.5017857142857144e-05,
+      "loss": 0.9165,
+      "step": 9518
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 1.5008928571428572e-05,
+      "loss": 0.8983,
+      "step": 9519
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 1.5e-05,
+      "loss": 0.9766,
+      "step": 9520
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 1.499107142857143e-05,
+      "loss": 0.9205,
+      "step": 9521
+    },
+    {
+      "epoch": 42.51,
+      "learning_rate": 1.4982142857142858e-05,
+      "loss": 0.9397,
+      "step": 9522
+    },
+    {
+      "epoch": 42.51,
+      "learning_rate": 1.4973214285714287e-05,
+      "loss": 0.8544,
+      "step": 9523
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 1.4964285714285714e-05,
+      "loss": 0.9362,
+      "step": 9524
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 1.4955357142857143e-05,
+      "loss": 0.93,
+      "step": 9525
+    },
+    {
+      "epoch": 42.53,
+      "learning_rate": 1.4946428571428572e-05,
+      "loss": 0.969,
+      "step": 9526
+    },
+    {
+      "epoch": 42.53,
+      "learning_rate": 1.4937500000000002e-05,
+      "loss": 0.9926,
+      "step": 9527
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 1.4928571428571431e-05,
+      "loss": 0.9897,
+      "step": 9528
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 1.4919642857142857e-05,
+      "loss": 0.9468,
+      "step": 9529
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 1.4910714285714285e-05,
+      "loss": 0.9185,
+      "step": 9530
+    },
+    {
+      "epoch": 42.55,
+      "learning_rate": 1.4901785714285716e-05,
+      "loss": 0.9157,
+      "step": 9531
+    },
+    {
+      "epoch": 42.55,
+      "learning_rate": 1.4892857142857145e-05,
+      "loss": 0.9204,
+      "step": 9532
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 1.488392857142857e-05,
+      "loss": 0.9185,
+      "step": 9533
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 1.4875e-05,
+      "loss": 0.9747,
+      "step": 9534
+    },
+    {
+      "epoch": 42.57,
+      "learning_rate": 1.486607142857143e-05,
+      "loss": 0.9562,
+      "step": 9535
+    },
+    {
+      "epoch": 42.57,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.9495,
+      "step": 9536
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 1.4848214285714287e-05,
+      "loss": 1.0228,
+      "step": 9537
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 1.4839285714285714e-05,
+      "loss": 0.9593,
+      "step": 9538
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 1.4830357142857143e-05,
+      "loss": 0.9827,
+      "step": 9539
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 1.4821428571428572e-05,
+      "loss": 0.9573,
+      "step": 9540
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 1.4812500000000001e-05,
+      "loss": 1.013,
+      "step": 9541
+    },
+    {
+      "epoch": 42.6,
+      "learning_rate": 1.480357142857143e-05,
+      "loss": 0.9061,
+      "step": 9542
+    },
+    {
+      "epoch": 42.6,
+      "learning_rate": 1.4794642857142857e-05,
+      "loss": 0.9561,
+      "step": 9543
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 1.4785714285714286e-05,
+      "loss": 0.926,
+      "step": 9544
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 1.4776785714285715e-05,
+      "loss": 0.9608,
+      "step": 9545
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 1.4767857142857145e-05,
+      "loss": 0.9792,
+      "step": 9546
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 1.475892857142857e-05,
+      "loss": 0.9879,
+      "step": 9547
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 1.475e-05,
+      "loss": 1.0819,
+      "step": 9548
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 1.474107142857143e-05,
+      "loss": 0.9396,
+      "step": 9549
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 1.4732142857142859e-05,
+      "loss": 0.9324,
+      "step": 9550
+    },
+    {
+      "epoch": 42.64,
+      "learning_rate": 1.4723214285714288e-05,
+      "loss": 0.9687,
+      "step": 9551
+    },
+    {
+      "epoch": 42.64,
+      "learning_rate": 1.4714285714285713e-05,
+      "loss": 0.9632,
+      "step": 9552
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 1.4705357142857144e-05,
+      "loss": 0.9516,
+      "step": 9553
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 1.4696428571428572e-05,
+      "loss": 0.9364,
+      "step": 9554
+    },
+    {
+      "epoch": 42.66,
+      "learning_rate": 1.4687500000000001e-05,
+      "loss": 0.9903,
+      "step": 9555
+    },
+    {
+      "epoch": 42.66,
+      "learning_rate": 1.467857142857143e-05,
+      "loss": 0.9309,
+      "step": 9556
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 1.4669642857142857e-05,
+      "loss": 0.9825,
+      "step": 9557
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 1.4660714285714286e-05,
+      "loss": 0.9236,
+      "step": 9558
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 1.4651785714285715e-05,
+      "loss": 0.9524,
+      "step": 9559
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.9854,
+      "step": 9560
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 1.4633928571428573e-05,
+      "loss": 0.9561,
+      "step": 9561
+    },
+    {
+      "epoch": 42.69,
+      "learning_rate": 1.4625e-05,
+      "loss": 0.9859,
+      "step": 9562
+    },
+    {
+      "epoch": 42.69,
+      "learning_rate": 1.4616071428571429e-05,
+      "loss": 1.0066,
+      "step": 9563
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 1.4607142857142857e-05,
+      "loss": 1.0071,
+      "step": 9564
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 1.4598214285714288e-05,
+      "loss": 0.9603,
+      "step": 9565
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 1.4589285714285713e-05,
+      "loss": 0.9076,
+      "step": 9566
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 1.4580357142857142e-05,
+      "loss": 0.9396,
+      "step": 9567
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 1.0167,
+      "step": 9568
+    },
+    {
+      "epoch": 42.72,
+      "learning_rate": 1.4562500000000002e-05,
+      "loss": 0.9638,
+      "step": 9569
+    },
+    {
+      "epoch": 42.72,
+      "learning_rate": 1.455357142857143e-05,
+      "loss": 0.9203,
+      "step": 9570
+    },
+    {
+      "epoch": 42.73,
+      "learning_rate": 1.4544642857142856e-05,
+      "loss": 0.9685,
+      "step": 9571
+    },
+    {
+      "epoch": 42.73,
+      "learning_rate": 1.4535714285714286e-05,
+      "loss": 0.8723,
+      "step": 9572
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 1.4526785714285715e-05,
+      "loss": 0.9822,
+      "step": 9573
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 1.4517857142857144e-05,
+      "loss": 0.9293,
+      "step": 9574
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 1.4508928571428573e-05,
+      "loss": 0.9925,
+      "step": 9575
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 1.45e-05,
+      "loss": 0.9055,
+      "step": 9576
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 1.4491071428571429e-05,
+      "loss": 0.9547,
+      "step": 9577
+    },
+    {
+      "epoch": 42.76,
+      "learning_rate": 1.4482142857142858e-05,
+      "loss": 0.8745,
+      "step": 9578
+    },
+    {
+      "epoch": 42.76,
+      "learning_rate": 1.4473214285714287e-05,
+      "loss": 0.9514,
+      "step": 9579
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 1.4464285714285717e-05,
+      "loss": 0.8978,
+      "step": 9580
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 1.4455357142857143e-05,
+      "loss": 0.8991,
+      "step": 9581
+    },
+    {
+      "epoch": 42.78,
+      "learning_rate": 1.4446428571428571e-05,
+      "loss": 0.9904,
+      "step": 9582
+    },
+    {
+      "epoch": 42.78,
+      "learning_rate": 1.44375e-05,
+      "loss": 0.8924,
+      "step": 9583
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 1.442857142857143e-05,
+      "loss": 0.9356,
+      "step": 9584
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 1.4419642857142856e-05,
+      "loss": 0.8834,
+      "step": 9585
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 1.4410714285714285e-05,
+      "loss": 0.9071,
+      "step": 9586
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 1.4401785714285716e-05,
+      "loss": 0.8955,
+      "step": 9587
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 1.4392857142857144e-05,
+      "loss": 0.9467,
+      "step": 9588
+    },
+    {
+      "epoch": 42.81,
+      "learning_rate": 1.4383928571428573e-05,
+      "loss": 0.8797,
+      "step": 9589
+    },
+    {
+      "epoch": 42.81,
+      "learning_rate": 1.4374999999999999e-05,
+      "loss": 0.9068,
+      "step": 9590
+    },
+    {
+      "epoch": 42.82,
+      "learning_rate": 1.436607142857143e-05,
+      "loss": 0.9497,
+      "step": 9591
+    },
+    {
+      "epoch": 42.82,
+      "learning_rate": 1.4357142857142858e-05,
+      "loss": 0.9679,
+      "step": 9592
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 1.4348214285714287e-05,
+      "loss": 0.9892,
+      "step": 9593
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 1.4339285714285716e-05,
+      "loss": 0.9888,
+      "step": 9594
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 1.4330357142857143e-05,
+      "loss": 0.9365,
+      "step": 9595
+    },
+    {
+      "epoch": 42.84,
+      "learning_rate": 1.4321428571428572e-05,
+      "loss": 0.9242,
+      "step": 9596
+    },
+    {
+      "epoch": 42.84,
+      "learning_rate": 1.43125e-05,
+      "loss": 0.9063,
+      "step": 9597
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 1.430357142857143e-05,
+      "loss": 0.9752,
+      "step": 9598
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 1.4294642857142857e-05,
+      "loss": 0.9835,
+      "step": 9599
+    },
+    {
+      "epoch": 42.86,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.9659,
+      "step": 9600
+    },
+    {
+      "epoch": 42.86,
+      "eval_loss": 1.1119632720947266,
+      "eval_runtime": 44.2815,
+      "eval_samples_per_second": 295.699,
+      "eval_steps_per_second": 1.558,
+      "step": 9600
+    },
+    {
+      "epoch": 42.86,
+      "learning_rate": 1.4276785714285714e-05,
+      "loss": 0.9934,
+      "step": 9601
+    },
+    {
+      "epoch": 42.87,
+      "learning_rate": 1.4267857142857143e-05,
+      "loss": 0.9195,
+      "step": 9602
+    },
+    {
+      "epoch": 42.87,
+      "learning_rate": 1.4258928571428574e-05,
+      "loss": 0.9229,
+      "step": 9603
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 0.9625,
+      "step": 9604
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 1.4241071428571428e-05,
+      "loss": 0.9531,
+      "step": 9605
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 1.4232142857142858e-05,
+      "loss": 0.9198,
+      "step": 9606
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 1.4223214285714287e-05,
+      "loss": 0.9207,
+      "step": 9607
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 1.4214285714285716e-05,
+      "loss": 0.8967,
+      "step": 9608
+    },
+    {
+      "epoch": 42.9,
+      "learning_rate": 1.4205357142857143e-05,
+      "loss": 0.9219,
+      "step": 9609
+    },
+    {
+      "epoch": 42.9,
+      "learning_rate": 1.4196428571428572e-05,
+      "loss": 1.021,
+      "step": 9610
+    },
+    {
+      "epoch": 42.91,
+      "learning_rate": 1.4187500000000001e-05,
+      "loss": 0.9262,
+      "step": 9611
+    },
+    {
+      "epoch": 42.91,
+      "learning_rate": 1.417857142857143e-05,
+      "loss": 0.9772,
+      "step": 9612
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 1.4169642857142859e-05,
+      "loss": 0.9039,
+      "step": 9613
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 1.4160714285714286e-05,
+      "loss": 0.923,
+      "step": 9614
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 1.4151785714285715e-05,
+      "loss": 0.9317,
+      "step": 9615
+    },
+    {
+      "epoch": 42.93,
+      "learning_rate": 1.4142857142857143e-05,
+      "loss": 0.9609,
+      "step": 9616
+    },
+    {
+      "epoch": 42.93,
+      "learning_rate": 1.4133928571428572e-05,
+      "loss": 0.999,
+      "step": 9617
+    },
+    {
+      "epoch": 42.94,
+      "learning_rate": 1.4125e-05,
+      "loss": 1.0291,
+      "step": 9618
+    },
+    {
+      "epoch": 42.94,
+      "learning_rate": 1.4116071428571428e-05,
+      "loss": 0.9466,
+      "step": 9619
+    },
+    {
+      "epoch": 42.95,
+      "learning_rate": 1.4107142857142857e-05,
+      "loss": 0.9336,
+      "step": 9620
+    },
+    {
+      "epoch": 42.95,
+      "learning_rate": 1.4098214285714288e-05,
+      "loss": 0.9045,
+      "step": 9621
+    },
+    {
+      "epoch": 42.96,
+      "learning_rate": 1.4089285714285716e-05,
+      "loss": 0.934,
+      "step": 9622
+    },
+    {
+      "epoch": 42.96,
+      "learning_rate": 1.4080357142857142e-05,
+      "loss": 0.9013,
+      "step": 9623
+    },
+    {
+      "epoch": 42.96,
+      "learning_rate": 1.407142857142857e-05,
+      "loss": 0.9334,
+      "step": 9624
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.912,
+      "step": 9625
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 1.405357142857143e-05,
+      "loss": 0.9321,
+      "step": 9626
+    },
+    {
+      "epoch": 42.98,
+      "learning_rate": 1.4044642857142859e-05,
+      "loss": 0.9645,
+      "step": 9627
+    },
+    {
+      "epoch": 42.98,
+      "learning_rate": 1.4035714285714286e-05,
+      "loss": 0.9753,
+      "step": 9628
+    },
+    {
+      "epoch": 42.99,
+      "learning_rate": 1.4026785714285715e-05,
+      "loss": 1.0058,
+      "step": 9629
+    },
+    {
+      "epoch": 42.99,
+      "learning_rate": 1.4017857142857144e-05,
+      "loss": 0.9849,
+      "step": 9630
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 1.4008928571428573e-05,
+      "loss": 0.949,
+      "step": 9631
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.007,
+      "step": 9632
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 1.3991071428571429e-05,
+      "loss": 0.9496,
+      "step": 9633
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 1.3982142857142857e-05,
+      "loss": 1.0676,
+      "step": 9634
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 1.3973214285714286e-05,
+      "loss": 0.924,
+      "step": 9635
+    },
+    {
+      "epoch": 43.02,
+      "learning_rate": 1.3964285714285715e-05,
+      "loss": 0.9632,
+      "step": 9636
+    },
+    {
+      "epoch": 43.02,
+      "learning_rate": 1.3955357142857142e-05,
+      "loss": 0.9731,
+      "step": 9637
+    },
+    {
+      "epoch": 43.03,
+      "learning_rate": 1.3946428571428571e-05,
+      "loss": 0.9512,
+      "step": 9638
+    },
+    {
+      "epoch": 43.03,
+      "learning_rate": 1.39375e-05,
+      "loss": 0.9495,
+      "step": 9639
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 1.392857142857143e-05,
+      "loss": 0.935,
+      "step": 9640
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 1.391964285714286e-05,
+      "loss": 0.8879,
+      "step": 9641
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 1.3910714285714285e-05,
+      "loss": 1.0206,
+      "step": 9642
+    },
+    {
+      "epoch": 43.05,
