diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18045 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9236453201970444,
+  "eval_steps": 1000,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999983553493e-05,
+      "loss": 3.4043,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993421397383e-05,
+      "loss": 3.0831,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998519814493e-05,
+      "loss": 3.0599,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999973685592996e-05,
+      "loss": 3.0168,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995888374311e-05,
+      "loss": 3.1918,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994079259722e-05,
+      "loss": 3.1395,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999194121577e-05,
+      "loss": 3.205,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989474242737e-05,
+      "loss": 3.1289,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986678340948e-05,
+      "loss": 3.0062,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983553510769e-05,
+      "loss": 3.0641,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980099752612e-05,
+      "loss": 3.2372,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976317066932e-05,
+      "loss": 3.2362,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972205454226e-05,
+      "loss": 3.1711,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999677649150356e-05,
+      "loss": 3.0935,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962995449945e-05,
+      "loss": 3.2991,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957897059582e-05,
+      "loss": 3.0897,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999524697446157e-05,
+      "loss": 3.1639,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946713505762e-05,
+      "loss": 3.0932,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940628343778e-05,
+      "loss": 3.2151,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934214259463e-05,
+      "loss": 3.2069,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927471253663e-05,
+      "loss": 3.0901,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999203993272625e-05,
+      "loss": 3.1199,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912998481195e-05,
+      "loss": 2.9998,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999052687164324e-05,
+      "loss": 3.1484,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897210033992e-05,
+      "loss": 3.225,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888822434934e-05,
+      "loss": 2.9661,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880105920363e-05,
+      "loss": 3.1064,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871060491424e-05,
+      "loss": 3.086,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998616861493084e-05,
+      "loss": 3.0362,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999851982895249e-05,
+      "loss": 3.2671,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841950730524e-05,
+      "loss": 2.927,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831589656452e-05,
+      "loss": 3.0513,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998208996743954e-05,
+      "loss": 3.1097,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998098807857625e-05,
+      "loss": 3.088,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798532992002e-05,
+      "loss": 3.1616,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997868562946076e-05,
+      "loss": 3.011,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997748506951145e-05,
+      "loss": 2.9767,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997625161951045e-05,
+      "loss": 2.867,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749852796198e-05,
+      "loss": 3.0862,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997368605000624e-05,
+      "loss": 3.093,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723539308406e-05,
+      "loss": 3.1318,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709889222984e-05,
+      "loss": 3.0994,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9996959102455896e-05,
+      "loss": 3.0203,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996816023780646e-05,
+      "loss": 3.1066,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996669656222896e-05,
+      "loss": 2.8783,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996519999801906e-05,
+      "loss": 3.079,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996367054537366e-05,
+      "loss": 3.0328,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999621082044941e-05,
+      "loss": 3.1569,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999605129755858e-05,
+      "loss": 3.0146,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995888485885875e-05,
+      "loss": 2.9727,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995722385452724e-05,
+      "loss": 3.0759,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555299628096e-05,
+      "loss": 3.0252,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999538031839288e-05,
+      "loss": 2.9176,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999520435181121e-05,
+      "loss": 2.9924,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99950250965591e-05,
+      "loss": 3.1252,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994842552660124e-05,
+      "loss": 3.1506,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999465672013831e-05,
+      "loss": 3.0059,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99944675990181e-05,
+      "loss": 3.2809,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427518932439e-05,
+      "loss": 3.1613,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999407949108249e-05,
+      "loss": 3.0878,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388050431813e-05,
+      "loss": 3.0991,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993678229057525e-05,
+      "loss": 2.8576,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347266532727e-05,
+      "loss": 3.0998,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993263813154414e-05,
+      "loss": 2.9551,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305167256643e-05,
+      "loss": 3.0857,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999283624359123e-05,
+      "loss": 3.0168,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992617526257165e-05,
+      "loss": 3.2864,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992395520593015e-05,
+      "loss": 2.9523,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992170226627985e-05,
+      "loss": 3.0138,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999194164439172e-05,
+      "loss": 3.0991,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991709773914294e-05,
+      "loss": 3.0625,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991474615226206e-05,
+      "loss": 3.0028,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991236168358405e-05,
+      "loss": 2.9602,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999099443334227e-05,
+      "loss": 3.1386,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999074941020959e-05,
+      "loss": 2.9219,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990501098992616e-05,
+      "loss": 3.0663,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990249499724016e-05,
+      "loss": 3.0705,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99899946124369e-05,
+      "loss": 2.9632,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973643716478e-05,
+      "loss": 3.1885,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998947497394165e-05,
+      "loss": 2.8849,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989210222801906e-05,
+      "loss": 3.0083,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998894218378037e-05,
+      "loss": 2.9712,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998867085691232e-05,
+      "loss": 3.0082,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998839624223346e-05,
+      "loss": 3.1991,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998811833977991e-05,
+      "loss": 2.8934,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998783714958823e-05,
+      "loss": 2.9454,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998755267169543e-05,
+      "loss": 3.228,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998726490613893e-05,
+      "loss": 2.958,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986973852956595e-05,
+      "loss": 2.8969,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998667951218673e-05,
+      "loss": 3.0094,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998638188386805e-05,
+      "loss": 2.9002,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986080968039706e-05,
+      "loss": 3.1021,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998577676474131e-05,
+      "loss": 2.877,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985469274012874e-05,
+      "loss": 3.0086,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985158495894865e-05,
+      "loss": 3.0713,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998484443042816e-05,
+      "loss": 3.0395,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984527077654086e-05,
+      "loss": 2.9055,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998420643761441e-05,
+      "loss": 2.9538,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99838825103513e-05,
+      "loss": 3.1361,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998355529590738e-05,
+      "loss": 3.0713,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998322479432571e-05,
+      "loss": 2.8894,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998289100564978e-05,
+      "loss": 2.9866,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998255392992349e-05,
+      "loss": 2.8606,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982213567191205e-05,
+      "loss": 3.0092,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998186991749769e-05,
+      "loss": 2.8533,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981522980888184e-05,
+      "loss": 3.0747,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981172757408315e-05,
+      "loss": 2.8565,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998081924710417e-05,
+      "loss": 3.1623,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998046245002225e-05,
+      "loss": 3.1386,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980102366209516e-05,
+      "loss": 3.1452,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997973899571334e-05,
+      "loss": 3.0869,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997937233858153e-05,
+      "loss": 2.9946,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997900239486232e-05,
+      "loss": 2.6081,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978629164604396e-05,
+      "loss": 3.1378,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997825264785686e-05,
+      "loss": 2.8166,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997787284466925e-05,
+      "loss": 3.0611,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977489755091536e-05,
+      "loss": 3.0264,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997710337917413e-05,
+      "loss": 2.9936,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997671371696786e-05,
+      "loss": 3.0462,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976320768524e-05,
+      "loss": 3.0403,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997592453389424e-05,
+      "loss": 3.0577,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997552501313073e-05,
+      "loss": 3.0858,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997512220628603e-05,
+      "loss": 3.0108,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997471611341312e-05,
+      "loss": 3.0456,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997430673456546e-05,
+      "loss": 2.9383,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99738940697969e-05,
+      "loss": 2.8973,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997347811916173e-05,
+      "loss": 2.9673,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997305888271469e-05,
+      "loss": 3.2665,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997263636051092e-05,
+      "loss": 3.0901,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9972210552606026e-05,
+      "loss": 3.1694,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997178145905602e-05,
+      "loss": 3.1717,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9971349079917384e-05,
+      "loss": 2.9002,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997091341524699e-05,
+      "loss": 2.8667,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997047446510216e-05,
+      "loss": 3.0491,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997003222954064e-05,
+      "loss": 2.8327,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969586708620633e-05,
+      "loss": 3.161,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996913790240075e-05,
+      "loss": 2.861,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968685810940034e-05,
+      "loss": 2.8602,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996823043429798e-05,
+      "loss": 3.108,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996777177253449e-05,
+      "loss": 3.0082,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996730982570992e-05,
+      "loss": 2.8616,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996684459388504e-05,
+      "loss": 2.9973,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996637607712108e-05,
+      "loss": 2.9611,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996590427547967e-05,
+      "loss": 3.0271,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965429189022884e-05,
+      "loss": 2.9866,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996495081781323e-05,
+      "loss": 2.957,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9964469161913656e-05,
+      "loss": 3.2091,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996398422138753e-05,
+      "loss": 2.9834,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996349599629866e-05,
+      "loss": 2.9675,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996300448671128e-05,
+      "loss": 3.1283,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996250969269006e-05,
+      "loss": 2.9858,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99620116143001e-05,
+      "loss": 3.0445,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996151025160692e-05,
+      "loss": 3.099,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996100560467651e-05,
+      "loss": 3.0977,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996049767357526e-05,
+      "loss": 2.8504,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9959986458369987e-05,
+      "loss": 2.9713,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995947195912797e-05,
+      "loss": 2.8969,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9958954175916886e-05,
+      "loss": 3.065,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9958433108804875e-05,
+      "loss": 2.9295,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995790875786048e-05,
+      "loss": 2.8974,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995738112315271e-05,
+      "loss": 2.9119,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995685020475097e-05,
+      "loss": 2.853,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995631600272512e-05,
+      "loss": 3.0124,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995577851714545e-05,
+      "loss": 2.9924,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9955237748082674e-05,
+      "loss": 2.7665,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9954693695607945e-05,
+      "loss": 2.962,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995414635979284e-05,
+      "loss": 2.7199,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995359574070937e-05,
+      "loss": 2.8501,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995304183842999e-05,
+      "loss": 2.9089,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995248465302759e-05,
+      "loss": 2.9998,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995192418457545e-05,
+      "loss": 2.9771,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995136043314733e-05,
+      "loss": 2.9403,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9950793398817396e-05,
+      "loss": 3.027,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995022308166026e-05,
+      "loss": 3.0418,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994964948175097e-05,
+      "loss": 2.7615,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9949072599164974e-05,
+      "loss": 2.9725,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994849243397819e-05,
+      "loss": 2.9498,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994790898626695e-05,
+      "loss": 2.9932,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9947322256107996e-05,
+      "loss": 2.9658,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994673224357856e-05,
+      "loss": 2.907,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9946138948756254e-05,
+      "loss": 2.9849,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994554237171914e-05,
+      "loss": 3.016,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994494251254571e-05,
+      "loss": 3.0119,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9944339371314886e-05,
+      "loss": 3.0532,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9943732948106036e-05,
+      "loss": 2.8489,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9943123242998935e-05,
+      "loss": 2.9061,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9942510256073814e-05,
+      "loss": 3.0304,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9941893987411316e-05,
+      "loss": 3.0288,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994127443709253e-05,
+      "loss": 2.9195,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940651605198976e-05,
+      "loss": 2.705,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940025491812586e-05,
+      "loss": 3.1535,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993939609701575e-05,
+      "loss": 2.8932,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993876342089129e-05,
+      "loss": 2.9281,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993812746352242e-05,
+      "loss": 2.9596,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993748822499284e-05,
+      "loss": 2.7809,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993684570538664e-05,
+      "loss": 3.0385,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9936199904788365e-05,
+      "loss": 2.8533,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9935550823282984e-05,
+      "loss": 2.9191,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9934898460955895e-05,
+      "loss": 2.8818,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993424281789294e-05,
+      "loss": 2.9768,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993358389418037e-05,
+      "loss": 2.7508,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993292168990489e-05,
+      "loss": 3.2728,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993225620515362e-05,
+      "loss": 3.174,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993158744001412e-05,
+      "loss": 2.9845,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9930915394574396e-05,
+      "loss": 2.8783,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993024006892285e-05,
+      "loss": 2.9485,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992956146314834e-05,
+      "loss": 2.8678,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992887957734017e-05,
+      "loss": 2.9201,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992819441158804e-05,
+      "loss": 3.1036,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99275059659821e-05,
+      "loss": 2.8722,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992681424061293e-05,
+      "loss": 2.7485,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9926119235571547e-05,
+      "loss": 3.1938,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9925420950949385e-05,
+      "loss": 3.2015,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992471938683834e-05,
+      "loss": 3.0266,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992401454333069e-05,
+      "loss": 2.7638,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992330642051919e-05,
+      "loss": 2.9698,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9922595018497006e-05,
+      "loss": 2.873,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992188033735773e-05,
+      "loss": 3.0474,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9921162377195415e-05,
+      "loss": 2.8571,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99204411381045e-05,
+      "loss": 3.1113,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99197166201799e-05,
+      "loss": 2.9413,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9918988823516936e-05,
+      "loss": 3.0436,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991825774821135e-05,
+      "loss": 2.7534,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991752339435935e-05,
+      "loss": 3.0493,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991678576205755e-05,
+      "loss": 3.0166,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9916044851403e-05,
+      "loss": 2.8793,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991530066249318e-05,
+      "loss": 2.8588,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9914553195426015e-05,
+      "loss": 3.1126,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913802450299854e-05,
+      "loss": 3.0037,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913048427213446e-05,
+      "loss": 2.9035,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991229112626603e-05,