+      "learning_rate": 1.3901785714285714e-05,
+      "loss": 0.9836,
+      "step": 9643
+    },
+    {
+      "epoch": 43.05,
+      "learning_rate": 1.3892857142857144e-05,
+      "loss": 0.9126,
+      "step": 9644
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 1.3883928571428573e-05,
+      "loss": 0.89,
+      "step": 9645
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 1.3875000000000002e-05,
+      "loss": 0.8825,
+      "step": 9646
+    },
+    {
+      "epoch": 43.07,
+      "learning_rate": 1.3866071428571429e-05,
+      "loss": 0.9992,
+      "step": 9647
+    },
+    {
+      "epoch": 43.07,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.9173,
+      "step": 9648
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 1.3848214285714287e-05,
+      "loss": 0.9357,
+      "step": 9649
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 1.3839285714285715e-05,
+      "loss": 1.0309,
+      "step": 9650
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 1.3830357142857144e-05,
+      "loss": 0.8914,
+      "step": 9651
+    },
+    {
+      "epoch": 43.09,
+      "learning_rate": 1.3821428571428571e-05,
+      "loss": 0.9391,
+      "step": 9652
+    },
+    {
+      "epoch": 43.09,
+      "learning_rate": 1.38125e-05,
+      "loss": 0.9836,
+      "step": 9653
+    },
+    {
+      "epoch": 43.1,
+      "learning_rate": 1.380357142857143e-05,
+      "loss": 1.0056,
+      "step": 9654
+    },
+    {
+      "epoch": 43.1,
+      "learning_rate": 1.3794642857142858e-05,
+      "loss": 0.8618,
+      "step": 9655
+    },
+    {
+      "epoch": 43.11,
+      "learning_rate": 1.3785714285714285e-05,
+      "loss": 0.9407,
+      "step": 9656
+    },
+    {
+      "epoch": 43.11,
+      "learning_rate": 1.3776785714285714e-05,
+      "loss": 0.9862,
+      "step": 9657
+    },
+    {
+      "epoch": 43.12,
+      "learning_rate": 1.3767857142857143e-05,
+      "loss": 0.949,
+      "step": 9658
+    },
+    {
+      "epoch": 43.12,
+      "learning_rate": 1.3758928571428573e-05,
+      "loss": 0.9063,
+      "step": 9659
+    },
+    {
+      "epoch": 43.12,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 0.9352,
+      "step": 9660
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 1.3741071428571428e-05,
+      "loss": 0.9716,
+      "step": 9661
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 1.3732142857142856e-05,
+      "loss": 0.9642,
+      "step": 9662
+    },
+    {
+      "epoch": 43.14,
+      "learning_rate": 1.3723214285714287e-05,
+      "loss": 0.886,
+      "step": 9663
+    },
+    {
+      "epoch": 43.14,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.9618,
+      "step": 9664
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 1.3705357142857145e-05,
+      "loss": 0.9561,
+      "step": 9665
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 1.3696428571428572e-05,
+      "loss": 0.9117,
+      "step": 9666
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 1.36875e-05,
+      "loss": 0.9849,
+      "step": 9667
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 1.367857142857143e-05,
+      "loss": 0.9507,
+      "step": 9668
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 1.3669642857142858e-05,
+      "loss": 0.9182,
+      "step": 9669
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 1.3660714285714285e-05,
+      "loss": 0.9325,
+      "step": 9670
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 1.3651785714285714e-05,
+      "loss": 0.9817,
+      "step": 9671
+    },
+    {
+      "epoch": 43.18,
+      "learning_rate": 1.3642857142857143e-05,
+      "loss": 0.9363,
+      "step": 9672
+    },
+    {
+      "epoch": 43.18,
+      "learning_rate": 1.3633928571428572e-05,
+      "loss": 0.9595,
+      "step": 9673
+    },
+    {
+      "epoch": 43.19,
+      "learning_rate": 1.3625e-05,
+      "loss": 0.8714,
+      "step": 9674
+    },
+    {
+      "epoch": 43.19,
+      "learning_rate": 1.3616071428571428e-05,
+      "loss": 0.9412,
+      "step": 9675
+    },
+    {
+      "epoch": 43.2,
+      "learning_rate": 1.3607142857142857e-05,
+      "loss": 0.9595,
+      "step": 9676
+    },
+    {
+      "epoch": 43.2,
+      "learning_rate": 1.3598214285714286e-05,
+      "loss": 0.9509,
+      "step": 9677
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 1.3589285714285716e-05,
+      "loss": 0.9269,
+      "step": 9678
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 1.3580357142857145e-05,
+      "loss": 1.0098,
+      "step": 9679
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 1.357142857142857e-05,
+      "loss": 0.8764,
+      "step": 9680
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 1.3562500000000001e-05,
+      "loss": 0.9186,
+      "step": 9681
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 1.355357142857143e-05,
+      "loss": 0.9702,
+      "step": 9682
+    },
+    {
+      "epoch": 43.23,
+      "learning_rate": 1.3544642857142859e-05,
+      "loss": 0.945,
+      "step": 9683
+    },
+    {
+      "epoch": 43.23,
+      "learning_rate": 1.3535714285714288e-05,
+      "loss": 0.9916,
+      "step": 9684
+    },
+    {
+      "epoch": 43.24,
+      "learning_rate": 1.3526785714285715e-05,
+      "loss": 0.9375,
+      "step": 9685
+    },
+    {
+      "epoch": 43.24,
+      "learning_rate": 1.3517857142857144e-05,
+      "loss": 0.8837,
+      "step": 9686
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 1.3508928571428572e-05,
+      "loss": 0.9535,
+      "step": 9687
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.9282,
+      "step": 9688
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 1.3491071428571428e-05,
+      "loss": 0.93,
+      "step": 9689
+    },
+    {
+      "epoch": 43.26,
+      "learning_rate": 1.3482142857142857e-05,
+      "loss": 1.0064,
+      "step": 9690
+    },
+    {
+      "epoch": 43.26,
+      "learning_rate": 1.3473214285714286e-05,
+      "loss": 0.8907,
+      "step": 9691
+    },
+    {
+      "epoch": 43.27,
+      "learning_rate": 1.3464285714285715e-05,
+      "loss": 0.966,
+      "step": 9692
+    },
+    {
+      "epoch": 43.27,
+      "learning_rate": 1.3455357142857145e-05,
+      "loss": 0.9322,
+      "step": 9693
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 1.344642857142857e-05,
+      "loss": 0.9946,
+      "step": 9694
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 1.34375e-05,
+      "loss": 0.9387,
+      "step": 9695
+    },
+    {
+      "epoch": 43.29,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.9796,
+      "step": 9696
+    },
+    {
+      "epoch": 43.29,
+      "learning_rate": 1.3419642857142859e-05,
+      "loss": 0.9172,
+      "step": 9697
+    },
+    {
+      "epoch": 43.29,
+      "learning_rate": 1.3410714285714288e-05,
+      "loss": 0.9817,
+      "step": 9698
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 1.3401785714285713e-05,
+      "loss": 0.9637,
+      "step": 9699
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 1.3392857142857144e-05,
+      "loss": 0.9851,
+      "step": 9700
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 1.3383928571428573e-05,
+      "loss": 0.9089,
+      "step": 9701
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 1.3375000000000002e-05,
+      "loss": 0.9028,
+      "step": 9702
+    },
+    {
+      "epoch": 43.32,
+      "learning_rate": 1.336607142857143e-05,
+      "loss": 0.9444,
+      "step": 9703
+    },
+    {
+      "epoch": 43.32,
+      "learning_rate": 1.3357142857142858e-05,
+      "loss": 0.8729,
+      "step": 9704
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 1.3348214285714286e-05,
+      "loss": 0.9732,
+      "step": 9705
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 1.3339285714285715e-05,
+      "loss": 0.9055,
+      "step": 9706
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 1.3330357142857144e-05,
+      "loss": 0.9619,
+      "step": 9707
+    },
+    {
+      "epoch": 43.34,
+      "learning_rate": 1.3321428571428571e-05,
+      "loss": 0.9194,
+      "step": 9708
+    },
+    {
+      "epoch": 43.34,
+      "learning_rate": 1.33125e-05,
+      "loss": 0.9155,
+      "step": 9709
+    },
+    {
+      "epoch": 43.35,
+      "learning_rate": 1.3303571428571429e-05,
+      "loss": 0.9754,
+      "step": 9710
+    },
+    {
+      "epoch": 43.35,
+      "learning_rate": 1.3294642857142858e-05,
+      "loss": 0.9893,
+      "step": 9711
+    },
+    {
+      "epoch": 43.36,
+      "learning_rate": 1.3285714285714288e-05,
+      "loss": 0.97,
+      "step": 9712
+    },
+    {
+      "epoch": 43.36,
+      "learning_rate": 1.3276785714285714e-05,
+      "loss": 0.9848,
+      "step": 9713
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 1.3267857142857143e-05,
+      "loss": 0.9456,
+      "step": 9714
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 1.3258928571428571e-05,
+      "loss": 1.0125,
+      "step": 9715
+    },
+    {
+      "epoch": 43.38,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 0.9185,
+      "step": 9716
+    },
+    {
+      "epoch": 43.38,
+      "learning_rate": 1.324107142857143e-05,
+      "loss": 0.9301,
+      "step": 9717
+    },
+    {
+      "epoch": 43.38,
+      "learning_rate": 1.3232142857142856e-05,
+      "loss": 0.9207,
+      "step": 9718
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 1.3223214285714287e-05,
+      "loss": 0.9541,
+      "step": 9719
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.912,
+      "step": 9720
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 1.3205357142857144e-05,
+      "loss": 0.9175,
+      "step": 9721
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 1.3196428571428572e-05,
+      "loss": 0.9265,
+      "step": 9722
+    },
+    {
+      "epoch": 43.41,
+      "learning_rate": 1.31875e-05,
+      "loss": 0.9215,
+      "step": 9723
+    },
+    {
+      "epoch": 43.41,
+      "learning_rate": 1.317857142857143e-05,
+      "loss": 0.9334,
+      "step": 9724
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 1.3169642857142858e-05,
+      "loss": 0.9098,
+      "step": 9725
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 1.3160714285714287e-05,
+      "loss": 0.9514,
+      "step": 9726
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 1.3151785714285714e-05,
+      "loss": 0.9816,
+      "step": 9727
+    },
+    {
+      "epoch": 43.43,
+      "learning_rate": 1.3142857142857143e-05,
+      "loss": 0.9703,
+      "step": 9728
+    },
+    {
+      "epoch": 43.43,
+      "learning_rate": 1.3133928571428572e-05,
+      "loss": 1.0305,
+      "step": 9729
+    },
+    {
+      "epoch": 43.44,
+      "learning_rate": 1.3125e-05,
+      "loss": 0.8984,
+      "step": 9730
+    },
+    {
+      "epoch": 43.44,
+      "learning_rate": 1.3116071428571431e-05,
+      "loss": 0.9646,
+      "step": 9731
+    },
+    {
+      "epoch": 43.45,
+      "learning_rate": 1.3107142857142857e-05,
+      "loss": 0.8757,
+      "step": 9732
+    },
+    {
+      "epoch": 43.45,
+      "learning_rate": 1.3098214285714285e-05,
+      "loss": 1.0243,
+      "step": 9733
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 1.3089285714285714e-05,
+      "loss": 0.9327,
+      "step": 9734
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 1.3080357142857145e-05,
+      "loss": 0.968,
+      "step": 9735
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 1.3071428571428574e-05,
+      "loss": 0.9439,
+      "step": 9736
+    },
+    {
+      "epoch": 43.47,
+      "learning_rate": 1.3062499999999999e-05,
+      "loss": 0.9164,
+      "step": 9737
+    },
+    {
+      "epoch": 43.47,
+      "learning_rate": 1.305357142857143e-05,
+      "loss": 0.948,
+      "step": 9738
+    },
+    {
+      "epoch": 43.48,
+      "learning_rate": 1.3044642857142858e-05,
+      "loss": 1.0513,
+      "step": 9739
+    },
+    {
+      "epoch": 43.48,
+      "learning_rate": 1.3035714285714287e-05,
+      "loss": 0.8949,
+      "step": 9740
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 1.3026785714285714e-05,
+      "loss": 0.9827,
+      "step": 9741
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 1.3017857142857143e-05,
+      "loss": 0.9233,
+      "step": 9742
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 1.3008928571428572e-05,
+      "loss": 0.9864,
+      "step": 9743
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.0002,
+      "step": 9744
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 1.299107142857143e-05,
+      "loss": 0.9246,
+      "step": 9745
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 1.2982142857142857e-05,
+      "loss": 0.9709,
+      "step": 9746
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 1.2973214285714286e-05,
+      "loss": 0.9733,
+      "step": 9747
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 1.2964285714285715e-05,
+      "loss": 0.9609,
+      "step": 9748
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 1.2955357142857143e-05,
+      "loss": 0.9234,
+      "step": 9749
+    },
+    {
+      "epoch": 43.53,
+      "learning_rate": 1.2946428571428574e-05,
+      "loss": 0.9341,
+      "step": 9750
+    },
+    {
+      "epoch": 43.53,
+      "learning_rate": 1.29375e-05,
+      "loss": 0.911,
+      "step": 9751
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 1.2928571428571428e-05,
+      "loss": 0.9075,
+      "step": 9752
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 1.2919642857142859e-05,
+      "loss": 0.9046,
+      "step": 9753
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 1.2910714285714288e-05,
+      "loss": 0.9382,
+      "step": 9754
+    },
+    {
+      "epoch": 43.55,
+      "learning_rate": 1.2901785714285716e-05,
+      "loss": 0.8447,
+      "step": 9755
+    },
+    {
+      "epoch": 43.55,
+      "learning_rate": 1.2892857142857142e-05,
+      "loss": 0.9224,
+      "step": 9756
+    },
+    {
+      "epoch": 43.56,
+      "learning_rate": 1.2883928571428572e-05,
+      "loss": 0.9187,
+      "step": 9757
+    },
+    {
+      "epoch": 43.56,
+      "learning_rate": 1.2875000000000001e-05,
+      "loss": 1.0154,
+      "step": 9758
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 1.286607142857143e-05,
+      "loss": 1.0137,
+      "step": 9759
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.9829,
+      "step": 9760
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 1.2848214285714286e-05,
+      "loss": 0.9728,
+      "step": 9761
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 1.2839285714285715e-05,
+      "loss": 0.971,
+      "step": 9762
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 1.2830357142857144e-05,
+      "loss": 0.9229,
+      "step": 9763
+    },
+    {
+      "epoch": 43.59,
+      "learning_rate": 1.2821428571428573e-05,
+      "loss": 0.9412,
+      "step": 9764
+    },
+    {
+      "epoch": 43.59,
+      "learning_rate": 1.28125e-05,
+      "loss": 0.9787,
+      "step": 9765
+    },
+    {
+      "epoch": 43.6,
+      "learning_rate": 1.2803571428571429e-05,
+      "loss": 0.8493,
+      "step": 9766
+    },
+    {
+      "epoch": 43.6,
+      "learning_rate": 1.2794642857142857e-05,
+      "loss": 0.9415,
+      "step": 9767
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 1.2785714285714286e-05,
+      "loss": 0.8822,
+      "step": 9768