+      "loss": 3.0389,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991153054755723e-05,
+      "loss": 2.8425,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991076669118713e-05,
+      "loss": 2.8784,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990999955725621e-05,
+      "loss": 2.9768,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990922914586543e-05,
+      "loss": 3.0366,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990845545711613e-05,
+      "loss": 3.1077,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990767849111012e-05,
+      "loss": 2.9409,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906898247949626e-05,
+      "loss": 3.0924,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906114727737306e-05,
+      "loss": 2.8688,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990532793057624e-05,
+      "loss": 2.8826,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9904537856569967e-05,
+      "loss": 3.1227,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990374450582242e-05,
+      "loss": 3.1255,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990294787843799e-05,
+      "loss": 2.8354,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9902147974521495e-05,
+      "loss": 3.0734,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9901344794178176e-05,
+      "loss": 2.7941,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99005383375137e-05,
+      "loss": 2.9046,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989972860463419e-05,
+      "loss": 3.0516,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989891559564617e-05,
+      "loss": 2.7806,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989809931065662e-05,
+      "loss": 3.0004,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989727974977294e-05,
+      "loss": 2.883,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989645691310295e-05,
+      "loss": 3.1284,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9895630800754935e-05,
+      "loss": 2.9112,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9894801412837555e-05,
+      "loss": 2.9988,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989396874945996e-05,
+      "loss": 2.9184,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98931328107317e-05,
+      "loss": 3.007,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989229359676276e-05,
+      "loss": 3.0694,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9891451107663553e-05,
+      "loss": 2.9575,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9890605343544935e-05,
+      "loss": 2.7877,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9889756304518174e-05,
+      "loss": 2.8886,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988890399069499e-05,
+      "loss": 3.0948,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9888048402187525e-05,
+      "loss": 3.0616,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988718953910834e-05,
+      "loss": 3.0531,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988632740157044e-05,
+      "loss": 2.8923,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988546198968726e-05,
+      "loss": 2.9304,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9884593303572665e-05,
+      "loss": 3.2415,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988372134334095e-05,
+      "loss": 2.9573,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9882846109106845e-05,
+      "loss": 2.8053,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98819676009855e-05,
+      "loss": 2.9791,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98810858190925e-05,
+      "loss": 2.8542,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9880200763543875e-05,
+      "loss": 2.809,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9879312434456055e-05,
+      "loss": 3.1602,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987842083194593e-05,
+      "loss": 2.8959,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987752595613082e-05,
+      "loss": 3.0802,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987662780712844e-05,
+      "loss": 2.8776,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987572638505699e-05,
+      "loss": 3.0885,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987482169003505e-05,
+      "loss": 2.9143,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9873913722181666e-05,
+      "loss": 2.9768,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9873002481616294e-05,
+      "loss": 2.8794,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9872087968458824e-05,
+      "loss": 2.8975,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98711701828296e-05,
+      "loss": 2.7598,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987024912484935e-05,
+      "loss": 3.0701,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9869324794639276e-05,
+      "loss": 2.777,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986839719232099e-05,
+      "loss": 2.7956,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9867466318016545e-05,
+      "loss": 3.0087,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986653217184841e-05,
+      "loss": 3.0138,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9865594753939496e-05,
+      "loss": 2.7282,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9864654064413133e-05,
+      "loss": 2.9702,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98637101033931e-05,
+      "loss": 2.8135,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9862762871003596e-05,
+      "loss": 2.9884,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9861812367369245e-05,
+      "loss": 3.1451,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98608585926151e-05,
+      "loss": 3.0401,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985990154686667e-05,
+      "loss": 2.8329,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9858941230249864e-05,
+      "loss": 2.8886,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985797764289103e-05,
+      "loss": 2.9768,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985701078491696e-05,
+      "loss": 2.8015,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985604065645485e-05,
+      "loss": 2.9142,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985506725763235e-05,
+      "loss": 3.0426,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9854090588577534e-05,
+      "loss": 3.0611,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98531106494189e-05,
+      "loss": 3.017,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985212744028539e-05,
+      "loss": 2.8556,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9851140961306356e-05,
+      "loss": 3.0554,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98501512126116e-05,
+      "loss": 3.0169,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984915819433133e-05,
+      "loss": 2.9016,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984816190659621e-05,
+      "loss": 3.0507,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9847162349537336e-05,
+      "loss": 2.8805,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98461595232862e-05,
+      "loss": 2.7568,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9845153427974753e-05,
+      "loss": 3.0457,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984414406373538e-05,
+      "loss": 2.9373,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9843131430700874e-05,
+      "loss": 3.0082,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984211552900447e-05,
+      "loss": 2.8896,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9841096358779835e-05,
+      "loss": 2.7628,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984007392016106e-05,
+      "loss": 2.9409,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9839048213282665e-05,
+      "loss": 2.9944,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983801923827962e-05,
+      "loss": 3.1713,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98369869952873e-05,
+      "loss": 2.9777,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9835951484441515e-05,
+      "loss": 2.8964,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9834912705878524e-05,
+      "loss": 2.7898,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983387065973498e-05,
+      "loss": 2.9286,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9832825346148e-05,
+      "loss": 3.1001,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9831776765255115e-05,
+      "loss": 2.9689,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9830724917194295e-05,
+      "loss": 2.811,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982966980210393e-05,
+      "loss": 2.8712,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982861142012284e-05,
+      "loss": 3.1317,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982754977139028e-05,
+      "loss": 2.8647,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982648485604593e-05,
+      "loss": 3.0162,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982541667422992e-05,
+      "loss": 3.0431,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982434522608277e-05,
+      "loss": 2.9169,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982327051174547e-05,
+      "loss": 2.7934,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9822192531359404e-05,
+      "loss": 2.8861,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9821111285066424e-05,
+      "loss": 2.96,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9820026773008785e-05,
+      "loss": 3.0378,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9818938995329174e-05,
+      "loss": 2.9494,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981784795217071e-05,
+      "loss": 2.5625,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9816753643676964e-05,
+      "loss": 2.9196,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981565606999189e-05,
+      "loss": 2.9865,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981455523125991e-05,
+      "loss": 2.7778,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981345112762586e-05,
+      "loss": 2.7235,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9812343759235016e-05,
+      "loss": 2.9119,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981123312623307e-05,
+      "loss": 2.883,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981011922876616e-05,
+      "loss": 2.8827,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9809002066980834e-05,
+      "loss": 2.6851,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980788164102408e-05,
+      "loss": 2.8943,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9806757951043316e-05,
+      "loss": 3.1971,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9805630997186395e-05,
+      "loss": 2.8274,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9804500779601577e-05,
+      "loss": 2.9544,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980336729843758e-05,
+      "loss": 2.8522,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9802230553843536e-05,
+      "loss": 2.9034,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980109054596901e-05,
+      "loss": 2.8108,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979994727496399e-05,
+      "loss": 2.907,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97988007409789e-05,
+      "loss": 3.0734,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97976509441646e-05,
+      "loss": 2.8572,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979649788467237e-05,
+      "loss": 2.899,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97953415626539e-05,
+      "loss": 2.9703,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9794181978261356e-05,
+      "loss": 2.7814,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979301913164729e-05,
+      "loss": 3.1088,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979185302296471e-05,
+      "loss": 2.9629,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979068365236703e-05,
+      "loss": 2.9571,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978951102000812e-05,
+      "loss": 2.8758,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9788335126042265e-05,
+      "loss": 2.7052,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9787155970624175e-05,
+      "loss": 3.1233,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978597355390899e-05,
+      "loss": 2.8085,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97847878760523e-05,
+      "loss": 2.8563,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9783598937210084e-05,
+      "loss": 2.9314,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978240673753879e-05,
+      "loss": 3.2052,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978121127719526e-05,
+      "loss": 2.8695,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978001255633681e-05,
+      "loss": 3.013,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977881057512114e-05,
+      "loss": 3.0662,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977760533370639e-05,
+      "loss": 2.8034,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977639683225116e-05,
+      "loss": 2.7808,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977518507091444e-05,
+      "loss": 2.8849,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977397004985567e-05,
+      "loss": 2.9698,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977275176923469e-05,
+      "loss": 3.0787,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977153022921183e-05,
+      "loss": 2.725,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9770305429947785e-05,
+      "loss": 2.9388,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976907737160371e-05,
+      "loss": 2.9218,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9767846054341176e-05,
+      "loss": 3.1943,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9766611478322205e-05,
+      "loss": 2.9209,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976537364370922e-05,
+      "loss": 2.8309,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97641325506651e-05,
+      "loss": 3.0997,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976288819935312e-05,
+      "loss": 2.8977,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976164058993701e-05,
+      "loss": 2.9344,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9760389722580926e-05,
+      "loss": 3.036,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975913559744944e-05,
+      "loss": 3.0241,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9757878214707565e-05,
+      "loss": 2.7318,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975661757452073e-05,
+      "loss": 2.9924,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9755353677054807e-05,
+      "loss": 3.1898,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9754086522476085e-05,
+      "loss": 2.9617,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975281611095129e-05,
+      "loss": 2.9447,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975154244264757e-05,
+      "loss": 3.23,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9750265517732505e-05,
+      "loss": 2.837,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97489853363741e-05,
+      "loss": 2.888,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97477018987408e-05,
+      "loss": 3.0481,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9746415205001465e-05,
+      "loss": 3.0216,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974512525532538e-05,
+      "loss": 3.0586,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974383204988227e-05,
+      "loss": 2.9727,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974253558884229e-05,
+      "loss": 2.8266,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974123587237602e-05,
+      "loss": 2.9315,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973993290065446e-05,
+      "loss": 2.9115,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973862667384904e-05,
+      "loss": 2.665,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973731719213164e-05,
+      "loss": 3.2225,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973600445567453e-05,
+      "loss": 2.9846,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973468846465045e-05,
+      "loss": 3.0866,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9733369219232526e-05,
+      "loss": 2.946,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973204671959435e-05,
+      "loss": 3.0069,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9730720965909925e-05,
+      "loss": 2.9422,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972939195835368e-05,
+      "loss": 2.97,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972805969710047e-05,
+      "loss": 3.1478,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972672418232559e-05,
+      "loss": 2.9433,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972538541420475e-05,
+      "loss": 2.6655,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972404339291411e-05,
+      "loss": 2.8952,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972269811863022e-05,
+      "loss": 2.8518,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9721349591530096e-05,
+      "loss": 2.8609,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971999781179117e-05,
+      "loss": 3.0171,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9718642779591286e-05,
+      "loss": 3.037,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971728449510873e-05,
+      "loss": 2.9468,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971592295852222e-05,
+      "loss": 2.9034,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9714558170010895e-05,
+      "loss": 3.1114,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9713190129754324e-05,
+      "loss": 2.8301,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9711818837932496e-05,
+      "loss": 2.9594,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9710444294725844e-05,
+      "loss": 2.873,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9709066500315205e-05,
+      "loss": 2.9304,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970768545488188e-05,
+      "loss": 2.8293,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970630115860756e-05,
+      "loss": 3.0366,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970491361167438e-05,
+      "loss": 2.8979,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970352281426491e-05,
+      "loss": 2.9648,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970212876656214e-05,
+      "loss": 2.9435,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970073146874947e-05,
+      "loss": 2.9463,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9699330921010774e-05,
+      "loss": 2.8158,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969792712353031e-05,
+      "loss": 2.8057,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969652007649277e-05,
+      "loss": 2.8759,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9695109780083296e-05,
+      "loss": 2.9698,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9693696234487435e-05,
+      "loss": 3.1211,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9692279439891173e-05,
+      "loss": 2.8682,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969085939648093e-05,
+      "loss": 2.9529,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968943610444353e-05,
+      "loss": 3.0069,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968800956396624e-05,
+      "loss": 2.6471,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968657977523676e-05,
+      "loss": 3.0954,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968514673844321e-05,
+      "loss": 2.9652,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968371045377413e-05,
+      "loss": 2.9336,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96822709214185e-05,
+      "loss": 2.9051,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9680828141565725e-05,
+      "loss": 2.829,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9679382114405625e-05,
+      "loss": 3.0488,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9677932840128465e-05,
+      "loss": 2.8147,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967648031892493e-05,
+      "loss": 3.017,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967502455098613e-05,
+      "loss": 3.2475,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967356553650359e-05,
+      "loss": 2.7062,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967210327566929e-05,
+      "loss": 2.9374,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967063776867563e-05,
+      "loss": 2.8939,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966916901571541e-05,
+      "loss": 2.9136,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966769701698189e-05,
+      "loss": 2.7856,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966622177266874e-05,
+      "loss": 2.6488,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966474328297006e-05,
+      "loss": 3.002,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966326154808038e-05,
+      "loss": 2.8571,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966177656819465e-05,
+      "loss": 3.0592,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9660288343508266e-05,
+      "loss": 2.9477,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9658796874217015e-05,
+      "loss": 2.9385,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965730216051715e-05,
+      "loss": 2.7078,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9655804202605325e-05,
+      "loss": 2.9347,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9654303000678634e-05,
+      "loss": 3.0276,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965279855493459e-05,
+      "loss": 2.8578,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965129086557114e-05,
+      "loss": 2.8638,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9649779932786654e-05,
+      "loss": 2.875,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964826575677992e-05,
+      "loss": 3.0968,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964674833775017e-05,
+      "loss": 3.0716,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9645227675897046e-05,
+      "loss": 3.0267,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643703771420636e-05,
+      "loss": 2.8706,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964217662452143e-05,
+      "loss": 2.8549,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964064623540037e-05,
+      "loss": 2.8304,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963911260425881e-05,
+      "loss": 2.898,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9637575731298525e-05,
+      "loss": 2.9633,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9636035616721734e-05,
+      "loss": 2.8922,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963449226073106e-05,
+      "loss": 3.1268,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963294566352958e-05,
+      "loss": 3.1805,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9631395825320776e-05,
+      "loss": 2.8552,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9629842746308565e-05,
+      "loss": 2.9504,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962828642669729e-05,
+      "loss": 3.0552,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9626726866691716e-05,
+      "loss": 2.9135,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9625164066497035e-05,
+      "loss": 2.8369,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962359802631888e-05,
+      "loss": 2.8293,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962202874636329e-05,
+      "loss": 2.7495,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9620456226836734e-05,
+      "loss": 2.9174,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9618880467946115e-05,
+      "loss": 2.8643,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961730146989876e-05,
+      "loss": 2.9258,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961571923290243e-05,
+      "loss": 3.1143,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961413375716528e-05,
+      "loss": 3.1304,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961254504289593e-05,
+      "loss": 3.1873,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961095309030341e-05,
+      "loss": 2.8018,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9609357899597185e-05,
+      "loss": 2.9553,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9607759470987125e-05,
+      "loss": 2.924,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960615780468353e-05,
+      "loss": 2.7676,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960455290089716e-05,
+      "loss": 2.9181,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960294475983915e-05,
+      "loss": 2.8166,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96013333817211e-05,
+      "loss": 2.8414,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9599718766755035e-05,
+      "loss": 2.9865,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959810091515336e-05,
+      "loss": 3.1324,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959647982712897e-05,
+      "loss": 2.818,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9594855502895135e-05,
+      "loss": 2.9822,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9593227942665574e-05,
+      "loss": 3.1369,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959159714665444e-05,
+      "loss": 2.8787,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958996311507629e-05,
+      "loss": 2.7887,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958832584814612e-05,
+      "loss": 2.9582,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958668534607935e-05,
+      "loss": 2.8886,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9585041609091824e-05,
+      "loss": 2.9402,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95833946373998e-05,
+      "loss": 2.9132,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958174443121999e-05,
+      "loss": 3.0441,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958009099076951e-05,
+      "loss": 3.0004,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95784343162659e-05,
+      "loss": 2.7652,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957677440792714e-05,
+      "loss": 2.9997,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957511126597163e-05,
+      "loss": 2.8073,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957344489061818e-05,
+      "loss": 2.8447,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957177528208605e-05,
+      "loss": 3.2068,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9570102440594915e-05,
+      "loss": 2.9505,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956842636636486e-05,
+      "loss": 2.7945,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9566747059616425e-05,
+      "loss": 2.9206,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9565064520570545e-05,
+      "loss": 3.0368,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956337874944861e-05,
+      "loss": 2.5747,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9561689746472417e-05,
+      "loss": 2.8075,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9559997511864183e-05,
+      "loss": 3.0357,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9558302045846575e-05,
+      "loss": 2.9246,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955660334864265e-05,
+      "loss": 2.9273,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955490142047592e-05,
+      "loss": 2.8131,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95531962615703e-05,
+      "loss": 3.1271,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955148787215016e-05,
+      "loss": 2.7091,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9549776252440266e-05,
+      "loss": 3.0261,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9548061402665817e-05,
+      "loss": 2.9961,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9546343323052445e-05,
+      "loss": 2.9714,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9544622013826193e-05,
+      "loss": 3.0858,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9542897475213545e-05,
+      "loss": 2.9669,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9541169707441404e-05,
+      "loss": 2.9375,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9539438710737085e-05,
+      "loss": 2.8682,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9537704485328347e-05,
+      "loss": 2.7688,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9535967031443365e-05,
+      "loss": 2.7604,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953422634931074e-05,
+      "loss": 3.0219,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953248243915949e-05,
+      "loss": 2.6915,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953073530121907e-05,
+      "loss": 2.9847,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952898493571936e-05,
+      "loss": 2.9537,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9527231342890654e-05,
+      "loss": 2.5326,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9525474522963665e-05,
+      "loss": 2.879,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9523714476169566e-05,
+      "loss": 3.0142,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9521951202739915e-05,
+      "loss": 2.8485,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9520184702906703e-05,
+      "loss": 2.951,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951841497690236e-05,
+      "loss": 2.7913,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951664202495974e-05,
+      "loss": 2.7681,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9514865847312096e-05,
+      "loss": 2.7605,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9513086444193145e-05,
+      "loss": 2.8749,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9511303815836984e-05,
+      "loss": 3.1045,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950951796247818e-05,
+      "loss": 2.7966,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950772888435168e-05,
+      "loss": 3.0784,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950593658169289e-05,
+      "loss": 2.7361,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950414105473762e-05,
+      "loss": 2.8551,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950234230372212e-05,
+      "loss": 2.874,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950054032888305e-05,
+      "loss": 2.8635,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94987351304575e-05,
+      "loss": 3.1227,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949692670868297e-05,
+      "loss": 2.8,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9495115063797425e-05,
+      "loss": 2.7885,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9493300196039205e-05,
+      "loss": 2.9434,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949148210564711e-05,
+      "loss": 2.7897,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948966079286034e-05,
+      "loss": 2.8533,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9487836257918526e-05,
+      "loss": 2.8374,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9486008501061746e-05,
+      "loss": 2.8544,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948417752253046e-05,
+      "loss": 2.9419,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948234332256558e-05,
+      "loss": 2.8239,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948050590140843e-05,
+      "loss": 2.8319,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947866525930078e-05,
+      "loss": 2.9916,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9476821396484804e-05,
+      "loss": 2.8293,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947497431320308e-05,
+      "loss": 2.849,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947312400969866e-05,
+      "loss": 2.9276,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9471270486214984e-05,
+      "loss": 3.0483,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946941374299591e-05,
+      "loss": 3.008,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946755378028575e-05,
+      "loss": 3.1335,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9465690598329216e-05,
+      "loss": 2.85,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946382419737145e-05,
+      "loss": 2.825,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946195457765802e-05,
+      "loss": 2.6513,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946008173943492e-05,
+      "loss": 3.0119,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945820568294856e-05,
+      "loss": 2.9571,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945632640844577e-05,
+      "loss": 3.0876,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945444391617382e-05,
+      "loss": 2.8397,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94525582063804e-05,
+      "loss": 2.9775,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9450669279313586e-05,
+      "loss": 2.8672,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9448777135221936e-05,
+      "loss": 2.7225,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9446881774354394e-05,
+      "loss": 3.0902,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944498319696034e-05,
+      "loss": 3.0401,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944308140328957e-05,
+      "loss": 2.9209,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9441176393592315e-05,
+      "loss": 3.07,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94392681681192e-05,
+      "loss": 2.8881,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943735672712132e-05,
+      "loss": 2.9015,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943544207085016e-05,
+      "loss": 3.092,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943352419955762e-05,
+      "loss": 2.9395,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943160311349605e-05,
+      "loss": 2.8626,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942967881291822e-05,
+      "loss": 2.5655,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94277512980773e-05,
+      "loss": 2.945,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94258205692269e-05,
+      "loss": 2.9563,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942388662662105e-05,
+      "loss": 2.8425,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942194947051421e-05,
+      "loss": 2.9675,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942000910116125e-05,
+      "loss": 2.8706,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9418065518817456e-05,
+      "loss": 2.6535,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941611872373858e-05,
+      "loss": 3.0167,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9414168716180734e-05,
+      "loss": 3.0754,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94122154964005e-05,
+      "loss": 2.9719,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941025906465488e-05,
+      "loss": 2.8991,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940829942120125e-05,
+      "loss": 2.9345,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9406336566297484e-05,
+      "loss": 2.7288,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940437050020181e-05,
+      "loss": 2.816,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9402401223172916e-05,
+      "loss": 2.9093,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9400428735469905e-05,
+      "loss": 2.9262,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939845303735231e-05,
+      "loss": 2.8787,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939647412908006e-05,
+      "loss": 2.788,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939449201091354e-05,
+      "loss": 3.0242,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9392506683113534e-05,
+      "loss": 2.9629,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939051814594125e-05,
+      "loss": 3.1534,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9388526399658335e-05,
+      "loss": 2.9088,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938653144452684e-05,
+      "loss": 2.7063,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9384533280809255e-05,
+      "loss": 3.022,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938253190876847e-05,
+      "loss": 2.7342,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938052732866781e-05,
+      "loss": 2.8447,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937851954077104e-05,
+      "loss": 2.9308,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93765085453423e-05,
+      "loss": 2.8603,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93744943426462e-05,
+      "loss": 3.1404,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937247693294776e-05,
+      "loss": 2.693,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937045631651239e-05,
+      "loss": 3.099,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936843249360597e-05,
+      "loss": 3.029,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936640546449477e-05,
+      "loss": 2.7375,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936437522944548e-05,
+      "loss": 3.1537,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9362341788725244e-05,
+      "loss": 2.8618,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9360305142601584e-05,
+      "loss": 3.0373,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935826529134247e-05,
+      "loss": 2.9235,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9356222235216306e-05,
+      "loss": 2.8463,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9354175974491884e-05,
+      "loss": 3.0148,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935212650943845e-05,
+      "loss": 2.8006,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9350073840325625e-05,
+      "loss": 2.831,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9348017967423525e-05,
+      "loss": 2.915,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934595889100262e-05,
+      "loss": 2.8858,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934389661133383e-05,
+      "loss": 2.9488,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9341831128688497e-05,
+      "loss": 2.9035,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933976244333838e-05,
+      "loss": 3.147,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933769055555566e-05,
+      "loss": 2.881,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933561546561293e-05,
+      "loss": 2.8133,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9333537173783235e-05,
+      "loss": 3.2405,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933145568034e-05,
+      "loss": 2.7055,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93293709855571e-05,
+      "loss": 2.9618,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.932728308970883e-05,
+      "loss": 3.0071,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9325191993069886e-05,
+      "loss": 2.8933,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93230976959154e-05,
+      "loss": 2.893,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9321000198520925e-05,
+      "loss": 3.1455,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9318899501162436e-05,
+      "loss": 2.7831,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931679560411632e-05,
+      "loss": 2.6851,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93146885076594e-05,
+      "loss": 2.9605,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931257821206891e-05,
+      "loss": 2.9105,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931046471762249e-05,
+      "loss": 2.6036,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930834802459823e-05,
+      "loss": 2.9051,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930622813327463e-05,
+      "loss": 3.0557,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9304105043930606e-05,
+      "loss": 3.0752,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93019787568455e-05,
+      "loss": 2.6726,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929984927229906e-05,
+      "loss": 2.9958,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9297716590571474e-05,
+      "loss": 3.0491,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9295580711943343e-05,
+      "loss": 2.8972,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9293441636695697e-05,
+      "loss": 2.7931,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9291299365109965e-05,
+      "loss": 2.7852,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928915389746802e-05,
+      "loss": 2.9242,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928700523405214e-05,
+      "loss": 2.9466,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9284853375145035e-05,
+      "loss": 2.9145,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928269832102982e-05,
+      "loss": 2.9325,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928054007199004e-05,
+      "loss": 2.835,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927837862830968e-05,
+      "loss": 2.7266,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92762139902731e-05,
+      "loss": 2.9221,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927404615816513e-05,
+      "loss": 2.8798,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9271875132270974e-05,
+      "loss": 3.0721,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926970091287628e-05,
+      "loss": 3.0072,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926752350026713e-05,
+      "loss": 2.9507,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926534289473e-05,
+      "loss": 2.8236,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9263159096551816e-05,
+      "loss": 2.9174,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926097210601987e-05,
+      "loss": 2.9513,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9258781923421934e-05,
+      "loss": 3.2124,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925658854904616e-05,
+      "loss": 2.9905,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925439198318115e-05,
+      "loss": 2.8358,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92521922261159e-05,
+      "loss": 2.5861,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9249989278139844e-05,
+      "loss": 2.8263,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924778313954282e-05,
+      "loss": 2.8853,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9245573810615106e-05,
+      "loss": 2.7928,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924336129164738e-05,
+      "loss": 2.8778,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924114558293074e-05,
+      "loss": 2.8041,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923892668475673e-05,
+      "loss": 2.9944,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923670459741727e-05,
+      "loss": 2.7958,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923447932120475e-05,
+      "loss": 3.0188,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923225085641194e-05,
+      "loss": 2.7916,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923001920333205e-05,
+      "loss": 2.6775,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92277843622587e-05,
+      "loss": 2.7756,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.922554633348593e-05,
+      "loss": 3.0174,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92233051173082e-05,
+      "loss": 2.9074,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.922106071402041e-05,
+      "loss": 3.013,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9218813123917836e-05,
+      "loss": 2.7563,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9216562347296216e-05,