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 1.2776785714285717e-05,
+      "loss": 0.9485,
+      "step": 9769
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 1.2767857142857142e-05,
+      "loss": 0.9683,
+      "step": 9770
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 1.2758928571428571e-05,
+      "loss": 0.9064,
+      "step": 9771
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 0.9464,
+      "step": 9772
+    },
+    {
+      "epoch": 43.63,
+      "learning_rate": 1.274107142857143e-05,
+      "loss": 0.9283,
+      "step": 9773
+    },
+    {
+      "epoch": 43.63,
+      "learning_rate": 1.2732142857142856e-05,
+      "loss": 0.9533,
+      "step": 9774
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 1.2723214285714285e-05,
+      "loss": 0.9784,
+      "step": 9775
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.9578,
+      "step": 9776
+    },
+    {
+      "epoch": 43.65,
+      "learning_rate": 1.2705357142857144e-05,
+      "loss": 1.0066,
+      "step": 9777
+    },
+    {
+      "epoch": 43.65,
+      "learning_rate": 1.2696428571428573e-05,
+      "loss": 0.9364,
+      "step": 9778
+    },
+    {
+      "epoch": 43.66,
+      "learning_rate": 1.26875e-05,
+      "loss": 0.934,
+      "step": 9779
+    },
+    {
+      "epoch": 43.66,
+      "learning_rate": 1.2678571428571429e-05,
+      "loss": 0.9532,
+      "step": 9780
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 1.2669642857142858e-05,
+      "loss": 0.9354,
+      "step": 9781
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 1.2660714285714287e-05,
+      "loss": 0.8799,
+      "step": 9782
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 1.2651785714285715e-05,
+      "loss": 0.9491,
+      "step": 9783
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 1.2642857142857143e-05,
+      "loss": 1.0182,
+      "step": 9784
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 1.2633928571428571e-05,
+      "loss": 0.9906,
+      "step": 9785
+    },
+    {
+      "epoch": 43.69,
+      "learning_rate": 1.2625e-05,
+      "loss": 0.8879,
+      "step": 9786
+    },
+    {
+      "epoch": 43.69,
+      "learning_rate": 1.2616071428571429e-05,
+      "loss": 0.9757,
+      "step": 9787
+    },
+    {
+      "epoch": 43.7,
+      "learning_rate": 1.260714285714286e-05,
+      "loss": 1.0286,
+      "step": 9788
+    },
+    {
+      "epoch": 43.7,
+      "learning_rate": 1.2598214285714285e-05,
+      "loss": 0.9593,
+      "step": 9789
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 1.2589285714285714e-05,
+      "loss": 0.9645,
+      "step": 9790
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 1.2580357142857144e-05,
+      "loss": 0.9984,
+      "step": 9791
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 1.2571428571428573e-05,
+      "loss": 0.978,
+      "step": 9792
+    },
+    {
+      "epoch": 43.72,
+      "learning_rate": 1.2562499999999999e-05,
+      "loss": 0.8222,
+      "step": 9793
+    },
+    {
+      "epoch": 43.72,
+      "learning_rate": 1.255357142857143e-05,
+      "loss": 0.9645,
+      "step": 9794
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 1.2544642857142858e-05,
+      "loss": 0.9174,
+      "step": 9795
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 1.2535714285714287e-05,
+      "loss": 0.9881,
+      "step": 9796
+    },
+    {
+      "epoch": 43.74,
+      "learning_rate": 1.2526785714285716e-05,
+      "loss": 0.9619,
+      "step": 9797
+    },
+    {
+      "epoch": 43.74,
+      "learning_rate": 1.2517857142857143e-05,
+      "loss": 0.9691,
+      "step": 9798
+    },
+    {
+      "epoch": 43.75,
+      "learning_rate": 1.2508928571428572e-05,
+      "loss": 0.9142,
+      "step": 9799
+    },
+    {
+      "epoch": 43.75,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9458,
+      "step": 9800
+    },
+    {
+      "epoch": 43.75,
+      "learning_rate": 1.249107142857143e-05,
+      "loss": 0.9577,
+      "step": 9801
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 1.2482142857142858e-05,
+      "loss": 0.9352,
+      "step": 9802
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 1.2473214285714287e-05,
+      "loss": 0.9433,
+      "step": 9803
+    },
+    {
+      "epoch": 43.77,
+      "learning_rate": 1.2464285714285714e-05,
+      "loss": 0.9654,
+      "step": 9804
+    },
+    {
+      "epoch": 43.77,
+      "learning_rate": 1.2455357142857143e-05,
+      "loss": 0.8998,
+      "step": 9805
+    },
+    {
+      "epoch": 43.78,
+      "learning_rate": 1.2446428571428572e-05,
+      "loss": 0.9104,
+      "step": 9806
+    },
+    {
+      "epoch": 43.78,
+      "learning_rate": 1.24375e-05,
+      "loss": 0.9544,
+      "step": 9807
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 1.0093,
+      "step": 9808
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 1.2419642857142857e-05,
+      "loss": 1.0319,
+      "step": 9809
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 1.2410714285714287e-05,
+      "loss": 0.9521,
+      "step": 9810
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 1.2401785714285714e-05,
+      "loss": 0.9673,
+      "step": 9811
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 1.2392857142857143e-05,
+      "loss": 1.0371,
+      "step": 9812
+    },
+    {
+      "epoch": 43.81,
+      "learning_rate": 1.2383928571428572e-05,
+      "loss": 0.9169,
+      "step": 9813
+    },
+    {
+      "epoch": 43.81,
+      "learning_rate": 1.2375000000000001e-05,
+      "loss": 0.9512,
+      "step": 9814
+    },
+    {
+      "epoch": 43.82,
+      "learning_rate": 1.236607142857143e-05,
+      "loss": 0.9981,
+      "step": 9815
+    },
+    {
+      "epoch": 43.82,
+      "learning_rate": 1.2357142857142857e-05,
+      "loss": 0.8926,
+      "step": 9816
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 1.2348214285714286e-05,
+      "loss": 0.9035,
+      "step": 9817
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 1.2339285714285715e-05,
+      "loss": 0.9675,
+      "step": 9818
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 1.2330357142857143e-05,
+      "loss": 0.939,
+      "step": 9819
+    },
+    {
+      "epoch": 43.84,
+      "learning_rate": 1.2321428571428572e-05,
+      "loss": 0.8686,
+      "step": 9820
+    },
+    {
+      "epoch": 43.84,
+      "learning_rate": 1.2312500000000001e-05,
+      "loss": 0.9145,
+      "step": 9821
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 1.230357142857143e-05,
+      "loss": 0.9643,
+      "step": 9822
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 1.2294642857142857e-05,
+      "loss": 0.9625,
+      "step": 9823
+    },
+    {
+      "epoch": 43.86,
+      "learning_rate": 1.2285714285714286e-05,
+      "loss": 0.9298,
+      "step": 9824
+    },
+    {
+      "epoch": 43.86,
+      "learning_rate": 1.2276785714285715e-05,
+      "loss": 0.8752,
+      "step": 9825
+    },
+    {
+      "epoch": 43.87,
+      "learning_rate": 1.2267857142857144e-05,
+      "loss": 0.9771,
+      "step": 9826
+    },
+    {
+      "epoch": 43.87,
+      "learning_rate": 1.2258928571428572e-05,
+      "loss": 0.9319,
+      "step": 9827
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 1.225e-05,
+      "loss": 0.9338,
+      "step": 9828
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 1.224107142857143e-05,
+      "loss": 0.9221,
+      "step": 9829
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 1.2232142857142857e-05,
+      "loss": 0.902,
+      "step": 9830
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 1.2223214285714286e-05,
+      "loss": 1.0282,
+      "step": 9831
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 1.2214285714285715e-05,
+      "loss": 0.9286,
+      "step": 9832
+    },
+    {
+      "epoch": 43.9,
+      "learning_rate": 1.2205357142857144e-05,
+      "loss": 0.9086,
+      "step": 9833
+    },
+    {
+      "epoch": 43.9,
+      "learning_rate": 1.2196428571428573e-05,
+      "loss": 0.9153,
+      "step": 9834
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 1.21875e-05,
+      "loss": 0.977,
+      "step": 9835
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 1.2178571428571429e-05,
+      "loss": 0.9497,
+      "step": 9836
+    },
+    {
+      "epoch": 43.92,
+      "learning_rate": 1.2169642857142857e-05,
+      "loss": 0.948,
+      "step": 9837
+    },
+    {
+      "epoch": 43.92,
+      "learning_rate": 1.2160714285714286e-05,
+      "loss": 0.9453,
+      "step": 9838
+    },
+    {
+      "epoch": 43.92,
+      "learning_rate": 1.2151785714285715e-05,
+      "loss": 0.9505,
+      "step": 9839
+    },
+    {
+      "epoch": 43.93,
+      "learning_rate": 1.2142857142857144e-05,
+      "loss": 0.9489,
+      "step": 9840
+    },
+    {
+      "epoch": 43.93,
+      "learning_rate": 1.2133928571428573e-05,
+      "loss": 0.9057,
+      "step": 9841
+    },
+    {
+      "epoch": 43.94,
+      "learning_rate": 1.2125e-05,
+      "loss": 0.9916,
+      "step": 9842
+    },
+    {
+      "epoch": 43.94,
+      "learning_rate": 1.2116071428571429e-05,
+      "loss": 0.999,
+      "step": 9843
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 1.2107142857142858e-05,
+      "loss": 0.9071,
+      "step": 9844
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 1.2098214285714286e-05,
+      "loss": 0.9254,
+      "step": 9845
+    },
+    {
+      "epoch": 43.96,
+      "learning_rate": 1.2089285714285715e-05,
+      "loss": 0.9242,
+      "step": 9846
+    },
+    {
+      "epoch": 43.96,
+      "learning_rate": 1.2080357142857142e-05,
+      "loss": 0.912,
+      "step": 9847
+    },
+    {
+      "epoch": 43.96,
+      "learning_rate": 1.2071428571428573e-05,
+      "loss": 0.9204,
+      "step": 9848
+    },
+    {
+      "epoch": 43.97,
+      "learning_rate": 1.20625e-05,
+      "loss": 1.0121,
+      "step": 9849
+    },
+    {
+      "epoch": 43.97,
+      "learning_rate": 1.2053571428571429e-05,
+      "loss": 0.9599,
+      "step": 9850
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 1.2044642857142858e-05,
+      "loss": 0.9121,
+      "step": 9851
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 1.2035714285714287e-05,
+      "loss": 0.9456,
+      "step": 9852
+    },
+    {
+      "epoch": 43.99,
+      "learning_rate": 1.2026785714285715e-05,
+      "loss": 0.9846,
+      "step": 9853
+    },
+    {
+      "epoch": 43.99,
+      "learning_rate": 1.2017857142857143e-05,
+      "loss": 1.0242,
+      "step": 9854
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 1.2008928571428573e-05,
+      "loss": 0.9546,
+      "step": 9855
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9477,
+      "step": 9856
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 1.1991071428571429e-05,
+      "loss": 0.9858,
+      "step": 9857
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 1.1982142857142856e-05,
+      "loss": 0.9681,
+      "step": 9858
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 1.1973214285714287e-05,
+      "loss": 0.9461,
+      "step": 9859
+    },
+    {
+      "epoch": 44.02,
+      "learning_rate": 1.1964285714285716e-05,
+      "loss": 0.9504,
+      "step": 9860
+    },
+    {
+      "epoch": 44.02,
+      "learning_rate": 1.1955357142857143e-05,
+      "loss": 0.9829,
+      "step": 9861
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 1.1946428571428572e-05,
+      "loss": 0.939,
+      "step": 9862
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 1.19375e-05,
+      "loss": 0.9767,
+      "step": 9863
+    },
+    {
+      "epoch": 44.04,
+      "learning_rate": 1.192857142857143e-05,
+      "loss": 0.9408,
+      "step": 9864
+    },
+    {
+      "epoch": 44.04,
+      "learning_rate": 1.1919642857142858e-05,
+      "loss": 0.9437,
+      "step": 9865
+    },
+    {
+      "epoch": 44.04,
+      "learning_rate": 1.1910714285714287e-05,
+      "loss": 0.9194,
+      "step": 9866
+    },
+    {
+      "epoch": 44.05,
+      "learning_rate": 1.1901785714285716e-05,
+      "loss": 0.8962,
+      "step": 9867
+    },
+    {
+      "epoch": 44.05,
+      "learning_rate": 1.1892857142857143e-05,
+      "loss": 0.962,
+      "step": 9868
+    },
+    {
+      "epoch": 44.06,
+      "learning_rate": 1.1883928571428572e-05,
+      "loss": 0.8659,
+      "step": 9869
+    },
+    {
+      "epoch": 44.06,
+      "learning_rate": 1.1875e-05,
+      "loss": 0.9787,
+      "step": 9870
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 1.186607142857143e-05,
+      "loss": 1.0093,
+      "step": 9871
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.9581,
+      "step": 9872
+    },
+    {
+      "epoch": 44.08,
+      "learning_rate": 1.1848214285714285e-05,
+      "loss": 0.9423,
+      "step": 9873
+    },
+    {
+      "epoch": 44.08,
+      "learning_rate": 1.1839285714285716e-05,
+      "loss": 0.9786,
+      "step": 9874
+    },
+    {
+      "epoch": 44.08,
+      "learning_rate": 1.1830357142857143e-05,
+      "loss": 0.9647,
+      "step": 9875
+    },
+    {
+      "epoch": 44.09,
+      "learning_rate": 1.1821428571428572e-05,
+      "loss": 0.9193,
+      "step": 9876
+    },
+    {
+      "epoch": 44.09,
+      "learning_rate": 1.1812499999999999e-05,
+      "loss": 0.9079,
+      "step": 9877
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 1.180357142857143e-05,
+      "loss": 0.8808,
+      "step": 9878
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 1.1794642857142859e-05,
+      "loss": 0.9179,
+      "step": 9879
+    },
+    {
+      "epoch": 44.11,
+      "learning_rate": 1.1785714285714286e-05,
+      "loss": 0.9356,
+      "step": 9880
+    },
+    {
+      "epoch": 44.11,
+      "learning_rate": 1.1776785714285714e-05,
+      "loss": 0.891,
+      "step": 9881
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 1.1767857142857143e-05,
+      "loss": 0.9331,
+      "step": 9882
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 1.1758928571428572e-05,
+      "loss": 0.9683,
+      "step": 9883
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 1.175e-05,
+      "loss": 0.9491,
+      "step": 9884
+    },
+    {
+      "epoch": 44.13,
+      "learning_rate": 1.174107142857143e-05,
+      "loss": 0.9865,
+      "step": 9885
+    },
+    {
+      "epoch": 44.13,
+      "learning_rate": 1.1732142857142859e-05,
+      "loss": 1.0011,
+      "step": 9886
+    },
+    {
+      "epoch": 44.14,
+      "learning_rate": 1.1723214285714286e-05,
+      "loss": 0.9723,
+      "step": 9887
+    },
+    {
+      "epoch": 44.14,
+      "learning_rate": 1.1714285714285715e-05,
+      "loss": 0.9956,
+      "step": 9888
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 1.1705357142857144e-05,
+      "loss": 0.9949,
+      "step": 9889
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 1.1696428571428572e-05,
+      "loss": 0.9371,
+      "step": 9890
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 1.1687500000000001e-05,
+      "loss": 0.9117,
+      "step": 9891
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 1.1678571428571428e-05,
+      "loss": 0.939,
+      "step": 9892
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 1.1669642857142859e-05,
+      "loss": 0.8843,
+      "step": 9893
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 1.1660714285714286e-05,