+      "loss": 2.8663,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9214308384451676e-05,
+      "loss": 2.7138,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921205123568079e-05,
+      "loss": 2.8653,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920979090128052e-05,
+      "loss": 2.8784,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920752738154826e-05,
+      "loss": 2.7811,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920526067678185e-05,
+      "loss": 3.0027,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92029907872795e-05,
+      "loss": 3.0318,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920071771333988e-05,
+      "loss": 2.8842,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919844145526207e-05,
+      "loss": 3.0735,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919616201334553e-05,
+      "loss": 3.0806,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91938793878902e-05,
+      "loss": 2.9399,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91915935791964e-05,
+      "loss": 2.7997,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.918930458756488e-05,
+      "loss": 2.815,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91870124132968e-05,
+      "loss": 2.8998,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9184717056693755e-05,
+      "loss": 2.813,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9182418518057746e-05,
+      "loss": 2.6865,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91801167976912e-05,
+      "loss": 2.7892,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9177811895896955e-05,
+      "loss": 2.692,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917550381297827e-05,
+      "loss": 2.7402,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9173192549238826e-05,
+      "loss": 3.0343,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9170878104982734e-05,
+      "loss": 2.6543,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916856048051448e-05,
+      "loss": 2.7838,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916623967613903e-05,
+      "loss": 2.8161,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916391569216172e-05,
+      "loss": 2.9989,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916158852888832e-05,
+      "loss": 2.968,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9159258186625035e-05,
+      "loss": 2.9092,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915692466567846e-05,
+      "loss": 3.0592,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915458796635563e-05,
+      "loss": 2.8295,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915224808896397e-05,
+      "loss": 2.8987,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914990503381136e-05,
+      "loss": 2.8665,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9147558801206075e-05,
+      "loss": 2.6523,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9145209391456815e-05,
+      "loss": 2.6424,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914285680487271e-05,
+      "loss": 2.7724,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914050104176326e-05,
+      "loss": 3.1032,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913814210243845e-05,
+      "loss": 2.8972,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913577998720864e-05,
+      "loss": 2.9003,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913341469638462e-05,
+      "loss": 2.9635,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913104623027759e-05,
+      "loss": 2.9314,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.912867458919917e-05,
+      "loss": 2.8461,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.912629977346142e-05,
+      "loss": 3.1408,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9123921783376785e-05,
+      "loss": 2.8835,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.912154061925815e-05,
+      "loss": 2.8202,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91191562814188e-05,
+      "loss": 2.7811,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911676877017246e-05,
+      "loss": 2.8217,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911437808583325e-05,
+      "loss": 2.9063,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911198422871571e-05,
+      "loss": 2.7935,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910958719913483e-05,
+      "loss": 2.9695,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9107186997405966e-05,
+      "loss": 2.9128,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910478362384493e-05,
+      "loss": 2.8967,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910237707876794e-05,
+      "loss": 2.62,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909996736249163e-05,
+      "loss": 2.8481,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909755447533304e-05,
+      "loss": 2.8639,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909513841760965e-05,
+      "loss": 2.7289,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9092719189639344e-05,
+      "loss": 2.765,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909029679174043e-05,
+      "loss": 2.8901,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9087871224231605e-05,
+      "loss": 2.9353,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908544248743204e-05,
+      "loss": 2.6597,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908301058166126e-05,
+      "loss": 2.7169,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9080575507239254e-05,
+      "loss": 2.6571,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90781372644864e-05,
+      "loss": 2.8156,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90756958537235e-05,
+      "loss": 2.9865,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907325127527178e-05,
+      "loss": 3.1188,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907080352945289e-05,
+      "loss": 2.9274,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906835261658887e-05,
+      "loss": 2.8994,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906589853700219e-05,
+      "loss": 2.8177,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906344129101574e-05,
+      "loss": 2.9258,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9060980878952844e-05,
+      "loss": 2.6357,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90585173011372e-05,
+      "loss": 2.9195,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9056050557892954e-05,
+      "loss": 2.9413,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905358064954466e-05,
+      "loss": 3.0365,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9051107576417286e-05,
+      "loss": 2.6864,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9048631338836226e-05,
+      "loss": 3.2054,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9046151937127286e-05,
+      "loss": 2.8012,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904366937161667e-05,
+      "loss": 3.1664,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9041183642631037e-05,
+      "loss": 2.7237,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903869475049742e-05,
+      "loss": 2.6128,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9036202695543295e-05,
+      "loss": 2.8975,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903370747809655e-05,
+      "loss": 2.9446,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903120909848549e-05,
+      "loss": 2.8582,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902870755703881e-05,
+      "loss": 2.7046,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902620285408566e-05,
+      "loss": 2.882,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9023694989955595e-05,
+      "loss": 2.8924,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902118396497857e-05,
+      "loss": 2.7466,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901866977948496e-05,
+      "loss": 3.004,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9016152433805575e-05,
+      "loss": 2.6572,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901363192827161e-05,
+      "loss": 2.7293,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9011108263214713e-05,
+      "loss": 2.8101,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9008581438966915e-05,
+      "loss": 2.7765,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900605145586069e-05,
+      "loss": 2.6712,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900351831422889e-05,
+      "loss": 2.7985,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900098201440482e-05,
+      "loss": 2.9476,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899844255672219e-05,
+      "loss": 2.8972,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899589994151511e-05,
+      "loss": 2.9627,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8993354169118136e-05,
+      "loss": 2.826,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899080523986619e-05,
+      "loss": 2.7098,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8988253154094664e-05,
+      "loss": 2.9315,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898569791213934e-05,
+      "loss": 2.8296,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89831395143364e-05,
+      "loss": 2.7253,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898057796102248e-05,
+      "loss": 2.705,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897801325253459e-05,
+      "loss": 3.0917,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8975445389210184e-05,
+      "loss": 2.8024,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897287437138712e-05,
+      "loss": 2.7358,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8970300199403666e-05,
+      "loss": 2.8548,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896772287359852e-05,
+      "loss": 2.7775,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896514239431078e-05,
+      "loss": 2.9609,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896255876187997e-05,
+      "loss": 2.7707,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895997197664601e-05,
+      "loss": 2.7579,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8957382038949276e-05,
+      "loss": 2.8753,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8954788949130505e-05,
+      "loss": 2.7707,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895219270753088e-05,
+      "loss": 2.9357,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894959331449201e-05,
+      "loss": 2.6666,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894699077035588e-05,
+      "loss": 2.7176,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8944385075464934e-05,
+      "loss": 3.0506,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894177623016199e-05,
+      "loss": 2.7188,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893916423479031e-05,
+      "loss": 3.1027,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893654908969356e-05,
+      "loss": 2.7862,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893393079521581e-05,
+      "loss": 3.1608,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8931309351701566e-05,
+      "loss": 2.9024,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892868475949573e-05,
+      "loss": 2.7566,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892605701894363e-05,
+      "loss": 2.7923,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892342613039099e-05,
+      "loss": 2.7907,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8920792094183984e-05,
+      "loss": 2.9328,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891815491066915e-05,
+      "loss": 2.7752,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89155145801935e-05,
+      "loss": 2.6691,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89128711031044e-05,
+      "loss": 2.9762,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891022447974967e-05,
+      "loss": 2.8923,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8907574710477536e-05,
+      "loss": 2.6812,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.890492179563662e-05,
+      "loss": 2.9333,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.890226573557597e-05,
+      "loss": 2.9702,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.889960653064507e-05,
+      "loss": 2.7238,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.889694418119378e-05,
+      "loss": 2.8867,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8894278687572405e-05,
+      "loss": 2.7485,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889161005013163e-05,
+      "loss": 2.8267,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8888938269222595e-05,
+      "loss": 2.8345,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888626334519682e-05,
+      "loss": 3.0633,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888358527840624e-05,
+      "loss": 2.8051,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888090406920323e-05,
+      "loss": 3.0613,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8878219717940546e-05,
+      "loss": 2.6794,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887553222497139e-05,
+      "loss": 3.0478,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887284159064936e-05,
+      "loss": 2.8895,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887014781532846e-05,
+      "loss": 2.6059,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886745089936312e-05,
+      "loss": 2.7392,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886475084310817e-05,
+      "loss": 2.7621,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886204764691888e-05,
+      "loss": 2.9635,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8859341311150906e-05,
+      "loss": 3.0328,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8856631836160324e-05,
+      "loss": 2.759,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885391922230362e-05,
+      "loss": 2.9105,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885120346993771e-05,
+      "loss": 3.053,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88484845794199e-05,
+      "loss": 2.6199,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8845762551107935e-05,
+      "loss": 2.7566,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884303738535994e-05,
+      "loss": 3.0222,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8840309082534476e-05,
+      "loss": 2.8604,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8837577642990515e-05,
+      "loss": 2.6304,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883484306708744e-05,
+      "loss": 2.8734,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883210535518504e-05,
+      "loss": 2.7457,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882936450764353e-05,
+      "loss": 2.8505,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882662052482352e-05,
+      "loss": 2.7969,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882387340708604e-05,
+      "loss": 2.7404,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882112315479255e-05,
+      "loss": 2.9217,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881836976830488e-05,
+      "loss": 2.7775,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881561324798532e-05,
+      "loss": 2.9243,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881285359419654e-05,
+      "loss": 2.7504,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881009080730164e-05,
+      "loss": 2.7318,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880732488766412e-05,
+      "loss": 2.7218,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880455583564791e-05,
+      "loss": 2.9615,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880178365161732e-05,
+      "loss": 2.69,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879900833593711e-05,
+      "loss": 2.8179,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879622988897242e-05,
+      "loss": 2.6877,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879344831108883e-05,
+      "loss": 2.8417,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8790663602652306e-05,
+      "loss": 2.7477,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878787576402924e-05,
+      "loss": 2.8242,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878508479558644e-05,
+      "loss": 2.8952,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878229069769111e-05,
+      "loss": 2.8154,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877949347071088e-05,
+      "loss": 2.5731,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87766931150138e-05,
+      "loss": 2.9183,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8773889630968294e-05,
+      "loss": 2.8404,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877108301894324e-05,
+      "loss": 2.7642,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87682732793079e-05,
+      "loss": 2.9021,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876546041243196e-05,
+      "loss": 2.6859,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876264441868552e-05,
+      "loss": 2.9475,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8759825298439084e-05,
+      "loss": 2.9764,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875700305206356e-05,
+      "loss": 2.8927,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875417767993029e-05,
+      "loss": 3.0331,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875134918241101e-05,
+      "loss": 2.7934,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874851755987786e-05,
+      "loss": 2.8708,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874568281270342e-05,
+      "loss": 3.1918,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8742844941260644e-05,
+      "loss": 2.7983,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874000394592294e-05,
+      "loss": 2.8928,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873715982706409e-05,
+      "loss": 2.8912,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87343125850583e-05,
+      "loss": 2.8167,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873146222028019e-05,
+      "loss": 2.9647,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872860873310479e-05,
+      "loss": 2.9312,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872575212390753e-05,
+      "loss": 2.7564,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872289239306427e-05,
+      "loss": 2.7543,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872002954095127e-05,
+      "loss": 2.9557,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87171635679452e-05,
+      "loss": 2.9535,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871429447442314e-05,
+      "loss": 2.9028,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871142226076258e-05,
+      "loss": 2.8896,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8708546927341424e-05,
+      "loss": 2.7095,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8705668474537994e-05,
+      "loss": 2.7078,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8702786902731014e-05,
+      "loss": 2.9111,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8699902212299596e-05,
+      "loss": 2.5601,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.869701440362331e-05,
+      "loss": 2.9931,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.869412347708211e-05,
+      "loss": 2.6531,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.869122943305634e-05,
+      "loss": 2.799,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8688332271926796e-05,
+      "loss": 2.9672,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868543199407465e-05,
+      "loss": 2.9447,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868252859988151e-05,
+      "loss": 2.866,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867962208972937e-05,
+      "loss": 2.6396,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867671246400066e-05,
+      "loss": 2.9273,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8673799723078185e-05,
+      "loss": 2.8085,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86708838673452e-05,
+      "loss": 2.9348,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866796489718533e-05,
+      "loss": 2.9958,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866504281298265e-05,
+      "loss": 2.9426,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866211761512161e-05,
+      "loss": 2.9056,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8659189303987094e-05,
+      "loss": 2.8238,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.865625787996438e-05,
+      "loss": 3.0781,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8653323343439165e-05,
+      "loss": 2.6716,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865038569479755e-05,
+      "loss": 2.8268,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864744493442605e-05,
+      "loss": 2.6275,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864450106271158e-05,
+      "loss": 2.876,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864155408004148e-05,
+      "loss": 2.6799,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863860398680349e-05,
+      "loss": 2.6395,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8635650783385756e-05,
+      "loss": 2.9097,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863269447017683e-05,
+      "loss": 3.0735,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862973504756569e-05,
+      "loss": 2.9105,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8626772515941724e-05,
+      "loss": 2.7327,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862380687569469e-05,
+      "loss": 2.92,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8620838127214805e-05,
+      "loss": 2.9735,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8617866270892664e-05,
+      "loss": 2.8037,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861489130711929e-05,