+      "loss": 0.9709,
+      "step": 9894
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 1.1651785714285715e-05,
+      "loss": 0.9301,
+      "step": 9895
+    },
+    {
+      "epoch": 44.18,
+      "learning_rate": 1.1642857142857144e-05,
+      "loss": 0.8986,
+      "step": 9896
+    },
+    {
+      "epoch": 44.18,
+      "learning_rate": 1.1633928571428573e-05,
+      "loss": 0.9085,
+      "step": 9897
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 1.1625000000000001e-05,
+      "loss": 0.9343,
+      "step": 9898
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 1.1616071428571428e-05,
+      "loss": 0.9704,
+      "step": 9899
+    },
+    {
+      "epoch": 44.2,
+      "learning_rate": 1.1607142857142857e-05,
+      "loss": 0.9197,
+      "step": 9900
+    },
+    {
+      "epoch": 44.2,
+      "eval_loss": 1.111343502998352,
+      "eval_runtime": 44.4777,
+      "eval_samples_per_second": 294.395,
+      "eval_steps_per_second": 1.551,
+      "step": 9900
+    },
+    {
+      "epoch": 44.2,
+      "learning_rate": 1.1598214285714286e-05,
+      "loss": 0.9545,
+      "step": 9901
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 1.1589285714285715e-05,
+      "loss": 0.9613,
+      "step": 9902
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 1.1580357142857142e-05,
+      "loss": 0.9792,
+      "step": 9903
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.9461,
+      "step": 9904
+    },
+    {
+      "epoch": 44.22,
+      "learning_rate": 1.1562500000000002e-05,
+      "loss": 0.9379,
+      "step": 9905
+    },
+    {
+      "epoch": 44.22,
+      "learning_rate": 1.1553571428571429e-05,
+      "loss": 0.9114,
+      "step": 9906
+    },
+    {
+      "epoch": 44.23,
+      "learning_rate": 1.1544642857142858e-05,
+      "loss": 1.0105,
+      "step": 9907
+    },
+    {
+      "epoch": 44.23,
+      "learning_rate": 1.1535714285714286e-05,
+      "loss": 0.85,
+      "step": 9908
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 1.1526785714285715e-05,
+      "loss": 0.9931,
+      "step": 9909
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 1.1517857142857142e-05,
+      "loss": 1.0167,
+      "step": 9910
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 1.1508928571428571e-05,
+      "loss": 0.9668,
+      "step": 9911
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.9161,
+      "step": 9912
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 1.1491071428571429e-05,
+      "loss": 0.9042,
+      "step": 9913
+    },
+    {
+      "epoch": 44.26,
+      "learning_rate": 1.1482142857142858e-05,
+      "loss": 0.9206,
+      "step": 9914
+    },
+    {
+      "epoch": 44.26,
+      "learning_rate": 1.1473214285714287e-05,
+      "loss": 0.8823,
+      "step": 9915
+    },
+    {
+      "epoch": 44.27,
+      "learning_rate": 1.1464285714285715e-05,
+      "loss": 0.9552,
+      "step": 9916
+    },
+    {
+      "epoch": 44.27,
+      "learning_rate": 1.1455357142857144e-05,
+      "loss": 0.9925,
+      "step": 9917
+    },
+    {
+      "epoch": 44.28,
+      "learning_rate": 1.1446428571428571e-05,
+      "loss": 0.9291,
+      "step": 9918
+    },
+    {
+      "epoch": 44.28,
+      "learning_rate": 1.14375e-05,
+      "loss": 0.9923,
+      "step": 9919
+    },
+    {
+      "epoch": 44.29,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.0032,
+      "step": 9920
+    },
+    {
+      "epoch": 44.29,
+      "learning_rate": 1.1419642857142858e-05,
+      "loss": 0.9464,
+      "step": 9921
+    },
+    {
+      "epoch": 44.29,
+      "learning_rate": 1.1410714285714285e-05,
+      "loss": 0.992,
+      "step": 9922
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 1.1401785714285716e-05,
+      "loss": 0.9625,
+      "step": 9923
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 1.1392857142857144e-05,
+      "loss": 0.9245,
+      "step": 9924
+    },
+    {
+      "epoch": 44.31,
+      "learning_rate": 1.1383928571428572e-05,
+      "loss": 0.8766,
+      "step": 9925
+    },
+    {
+      "epoch": 44.31,
+      "learning_rate": 1.1375e-05,
+      "loss": 0.9531,
+      "step": 9926
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 1.136607142857143e-05,
+      "loss": 0.9815,
+      "step": 9927
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 1.1357142857142858e-05,
+      "loss": 0.9581,
+      "step": 9928
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 1.1348214285714285e-05,
+      "loss": 0.9412,
+      "step": 9929
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 1.1339285714285714e-05,
+      "loss": 1.0131,
+      "step": 9930
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 1.1330357142857145e-05,
+      "loss": 0.9417,
+      "step": 9931
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 1.1321428571428572e-05,
+      "loss": 0.9629,
+      "step": 9932
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 1.13125e-05,
+      "loss": 0.9734,
+      "step": 9933
+    },
+    {
+      "epoch": 44.35,
+      "learning_rate": 1.130357142857143e-05,
+      "loss": 0.863,
+      "step": 9934
+    },
+    {
+      "epoch": 44.35,
+      "learning_rate": 1.1294642857142858e-05,
+      "loss": 0.8936,
+      "step": 9935
+    },
+    {
+      "epoch": 44.36,
+      "learning_rate": 1.1285714285714285e-05,
+      "loss": 0.9845,
+      "step": 9936
+    },
+    {
+      "epoch": 44.36,
+      "learning_rate": 1.1276785714285714e-05,
+      "loss": 0.9513,
+      "step": 9937
+    },
+    {
+      "epoch": 44.37,
+      "learning_rate": 1.1267857142857145e-05,
+      "loss": 1.0005,
+      "step": 9938
+    },
+    {
+      "epoch": 44.37,
+      "learning_rate": 1.1258928571428572e-05,
+      "loss": 0.916,
+      "step": 9939
+    },
+    {
+      "epoch": 44.38,
+      "learning_rate": 1.125e-05,
+      "loss": 0.9211,
+      "step": 9940
+    },
+    {
+      "epoch": 44.38,
+      "learning_rate": 1.1241071428571428e-05,
+      "loss": 0.9269,
+      "step": 9941
+    },
+    {
+      "epoch": 44.38,
+      "learning_rate": 1.1232142857142858e-05,
+      "loss": 0.8914,
+      "step": 9942
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 1.1223214285714287e-05,
+      "loss": 0.9194,
+      "step": 9943
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 1.1214285714285714e-05,
+      "loss": 1.0158,
+      "step": 9944
+    },
+    {
+      "epoch": 44.4,
+      "learning_rate": 1.1205357142857143e-05,
+      "loss": 0.9557,
+      "step": 9945
+    },
+    {
+      "epoch": 44.4,
+      "learning_rate": 1.1196428571428572e-05,
+      "loss": 0.9414,
+      "step": 9946
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 1.1187500000000001e-05,
+      "loss": 0.9503,
+      "step": 9947
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 1.1178571428571428e-05,
+      "loss": 0.9349,
+      "step": 9948
+    },
+    {
+      "epoch": 44.42,
+      "learning_rate": 1.1169642857142857e-05,
+      "loss": 0.9375,
+      "step": 9949
+    },
+    {
+      "epoch": 44.42,
+      "learning_rate": 1.1160714285714287e-05,
+      "loss": 0.9743,
+      "step": 9950
+    },
+    {
+      "epoch": 44.42,
+      "learning_rate": 1.1151785714285715e-05,
+      "loss": 0.9281,
+      "step": 9951
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.9504,
+      "step": 9952
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 1.1133928571428572e-05,
+      "loss": 0.9582,
+      "step": 9953
+    },
+    {
+      "epoch": 44.44,
+      "learning_rate": 1.1125000000000001e-05,
+      "loss": 0.9817,
+      "step": 9954
+    },
+    {
+      "epoch": 44.44,
+      "learning_rate": 1.1116071428571428e-05,
+      "loss": 0.9369,
+      "step": 9955
+    },
+    {
+      "epoch": 44.45,
+      "learning_rate": 1.1107142857142857e-05,
+      "loss": 0.8627,
+      "step": 9956
+    },
+    {
+      "epoch": 44.45,
+      "learning_rate": 1.1098214285714288e-05,
+      "loss": 0.9648,
+      "step": 9957
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 1.1089285714285715e-05,
+      "loss": 1.0211,
+      "step": 9958
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 1.1080357142857144e-05,
+      "loss": 0.9522,
+      "step": 9959
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 1.107142857142857e-05,
+      "loss": 0.9715,
+      "step": 9960
+    },
+    {
+      "epoch": 44.47,
+      "learning_rate": 1.1062500000000001e-05,
+      "loss": 0.9438,
+      "step": 9961
+    },
+    {
+      "epoch": 44.47,
+      "learning_rate": 1.1053571428571428e-05,
+      "loss": 1.0109,
+      "step": 9962
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 1.1044642857142857e-05,
+      "loss": 0.8792,
+      "step": 9963
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 1.1035714285714286e-05,
+      "loss": 0.9775,
+      "step": 9964
+    },
+    {
+      "epoch": 44.49,
+      "learning_rate": 1.1026785714285715e-05,
+      "loss": 0.9396,
+      "step": 9965
+    },
+    {
+      "epoch": 44.49,
+      "learning_rate": 1.1017857142857144e-05,
+      "loss": 0.9173,
+      "step": 9966
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 1.1008928571428571e-05,
+      "loss": 0.8927,
+      "step": 9967
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.9065,
+      "step": 9968
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 1.099107142857143e-05,
+      "loss": 0.9187,
+      "step": 9969
+    },
+    {
+      "epoch": 44.51,
+      "learning_rate": 1.0982142857142857e-05,
+      "loss": 0.9786,
+      "step": 9970
+    },
+    {
+      "epoch": 44.51,
+      "learning_rate": 1.0973214285714286e-05,
+      "loss": 0.9572,
+      "step": 9971
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 1.0964285714285715e-05,
+      "loss": 0.952,
+      "step": 9972
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 1.0955357142857144e-05,
+      "loss": 0.9724,
+      "step": 9973
+    },
+    {
+      "epoch": 44.53,
+      "learning_rate": 1.0946428571428571e-05,
+      "loss": 0.9971,
+      "step": 9974
+    },
+    {
+      "epoch": 44.53,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.9802,
+      "step": 9975
+    },
+    {
+      "epoch": 44.54,
+      "learning_rate": 1.092857142857143e-05,
+      "loss": 0.9716,
+      "step": 9976
+    },
+    {
+      "epoch": 44.54,
+      "learning_rate": 1.0919642857142858e-05,
+      "loss": 0.944,
+      "step": 9977
+    },
+    {
+      "epoch": 44.54,
+      "learning_rate": 1.0910714285714286e-05,
+      "loss": 0.9812,
+      "step": 9978
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 1.0901785714285714e-05,
+      "loss": 1.0051,
+      "step": 9979
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 1.0892857142857144e-05,
+      "loss": 1.0157,
+      "step": 9980
+    },
+    {
+      "epoch": 44.56,
+      "learning_rate": 1.0883928571428571e-05,
+      "loss": 0.977,
+      "step": 9981
+    },
+    {
+      "epoch": 44.56,
+      "learning_rate": 1.0875e-05,
+      "loss": 0.9159,
+      "step": 9982
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 1.0866071428571429e-05,
+      "loss": 0.9893,
+      "step": 9983
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.9608,
+      "step": 9984
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 1.0848214285714287e-05,
+      "loss": 0.9331,
+      "step": 9985
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 1.0839285714285714e-05,
+      "loss": 0.8947,
+      "step": 9986
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 1.0830357142857144e-05,
+      "loss": 0.8944,
+      "step": 9987
+    },
+    {
+      "epoch": 44.59,
+      "learning_rate": 1.0821428571428573e-05,
+      "loss": 0.9172,
+      "step": 9988
+    },
+    {
+      "epoch": 44.59,
+      "learning_rate": 1.08125e-05,
+      "loss": 0.9303,
+      "step": 9989
+    },
+    {
+      "epoch": 44.6,
+      "learning_rate": 1.0803571428571429e-05,
+      "loss": 0.984,
+      "step": 9990
+    },
+    {
+      "epoch": 44.6,
+      "learning_rate": 1.0794642857142858e-05,
+      "loss": 0.9389,
+      "step": 9991
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 1.0785714285714287e-05,
+      "loss": 0.9428,
+      "step": 9992
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 1.0776785714285714e-05,
+      "loss": 0.9476,
+      "step": 9993
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 1.0767857142857143e-05,
+      "loss": 0.9602,
+      "step": 9994
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 1.0758928571428573e-05,
+      "loss": 0.8932,
+      "step": 9995
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 1.075e-05,
+      "loss": 0.9564,
+      "step": 9996
+    },
+    {
+      "epoch": 44.63,
+      "learning_rate": 1.074107142857143e-05,
+      "loss": 0.9908,
+      "step": 9997
+    },
+    {
+      "epoch": 44.63,
+      "learning_rate": 1.0732142857142858e-05,
+      "loss": 0.941,
+      "step": 9998
+    },
+    {
+      "epoch": 44.64,
+      "learning_rate": 1.0723214285714287e-05,
+      "loss": 0.9196,
+      "step": 9999
+    },
+    {
+      "epoch": 44.64,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 1.016,
+      "step": 10000
+    },
+    {
+      "epoch": 44.65,
+      "learning_rate": 1.0705357142857143e-05,
+      "loss": 0.9151,
+      "step": 10001
+    },
+    {
+      "epoch": 44.65,
+      "learning_rate": 1.0696428571428572e-05,
+      "loss": 0.8965,
+      "step": 10002
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 1.06875e-05,
+      "loss": 0.9791,
+      "step": 10003
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 1.067857142857143e-05,
+      "loss": 0.9148,
+      "step": 10004
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 1.0669642857142857e-05,
+      "loss": 1.0229,
+      "step": 10005
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 1.0660714285714287e-05,
+      "loss": 0.9286,
+      "step": 10006
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 1.0651785714285714e-05,
+      "loss": 0.8846,
+      "step": 10007
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 1.0642857142857143e-05,
+      "loss": 0.9372,
+      "step": 10008
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 1.0633928571428572e-05,
+      "loss": 0.9254,
+      "step": 10009
+    },
+    {
+      "epoch": 44.69,
+      "learning_rate": 1.0625e-05,
+      "loss": 0.9415,
+      "step": 10010
+    },
+    {
+      "epoch": 44.69,
+      "learning_rate": 1.061607142857143e-05,
+      "loss": 0.9085,
+      "step": 10011
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 1.0607142857142857e-05,
+      "loss": 0.9793,
+      "step": 10012
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 1.0598214285714286e-05,
+      "loss": 0.9528,
+      "step": 10013
+    },
+    {
+      "epoch": 44.71,
+      "learning_rate": 1.0589285714285716e-05,
+      "loss": 1.0702,
+      "step": 10014
+    },
+    {
+      "epoch": 44.71,
+      "learning_rate": 1.0580357142857143e-05,
+      "loss": 0.893,
+      "step": 10015
+    },
+    {
+      "epoch": 44.71,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.9253,
+      "step": 10016
+    },
+    {
+      "epoch": 44.72,
+      "learning_rate": 1.0562500000000001e-05,
+      "loss": 0.9171,
+      "step": 10017
+    },
+    {
+      "epoch": 44.72,
+      "learning_rate": 1.055357142857143e-05,
+      "loss": 0.9255,
+      "step": 10018
+    },
+    {
+      "epoch": 44.73,
+      "learning_rate": 1.0544642857142857e-05,
+      "loss": 0.9524,
+      "step": 10019
+    },
+    {
+      "epoch": 44.73,
+      "learning_rate": 1.0535714285714286e-05,
+      "loss": 0.9864,
+      "step": 10020
+    },
+    {
+      "epoch": 44.74,
+      "learning_rate": 1.0526785714285715e-05,
+      "loss": 0.8929,
+      "step": 10021
+    },
+    {
+      "epoch": 44.74,
+      "learning_rate": 1.0517857142857143e-05,
+      "loss": 0.9571,
+      "step": 10022
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 1.0508928571428572e-05,
+      "loss": 0.9953,
+      "step": 10023
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 1.05e-05,
+      "loss": 0.9484,
+      "step": 10024
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 1.049107142857143e-05,
+      "loss": 0.9809,
+      "step": 10025
+    },
+    {
+      "epoch": 44.76,
+      "learning_rate": 1.0482142857142857e-05,
+      "loss": 0.9479,
+      "step": 10026
+    },
+    {
+      "epoch": 44.76,
+      "learning_rate": 1.0473214285714286e-05,
+      "loss": 0.9597,
+      "step": 10027
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 1.0464285714285715e-05,
+      "loss": 0.9552,
+      "step": 10028
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 1.0455357142857144e-05,
+      "loss": 0.9508,
+      "step": 10029
+    },
+    {
+      "epoch": 44.78,
+      "learning_rate": 1.0446428571428572e-05,
+      "loss": 0.8772,
+      "step": 10030
+    },
+    {
+      "epoch": 44.78,
+      "learning_rate": 1.04375e-05,
+      "loss": 0.916,
+      "step": 10031
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 1.0428571428571428e-05,
+      "loss": 0.9069,
+      "step": 10032
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 1.0419642857142857e-05,
+      "loss": 0.9743,
+      "step": 10033
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 1.0410714285714286e-05,
+      "loss": 0.9247,
+      "step": 10034
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 1.0401785714285715e-05,
+      "loss": 0.9729,
+      "step": 10035
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 1.0392857142857144e-05,
+      "loss": 0.9166,
+      "step": 10036
+    },
+    {
+      "epoch": 44.81,
+      "learning_rate": 1.0383928571428573e-05,
+      "loss": 0.9138,
+      "step": 10037
+    },
+    {
+      "epoch": 44.81,
+      "learning_rate": 1.0375e-05,
+      "loss": 0.8878,
+      "step": 10038
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 1.0366071428571429e-05,
+      "loss": 0.8856,
+      "step": 10039
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.9653,
+      "step": 10040
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 1.0348214285714286e-05,
+      "loss": 0.9997,
+      "step": 10041
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 1.0339285714285715e-05,
+      "loss": 0.9085,
+      "step": 10042
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 1.0330357142857142e-05,
+      "loss": 0.9068,
+      "step": 10043
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 1.0321428571428573e-05,
+      "loss": 1.0106,
+      "step": 10044
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 1.03125e-05,
+      "loss": 0.9208,
+      "step": 10045
+    },
+    {
+      "epoch": 44.85,
+      "learning_rate": 1.0303571428571429e-05,
+      "loss": 0.9575,
+      "step": 10046
+    },
+    {
+      "epoch": 44.85,
+      "learning_rate": 1.0294642857142858e-05,
+      "loss": 0.9495,
+      "step": 10047
+    },
+    {
+      "epoch": 44.86,
+      "learning_rate": 1.0285714285714286e-05,
+      "loss": 0.9874,
+      "step": 10048
+    },
+    {
+      "epoch": 44.86,
+      "learning_rate": 1.0276785714285715e-05,
+      "loss": 0.9342,
+      "step": 10049
+    },
+    {
+      "epoch": 44.87,
+      "learning_rate": 1.0267857142857142e-05,
+      "loss": 0.9183,
+      "step": 10050
+    },
+    {
+      "epoch": 44.87,
+      "learning_rate": 1.0258928571428571e-05,
+      "loss": 0.9259,
+      "step": 10051
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 1.025e-05,
+      "loss": 0.9726,
+      "step": 10052
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 1.0241071428571429e-05,
+      "loss": 0.9433,
+      "step": 10053
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 1.0232142857142858e-05,
+      "loss": 0.9693,
+      "step": 10054
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 1.0223214285714287e-05,
+      "loss": 0.9158,
+      "step": 10055
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 1.0214285714285715e-05,
+      "loss": 0.938,
+      "step": 10056
+    },
+    {
+      "epoch": 44.9,
+      "learning_rate": 1.0205357142857143e-05,
+      "loss": 0.9347,
+      "step": 10057
+    },
+    {
+      "epoch": 44.9,
+      "learning_rate": 1.0196428571428571e-05,
+      "loss": 0.9321,
+      "step": 10058
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 1.01875e-05,
+      "loss": 0.9271,
+      "step": 10059
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 1.0178571428571429e-05,
+      "loss": 0.9533,
+      "step": 10060
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 1.0169642857142858e-05,
+      "loss": 0.9202,
+      "step": 10061
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 1.0160714285714285e-05,
+      "loss": 0.9215,
+      "step": 10062
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 1.0151785714285716e-05,
+      "loss": 0.8915,
+      "step": 10063
+    },
+    {
+      "epoch": 44.93,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.9639,
+      "step": 10064
+    },
+    {
+      "epoch": 44.93,
+      "learning_rate": 1.0133928571428572e-05,
+      "loss": 0.9363,
+      "step": 10065
+    },
+    {
+      "epoch": 44.94,
+      "learning_rate": 1.0125e-05,
+      "loss": 0.9075,
+      "step": 10066
+    },
+    {
+      "epoch": 44.94,
+      "learning_rate": 1.011607142857143e-05,
+      "loss": 0.9302,
+      "step": 10067
+    },
+    {
+      "epoch": 44.95,
+      "learning_rate": 1.0107142857142858e-05,
+      "loss": 0.9711,
+      "step": 10068
+    },
+    {
+      "epoch": 44.95,
+      "learning_rate": 1.0098214285714285e-05,
+      "loss": 0.9441,
+      "step": 10069
+    },
+    {
+      "epoch": 44.96,
+      "learning_rate": 1.0089285714285716e-05,
+      "loss": 0.9906,
+      "step": 10070
+    },
+    {
+      "epoch": 44.96,
+      "learning_rate": 1.0080357142857143e-05,
+      "loss": 0.8769,
+      "step": 10071
+    },
+    {
+      "epoch": 44.96,
+      "learning_rate": 1.0071428571428572e-05,
+      "loss": 0.9357,
+      "step": 10072
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 1.00625e-05,
+      "loss": 0.9022,
+      "step": 10073
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 1.005357142857143e-05,
+      "loss": 0.9695,
+      "step": 10074
+    },
+    {
+      "epoch": 44.98,
+      "learning_rate": 1.0044642857142858e-05,
+      "loss": 1.0212,
+      "step": 10075
+    },
+    {
+      "epoch": 44.98,
+      "learning_rate": 1.0035714285714285e-05,
+      "loss": 0.894,
+      "step": 10076
+    },
+    {
+      "epoch": 44.99,
+      "learning_rate": 1.0026785714285714e-05,
+      "loss": 0.9163,
+      "step": 10077
+    },
+    {
+      "epoch": 44.99,
+      "learning_rate": 1.0017857142857143e-05,
+      "loss": 0.9562,
+      "step": 10078
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 1.0008928571428572e-05,
+      "loss": 0.9228,
+      "step": 10079
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 1e-05,
+      "loss": 0.9697,
+      "step": 10080
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 9.991071428571428e-06,
+      "loss": 0.9186,
+      "step": 10081
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 9.982142857142858e-06,
+      "loss": 0.9467,
+      "step": 10082
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 9.973214285714286e-06,
+      "loss": 0.9746,
+      "step": 10083
+    },
+    {
+      "epoch": 45.02,
+      "learning_rate": 9.964285714285714e-06,
+      "loss": 0.9566,
+      "step": 10084
+    },
+    {
+      "epoch": 45.02,
+      "learning_rate": 9.955357142857143e-06,
+      "loss": 0.9101,
+      "step": 10085
+    },
+    {
+      "epoch": 45.03,
+      "learning_rate": 9.946428571428572e-06,
+      "loss": 0.9351,
+      "step": 10086
+    },
+    {
+      "epoch": 45.03,
+      "learning_rate": 9.937500000000001e-06,
+      "loss": 1.0212,
+      "step": 10087
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 9.928571428571428e-06,
+      "loss": 1.0084,
+      "step": 10088
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 9.919642857142859e-06,
+      "loss": 0.9255,
+      "step": 10089
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 9.910714285714286e-06,
+      "loss": 1.0016,
+      "step": 10090
+    },
+    {
+      "epoch": 45.05,
+      "learning_rate": 9.901785714285715e-06,
+      "loss": 0.917,
+      "step": 10091
+    },
+    {
+      "epoch": 45.05,
+      "learning_rate": 9.892857142857143e-06,
+      "loss": 0.9195,
+      "step": 10092
+    },
+    {
+      "epoch": 45.06,
+      "learning_rate": 9.883928571428572e-06,
+      "loss": 0.9838,
+      "step": 10093
+    },
+    {
+      "epoch": 45.06,
+      "learning_rate": 9.875000000000001e-06,
+      "loss": 0.9461,
+      "step": 10094
+    },
+    {
+      "epoch": 45.07,
+      "learning_rate": 9.866071428571428e-06,
+      "loss": 0.979,
+      "step": 10095
+    },
+    {
+      "epoch": 45.07,
+      "learning_rate": 9.857142857142857e-06,
+      "loss": 0.9578,
+      "step": 10096
+    },
+    {
+      "epoch": 45.08,
+      "learning_rate": 9.848214285714286e-06,
+      "loss": 0.9165,
+      "step": 10097
+    },
+    {
+      "epoch": 45.08,
+      "learning_rate": 9.839285714285715e-06,
+      "loss": 0.9294,
+      "step": 10098
+    },
+    {
+      "epoch": 45.08,
+      "learning_rate": 9.830357142857144e-06,
+      "loss": 0.9468,
+      "step": 10099
+    },
+    {
+      "epoch": 45.09,
+      "learning_rate": 9.821428571428573e-06,
+      "loss": 0.9723,
+      "step": 10100
+    },
+    {
+      "epoch": 45.09,
+      "learning_rate": 9.812500000000001e-06,
+      "loss": 0.941,
+      "step": 10101
+    },
+    {
+      "epoch": 45.1,
+      "learning_rate": 9.803571428571428e-06,
+      "loss": 0.9494,
+      "step": 10102
+    },
+    {
+      "epoch": 45.1,
+      "learning_rate": 9.794642857142857e-06,
+      "loss": 0.9837,
+      "step": 10103
+    },
+    {
+      "epoch": 45.11,
+      "learning_rate": 9.785714285714286e-06,
+      "loss": 0.8899,
+      "step": 10104
+    },
+    {
+      "epoch": 45.11,
+      "learning_rate": 9.776785714285715e-06,
+      "loss": 0.9028,
+      "step": 10105
+    },
+    {
+      "epoch": 45.12,
+      "learning_rate": 9.767857142857144e-06,
+      "loss": 0.9116,
+      "step": 10106
+    },
+    {
+      "epoch": 45.12,
+      "learning_rate": 9.758928571428571e-06,
+      "loss": 0.9542,
+      "step": 10107
+    },
+    {
+      "epoch": 45.12,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 0.902,
+      "step": 10108
+    },
+    {
+      "epoch": 45.13,
+      "learning_rate": 9.741071428571429e-06,
+      "loss": 0.9668,
+      "step": 10109
+    },
+    {
+      "epoch": 45.13,
+      "learning_rate": 9.732142857142858e-06,
+      "loss": 0.962,
+      "step": 10110
+    },
+    {
+      "epoch": 45.14,
+      "learning_rate": 9.723214285714286e-06,
+      "loss": 0.8178,
+      "step": 10111
+    },
+    {
+      "epoch": 45.14,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.9673,
+      "step": 10112
+    },
+    {
+      "epoch": 45.15,
+      "learning_rate": 9.705357142857144e-06,
+      "loss": 0.9581,
+      "step": 10113
+    },
+    {
+      "epoch": 45.15,
+      "learning_rate": 9.696428571428571e-06,
+      "loss": 0.9791,
+      "step": 10114
+    },
+    {
+      "epoch": 45.16,
+      "learning_rate": 9.6875e-06,
+      "loss": 0.9268,
+      "step": 10115
+    },
+    {
+      "epoch": 45.16,
+      "learning_rate": 9.678571428571429e-06,
+      "loss": 0.9829,
+      "step": 10116
+    },
+    {
+      "epoch": 45.17,
+      "learning_rate": 9.669642857142858e-06,
+      "loss": 0.9792,
+      "step": 10117
+    },
+    {
+      "epoch": 45.17,
+      "learning_rate": 9.660714285714287e-06,
+      "loss": 0.9508,
+      "step": 10118
+    },
+    {
+      "epoch": 45.17,
+      "learning_rate": 9.651785714285715e-06,
+      "loss": 0.955,
+      "step": 10119
+    },
+    {
+      "epoch": 45.18,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.8858,
+      "step": 10120
+    },
+    {
+      "epoch": 45.18,
+      "learning_rate": 9.633928571428571e-06,
+      "loss": 1.0069,
+      "step": 10121
+    },
+    {
+      "epoch": 45.19,
+      "learning_rate": 9.625e-06,
+      "loss": 0.9311,
+      "step": 10122
+    },
+    {
+      "epoch": 45.19,
+      "learning_rate": 9.616071428571429e-06,
+      "loss": 0.9184,
+      "step": 10123
+    },
+    {
+      "epoch": 45.2,
+      "learning_rate": 9.607142857142858e-06,
+      "loss": 0.922,
+      "step": 10124
+    },
+    {
+      "epoch": 45.2,
+      "learning_rate": 9.598214285714287e-06,
+      "loss": 0.9693,
+      "step": 10125
+    },
+    {
+      "epoch": 45.21,
+      "learning_rate": 9.589285714285714e-06,
+      "loss": 0.9699,
+      "step": 10126
+    },
+    {
+      "epoch": 45.21,
+      "learning_rate": 9.580357142857144e-06,
+      "loss": 0.9185,
+      "step": 10127
+    },
+    {
+      "epoch": 45.21,
+      "learning_rate": 9.571428571428572e-06,
+      "loss": 0.9938,
+      "step": 10128
+    },
+    {
+      "epoch": 45.22,
+      "learning_rate": 9.5625e-06,
+      "loss": 0.9492,
+      "step": 10129
+    },
+    {
+      "epoch": 45.22,
+      "learning_rate": 9.55357142857143e-06,
+      "loss": 0.9645,
+      "step": 10130
+    },
+    {
+      "epoch": 45.23,
+      "learning_rate": 9.544642857142858e-06,
+      "loss": 0.9138,
+      "step": 10131
+    },
+    {
+      "epoch": 45.23,
+      "learning_rate": 9.535714285714287e-06,
+      "loss": 0.9415,
+      "step": 10132
+    },
+    {
+      "epoch": 45.24,
+      "learning_rate": 9.526785714285714e-06,
+      "loss": 0.9255,
+      "step": 10133
+    },
+    {
+      "epoch": 45.24,
+      "learning_rate": 9.517857142857143e-06,
+      "loss": 0.9584,
+      "step": 10134
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 9.508928571428572e-06,
+      "loss": 0.9855,
+      "step": 10135
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 9.5e-06,
+      "loss": 0.9025,
+      "step": 10136
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 9.491071428571428e-06,
+      "loss": 0.9205,
+      "step": 10137
+    },
+    {
+      "epoch": 45.26,
+      "learning_rate": 9.482142857142858e-06,
+      "loss": 0.9546,
+      "step": 10138
+    },
+    {
+      "epoch": 45.26,
+      "learning_rate": 9.473214285714287e-06,
+      "loss": 1.0095,
+      "step": 10139
+    },
+    {
+      "epoch": 45.27,
+      "learning_rate": 9.464285714285714e-06,
+      "loss": 0.9375,
+      "step": 10140
+    },
+    {
+      "epoch": 45.27,
+      "learning_rate": 9.455357142857143e-06,
+      "loss": 0.9621,