+      "loss": 2.9072,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861191323628609e-05,
+      "loss": 2.8639,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860893205878491e-05,
+      "loss": 2.7355,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8605947775007984e-05,
+      "loss": 2.7113,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860296038534796e-05,
+      "loss": 2.8983,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8599969890197886e-05,
+      "loss": 2.8838,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8596976289951247e-05,
+      "loss": 2.6814,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85939795850019e-05,
+      "loss": 2.9905,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859097977574414e-05,
+      "loss": 2.8253,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8587976862572645e-05,
+      "loss": 2.8402,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858497084588252e-05,
+      "loss": 2.9113,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858196172606928e-05,
+      "loss": 2.7455,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8578949503528836e-05,
+      "loss": 2.949,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857593417865751e-05,
+      "loss": 2.7546,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8572915751852036e-05,
+      "loss": 2.986,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8569894223509546e-05,
+      "loss": 2.5092,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856686959402761e-05,
+      "loss": 2.7091,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856384186380417e-05,
+      "loss": 2.7898,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856081103323759e-05,
+      "loss": 2.8438,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855777710272664e-05,
+      "loss": 2.8664,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855474007267052e-05,
+      "loss": 2.878,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8551699943468795e-05,
+      "loss": 2.991,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8548656715521464e-05,
+      "loss": 2.7532,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854561038922895e-05,
+      "loss": 2.8607,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8542560964992045e-05,
+      "loss": 3.1002,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8539508443211975e-05,
+      "loss": 2.7825,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853645282429037e-05,
+      "loss": 3.0496,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853339410862926e-05,
+      "loss": 2.8152,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8530332296631087e-05,
+      "loss": 3.0245,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.852726738869869e-05,
+      "loss": 2.8254,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8524199385235344e-05,
+      "loss": 2.9164,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85211282866447e-05,
+      "loss": 2.7883,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851805409333083e-05,
+      "loss": 2.8048,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851497680569822e-05,
+      "loss": 2.6514,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851189642415175e-05,
+      "loss": 2.7966,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85088129490967e-05,
+      "loss": 2.7856,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8505726380938794e-05,
+      "loss": 2.9371,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.850263672008412e-05,
+      "loss": 2.7826,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8499543966939185e-05,
+      "loss": 2.6613,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849644812191093e-05,
+      "loss": 2.8921,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8493349185406675e-05,
+      "loss": 2.5899,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849024715783414e-05,
+      "loss": 2.9115,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848714203960148e-05,
+      "loss": 2.8891,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8484033831117234e-05,
+      "loss": 2.8336,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848092253279036e-05,
+      "loss": 2.8835,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8477808145030215e-05,
+      "loss": 2.7803,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847469066824657e-05,
+      "loss": 2.8069,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847157010284958e-05,
+      "loss": 2.708,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8468446449249854e-05,
+      "loss": 2.7459,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8465319707858356e-05,
+      "loss": 2.7935,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8462189879086485e-05,
+      "loss": 2.7061,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845905696334604e-05,
+      "loss": 2.7464,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845592096104922e-05,
+      "loss": 2.7909,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845278187260863e-05,
+      "loss": 2.8043,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8449639698437305e-05,
+      "loss": 2.6751,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8446494438948656e-05,
+      "loss": 3.0652,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.844334609455651e-05,
+      "loss": 2.8671,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.84401946656751e-05,
+      "loss": 3.089,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.843704015271907e-05,
+      "loss": 2.8624,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8433882556103464e-05,
+      "loss": 2.8756,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.843072187624373e-05,
+      "loss": 2.8213,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8427558113555746e-05,
+      "loss": 2.9839,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.842439126845575e-05,
+      "loss": 2.7018,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.842122134136041e-05,
+      "loss": 2.9717,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8418048332686814e-05,
+      "loss": 2.7795,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8414872242852436e-05,
+      "loss": 2.962,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.841169307227516e-05,
+      "loss": 2.8948,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840851082137328e-05,
+      "loss": 3.0479,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840532549056549e-05,
+      "loss": 3.0188,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840213708027088e-05,
+      "loss": 2.8014,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8398945590908966e-05,
+      "loss": 2.988,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839575102289966e-05,
+      "loss": 2.9206,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839255337666327e-05,
+      "loss": 2.7706,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8389352652620525e-05,
+      "loss": 2.8448,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.838614885119255e-05,
+      "loss": 2.6303,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8382941972800866e-05,
+      "loss": 2.6367,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8379732017867416e-05,
+      "loss": 2.7189,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837651898681455e-05,
+      "loss": 2.7311,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837330288006499e-05,
+      "loss": 2.7791,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.8719632625579834,
+      "eval_runtime": 1.3589,
+      "eval_samples_per_second": 73.591,
+      "eval_steps_per_second": 12.51,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8370083698041905e-05,
+      "loss": 2.8097,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836686144116884e-05,
+      "loss": 2.8174,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836363610986976e-05,
+      "loss": 2.8915,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836040770456903e-05,
+      "loss": 2.7866,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83571762256914e-05,
+      "loss": 2.7061,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835394167366207e-05,
+      "loss": 2.8611,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83507040489066e-05,
+      "loss": 2.5442,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834746335185098e-05,
+      "loss": 2.5989,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834421958292157e-05,
+      "loss": 2.8494,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8340972742545186e-05,
+      "loss": 2.8968,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833772283114902e-05,
+      "loss": 2.9116,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833446984916066e-05,
+      "loss": 2.6002,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8331213797008114e-05,
+      "loss": 2.7758,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8327954675119773e-05,
+      "loss": 2.7342,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8324692483924464e-05,
+      "loss": 2.8659,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8321427223851396e-05,
+      "loss": 3.0383,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8318158895330176e-05,
+      "loss": 2.849,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.831488749879084e-05,
+      "loss": 2.819,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83116130346638e-05,
+      "loss": 2.7092,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83083355033799e-05,
+      "loss": 3.1025,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830505490537035e-05,
+      "loss": 2.9126,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8301771241066796e-05,
+      "loss": 2.933,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829848451090128e-05,
+      "loss": 2.9446,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8295194715306234e-05,
+      "loss": 2.6669,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8291901854714506e-05,
+      "loss": 2.8231,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828860592955936e-05,
+      "loss": 2.8214,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828530694027442e-05,
+      "loss": 2.8348,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8282004887293765e-05,
+      "loss": 2.7278,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827869977105184e-05,
+      "loss": 3.0531,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827539159198351e-05,
+      "loss": 2.9349,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827208035052404e-05,
+      "loss": 2.8583,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8268766047109094e-05,
+      "loss": 2.994,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.826544868217474e-05,
+      "loss": 2.85,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.826212825615746e-05,
+      "loss": 2.6736,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.825880476949412e-05,
+      "loss": 3.0045,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8255478222621996e-05,
+      "loss": 2.9502,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.825214861597878e-05,
+      "loss": 2.8977,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.824881595000255e-05,
+      "loss": 2.8228,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8245480225131785e-05,
+      "loss": 3.106,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8242141441805374e-05,
+      "loss": 2.8343,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8238799600462626e-05,
+      "loss": 2.7627,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823545470154322e-05,
+      "loss": 2.7861,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8232106745487234e-05,
+      "loss": 2.7522,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8228755732735194e-05,
+      "loss": 2.8765,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822540166372799e-05,
+      "loss": 2.7694,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822204453890691e-05,
+      "loss": 2.7019,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8218684358713685e-05,
+      "loss": 2.8084,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.82153211235904e-05,
+      "loss": 2.7816,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8211954833979566e-05,
+      "loss": 2.7281,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8208585490324093e-05,
+      "loss": 2.9934,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.82052130930673e-05,
+      "loss": 2.8498,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8201837642652894e-05,
+      "loss": 2.7221,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8198459139524985e-05,
+      "loss": 2.8578,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81950775841281e-05,
+      "loss": 3.0098,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819169297690715e-05,
+      "loss": 2.601,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818830531830747e-05,
+      "loss": 2.8069,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8184914608774754e-05,
+      "loss": 2.7879,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818152084875515e-05,
+      "loss": 2.82,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817812403869516e-05,
+      "loss": 2.7189,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817472417904173e-05,
+      "loss": 2.8638,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817132127024217e-05,
+      "loss": 2.8179,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816791531274421e-05,
+      "loss": 2.6364,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8164506306996e-05,
+      "loss": 2.8112,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816109425344604e-05,
+      "loss": 2.8316,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815767915254328e-05,
+      "loss": 2.8075,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815426100473704e-05,
+      "loss": 2.5442,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8150839810477065e-05,
+      "loss": 3.1016,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8147415570213487e-05,
+      "loss": 2.7625,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814398828439683e-05,
+      "loss": 2.8248,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814055795347803e-05,
+      "loss": 2.8146,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813712457790843e-05,
+      "loss": 3.1639,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8133688158139774e-05,
+      "loss": 2.6815,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813024869462418e-05,
+      "loss": 2.9058,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812680618781419e-05,
+      "loss": 2.9016,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812336063816275e-05,
+      "loss": 3.0072,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811991204612319e-05,
+      "loss": 2.9507,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811646041214927e-05,
+      "loss": 2.7407,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8113005736695093e-05,
+      "loss": 2.9572,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8109548020215216e-05,
+      "loss": 2.854,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810608726316458e-05,
+      "loss": 2.9705,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810262346599852e-05,
+      "loss": 2.8427,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809915662917278e-05,
+      "loss": 2.6473,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80956867531435e-05,
+      "loss": 2.8541,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80922138383672e-05,
+      "loss": 2.8949,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808873788530084e-05,
+      "loss": 3.0295,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8085258894401744e-05,
+      "loss": 2.8712,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8081776866127664e-05,
+      "loss": 2.8941,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807829180093673e-05,
+      "loss": 2.615,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8074803699287475e-05,
+      "loss": 2.8562,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8071312561638844e-05,
+      "loss": 2.6365,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806781838845017e-05,
+      "loss": 2.5101,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8064321180181185e-05,
+      "loss": 2.6805,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806082093729202e-05,
+      "loss": 2.81,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805731766024322e-05,
+      "loss": 2.8479,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805381134949572e-05,
+      "loss": 2.6902,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805030200551084e-05,
+      "loss": 2.919,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804678962875032e-05,
+      "loss": 2.9433,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804327421967629e-05,
+      "loss": 2.722,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803975577875128e-05,
+      "loss": 2.5575,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8036234306438217e-05,
+      "loss": 2.8252,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8032709803200425e-05,
+      "loss": 2.9957,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802918226950164e-05,
+      "loss": 2.8756,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8025651705805974e-05,
+      "loss": 2.7355,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8022118112577954e-05,
+      "loss": 2.8776,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8018581490282513e-05,
+      "loss": 2.7768,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8015041839384964e-05,
+      "loss": 2.8389,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801149916035102e-05,
+      "loss": 2.9276,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80079534536468e-05,
+      "loss": 2.6412,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800440471973884e-05,
+      "loss": 2.7164,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800085295909402e-05,
+      "loss": 2.8269,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7997298172179684e-05,
+      "loss": 3.0499,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799374035946352e-05,
+      "loss": 2.9428,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799017952141366e-05,
+      "loss": 2.8076,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798661565849859e-05,
+      "loss": 3.0934,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798304877118722e-05,
+      "loss": 2.7814,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7979478859948855e-05,
+      "loss": 2.7183,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79759059252532e-05,
+      "loss": 2.7838,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797232996757034e-05,
+      "loss": 2.7496,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796875098737078e-05,
+      "loss": 2.8835,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796516898512542e-05,
+      "loss": 2.6436,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7961583961305544e-05,
+      "loss": 2.8942,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795799591638284e-05,
+      "loss": 2.8244,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795440485082939e-05,
+      "loss": 2.879,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7950810765117694e-05,
+      "loss": 2.6541,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7947213659720626e-05,
+      "loss": 2.6059,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7943613535111453e-05,
+      "loss": 2.6244,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7940010391763864e-05,
+      "loss": 2.7614,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793640423015192e-05,
+      "loss": 2.7344,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793279505075011e-05,
+      "loss": 2.6342,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792918285403328e-05,
+      "loss": 2.9939,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792556764047671e-05,
+      "loss": 2.7052,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7921949410556056e-05,
+      "loss": 3.0272,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7918328164747374e-05,
+      "loss": 2.9007,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7914703903527116e-05,
+      "loss": 2.5637,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791107662737214e-05,
+      "loss": 2.772,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79074463367597e-05,
+      "loss": 2.873,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790381303216742e-05,
+      "loss": 2.7876,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7900176714073366e-05,
+      "loss": 2.7164,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7896537382955954e-05,
+      "loss": 3.1197,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789289503929403e-05,
+      "loss": 2.947,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788924968356682e-05,
+      "loss": 2.7308,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788560131625395e-05,
+      "loss": 2.952,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7881949937835456e-05,
+      "loss": 2.6109,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7878295548791744e-05,
+      "loss": 2.8493,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787463814960363e-05,
+      "loss": 2.7528,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787097774075234e-05,
+      "loss": 2.7619,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786731432271946e-05,
+      "loss": 2.9728,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786364789598701e-05,
+      "loss": 2.7246,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785997846103738e-05,
+      "loss": 2.7589,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785630601835337e-05,
+      "loss": 2.8464,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785263056841817e-05,