+      "step": 10141
+    },
+    {
+      "epoch": 45.28,
+      "learning_rate": 9.446428571428572e-06,
+      "loss": 0.9795,
+      "step": 10142
+    },
+    {
+      "epoch": 45.28,
+      "learning_rate": 9.4375e-06,
+      "loss": 0.9662,
+      "step": 10143
+    },
+    {
+      "epoch": 45.29,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.9502,
+      "step": 10144
+    },
+    {
+      "epoch": 45.29,
+      "learning_rate": 9.419642857142857e-06,
+      "loss": 0.9903,
+      "step": 10145
+    },
+    {
+      "epoch": 45.29,
+      "learning_rate": 9.410714285714287e-06,
+      "loss": 0.9575,
+      "step": 10146
+    },
+    {
+      "epoch": 45.3,
+      "learning_rate": 9.401785714285714e-06,
+      "loss": 0.9443,
+      "step": 10147
+    },
+    {
+      "epoch": 45.3,
+      "learning_rate": 9.392857142857143e-06,
+      "loss": 0.9484,
+      "step": 10148
+    },
+    {
+      "epoch": 45.31,
+      "learning_rate": 9.383928571428572e-06,
+      "loss": 0.8918,
+      "step": 10149
+    },
+    {
+      "epoch": 45.31,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.9198,
+      "step": 10150
+    },
+    {
+      "epoch": 45.32,
+      "learning_rate": 9.36607142857143e-06,
+      "loss": 0.9233,
+      "step": 10151
+    },
+    {
+      "epoch": 45.32,
+      "learning_rate": 9.357142857142857e-06,
+      "loss": 0.9778,
+      "step": 10152
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 9.348214285714286e-06,
+      "loss": 0.9179,
+      "step": 10153
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 9.339285714285715e-06,
+      "loss": 0.9935,
+      "step": 10154
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 9.330357142857143e-06,
+      "loss": 0.8898,
+      "step": 10155
+    },
+    {
+      "epoch": 45.34,
+      "learning_rate": 9.32142857142857e-06,
+      "loss": 0.9557,
+      "step": 10156
+    },
+    {
+      "epoch": 45.34,
+      "learning_rate": 9.312500000000001e-06,
+      "loss": 0.8846,
+      "step": 10157
+    },
+    {
+      "epoch": 45.35,
+      "learning_rate": 9.30357142857143e-06,
+      "loss": 0.961,
+      "step": 10158
+    },
+    {
+      "epoch": 45.35,
+      "learning_rate": 9.294642857142857e-06,
+      "loss": 1.015,
+      "step": 10159
+    },
+    {
+      "epoch": 45.36,
+      "learning_rate": 9.285714285714286e-06,
+      "loss": 0.9097,
+      "step": 10160
+    },
+    {
+      "epoch": 45.36,
+      "learning_rate": 9.276785714285715e-06,
+      "loss": 0.9138,
+      "step": 10161
+    },
+    {
+      "epoch": 45.37,
+      "learning_rate": 9.267857142857144e-06,
+      "loss": 0.9795,
+      "step": 10162
+    },
+    {
+      "epoch": 45.37,
+      "learning_rate": 9.258928571428572e-06,
+      "loss": 0.959,
+      "step": 10163
+    },
+    {
+      "epoch": 45.38,
+      "learning_rate": 9.25e-06,
+      "loss": 0.943,
+      "step": 10164
+    },
+    {
+      "epoch": 45.38,
+      "learning_rate": 9.24107142857143e-06,
+      "loss": 0.9101,
+      "step": 10165
+    },
+    {
+      "epoch": 45.38,
+      "learning_rate": 9.232142857142857e-06,
+      "loss": 0.9909,
+      "step": 10166
+    },
+    {
+      "epoch": 45.39,
+      "learning_rate": 9.223214285714286e-06,
+      "loss": 0.9063,
+      "step": 10167
+    },
+    {
+      "epoch": 45.39,
+      "learning_rate": 9.214285714285715e-06,
+      "loss": 0.9608,
+      "step": 10168
+    },
+    {
+      "epoch": 45.4,
+      "learning_rate": 9.205357142857144e-06,
+      "loss": 0.9699,
+      "step": 10169
+    },
+    {
+      "epoch": 45.4,
+      "learning_rate": 9.196428571428573e-06,
+      "loss": 0.9801,
+      "step": 10170
+    },
+    {
+      "epoch": 45.41,
+      "learning_rate": 9.1875e-06,
+      "loss": 0.9427,
+      "step": 10171
+    },
+    {
+      "epoch": 45.41,
+      "learning_rate": 9.17857142857143e-06,
+      "loss": 1.002,
+      "step": 10172
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 9.169642857142857e-06,
+      "loss": 0.9553,
+      "step": 10173
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 9.160714285714286e-06,
+      "loss": 0.904,
+      "step": 10174
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 9.151785714285713e-06,
+      "loss": 0.9595,
+      "step": 10175
+    },
+    {
+      "epoch": 45.43,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.9151,
+      "step": 10176
+    },
+    {
+      "epoch": 45.43,
+      "learning_rate": 9.133928571428573e-06,
+      "loss": 0.9948,
+      "step": 10177
+    },
+    {
+      "epoch": 45.44,
+      "learning_rate": 9.125e-06,
+      "loss": 1.0105,
+      "step": 10178
+    },
+    {
+      "epoch": 45.44,
+      "learning_rate": 9.116071428571429e-06,
+      "loss": 0.9291,
+      "step": 10179
+    },
+    {
+      "epoch": 45.45,
+      "learning_rate": 9.107142857142858e-06,
+      "loss": 0.8921,
+      "step": 10180
+    },
+    {
+      "epoch": 45.45,
+      "learning_rate": 9.098214285714286e-06,
+      "loss": 0.8911,
+      "step": 10181
+    },
+    {
+      "epoch": 45.46,
+      "learning_rate": 9.089285714285714e-06,
+      "loss": 0.9375,
+      "step": 10182
+    },
+    {
+      "epoch": 45.46,
+      "learning_rate": 9.080357142857144e-06,
+      "loss": 0.9482,
+      "step": 10183
+    },
+    {
+      "epoch": 45.46,
+      "learning_rate": 9.071428571428573e-06,
+      "loss": 0.917,
+      "step": 10184
+    },
+    {
+      "epoch": 45.47,
+      "learning_rate": 9.0625e-06,
+      "loss": 0.9402,
+      "step": 10185
+    },
+    {
+      "epoch": 45.47,
+      "learning_rate": 9.053571428571429e-06,
+      "loss": 0.9384,
+      "step": 10186
+    },
+    {
+      "epoch": 45.48,
+      "learning_rate": 9.044642857142858e-06,
+      "loss": 0.954,
+      "step": 10187
+    },
+    {
+      "epoch": 45.48,
+      "learning_rate": 9.035714285714287e-06,
+      "loss": 0.8868,
+      "step": 10188
+    },
+    {
+      "epoch": 45.49,
+      "learning_rate": 9.026785714285715e-06,
+      "loss": 0.996,
+      "step": 10189
+    },
+    {
+      "epoch": 45.49,
+      "learning_rate": 9.017857142857143e-06,
+      "loss": 0.9899,
+      "step": 10190
+    },
+    {
+      "epoch": 45.5,
+      "learning_rate": 9.008928571428573e-06,
+      "loss": 0.9536,
+      "step": 10191
+    },
+    {
+      "epoch": 45.5,
+      "learning_rate": 9e-06,
+      "loss": 0.8959,
+      "step": 10192
+    },
+    {
+      "epoch": 45.5,
+      "learning_rate": 8.991071428571429e-06,
+      "loss": 0.907,
+      "step": 10193
+    },
+    {
+      "epoch": 45.51,
+      "learning_rate": 8.982142857142856e-06,
+      "loss": 0.9295,
+      "step": 10194
+    },
+    {
+      "epoch": 45.51,
+      "learning_rate": 8.973214285714287e-06,
+      "loss": 0.9356,
+      "step": 10195
+    },
+    {
+      "epoch": 45.52,
+      "learning_rate": 8.964285714285716e-06,
+      "loss": 1.0161,
+      "step": 10196
+    },
+    {
+      "epoch": 45.52,
+      "learning_rate": 8.955357142857143e-06,
+      "loss": 0.9541,
+      "step": 10197
+    },
+    {
+      "epoch": 45.53,
+      "learning_rate": 8.946428571428572e-06,
+      "loss": 1.0012,
+      "step": 10198
+    },
+    {
+      "epoch": 45.53,
+      "learning_rate": 8.9375e-06,
+      "loss": 0.983,
+      "step": 10199
+    },
+    {
+      "epoch": 45.54,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.9198,
+      "step": 10200
+    },
+    {
+      "epoch": 45.54,
+      "eval_loss": 1.1106387376785278,
+      "eval_runtime": 44.4792,
+      "eval_samples_per_second": 294.385,
+      "eval_steps_per_second": 1.551,
+      "step": 10200
+    },
+    {
+      "epoch": 45.54,
+      "learning_rate": 8.919642857142856e-06,
+      "loss": 0.9364,
+      "step": 10201
+    },
+    {
+      "epoch": 45.54,
+      "learning_rate": 8.910714285714287e-06,
+      "loss": 0.9801,
+      "step": 10202
+    },
+    {
+      "epoch": 45.55,
+      "learning_rate": 8.901785714285716e-06,
+      "loss": 0.9486,
+      "step": 10203
+    },
+    {
+      "epoch": 45.55,
+      "learning_rate": 8.892857142857143e-06,
+      "loss": 0.9599,
+      "step": 10204
+    },
+    {
+      "epoch": 45.56,
+      "learning_rate": 8.883928571428572e-06,
+      "loss": 0.9636,
+      "step": 10205
+    },
+    {
+      "epoch": 45.56,
+      "learning_rate": 8.875e-06,
+      "loss": 0.9297,
+      "step": 10206
+    },
+    {
+      "epoch": 45.57,
+      "learning_rate": 8.86607142857143e-06,
+      "loss": 0.9205,
+      "step": 10207
+    },
+    {
+      "epoch": 45.57,
+      "learning_rate": 8.857142857142857e-06,
+      "loss": 0.8811,
+      "step": 10208
+    },
+    {
+      "epoch": 45.58,
+      "learning_rate": 8.848214285714285e-06,
+      "loss": 0.9555,
+      "step": 10209
+    },
+    {
+      "epoch": 45.58,
+      "learning_rate": 8.839285714285716e-06,
+      "loss": 0.9324,
+      "step": 10210
+    },
+    {
+      "epoch": 45.58,
+      "learning_rate": 8.830357142857143e-06,
+      "loss": 0.9361,
+      "step": 10211
+    },
+    {
+      "epoch": 45.59,
+      "learning_rate": 8.821428571428572e-06,
+      "loss": 0.9691,
+      "step": 10212
+    },
+    {
+      "epoch": 45.59,
+      "learning_rate": 8.8125e-06,
+      "loss": 0.9491,
+      "step": 10213
+    },
+    {
+      "epoch": 45.6,
+      "learning_rate": 8.80357142857143e-06,
+      "loss": 0.9664,
+      "step": 10214
+    },
+    {
+      "epoch": 45.6,
+      "learning_rate": 8.794642857142858e-06,
+      "loss": 0.9656,
+      "step": 10215
+    },
+    {
+      "epoch": 45.61,
+      "learning_rate": 8.785714285714286e-06,
+      "loss": 0.9586,
+      "step": 10216
+    },
+    {
+      "epoch": 45.61,
+      "learning_rate": 8.776785714285714e-06,
+      "loss": 0.9665,
+      "step": 10217
+    },
+    {
+      "epoch": 45.62,
+      "learning_rate": 8.767857142857143e-06,
+      "loss": 0.9779,
+      "step": 10218
+    },
+    {
+      "epoch": 45.62,
+      "learning_rate": 8.758928571428572e-06,
+      "loss": 0.9092,
+      "step": 10219
+    },
+    {
+      "epoch": 45.62,
+      "learning_rate": 8.75e-06,
+      "loss": 0.9818,
+      "step": 10220
+    },
+    {
+      "epoch": 45.63,
+      "learning_rate": 8.74107142857143e-06,
+      "loss": 0.9831,
+      "step": 10221
+    },
+    {
+      "epoch": 45.63,
+      "learning_rate": 8.732142857142859e-06,
+      "loss": 0.8657,
+      "step": 10222
+    },
+    {
+      "epoch": 45.64,
+      "learning_rate": 8.723214285714286e-06,
+      "loss": 0.9733,
+      "step": 10223
+    },
+    {
+      "epoch": 45.64,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 0.8848,
+      "step": 10224
+    },
+    {
+      "epoch": 45.65,
+      "learning_rate": 8.705357142857143e-06,
+      "loss": 0.8808,
+      "step": 10225
+    },
+    {
+      "epoch": 45.65,
+      "learning_rate": 8.696428571428572e-06,
+      "loss": 0.8911,
+      "step": 10226
+    },
+    {
+      "epoch": 45.66,
+      "learning_rate": 8.6875e-06,
+      "loss": 0.9865,
+      "step": 10227
+    },
+    {
+      "epoch": 45.66,
+      "learning_rate": 8.678571428571428e-06,
+      "loss": 0.9814,
+      "step": 10228
+    },
+    {
+      "epoch": 45.67,
+      "learning_rate": 8.669642857142859e-06,
+      "loss": 0.9323,
+      "step": 10229
+    },
+    {
+      "epoch": 45.67,
+      "learning_rate": 8.660714285714286e-06,
+      "loss": 0.9291,
+      "step": 10230
+    },
+    {
+      "epoch": 45.67,
+      "learning_rate": 8.651785714285715e-06,
+      "loss": 0.9626,
+      "step": 10231
+    },
+    {
+      "epoch": 45.68,
+      "learning_rate": 8.642857142857144e-06,
+      "loss": 0.9501,
+      "step": 10232
+    },
+    {
+      "epoch": 45.68,
+      "learning_rate": 8.633928571428572e-06,
+      "loss": 0.9432,
+      "step": 10233
+    },
+    {
+      "epoch": 45.69,
+      "learning_rate": 8.625e-06,
+      "loss": 0.9729,
+      "step": 10234
+    },
+    {
+      "epoch": 45.69,
+      "learning_rate": 8.616071428571428e-06,
+      "loss": 0.9902,
+      "step": 10235
+    },
+    {
+      "epoch": 45.7,
+      "learning_rate": 8.607142857142857e-06,
+      "loss": 0.9542,
+      "step": 10236
+    },
+    {
+      "epoch": 45.7,
+      "learning_rate": 8.598214285714286e-06,
+      "loss": 0.9098,
+      "step": 10237
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 8.589285714285715e-06,
+      "loss": 0.9139,
+      "step": 10238
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 8.580357142857142e-06,
+      "loss": 0.9267,
+      "step": 10239
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 8.571428571428573e-06,
+      "loss": 0.9119,
+      "step": 10240
+    },
+    {
+      "epoch": 45.72,
+      "learning_rate": 8.562500000000001e-06,
+      "loss": 0.9403,
+      "step": 10241
+    },
+    {
+      "epoch": 45.72,
+      "learning_rate": 8.553571428571429e-06,
+      "loss": 0.9797,
+      "step": 10242
+    },
+    {
+      "epoch": 45.73,
+      "learning_rate": 8.544642857142857e-06,
+      "loss": 0.9051,
+      "step": 10243
+    },
+    {
+      "epoch": 45.73,
+      "learning_rate": 8.535714285714286e-06,
+      "loss": 0.8855,
+      "step": 10244
+    },
+    {
+      "epoch": 45.74,
+      "learning_rate": 8.526785714285715e-06,
+      "loss": 0.9656,
+      "step": 10245
+    },
+    {
+      "epoch": 45.74,
+      "learning_rate": 8.517857142857142e-06,
+      "loss": 0.9648,
+      "step": 10246
+    },
+    {
+      "epoch": 45.75,
+      "learning_rate": 8.508928571428571e-06,
+      "loss": 0.931,
+      "step": 10247
+    },
+    {
+      "epoch": 45.75,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.9121,
+      "step": 10248
+    },
+    {
+      "epoch": 45.75,
+      "learning_rate": 8.491071428571429e-06,
+      "loss": 0.8773,
+      "step": 10249
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 8.482142857142858e-06,
+      "loss": 1.0193,
+      "step": 10250
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 8.473214285714286e-06,
+      "loss": 1.0033,
+      "step": 10251
+    },
+    {
+      "epoch": 45.77,
+      "learning_rate": 8.464285714285715e-06,
+      "loss": 0.9575,
+      "step": 10252
+    },
+    {
+      "epoch": 45.77,
+      "learning_rate": 8.455357142857142e-06,
+      "loss": 0.8874,
+      "step": 10253
+    },
+    {
+      "epoch": 45.78,
+      "learning_rate": 8.446428571428571e-06,
+      "loss": 0.9881,
+      "step": 10254
+    },
+    {
+      "epoch": 45.78,
+      "learning_rate": 8.437500000000002e-06,
+      "loss": 0.8958,
+      "step": 10255
+    },
+    {
+      "epoch": 45.79,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.9171,
+      "step": 10256
+    },
+    {
+      "epoch": 45.79,
+      "learning_rate": 8.419642857142858e-06,
+      "loss": 0.9255,
+      "step": 10257
+    },
+    {
+      "epoch": 45.79,
+      "learning_rate": 8.410714285714285e-06,
+      "loss": 0.9758,
+      "step": 10258
+    },
+    {
+      "epoch": 45.8,
+      "learning_rate": 8.401785714285715e-06,
+      "loss": 0.9775,
+      "step": 10259
+    },
+    {
+      "epoch": 45.8,
+      "learning_rate": 8.392857142857143e-06,
+      "loss": 0.9172,
+      "step": 10260
+    },
+    {
+      "epoch": 45.81,
+      "learning_rate": 8.383928571428571e-06,
+      "loss": 0.9731,
+      "step": 10261
+    },
+    {
+      "epoch": 45.81,
+      "learning_rate": 8.375e-06,
+      "loss": 0.9994,
+      "step": 10262
+    },
+    {
+      "epoch": 45.82,
+      "learning_rate": 8.366071428571429e-06,
+      "loss": 1.0036,
+      "step": 10263
+    },
+    {
+      "epoch": 45.82,
+      "learning_rate": 8.357142857142858e-06,
+      "loss": 0.9213,
+      "step": 10264
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 8.348214285714285e-06,
+      "loss": 0.9609,
+      "step": 10265
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 8.339285714285714e-06,
+      "loss": 1.0089,
+      "step": 10266
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 8.330357142857144e-06,
+      "loss": 1.0004,
+      "step": 10267
+    },
+    {
+      "epoch": 45.84,
+      "learning_rate": 8.321428571428572e-06,
+      "loss": 0.9216,
+      "step": 10268
+    },
+    {
+      "epoch": 45.84,
+      "learning_rate": 8.3125e-06,
+      "loss": 0.9693,
+      "step": 10269
+    },
+    {
+      "epoch": 45.85,
+      "learning_rate": 8.30357142857143e-06,
+      "loss": 0.978,
+      "step": 10270
+    },
+    {
+      "epoch": 45.85,
+      "learning_rate": 8.294642857142858e-06,
+      "loss": 0.9351,
+      "step": 10271
+    },
+    {
+      "epoch": 45.86,
+      "learning_rate": 8.285714285714285e-06,
+      "loss": 0.9416,
+      "step": 10272
+    },
+    {
+      "epoch": 45.86,
+      "learning_rate": 8.276785714285714e-06,
+      "loss": 0.9327,
+      "step": 10273
+    },
+    {
+      "epoch": 45.87,
+      "learning_rate": 8.267857142857145e-06,
+      "loss": 0.9449,
+      "step": 10274
+    },
+    {
+      "epoch": 45.87,
+      "learning_rate": 8.258928571428572e-06,
+      "loss": 0.9697,
+      "step": 10275
+    },
+    {
+      "epoch": 45.88,
+      "learning_rate": 8.25e-06,
+      "loss": 0.9348,
+      "step": 10276
+    },
+    {
+      "epoch": 45.88,
+      "learning_rate": 8.241071428571428e-06,
+      "loss": 0.8848,
+      "step": 10277
+    },
+    {
+      "epoch": 45.88,
+      "learning_rate": 8.232142857142858e-06,
+      "loss": 0.9511,
+      "step": 10278
+    },
+    {
+      "epoch": 45.89,
+      "learning_rate": 8.223214285714285e-06,
+      "loss": 0.9965,
+      "step": 10279
+    },
+    {
+      "epoch": 45.89,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.9717,
+      "step": 10280
+    },
+    {
+      "epoch": 45.9,
+      "learning_rate": 8.205357142857143e-06,
+      "loss": 0.9841,
+      "step": 10281
+    },
+    {
+      "epoch": 45.9,
+      "learning_rate": 8.196428571428572e-06,
+      "loss": 0.9256,
+      "step": 10282
+    },
+    {
+      "epoch": 45.91,
+      "learning_rate": 8.1875e-06,
+      "loss": 0.992,
+      "step": 10283
+    },
+    {
+      "epoch": 45.91,
+      "learning_rate": 8.178571428571428e-06,
+      "loss": 0.9537,
+      "step": 10284
+    },
+    {
+      "epoch": 45.92,
+      "learning_rate": 8.169642857142858e-06,
+      "loss": 0.9529,
+      "step": 10285
+    },
+    {
+      "epoch": 45.92,
+      "learning_rate": 8.160714285714286e-06,
+      "loss": 0.9232,
+      "step": 10286
+    },
+    {
+      "epoch": 45.92,
+      "learning_rate": 8.151785714285714e-06,
+      "loss": 0.9239,
+      "step": 10287
+    },
+    {
+      "epoch": 45.93,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.9312,
+      "step": 10288
+    },
+    {
+      "epoch": 45.93,
+      "learning_rate": 8.133928571428572e-06,
+      "loss": 0.8335,
+      "step": 10289
+    },
+    {
+      "epoch": 45.94,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.9607,
+      "step": 10290
+    },
+    {
+      "epoch": 45.94,
+      "learning_rate": 8.116071428571428e-06,
+      "loss": 0.9254,
+      "step": 10291
+    },
+    {
+      "epoch": 45.95,
+      "learning_rate": 8.107142857142857e-06,
+      "loss": 0.8988,
+      "step": 10292
+    },
+    {
+      "epoch": 45.95,
+      "learning_rate": 8.098214285714287e-06,
+      "loss": 0.9763,
+      "step": 10293
+    },
+    {
+      "epoch": 45.96,
+      "learning_rate": 8.089285714285715e-06,
+      "loss": 0.9112,
+      "step": 10294
+    },
+    {
+      "epoch": 45.96,
+      "learning_rate": 8.080357142857143e-06,
+      "loss": 0.9348,
+      "step": 10295
+    },
+    {
+      "epoch": 45.96,
+      "learning_rate": 8.07142857142857e-06,
+      "loss": 0.9402,
+      "step": 10296
+    },
+    {
+      "epoch": 45.97,
+      "learning_rate": 8.062500000000001e-06,
+      "loss": 0.9452,
+      "step": 10297
+    },
+    {
+      "epoch": 45.97,
+      "learning_rate": 8.053571428571428e-06,
+      "loss": 0.8818,
+      "step": 10298
+    },
+    {
+      "epoch": 45.98,
+      "learning_rate": 8.044642857142857e-06,
+      "loss": 0.9151,
+      "step": 10299
+    },
+    {
+      "epoch": 45.98,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.9563,
+      "step": 10300
+    },
+    {
+      "epoch": 45.99,
+      "learning_rate": 8.026785714285715e-06,
+      "loss": 0.8952,
+      "step": 10301
+    },
+    {
+      "epoch": 45.99,
+      "learning_rate": 8.017857142857144e-06,
+      "loss": 0.9203,
+      "step": 10302
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 8.00892857142857e-06,
+      "loss": 0.9189,
+      "step": 10303
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.869,
+      "step": 10304
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 7.991071428571428e-06,
+      "loss": 0.8794,
+      "step": 10305
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 7.982142857142857e-06,
+      "loss": 0.8668,
+      "step": 10306
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 7.973214285714286e-06,
+      "loss": 0.9297,
+      "step": 10307
+    },
+    {
+      "epoch": 46.02,
+      "learning_rate": 7.964285714285715e-06,
+      "loss": 0.9375,
+      "step": 10308
+    },
+    {
+      "epoch": 46.02,
+      "learning_rate": 7.955357142857144e-06,
+      "loss": 0.9738,
+      "step": 10309
+    },
+    {
+      "epoch": 46.03,
+      "learning_rate": 7.946428571428571e-06,
+      "loss": 0.8963,
+      "step": 10310
+    },
+    {
+      "epoch": 46.03,
+      "learning_rate": 7.9375e-06,
+      "loss": 0.9086,
+      "step": 10311
+    },
+    {
+      "epoch": 46.04,
+      "learning_rate": 7.928571428571429e-06,
+      "loss": 0.9598,
+      "step": 10312
+    },
+    {
+      "epoch": 46.04,
+      "learning_rate": 7.919642857142857e-06,
+      "loss": 0.9091,
+      "step": 10313
+    },
+    {
+      "epoch": 46.04,
+      "learning_rate": 7.910714285714286e-06,
+      "loss": 1.0036,
+      "step": 10314
+    },
+    {
+      "epoch": 46.05,
+      "learning_rate": 7.901785714285715e-06,
+      "loss": 0.963,
+      "step": 10315
+    },
+    {
+      "epoch": 46.05,
+      "learning_rate": 7.892857142857144e-06,
+      "loss": 0.9114,
+      "step": 10316
+    },
+    {
+      "epoch": 46.06,
+      "learning_rate": 7.883928571428571e-06,
+      "loss": 0.9682,
+      "step": 10317
+    },
+    {
+      "epoch": 46.06,
+      "learning_rate": 7.875e-06,
+      "loss": 0.9998,
+      "step": 10318
+    },
+    {
+      "epoch": 46.07,
+      "learning_rate": 7.866071428571429e-06,
+      "loss": 0.9869,
+      "step": 10319
+    },
+    {
+      "epoch": 46.07,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.9537,
+      "step": 10320
+    },
+    {
+      "epoch": 46.08,
+      "learning_rate": 7.848214285714287e-06,
+      "loss": 0.9664,
+      "step": 10321
+    },
+    {
+      "epoch": 46.08,
+      "learning_rate": 7.839285714285714e-06,
+      "loss": 0.9123,
+      "step": 10322
+    },
+    {
+      "epoch": 46.08,
+      "learning_rate": 7.830357142857144e-06,
+      "loss": 0.9231,
+      "step": 10323
+    },
+    {
+      "epoch": 46.09,
+      "learning_rate": 7.821428571428571e-06,
+      "loss": 0.959,
+      "step": 10324
+    },
+    {
+      "epoch": 46.09,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.8572,
+      "step": 10325
+    },
+    {
+      "epoch": 46.1,
+      "learning_rate": 7.803571428571429e-06,
+      "loss": 0.8797,
+      "step": 10326
+    },
+    {
+      "epoch": 46.1,
+      "learning_rate": 7.794642857142858e-06,
+      "loss": 0.9148,
+      "step": 10327
+    },
+    {
+      "epoch": 46.11,
+      "learning_rate": 7.785714285714287e-06,
+      "loss": 0.9027,
+      "step": 10328
+    },
+    {
+      "epoch": 46.11,
+      "learning_rate": 7.776785714285714e-06,
+      "loss": 0.9908,
+      "step": 10329
+    },
+    {
+      "epoch": 46.12,
+      "learning_rate": 7.767857142857143e-06,
+      "loss": 1.0364,
+      "step": 10330
+    },
+    {
+      "epoch": 46.12,
+      "learning_rate": 7.758928571428571e-06,
+      "loss": 0.9181,
+      "step": 10331
+    },
+    {
+      "epoch": 46.12,
+      "learning_rate": 7.75e-06,
+      "loss": 0.9464,
+      "step": 10332
+    },
+    {
+      "epoch": 46.13,
+      "learning_rate": 7.74107142857143e-06,
+      "loss": 0.9744,
+      "step": 10333
+    },
+    {
+      "epoch": 46.13,
+      "learning_rate": 7.732142857142858e-06,
+      "loss": 1.0319,
+      "step": 10334
+    },
+    {
+      "epoch": 46.14,
+      "learning_rate": 7.723214285714287e-06,
+      "loss": 0.9927,
+      "step": 10335
+    },
+    {
+      "epoch": 46.14,
+      "learning_rate": 7.714285714285714e-06,
+      "loss": 0.9083,
+      "step": 10336
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 7.705357142857143e-06,
+      "loss": 0.9598,
+      "step": 10337
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 7.696428571428572e-06,
+      "loss": 0.8948,
+      "step": 10338
+    },
+    {
+      "epoch": 46.16,
+      "learning_rate": 7.6875e-06,
+      "loss": 0.9794,
+      "step": 10339
+    },
+    {
+      "epoch": 46.16,
+      "learning_rate": 7.67857142857143e-06,
+      "loss": 0.9323,
+      "step": 10340
+    },
+    {
+      "epoch": 46.17,
+      "learning_rate": 7.669642857142856e-06,
+      "loss": 1.0063,
+      "step": 10341
+    },
+    {
+      "epoch": 46.17,
+      "learning_rate": 7.660714285714287e-06,
+      "loss": 0.9172,
+      "step": 10342
+    },
+    {
+      "epoch": 46.17,
+      "learning_rate": 7.651785714285714e-06,
+      "loss": 0.9717,
+      "step": 10343
+    },
+    {
+      "epoch": 46.18,
+      "learning_rate": 7.642857142857143e-06,
+      "loss": 0.9449,
+      "step": 10344
+    },
+    {
+      "epoch": 46.18,
+      "learning_rate": 7.633928571428572e-06,
+      "loss": 0.9715,
+      "step": 10345
+    },
+    {
+      "epoch": 46.19,
+      "learning_rate": 7.625e-06,
+      "loss": 0.8758,
+      "step": 10346
+    },
+    {
+      "epoch": 46.19,
+      "learning_rate": 7.6160714285714295e-06,
+      "loss": 0.8838,
+      "step": 10347
+    },
+    {
+      "epoch": 46.2,
+      "learning_rate": 7.6071428571428575e-06,
+      "loss": 0.9335,
+      "step": 10348
+    },
+    {
+      "epoch": 46.2,
+      "learning_rate": 7.598214285714286e-06,
+      "loss": 0.9756,
+      "step": 10349
+    },
+    {
+      "epoch": 46.21,
+      "learning_rate": 7.589285714285714e-06,
+      "loss": 0.9319,
+      "step": 10350
+    },
+    {
+      "epoch": 46.21,
+      "learning_rate": 7.580357142857143e-06,
+      "loss": 0.9544,
+      "step": 10351
+    },
+    {
+      "epoch": 46.21,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.8896,
+      "step": 10352
+    },
+    {
+      "epoch": 46.22,
+      "learning_rate": 7.5625e-06,
+      "loss": 0.8854,
+      "step": 10353
+    },
+    {
+      "epoch": 46.22,
+      "learning_rate": 7.55357142857143e-06,
+      "loss": 0.9453,
+      "step": 10354
+    },
+    {
+      "epoch": 46.23,
+      "learning_rate": 7.544642857142857e-06,
+      "loss": 0.9413,
+      "step": 10355
+    },
+    {
+      "epoch": 46.23,
+      "learning_rate": 7.5357142857142865e-06,
+      "loss": 1.0237,
+      "step": 10356
+    },
+    {
+      "epoch": 46.24,
+      "learning_rate": 7.526785714285714e-06,
+      "loss": 0.8547,
+      "step": 10357
+    },
+    {
+      "epoch": 46.24,
+      "learning_rate": 7.517857142857143e-06,
+      "loss": 0.9445,
+      "step": 10358
+    },
+    {
+      "epoch": 46.25,
+      "learning_rate": 7.508928571428572e-06,
+      "loss": 0.9415,
+      "step": 10359
+    },
+    {
+      "epoch": 46.25,
+      "learning_rate": 7.5e-06,
+      "loss": 1.0119,
+      "step": 10360
+    },
+    {
+      "epoch": 46.25,
+      "learning_rate": 7.491071428571429e-06,
+      "loss": 0.9296,
+      "step": 10361
+    },
+    {
+      "epoch": 46.26,
+      "learning_rate": 7.482142857142857e-06,
+      "loss": 0.9127,
+      "step": 10362
+    },
+    {
+      "epoch": 46.26,
+      "learning_rate": 7.473214285714286e-06,
+      "loss": 1.0144,
+      "step": 10363
+    },
+    {
+      "epoch": 46.27,
+      "learning_rate": 7.4642857142857155e-06,
+      "loss": 0.9553,
+      "step": 10364
+    },
+    {
+      "epoch": 46.27,
+      "learning_rate": 7.455357142857143e-06,
+      "loss": 0.9444,
+      "step": 10365
+    },
+    {
+      "epoch": 46.28,
+      "learning_rate": 7.446428571428572e-06,
+      "loss": 0.9608,
+      "step": 10366
+    },
+    {
+      "epoch": 46.28,
+      "learning_rate": 7.4375e-06,
+      "loss": 0.9285,
+      "step": 10367
+    },
+    {
+      "epoch": 46.29,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.9531,
+      "step": 10368
+    },
+    {
+      "epoch": 46.29,
+      "learning_rate": 7.419642857142857e-06,
+      "loss": 0.9208,
+      "step": 10369
+    },
+    {
+      "epoch": 46.29,
+      "learning_rate": 7.410714285714286e-06,
+      "loss": 0.9744,
+      "step": 10370
+    },
+    {
+      "epoch": 46.3,
+      "learning_rate": 7.401785714285715e-06,
+      "loss": 0.9352,
+      "step": 10371
+    },
+    {
+      "epoch": 46.3,
+      "learning_rate": 7.392857142857143e-06,
+      "loss": 0.9171,
+      "step": 10372
+    },
+    {
+      "epoch": 46.31,
+      "learning_rate": 7.3839285714285725e-06,
+      "loss": 0.97,
+      "step": 10373
+    },
+    {
+      "epoch": 46.31,
+      "learning_rate": 7.375e-06,
+      "loss": 0.9309,
+      "step": 10374
+    },
+    {
+      "epoch": 46.32,
+      "learning_rate": 7.366071428571429e-06,
+      "loss": 0.9561,
+      "step": 10375
+    },
+    {
+      "epoch": 46.32,
+      "learning_rate": 7.3571428571428565e-06,
+      "loss": 0.9092,
+      "step": 10376
+    },
+    {
+      "epoch": 46.33,
+      "learning_rate": 7.348214285714286e-06,
+      "loss": 0.9282,
+      "step": 10377
+    },
+    {
+      "epoch": 46.33,
+      "learning_rate": 7.339285714285715e-06,
+      "loss": 0.9305,
+      "step": 10378
+    },
+    {
+      "epoch": 46.33,
+      "learning_rate": 7.330357142857143e-06,
+      "loss": 0.959,
+      "step": 10379
+    },
+    {
+      "epoch": 46.34,
+      "learning_rate": 7.321428571428572e-06,
+      "loss": 0.97,
+      "step": 10380
+    },
+    {
+      "epoch": 46.34,
+      "learning_rate": 7.3125e-06,
+      "loss": 0.8809,
+      "step": 10381
+    },
+    {
+      "epoch": 46.35,
+      "learning_rate": 7.303571428571429e-06,
+      "loss": 0.9306,
+      "step": 10382
+    },
+    {
+      "epoch": 46.35,
+      "learning_rate": 7.294642857142857e-06,
+      "loss": 0.9502,
+      "step": 10383
+    },
+    {