+      "loss": 2.8749,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784895211171536e-05,
+      "loss": 2.8629,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784527064872893e-05,
+      "loss": 2.9779,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784158617994327e-05,
+      "loss": 2.8408,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783789870584312e-05,
+      "loss": 2.8369,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7834208226913674e-05,
+      "loss": 2.9709,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783051474364049e-05,
+      "loss": 2.7941,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7826818256509525e-05,
+      "loss": 2.8528,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782311876600713e-05,
+      "loss": 2.6631,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781941627262006e-05,
+      "loss": 2.6307,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7815710776835476e-05,
+      "loss": 2.7387,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781200227914089e-05,
+      "loss": 2.4894,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780829078002424e-05,
+      "loss": 2.8424,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7804576279973867e-05,
+      "loss": 2.8886,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7800858779478495e-05,
+      "loss": 2.8385,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7797138279027236e-05,
+      "loss": 2.7258,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7793414779109606e-05,
+      "loss": 2.823,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7789688280215516e-05,
+      "loss": 2.8121,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7785958782835275e-05,
+      "loss": 2.708,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7782226287459566e-05,
+      "loss": 2.8642,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7778490794579486e-05,
+      "loss": 2.7792,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777475230468653e-05,
+      "loss": 2.7523,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7771010818272575e-05,
+      "loss": 2.9374,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7767266335829895e-05,
+      "loss": 2.984,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776351885785115e-05,
+      "loss": 2.937,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775976838482942e-05,
+      "loss": 2.8507,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775601491725815e-05,
+      "loss": 2.8795,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77522584556312e-05,
+      "loss": 2.8966,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7748499000442804e-05,
+      "loss": 2.6907,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774473655218762e-05,
+      "loss": 3.0099,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774097111136066e-05,
+      "loss": 2.7693,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773720267845736e-05,
+      "loss": 2.8193,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7733431253973536e-05,
+      "loss": 2.6987,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772965683840542e-05,
+      "loss": 2.7335,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77258794322496e-05,
+      "loss": 2.7185,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772209903600308e-05,
+      "loss": 3.2117,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771831565016326e-05,
+      "loss": 2.8147,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7714529275227925e-05,
+      "loss": 2.7835,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771073991169526e-05,
+      "loss": 2.796,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7706947560063834e-05,
+      "loss": 2.7185,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770315222083262e-05,
+      "loss": 2.8671,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7699353894500976e-05,
+      "loss": 2.7616,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769555258156865e-05,
+      "loss": 2.8428,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7691748282535796e-05,
+      "loss": 2.7776,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768794099790295e-05,
+      "loss": 2.7533,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768413072817105e-05,
+      "loss": 2.8387,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768031747384141e-05,
+      "loss": 2.9859,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767650123541576e-05,
+      "loss": 2.9054,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76726820133962e-05,
+      "loss": 2.9472,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766885980828523e-05,
+      "loss": 2.7688,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766503462058576e-05,
+      "loss": 2.8276,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766120645080108e-05,
+      "loss": 2.7933,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765737529943485e-05,
+      "loss": 2.8069,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765354116699115e-05,
+      "loss": 2.9392,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7649704053974454e-05,
+      "loss": 3.0056,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7645863960889615e-05,
+      "loss": 2.9402,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764202088824187e-05,
+      "loss": 3.1121,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763817483653688e-05,
+      "loss": 2.9471,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7634325806280655e-05,
+      "loss": 3.1385,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7630473797979647e-05,
+      "loss": 2.7851,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762661881214064e-05,
+      "loss": 2.7991,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7622760849270874e-05,
+      "loss": 2.9485,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7618899909877936e-05,
+      "loss": 2.9745,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761503599446981e-05,
+      "loss": 2.6703,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76111691035549e-05,
+      "loss": 2.6957,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7607299237641956e-05,
+      "loss": 2.8697,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760342639724016e-05,
+      "loss": 3.1201,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759955058285906e-05,
+      "loss": 2.9444,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7595671795008614e-05,
+      "loss": 2.9032,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759179003419916e-05,
+      "loss": 2.7997,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7587905300941413e-05,
+      "loss": 2.5712,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.758401759574652e-05,
+      "loss": 2.9208,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.758012691912599e-05,
+      "loss": 2.5988,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757623327159171e-05,
+      "loss": 2.8996,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7572336653656e-05,
+      "loss": 2.8672,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756843706583152e-05,
+      "loss": 2.8087,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756453450863137e-05,
+      "loss": 2.9397,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7560628982569e-05,
+      "loss": 2.8166,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7556720488158274e-05,
+      "loss": 2.9367,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755280902591345e-05,
+      "loss": 2.7516,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754889459634915e-05,
+      "loss": 2.9157,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754497719998041e-05,
+      "loss": 3.0504,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754105683732266e-05,
+      "loss": 2.7591,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7537133508891695e-05,
+      "loss": 2.9187,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7533207215203736e-05,
+      "loss": 2.9223,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7529277956775345e-05,
+      "loss": 2.945,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.752534573412353e-05,
+      "loss": 2.8964,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7521410547765646e-05,
+      "loss": 2.8791,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7517472398219445e-05,
+      "loss": 2.7406,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7513531286003104e-05,
+      "loss": 2.8013,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.750958721163514e-05,
+      "loss": 2.9551,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.75056401756345e-05,
+      "loss": 3.0294,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7501690178520493e-05,
+      "loss": 2.6659,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749773722081283e-05,
+      "loss": 2.6649,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7493781303031604e-05,
+      "loss": 2.9317,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748982242569732e-05,
+      "loss": 3.064,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748586058933084e-05,
+      "loss": 2.9507,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7481895794453434e-05,
+      "loss": 2.8223,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747792804158677e-05,
+      "loss": 2.9154,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747395733125287e-05,
+      "loss": 2.9238,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746998366397419e-05,
+      "loss": 2.813,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7466007040273545e-05,
+      "loss": 2.7776,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746202746067415e-05,
+      "loss": 2.655,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7458044925699607e-05,
+      "loss": 2.7624,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7454059435873896e-05,
+      "loss": 2.8976,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7450070991721414e-05,
+      "loss": 2.931,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7446079593766913e-05,
+      "loss": 2.9028,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744208524253556e-05,
+      "loss": 2.9402,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7438087938552885e-05,
+      "loss": 2.9779,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7434087682344844e-05,
+      "loss": 2.9596,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7430084474437744e-05,
+      "loss": 2.8227,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7426078315358304e-05,
+      "loss": 2.6827,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742206920563361e-05,
+      "loss": 2.9171,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741805714579115e-05,
+      "loss": 2.9728,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741404213635882e-05,
+      "loss": 2.7821,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741002417786485e-05,
+      "loss": 2.8869,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7406003270837926e-05,
+      "loss": 2.9065,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7401979415807065e-05,
+      "loss": 2.7999,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739795261330169e-05,
+      "loss": 2.6985,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739392286385163e-05,
+      "loss": 2.7951,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738989016798708e-05,
+      "loss": 2.4815,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7385854526238636e-05,
+      "loss": 2.8254,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738181593913727e-05,
+      "loss": 2.7456,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737777440721435e-05,
+      "loss": 2.7911,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737372993100162e-05,
+      "loss": 3.0156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736968251103124e-05,
+      "loss": 2.6429,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7365632147835717e-05,
+      "loss": 3.1117,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7361578841947975e-05,
+      "loss": 3.0847,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7357522593901314e-05,
+      "loss": 2.8106,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7353463404229424e-05,
+      "loss": 2.8917,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7349401273466376e-05,
+      "loss": 2.7336,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734533620214664e-05,
+      "loss": 2.802,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734126819080507e-05,
+      "loss": 3.0638,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7337197239976885e-05,
+      "loss": 2.703,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.733312335019773e-05,
+      "loss": 2.8291,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7329046522003596e-05,
+      "loss": 2.7202,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732496675593088e-05,
+      "loss": 2.9351,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732088405251639e-05,
+      "loss": 2.8511,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731679841229727e-05,
+      "loss": 2.7853,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731270983581108e-05,
+      "loss": 2.9279,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730861832359577e-05,
+      "loss": 2.9049,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730452387618967e-05,
+      "loss": 2.6434,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7300426494131486e-05,
+      "loss": 2.8168,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729632617796033e-05,
+      "loss": 2.9398,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7292222928215676e-05,
+      "loss": 2.9474,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7288116745437406e-05,
+      "loss": 2.9423,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.728400763016577e-05,
+      "loss": 2.8668,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.727989558294143e-05,
+      "loss": 2.876,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7275780604305394e-05,
+      "loss": 2.9592,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.72716626947991e-05,
+      "loss": 2.8336,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726754185496434e-05,
+      "loss": 3.0202,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7263418085343294e-05,
+      "loss": 2.8485,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725929138647855e-05,
+      "loss": 2.7281,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7255161758913055e-05,
+      "loss": 2.8642,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725102920319016e-05,
+      "loss": 3.0096,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.724689371985358e-05,
+      "loss": 3.1284,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.724275530944744e-05,
+      "loss": 2.8375,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7238613972516244e-05,
+      "loss": 2.6045,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7234469709604865e-05,
+      "loss": 2.8836,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.723032252125858e-05,
+      "loss": 2.845,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722617240802304e-05,
+      "loss": 3.0032,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7222019370444284e-05,
+      "loss": 2.6919,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7217863409068734e-05,
+      "loss": 2.7637,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.72137045244432e-05,
+      "loss": 2.9319,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720954271711488e-05,
+      "loss": 2.9124,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720537798763135e-05,
+      "loss": 2.9069,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720121033654056e-05,
+      "loss": 2.7682,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.719703976439087e-05,
+      "loss": 2.9375,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.719286627173101e-05,
+      "loss": 3.1549,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7188689859110086e-05,
+      "loss": 2.9143,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7184510527077604e-05,
+      "loss": 2.8311,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.718032827618344e-05,
+      "loss": 2.8724,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717614310697788e-05,
+      "loss": 2.7619,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717195502001155e-05,
+      "loss": 2.8245,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7167764015835505e-05,
+      "loss": 2.7618,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.716357009500115e-05,
+      "loss": 2.8027,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7159373258060304e-05,
+      "loss": 3.097,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.715517350556513e-05,
+      "loss": 2.9714,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.715097083806822e-05,
+      "loss": 2.8076,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.714676525612253e-05,
+      "loss": 3.0274,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.714255676028137e-05,
+      "loss": 2.9944,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7138345351098476e-05,
+      "loss": 2.8797,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.713413102912797e-05,
+      "loss": 2.6941,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712991379492431e-05,
+      "loss": 2.6939,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712569364904238e-05,
+      "loss": 2.864,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712147059203743e-05,
+      "loss": 2.8419,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7117244624465103e-05,
+      "loss": 2.7389,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.711301574688141e-05,
+      "loss": 2.673,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.710878395984275e-05,
+      "loss": 2.6876,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.710454926390592e-05,
+      "loss": 2.784,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7100311659628084e-05,
+      "loss": 2.8553,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7096071147566785e-05,
+      "loss": 2.938,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.709182772827996e-05,
+      "loss": 2.8637,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708758140232593e-05,
+      "loss": 2.807,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7083332170263385e-05,
+      "loss": 2.7972,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7079080032651406e-05,
+      "loss": 2.8723,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707482499004946e-05,
+      "loss": 3.0062,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707056704301739e-05,
+      "loss": 2.8532,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7066306192115424e-05,
+      "loss": 3.023,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706204243790416e-05,
+      "loss": 3.0162,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705777578094461e-05,
+      "loss": 2.6646,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705350622179813e-05,
+      "loss": 2.9677,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7049233761026476e-05,
+      "loss": 2.7543,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7044958399191794e-05,
+      "loss": 2.7364,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7040680136856594e-05,
+      "loss": 2.7483,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703639897458378e-05,
+      "loss": 2.7827,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703211491293663e-05,
+      "loss": 2.9273,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70278279524788e-05,
+      "loss": 2.8156,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702353809377435e-05,
+      "loss": 2.8911,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70192453373877e-05,
+      "loss": 2.8495,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701494968388365e-05,
+      "loss": 2.6836,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70106511338274e-05,
+      "loss": 2.9819,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7006349687784504e-05,
+      "loss": 2.6925,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7002045346320925e-05,
+      "loss": 2.8883,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.699773811000298e-05,
+      "loss": 2.8014,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69934279793974e-05,
+      "loss": 3.0918,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698911495507125e-05,
+      "loss": 2.6728,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698479903759204e-05,
+      "loss": 2.9387,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69804802275276e-05,
+      "loss": 2.5728,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6976158525446166e-05,
+      "loss": 2.8917,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697183393191636e-05,
+      "loss": 2.9043,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696750644750718e-05,
+      "loss": 2.8453,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696317607278799e-05,
+      "loss": 2.725,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695884280832856e-05,
+      "loss": 2.9978,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6954506654699015e-05,
+      "loss": 2.7955,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695016761246987e-05,
+      "loss": 2.8757,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6945825682212034e-05,
+      "loss": 2.6194,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694148086449678e-05,
+      "loss": 2.9828,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693713315989575e-05,
+      "loss": 2.9085,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693278256898101e-05,
+      "loss": 2.9611,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692842909232495e-05,
+      "loss": 2.7269,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692407273050038e-05,
+      "loss": 2.8841,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.691971348408046e-05,
+      "loss": 2.9067,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.691535135363876e-05,
+      "loss": 2.9075,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.691098633974921e-05,
+      "loss": 3.1361,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.690661844298613e-05,
+      "loss": 2.9374,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69022476639242e-05,
+      "loss": 2.9115,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.68978740031385e-05,
+      "loss": 2.6648,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6893497461204486e-05,
+      "loss": 2.8754,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.688911803869797e-05,
+      "loss": 2.6137,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.688473573619519e-05,
+      "loss": 2.7786,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6880350554272715e-05,
+      "loss": 2.9621,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.687596249350751e-05,
+      "loss": 2.8303,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.687157155447693e-05,
+      "loss": 2.7817,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.68671777377587e-05,
+      "loss": 2.792,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6862781043930926e-05,
+      "loss": 2.743,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6858381473572075e-05,
+      "loss": 2.8136,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685397902726103e-05,
+      "loss": 2.6104,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684957370557701e-05,
+      "loss": 2.7617,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684516550909964e-05,
+      "loss": 2.8864,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684075443840892e-05,
+      "loss": 2.8346,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6836340494085215e-05,
+      "loss": 2.8175,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.683192367670929e-05,
+      "loss": 2.8221,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682750398686225e-05,
+      "loss": 2.7319,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682308142512564e-05,
+      "loss": 2.7036,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6818655992081305e-05,
+      "loss": 2.7728,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.681422768831153e-05,
+      "loss": 2.7118,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6809796514398965e-05,
+      "loss": 2.8536,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680536247092661e-05,
+      "loss": 2.7047,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680092555847787e-05,
+      "loss": 2.93,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.679648577763651e-05,
+      "loss": 2.7527,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.67920431289867e-05,
+      "loss": 2.9044,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6787597613112945e-05,
+      "loss": 2.779,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.678314923060016e-05,
+      "loss": 2.7102,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.677869798203364e-05,
+      "loss": 2.8129,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.677424386799903e-05,
+      "loss": 2.8747,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6769786889082376e-05,
+      "loss": 2.6295,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6765327045870085e-05,
+      "loss": 2.8752,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.676086433894895e-05,
+      "loss": 2.7476,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.675639876890614e-05,
+      "loss": 2.7429,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6751930336329195e-05,
+      "loss": 2.7949,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.674745904180604e-05,
+      "loss": 2.8486,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.674298488592497e-05,
+      "loss": 2.461,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.673850786927467e-05,
+      "loss": 2.8047,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6734027992444175e-05,
+      "loss": 2.8504,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.672954525602291e-05,
+      "loss": 2.5995,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6725059660600696e-05,
+      "loss": 2.756,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.67205712067677e-05,
+      "loss": 2.723,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.671607989511447e-05,
+      "loss": 2.9432,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.671158572623195e-05,
+      "loss": 2.7917,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670708870071144e-05,
+      "loss": 2.9594,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6702588819144634e-05,
+      "loss": 2.6357,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669808608212357e-05,
+      "loss": 2.9347,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66935804902407e-05,
+      "loss": 3.0091,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6689072044088834e-05,
+      "loss": 3.0559,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668456074426114e-05,
+      "loss": 2.9414,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66800465913512e-05,
+      "loss": 2.7751,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667552958595294e-05,
+      "loss": 2.8454,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667100972866067e-05,
+      "loss": 2.7501,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6666487020069086e-05,
+      "loss": 3.1382,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666196146077323e-05,
+      "loss": 2.7887,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665743305136857e-05,
+      "loss": 2.761,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665290179245089e-05,
+      "loss": 2.7421,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.664836768461639e-05,
+      "loss": 2.7411,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6643830728461634e-05,
+      "loss": 2.7315,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6639290924583555e-05,
+      "loss": 2.8641,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6634748273579466e-05,
+      "loss": 3.0045,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663020277604704e-05,
+      "loss": 2.6508,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662565443258435e-05,
+      "loss": 2.7304,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6621103243789846e-05,
+      "loss": 2.9365,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6616549210262306e-05,
+      "loss": 2.752,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.661199233260093e-05,
+      "loss": 2.9958,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660743261140528e-05,
+      "loss": 2.9533,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660287004727528e-05,
+      "loss": 2.8861,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659830464081123e-05,
+      "loss": 2.6643,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659373639261383e-05,
+      "loss": 2.7208,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.658916530328411e-05,
+      "loss": 2.9131,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.658459137342351e-05,
+      "loss": 2.7551,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6580014603633845e-05,
+      "loss": 2.9443,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657543499451726e-05,
+      "loss": 2.8976,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6570852546676325e-05,
+      "loss": 2.9164,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656626726071396e-05,
+      "loss": 2.9009,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656167913723345e-05,
+      "loss": 2.8147,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655708817683848e-05,
+      "loss": 2.6477,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6552494380133075e-05,
+      "loss": 2.7483,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6547897747721655e-05,
+      "loss": 2.8365,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6543298280209016e-05,
+      "loss": 2.9149,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653869597820032e-05,
+      "loss": 2.6807,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653409084230109e-05,
+      "loss": 3.1061,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652948287311724e-05,
+      "loss": 2.5661,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652487207125504e-05,
+      "loss": 2.81,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652025843732117e-05,
+      "loss": 2.838,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.651564197192263e-05,
+      "loss": 2.8525,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6511022675666814e-05,
+      "loss": 2.6987,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6506400549161525e-05,
+      "loss": 2.8935,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.650177559301487e-05,
+      "loss": 2.966,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.649714780783538e-05,
+      "loss": 2.8109,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6492517194231935e-05,
+      "loss": 2.9282,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.648788375281381e-05,
+      "loss": 2.998,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.648324748419062e-05,
+      "loss": 2.9009,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.647860838897238e-05,
+      "loss": 2.6681,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6473966467769455e-05,
+      "loss": 2.6917,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.64693217211926e-05,
+      "loss": 2.7503,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.646467414985293e-05,
+      "loss": 2.9372,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6460023754361934e-05,
+      "loss": 2.8442,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645537053533148e-05,
+      "loss": 2.7211,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645071449337379e-05,
+      "loss": 2.6741,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.644605562910148e-05,
+      "loss": 2.9732,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6441393943127534e-05,
+      "loss": 2.9014,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6436729436065275e-05,
+      "loss": 3.0238,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.643206210852844e-05,
+      "loss": 2.71,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6427391961131115e-05,
+      "loss": 2.6818,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642271899448777e-05,
+      "loss": 2.9235,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.641804320921322e-05,
+      "loss": 2.9088,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6413364605922674e-05,
+      "loss": 2.6615,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.640868318523171e-05,
+      "loss": 2.7213,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6403998947756264e-05,
+      "loss": 2.6182,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6399311894112656e-05,
+      "loss": 2.9742,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.639462202491757e-05,
+      "loss": 2.7564,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.638992934078806e-05,
+      "loss": 2.9326,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.638523384234157e-05,
+      "loss": 2.8758,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6380535530195865e-05,
+      "loss": 3.0347,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.637583440496913e-05,
+      "loss": 2.6193,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6371130467279904e-05,
+      "loss": 2.9773,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.636642371774709e-05,
+      "loss": 2.8565,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.636171415698996e-05,
+      "loss": 2.6643,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.635700178562816e-05,
+      "loss": 3.034,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.635228660428172e-05,
+      "loss": 2.8238,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.634756861357101e-05,
+      "loss": 2.5518,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6342847814116794e-05,
+      "loss": 2.9471,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6338124206540204e-05,
+      "loss": 3.0534,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.633339779146272e-05,
+      "loss": 2.932,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.632866856950621e-05,
+      "loss": 2.7324,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.632393654129291e-05,
+      "loss": 2.6846,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.631920170744543e-05,
+      "loss": 2.9703,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6314464068586724e-05,
+      "loss": 2.7001,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.630972362534016e-05,
+      "loss": 2.7996,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6304980378329424e-05,
+      "loss": 2.8509,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.630023432817859e-05,
+      "loss": 2.9106,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6295485475512136e-05,
+      "loss": 2.7627,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629073382095485e-05,
+      "loss": 2.5678,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.628597936513193e-05,
+      "loss": 2.8849,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.628122210866892e-05,
+      "loss": 2.8099,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627646205219176e-05,
+      "loss": 3.0162,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627169919632672e-05,
+      "loss": 2.8637,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626693354170047e-05,
+      "loss": 2.6061,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626216508894004e-05,
+      "loss": 2.7583,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6257393838672816e-05,
+      "loss": 2.7867,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6252619791526563e-05,
+      "loss": 2.8593,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624784294812942e-05,
+      "loss": 2.7017,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624306330910988e-05,
+      "loss": 2.8216,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623828087509681e-05,
+      "loss": 2.6944,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623349564671944e-05,
+      "loss": 2.7091,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622870762460738e-05,
+      "loss": 2.916,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.62239168093906e-05,
+      "loss": 2.7601,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621912320169943e-05,
+      "loss": 2.8019,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621432680216458e-05,
+      "loss": 2.61,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6209527611417125e-05,
+      "loss": 2.8314,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.62047256300885e-05,
+      "loss": 3.1245,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.619992085881051e-05,
+      "loss": 3.0585,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6195113298215333e-05,
+      "loss": 2.7637,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.619030294893551e-05,
+      "loss": 2.7387,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618548981160395e-05,
+      "loss": 2.7753,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618067388685392e-05,
+      "loss": 2.7005,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617585517531906e-05,
+      "loss": 2.9796,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617103367763339e-05,
+      "loss": 2.9062,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6166209394431276e-05,
+      "loss": 2.6119,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6161382326347455e-05,
+      "loss": 2.906,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6156552474017045e-05,
+      "loss": 2.7996,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615171983807551e-05,
+      "loss": 2.9017,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.61468844191587e-05,
+      "loss": 3.066,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.614204621790281e-05,
+      "loss": 3.101,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613720523494441e-05,
+      "loss": 2.7523,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613236147092046e-05,
+      "loss": 2.7336,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612751492646824e-05,
+      "loss": 2.7703,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612266560222543e-05,
+      "loss": 2.9769,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6117813498830066e-05,
+      "loss": 2.9266,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6112958616920546e-05,
+      "loss": 2.9238,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6108100957135636e-05,
+      "loss": 2.8239,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6103240520114475e-05,
+      "loss": 2.8158,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.609837730649656e-05,
+      "loss": 2.6834,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.609351131692173e-05,
+      "loss": 2.8864,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6088642552030257e-05,
+      "loss": 2.9001,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.60837710124627e-05,
+      "loss": 2.7623,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607889669886003e-05,
+      "loss": 2.7866,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607401961186357e-05,
+      "loss": 2.839,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.606913975211501e-05,
+      "loss": 2.8518,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.606425712025639e-05,
+      "loss": 2.8504,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.605937171693015e-05,
+      "loss": 2.7094,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6054483542779056e-05,
+      "loss": 3.2475,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.604959259844626e-05,
+      "loss": 2.9181,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6044698884575275e-05,
+      "loss": 2.8209,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.603980240180998e-05,
+      "loss": 2.6888,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.603490315079461e-05,
+      "loss": 2.8623,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.603000113217376e-05,
+      "loss": 2.9369,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.602509634659242e-05,
+      "loss": 2.6409,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6020188794695904e-05,
+      "loss": 2.9397,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6015278477129925e-05,
+      "loss": 2.9756,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6010365394540534e-05,
+      "loss": 2.8435,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.600544954757415e-05,
+      "loss": 2.5279,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.600053093687757e-05,
+      "loss": 2.8527,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5995609563097944e-05,
+      "loss": 2.7136,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.599068542688279e-05,
+      "loss": 2.9273,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.598575852887999e-05,
+      "loss": 2.7724,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.598082886973776e-05,
+      "loss": 2.9766,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597589645010474e-05,
+      "loss": 2.7797,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597096127062988e-05,
+      "loss": 2.9004,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5966023331962515e-05,
+      "loss": 2.7798,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.596108263475234e-05,
+      "loss": 2.8909,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.595613917964941e-05,
+      "loss": 3.0948,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.595119296730415e-05,
+      "loss": 3.0097,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5946243998367345e-05,
+      "loss": 2.893,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.594129227349013e-05,
+      "loss": 2.6734,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.593633779332402e-05,
+      "loss": 3.0493,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.59313805585209e-05,
+      "loss": 2.7847,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5926420569732984e-05,
+      "loss": 2.6036,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.592145782761288e-05,
+      "loss": 2.8994,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5916492332813534e-05,
+      "loss": 2.7678,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.591152408598828e-05,
+      "loss": 2.6251,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.590655308779078e-05,
+      "loss": 2.8785,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5901579338875104e-05,
+      "loss": 2.8921,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.589660283989565e-05,
+      "loss": 2.6934,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.589162359150718e-05,
+      "loss": 2.7858,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588664159436482e-05,
+      "loss": 2.808,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588165684912407e-05,
+      "loss": 2.767,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587666935644079e-05,
+      "loss": 3.1453,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587167911697118e-05,
+      "loss": 2.6833,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5866686131371816e-05,
+      "loss": 2.8935,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5861690400299654e-05,
+      "loss": 2.7143,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5856691924411975e-05,