+      "epoch": 46.36,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 0.9589,
+      "step": 10384
+    },
+    {
+      "epoch": 46.36,
+      "learning_rate": 7.276785714285715e-06,
+      "loss": 1.0195,
+      "step": 10385
+    },
+    {
+      "epoch": 46.37,
+      "learning_rate": 7.267857142857143e-06,
+      "loss": 0.9724,
+      "step": 10386
+    },
+    {
+      "epoch": 46.37,
+      "learning_rate": 7.258928571428572e-06,
+      "loss": 0.9467,
+      "step": 10387
+    },
+    {
+      "epoch": 46.38,
+      "learning_rate": 7.25e-06,
+      "loss": 0.918,
+      "step": 10388
+    },
+    {
+      "epoch": 46.38,
+      "learning_rate": 7.241071428571429e-06,
+      "loss": 0.9933,
+      "step": 10389
+    },
+    {
+      "epoch": 46.38,
+      "learning_rate": 7.2321428571428586e-06,
+      "loss": 0.9707,
+      "step": 10390
+    },
+    {
+      "epoch": 46.39,
+      "learning_rate": 7.223214285714286e-06,
+      "loss": 0.919,
+      "step": 10391
+    },
+    {
+      "epoch": 46.39,
+      "learning_rate": 7.214285714285715e-06,
+      "loss": 0.9125,
+      "step": 10392
+    },
+    {
+      "epoch": 46.4,
+      "learning_rate": 7.2053571428571425e-06,
+      "loss": 0.9806,
+      "step": 10393
+    },
+    {
+      "epoch": 46.4,
+      "learning_rate": 7.196428571428572e-06,
+      "loss": 0.9072,
+      "step": 10394
+    },
+    {
+      "epoch": 46.41,
+      "learning_rate": 7.187499999999999e-06,
+      "loss": 0.9252,
+      "step": 10395
+    },
+    {
+      "epoch": 46.41,
+      "learning_rate": 7.178571428571429e-06,
+      "loss": 1.0194,
+      "step": 10396
+    },
+    {
+      "epoch": 46.42,
+      "learning_rate": 7.169642857142858e-06,
+      "loss": 0.9336,
+      "step": 10397
+    },
+    {
+      "epoch": 46.42,
+      "learning_rate": 7.160714285714286e-06,
+      "loss": 0.9736,
+      "step": 10398
+    },
+    {
+      "epoch": 46.42,
+      "learning_rate": 7.151785714285715e-06,
+      "loss": 0.9172,
+      "step": 10399
+    },
+    {
+      "epoch": 46.43,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.9336,
+      "step": 10400
+    },
+    {
+      "epoch": 46.43,
+      "learning_rate": 7.1339285714285716e-06,
+      "loss": 0.9355,
+      "step": 10401
+    },
+    {
+      "epoch": 46.44,
+      "learning_rate": 7.1249999999999995e-06,
+      "loss": 0.983,
+      "step": 10402
+    },
+    {
+      "epoch": 46.44,
+      "learning_rate": 7.116071428571429e-06,
+      "loss": 1.01,
+      "step": 10403
+    },
+    {
+      "epoch": 46.45,
+      "learning_rate": 7.107142857142858e-06,
+      "loss": 0.9548,
+      "step": 10404
+    },
+    {
+      "epoch": 46.45,
+      "learning_rate": 7.098214285714286e-06,
+      "loss": 0.9681,
+      "step": 10405
+    },
+    {
+      "epoch": 46.46,
+      "learning_rate": 7.089285714285715e-06,
+      "loss": 0.9399,
+      "step": 10406
+    },
+    {
+      "epoch": 46.46,
+      "learning_rate": 7.080357142857143e-06,
+      "loss": 0.9598,
+      "step": 10407
+    },
+    {
+      "epoch": 46.46,
+      "learning_rate": 7.071428571428572e-06,
+      "loss": 0.9053,
+      "step": 10408
+    },
+    {
+      "epoch": 46.47,
+      "learning_rate": 7.0625e-06,
+      "loss": 0.9279,
+      "step": 10409
+    },
+    {
+      "epoch": 46.47,
+      "learning_rate": 7.0535714285714286e-06,
+      "loss": 0.9742,
+      "step": 10410
+    },
+    {
+      "epoch": 46.48,
+      "learning_rate": 7.044642857142858e-06,
+      "loss": 0.9962,
+      "step": 10411
+    },
+    {
+      "epoch": 46.48,
+      "learning_rate": 7.035714285714285e-06,
+      "loss": 0.9862,
+      "step": 10412
+    },
+    {
+      "epoch": 46.49,
+      "learning_rate": 7.026785714285715e-06,
+      "loss": 0.8847,
+      "step": 10413
+    },
+    {
+      "epoch": 46.49,
+      "learning_rate": 7.017857142857143e-06,
+      "loss": 0.9451,
+      "step": 10414
+    },
+    {
+      "epoch": 46.5,
+      "learning_rate": 7.008928571428572e-06,
+      "loss": 0.9115,
+      "step": 10415
+    },
+    {
+      "epoch": 46.5,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.0041,
+      "step": 10416
+    },
+    {
+      "epoch": 46.5,
+      "learning_rate": 6.991071428571429e-06,
+      "loss": 0.8752,
+      "step": 10417
+    },
+    {
+      "epoch": 46.51,
+      "learning_rate": 6.9821428571428576e-06,
+      "loss": 0.9878,
+      "step": 10418
+    },
+    {
+      "epoch": 46.51,
+      "learning_rate": 6.9732142857142856e-06,
+      "loss": 0.9551,
+      "step": 10419
+    },
+    {
+      "epoch": 46.52,
+      "learning_rate": 6.964285714285715e-06,
+      "loss": 0.9222,
+      "step": 10420
+    },
+    {
+      "epoch": 46.52,
+      "learning_rate": 6.955357142857142e-06,
+      "loss": 0.9567,
+      "step": 10421
+    },
+    {
+      "epoch": 46.53,
+      "learning_rate": 6.946428571428572e-06,
+      "loss": 0.9597,
+      "step": 10422
+    },
+    {
+      "epoch": 46.53,
+      "learning_rate": 6.937500000000001e-06,
+      "loss": 0.9622,
+      "step": 10423
+    },
+    {
+      "epoch": 46.54,
+      "learning_rate": 6.928571428571429e-06,
+      "loss": 0.9858,
+      "step": 10424
+    },
+    {
+      "epoch": 46.54,
+      "learning_rate": 6.919642857142858e-06,
+      "loss": 0.9874,
+      "step": 10425
+    },
+    {
+      "epoch": 46.54,
+      "learning_rate": 6.910714285714286e-06,
+      "loss": 0.9089,
+      "step": 10426
+    },
+    {
+      "epoch": 46.55,
+      "learning_rate": 6.901785714285715e-06,
+      "loss": 0.9837,
+      "step": 10427
+    },
+    {
+      "epoch": 46.55,
+      "learning_rate": 6.8928571428571426e-06,
+      "loss": 0.9037,
+      "step": 10428
+    },
+    {
+      "epoch": 46.56,
+      "learning_rate": 6.883928571428571e-06,
+      "loss": 0.9406,
+      "step": 10429
+    },
+    {
+      "epoch": 46.56,
+      "learning_rate": 6.875000000000001e-06,
+      "loss": 0.9246,
+      "step": 10430
+    },
+    {
+      "epoch": 46.57,
+      "learning_rate": 6.866071428571428e-06,
+      "loss": 0.9176,
+      "step": 10431
+    },
+    {
+      "epoch": 46.57,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.9788,
+      "step": 10432
+    },
+    {
+      "epoch": 46.58,
+      "learning_rate": 6.848214285714286e-06,
+      "loss": 0.9878,
+      "step": 10433
+    },
+    {
+      "epoch": 46.58,
+      "learning_rate": 6.839285714285715e-06,
+      "loss": 0.9307,
+      "step": 10434
+    },
+    {
+      "epoch": 46.58,
+      "learning_rate": 6.830357142857143e-06,
+      "loss": 0.9407,
+      "step": 10435
+    },
+    {
+      "epoch": 46.59,
+      "learning_rate": 6.821428571428572e-06,
+      "loss": 0.8822,
+      "step": 10436
+    },
+    {
+      "epoch": 46.59,
+      "learning_rate": 6.8125e-06,
+      "loss": 0.9119,
+      "step": 10437
+    },
+    {
+      "epoch": 46.6,
+      "learning_rate": 6.803571428571428e-06,
+      "loss": 1.007,
+      "step": 10438
+    },
+    {
+      "epoch": 46.6,
+      "learning_rate": 6.794642857142858e-06,
+      "loss": 0.9894,
+      "step": 10439
+    },
+    {
+      "epoch": 46.61,
+      "learning_rate": 6.785714285714285e-06,
+      "loss": 0.9541,
+      "step": 10440
+    },
+    {
+      "epoch": 46.61,
+      "learning_rate": 6.776785714285715e-06,
+      "loss": 0.8845,
+      "step": 10441
+    },
+    {
+      "epoch": 46.62,
+      "learning_rate": 6.767857142857144e-06,
+      "loss": 0.9564,
+      "step": 10442
+    },
+    {
+      "epoch": 46.62,
+      "learning_rate": 6.758928571428572e-06,
+      "loss": 0.9291,
+      "step": 10443
+    },
+    {
+      "epoch": 46.62,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 0.9543,
+      "step": 10444
+    },
+    {
+      "epoch": 46.63,
+      "learning_rate": 6.741071428571429e-06,
+      "loss": 0.9445,
+      "step": 10445
+    },
+    {
+      "epoch": 46.63,
+      "learning_rate": 6.7321428571428574e-06,
+      "loss": 0.9684,
+      "step": 10446
+    },
+    {
+      "epoch": 46.64,
+      "learning_rate": 6.723214285714285e-06,
+      "loss": 0.8862,
+      "step": 10447
+    },
+    {
+      "epoch": 46.64,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 0.961,
+      "step": 10448
+    },
+    {
+      "epoch": 46.65,
+      "learning_rate": 6.705357142857144e-06,
+      "loss": 0.9109,
+      "step": 10449
+    },
+    {
+      "epoch": 46.65,
+      "learning_rate": 6.696428571428572e-06,
+      "loss": 0.9136,
+      "step": 10450
+    },
+    {
+      "epoch": 46.66,
+      "learning_rate": 6.687500000000001e-06,
+      "loss": 0.9813,
+      "step": 10451
+    },
+    {
+      "epoch": 46.66,
+      "learning_rate": 6.678571428571429e-06,
+      "loss": 0.996,
+      "step": 10452
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 6.669642857142858e-06,
+      "loss": 0.8909,
+      "step": 10453
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 6.660714285714286e-06,
+      "loss": 0.9373,
+      "step": 10454
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 6.6517857142857144e-06,
+      "loss": 0.9688,
+      "step": 10455
+    },
+    {
+      "epoch": 46.68,
+      "learning_rate": 6.642857142857144e-06,
+      "loss": 0.9847,
+      "step": 10456
+    },
+    {
+      "epoch": 46.68,
+      "learning_rate": 6.633928571428571e-06,
+      "loss": 0.9287,
+      "step": 10457
+    },
+    {
+      "epoch": 46.69,
+      "learning_rate": 6.625000000000001e-06,
+      "loss": 0.9219,
+      "step": 10458
+    },
+    {
+      "epoch": 46.69,
+      "learning_rate": 6.616071428571428e-06,
+      "loss": 0.9206,
+      "step": 10459
+    },
+    {
+      "epoch": 46.7,
+      "learning_rate": 6.607142857142858e-06,
+      "loss": 0.9839,
+      "step": 10460
+    },
+    {
+      "epoch": 46.7,
+      "learning_rate": 6.598214285714286e-06,
+      "loss": 0.9903,
+      "step": 10461
+    },
+    {
+      "epoch": 46.71,
+      "learning_rate": 6.589285714285715e-06,
+      "loss": 0.9949,
+      "step": 10462
+    },
+    {
+      "epoch": 46.71,
+      "learning_rate": 6.5803571428571434e-06,
+      "loss": 0.9459,
+      "step": 10463
+    },
+    {
+      "epoch": 46.71,
+      "learning_rate": 6.5714285714285714e-06,
+      "loss": 0.9152,
+      "step": 10464
+    },
+    {
+      "epoch": 46.72,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.9057,
+      "step": 10465
+    },
+    {
+      "epoch": 46.72,
+      "learning_rate": 6.553571428571428e-06,
+      "loss": 0.9313,
+      "step": 10466
+    },
+    {
+      "epoch": 46.73,
+      "learning_rate": 6.544642857142857e-06,
+      "loss": 0.8659,
+      "step": 10467
+    },
+    {
+      "epoch": 46.73,
+      "learning_rate": 6.535714285714287e-06,
+      "loss": 1.0041,
+      "step": 10468
+    },
+    {
+      "epoch": 46.74,
+      "learning_rate": 6.526785714285715e-06,
+      "loss": 0.9438,
+      "step": 10469
+    },
+    {
+      "epoch": 46.74,
+      "learning_rate": 6.517857142857144e-06,
+      "loss": 0.9567,
+      "step": 10470
+    },
+    {
+      "epoch": 46.75,
+      "learning_rate": 6.508928571428572e-06,
+      "loss": 0.8585,
+      "step": 10471
+    },
+    {
+      "epoch": 46.75,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.0017,
+      "step": 10472
+    },
+    {
+      "epoch": 46.75,
+      "learning_rate": 6.4910714285714284e-06,
+      "loss": 1.046,
+      "step": 10473
+    },
+    {
+      "epoch": 46.76,
+      "learning_rate": 6.482142857142857e-06,
+      "loss": 0.9386,
+      "step": 10474
+    },
+    {
+      "epoch": 46.76,
+      "learning_rate": 6.473214285714287e-06,
+      "loss": 1.0263,
+      "step": 10475
+    },
+    {
+      "epoch": 46.77,
+      "learning_rate": 6.464285714285714e-06,
+      "loss": 0.9341,
+      "step": 10476
+    },
+    {
+      "epoch": 46.77,
+      "learning_rate": 6.455357142857144e-06,
+      "loss": 0.9086,
+      "step": 10477
+    },
+    {
+      "epoch": 46.78,
+      "learning_rate": 6.446428571428571e-06,
+      "loss": 0.971,
+      "step": 10478
+    },
+    {
+      "epoch": 46.78,
+      "learning_rate": 6.437500000000001e-06,
+      "loss": 1.0152,
+      "step": 10479
+    },
+    {
+      "epoch": 46.79,
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 0.9421,
+      "step": 10480
+    },
+    {
+      "epoch": 46.79,
+      "learning_rate": 6.4196428571428575e-06,
+      "loss": 0.934,
+      "step": 10481
+    },
+    {
+      "epoch": 46.79,
+      "learning_rate": 6.410714285714286e-06,
+      "loss": 0.8898,
+      "step": 10482
+    },
+    {
+      "epoch": 46.8,
+      "learning_rate": 6.401785714285714e-06,
+      "loss": 0.9277,
+      "step": 10483
+    },
+    {
+      "epoch": 46.8,
+      "learning_rate": 6.392857142857143e-06,
+      "loss": 0.9739,
+      "step": 10484
+    },
+    {
+      "epoch": 46.81,
+      "learning_rate": 6.383928571428571e-06,
+      "loss": 0.9415,
+      "step": 10485
+    },
+    {
+      "epoch": 46.81,
+      "learning_rate": 6.375000000000001e-06,
+      "loss": 0.9472,
+      "step": 10486
+    },
+    {
+      "epoch": 46.82,
+      "learning_rate": 6.366071428571428e-06,
+      "loss": 0.9137,
+      "step": 10487
+    },
+    {
+      "epoch": 46.82,
+      "learning_rate": 6.357142857142858e-06,
+      "loss": 0.9401,
+      "step": 10488
+    },
+    {
+      "epoch": 46.83,
+      "learning_rate": 6.3482142857142865e-06,
+      "loss": 0.9017,
+      "step": 10489
+    },
+    {
+      "epoch": 46.83,
+      "learning_rate": 6.3392857142857145e-06,
+      "loss": 0.9558,
+      "step": 10490
+    },
+    {
+      "epoch": 46.83,
+      "learning_rate": 6.330357142857143e-06,
+      "loss": 0.9105,
+      "step": 10491
+    },
+    {
+      "epoch": 46.84,
+      "learning_rate": 6.321428571428571e-06,
+      "loss": 0.8917,
+      "step": 10492
+    },
+    {
+      "epoch": 46.84,
+      "learning_rate": 6.3125e-06,
+      "loss": 0.8952,
+      "step": 10493
+    },
+    {
+      "epoch": 46.85,
+      "learning_rate": 6.30357142857143e-06,
+      "loss": 1.0182,
+      "step": 10494
+    },
+    {
+      "epoch": 46.85,
+      "learning_rate": 6.294642857142857e-06,
+      "loss": 1.0093,
+      "step": 10495
+    },
+    {
+      "epoch": 46.86,
+      "learning_rate": 6.285714285714287e-06,
+      "loss": 0.8734,
+      "step": 10496
+    },
+    {
+      "epoch": 46.86,
+      "learning_rate": 6.276785714285715e-06,
+      "loss": 0.9765,
+      "step": 10497
+    },
+    {
+      "epoch": 46.87,
+      "learning_rate": 6.2678571428571435e-06,
+      "loss": 0.953,
+      "step": 10498
+    },
+    {
+      "epoch": 46.87,
+      "learning_rate": 6.2589285714285715e-06,
+      "loss": 0.9796,
+      "step": 10499
+    },
+    {
+      "epoch": 46.88,
+      "learning_rate": 6.25e-06,
+      "loss": 0.9316,
+      "step": 10500
+    },
+    {
+      "epoch": 46.88,
+      "eval_loss": 1.1098241806030273,
+      "eval_runtime": 44.5486,
+      "eval_samples_per_second": 293.926,
+      "eval_steps_per_second": 1.549,
+      "step": 10500
     }
   ],
-  "max_steps": 5070,
-  "num_train_epochs": 15,
-  "total_flos": 4.898288883108741e+17,
+  "logging_steps": 1.0,
+  "max_steps": 11200,
+  "num_train_epochs": 50,
+  "save_steps": 300,
+  "total_flos": 1.02660498633071e+18,
   "trial_name": null,
   "trial_params": null
 }