+      "loss": 2.801,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.585169070436645e-05,
+      "loss": 2.9462,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5846686740821085e-05,
+      "loss": 2.9145,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.584168003443428e-05,
+      "loss": 2.7779,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583667058586476e-05,
+      "loss": 2.8321,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583165839577164e-05,
+      "loss": 2.8629,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5826643464814375e-05,
+      "loss": 2.7269,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58216257936528e-05,
+      "loss": 2.8382,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58166053829471e-05,
+      "loss": 2.9263,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5811582233357805e-05,
+      "loss": 2.8192,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5806556345545847e-05,
+      "loss": 2.9068,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5801527720172476e-05,
+      "loss": 2.9302,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579649635789932e-05,
+      "loss": 2.5219,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579146225938836e-05,
+      "loss": 2.6421,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.578642542530196e-05,
+      "loss": 2.7975,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.57813858563028e-05,
+      "loss": 2.6354,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577634355305398e-05,
+      "loss": 2.5065,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5771298516218905e-05,
+      "loss": 2.7978,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576625074646136e-05,
+      "loss": 2.6666,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5761200244445494e-05,
+      "loss": 2.628,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575614701083582e-05,
+      "loss": 2.7789,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575109104629719e-05,
+      "loss": 2.9315,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574603235149483e-05,
+      "loss": 3.1568,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574097092709433e-05,
+      "loss": 2.6354,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573590677376163e-05,
+      "loss": 2.692,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573083989216302e-05,
+      "loss": 2.7434,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572577028296517e-05,
+      "loss": 2.6617,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572069794683511e-05,
+      "loss": 2.646,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571562288444019e-05,
+      "loss": 2.8579,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571054509644817e-05,
+      "loss": 2.8353,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5705464583527136e-05,
+      "loss": 2.7368,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5700381346345547e-05,
+      "loss": 2.7153,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.56952953855722e-05,
+      "loss": 2.8015,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569020670187628e-05,
+      "loss": 2.6434,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.568511529592732e-05,
+      "loss": 2.7215,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5680021168395185e-05,
+      "loss": 2.9656,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.567492431995014e-05,
+      "loss": 2.8879,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566982475126278e-05,
+      "loss": 2.8759,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566472246300407e-05,
+      "loss": 2.6323,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5659617455845315e-05,
+      "loss": 2.7244,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5654509730458214e-05,
+      "loss": 2.8109,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5649399287514784e-05,
+      "loss": 2.7045,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.564428612768741e-05,
+      "loss": 2.6645,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.563917025164886e-05,
+      "loss": 2.8389,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.563405166007223e-05,
+      "loss": 2.7944,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562893035363099e-05,
+      "loss": 3.07,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562380633299895e-05,
+      "loss": 2.8479,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.56186795988503e-05,
+      "loss": 2.8231,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.561355015185957e-05,
+      "loss": 2.9422,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5608417992701643e-05,
+      "loss": 2.9621,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.560328312205179e-05,
+      "loss": 2.6117,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559814554058559e-05,
+      "loss": 3.1272,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559300524897903e-05,
+      "loss": 2.7486,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5587862247908404e-05,
+      "loss": 2.822,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55827165380504e-05,
+      "loss": 2.9283,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.557756812008206e-05,
+      "loss": 2.6806,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.557241699468075e-05,
+      "loss": 2.7897,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.556726316252423e-05,
+      "loss": 2.8255,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5562106624290596e-05,
+      "loss": 2.7427,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55569473806583e-05,
+      "loss": 2.8064,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.555178543230616e-05,
+      "loss": 3.0573,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.554662077991335e-05,
+      "loss": 2.8414,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5541453424159374e-05,
+      "loss": 2.7897,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5536283365724134e-05,
+      "loss": 2.4694,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.553111060528785e-05,
+      "loss": 2.6749,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5525935143531115e-05,
+      "loss": 3.0183,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.552075698113489e-05,
+      "loss": 2.8258,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5515576118780456e-05,
+      "loss": 2.6904,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5510392557149476e-05,
+      "loss": 2.6689,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5505206296923967e-05,
+      "loss": 2.8874,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.550001733878629e-05,
+      "loss": 2.8494,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5494825683419174e-05,
+      "loss": 2.8488,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5489631331505694e-05,
+      "loss": 3.1338,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.548443428372928e-05,
+      "loss": 2.6935,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.547923454077371e-05,
+      "loss": 2.7243,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.547403210332314e-05,
+      "loss": 2.8311,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.546882697206205e-05,
+      "loss": 2.9744,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5463619147675303e-05,
+      "loss": 2.729,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.545840863084809e-05,
+      "loss": 2.8785,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.545319542226599e-05,
+      "loss": 2.7381,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5447979522614895e-05,
+      "loss": 2.9157,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544276093258108e-05,
+      "loss": 2.8692,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5437539652851166e-05,
+      "loss": 2.9951,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5432315684112126e-05,
+      "loss": 2.8034,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542708902705129e-05,
+      "loss": 2.9228,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542185968235634e-05,
+      "loss": 2.7494,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541662765071531e-05,
+      "loss": 2.6675,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541139293281659e-05,
+      "loss": 2.8494,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540615552934891e-05,
+      "loss": 2.7628,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5400915441001404e-05,
+      "loss": 2.648,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5395672668463474e-05,
+      "loss": 2.8568,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5390427212424945e-05,
+      "loss": 2.8421,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538517907357598e-05,
+      "loss": 3.0487,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537992825260707e-05,
+      "loss": 2.7867,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53746747502091e-05,
+      "loss": 2.8304,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.536941856707326e-05,
+      "loss": 2.892,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.536415970389112e-05,
+      "loss": 2.7874,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535889816135461e-05,
+      "loss": 2.6022,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5353633940156e-05,
+      "loss": 2.7561,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534836704098791e-05,
+      "loss": 2.9562,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5343097464543315e-05,
+      "loss": 2.9617,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.533782521151556e-05,
+      "loss": 2.673,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53325502825983e-05,
+      "loss": 2.837,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532727267848559e-05,
+      "loss": 2.6382,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53219923998718e-05,
+      "loss": 2.8831,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531670944745168e-05,
+      "loss": 2.6538,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531142382192031e-05,
+      "loss": 2.8193,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530613552397314e-05,
+      "loss": 2.778,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530084455430594e-05,
+      "loss": 3.1276,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5295550913614884e-05,
+      "loss": 2.7659,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529025460259645e-05,
+      "loss": 2.9585,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528495562194749e-05,
+      "loss": 2.6006,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.52796539723652e-05,
+      "loss": 2.9381,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5274349654547124e-05,
+      "loss": 2.7445,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526904266919116e-05,
+      "loss": 2.9787,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526373301699558e-05,
+      "loss": 2.6909,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525842069865897e-05,
+      "loss": 3.149,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525310571488027e-05,
+      "loss": 2.8396,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.52477880663588e-05,
+      "loss": 2.897,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.524246775379422e-05,
+      "loss": 2.9299,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523714477788652e-05,
+      "loss": 2.7059,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523181913933606e-05,
+      "loss": 2.7462,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522649083884355e-05,
+      "loss": 2.6583,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5221159877110045e-05,
+      "loss": 2.5439,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5215826254836936e-05,
+      "loss": 2.8356,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.521048997272599e-05,
+      "loss": 2.7608,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.520515103147932e-05,
+      "loss": 2.5878,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519980943179937e-05,
+      "loss": 2.8384,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5194465174388946e-05,
+      "loss": 2.7799,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.51891182599512e-05,
+      "loss": 2.8181,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.518376868918966e-05,
+      "loss": 2.8286,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.517841646280815e-05,
+      "loss": 2.846,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.517306158151088e-05,
+      "loss": 2.9575,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.516770404600242e-05,
+      "loss": 2.7886,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5162343856987655e-05,
+      "loss": 2.9213,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.515698101517184e-05,
+      "loss": 2.9154,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5151615521260583e-05,
+      "loss": 2.9769,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5146247375959816e-05,
+      "loss": 2.7532,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5140876579975854e-05,
+      "loss": 3.0397,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513550313401535e-05,
+      "loss": 2.698,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513012703878526e-05,
+      "loss": 2.9242,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5124748294992976e-05,
+      "loss": 2.7938,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.511936690334616e-05,
+      "loss": 2.7359,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.511398286455286e-05,
+      "loss": 2.9141,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510859617932147e-05,
+      "loss": 2.8209,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5103206848360724e-05,
+      "loss": 2.8195,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.509781487237971e-05,
+      "loss": 2.7302,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.509242025208785e-05,
+      "loss": 2.8837,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5087022988194936e-05,
+      "loss": 2.7395,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.50816230814111e-05,
+      "loss": 2.7835,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.507622053244682e-05,
+      "loss": 2.6499,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.507081534201291e-05,
+      "loss": 2.7553,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5065407510820535e-05,
+      "loss": 2.8209,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.505999703958124e-05,
+      "loss": 2.9088,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5054583929006865e-05,
+      "loss": 2.7685,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.504916817980964e-05,
+      "loss": 2.7777,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.504374979270213e-05,
+      "loss": 2.715,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.503832876839723e-05,
+      "loss": 3.0862,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5032905107608205e-05,
+      "loss": 2.6647,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.502747881104865e-05,
+      "loss": 2.7846,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5022049879432524e-05,
+      "loss": 2.9389,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5016618313474115e-05,
+      "loss": 2.867,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.501118411388807e-05,
+      "loss": 2.955,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.500574728138936e-05,
+      "loss": 2.9352,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5000307816693355e-05,
+      "loss": 2.7354,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.49948657205157e-05,
+      "loss": 2.9211,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4989420993572454e-05,
+      "loss": 2.7337,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498397363657996e-05,
+      "loss": 2.8245,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497852365025497e-05,
+      "loss": 2.9281,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4973071035314523e-05,
+      "loss": 2.9589,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496761579247605e-05,
+      "loss": 2.7471,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496215792245729e-05,
+      "loss": 2.7791,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495669742597636e-05,
+      "loss": 2.7701,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495123430375171e-05,
+      "loss": 2.9654,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494576855650213e-05,
+      "loss": 3.0029,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494030018494675e-05,
+      "loss": 2.8935,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4934829189805076e-05,
+      "loss": 2.7328,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492935557179692e-05,
+      "loss": 2.6922,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492387933164246e-05,
+      "loss": 2.727,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491840047006223e-05,
+      "loss": 2.5635,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491291898777708e-05,
+      "loss": 2.8066,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4907434885508215e-05,
+      "loss": 2.712,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4901948163977206e-05,
+      "loss": 2.7115,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4896458823905944e-05,
+      "loss": 2.6193,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489096686601668e-05,
+      "loss": 2.8212,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.488547229103199e-05,
+      "loss": 2.9003,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487997509967481e-05,
+      "loss": 3.0374,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487447529266843e-05,
+      "loss": 2.7415,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.486897287073645e-05,
+      "loss": 2.7416,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.486346783460285e-05,
+      "loss": 2.8502,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4857960184991935e-05,
+      "loss": 2.6055,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.485244992262836e-05,
+      "loss": 3.0998,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.484693704823712e-05,
+      "loss": 2.9044,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.484142156254354e-05,
+      "loss": 2.8127,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.483590346627333e-05,
+      "loss": 2.9604,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.48303827601525e-05,
+      "loss": 2.8465,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482485944490743e-05,
+      "loss": 3.1304,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.481933352126483e-05,
+      "loss": 2.6252,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4813804989951744e-05,
+      "loss": 2.8048,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.48082738516956e-05,
+      "loss": 2.8965,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4802740107224125e-05,
+      "loss": 2.8936,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4797203757265405e-05,
+      "loss": 2.79,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479166480254786e-05,
+      "loss": 2.9136,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4786123243800285e-05,
+      "loss": 2.9625,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.478057908175177e-05,
+      "loss": 2.6808,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.477503231713179e-05,
+      "loss": 2.883,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476948295067014e-05,
+      "loss": 2.7861,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476393098309696e-05,
+      "loss": 2.8816,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475837641514272e-05,
+      "loss": 2.747,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475281924753827e-05,
+      "loss": 2.6906,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.474725948101477e-05,
+      "loss": 2.6277,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.474169711630371e-05,
+      "loss": 2.7528,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.473613215413697e-05,
+      "loss": 2.687,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4730564595246735e-05,
+      "loss": 2.7486,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.472499444036553e-05,
+      "loss": 2.8049,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.471942169022624e-05,
+      "loss": 2.8727,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4713846345562085e-05,
+      "loss": 2.7745,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.470826840710661e-05,
+      "loss": 2.7777,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.470268787559374e-05,
+      "loss": 2.9502,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.46971047517577e-05,
+      "loss": 2.7359,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.469151903633308e-05,
+      "loss": 3.0222,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.46859307300548e-05,
+      "loss": 2.7672,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.468033983365813e-05,
+      "loss": 2.7779,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.467474634787866e-05,
+      "loss": 2.719,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4669150273452355e-05,
+      "loss": 2.6672,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.46635516111155e-05,
+      "loss": 2.8228,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.465795036160471e-05,
+      "loss": 2.7332,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.465234652565697e-05,
+      "loss": 2.6815,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.464674010400958e-05,
+      "loss": 2.7086,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4641131097400176e-05,
+      "loss": 2.9731,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.463551950656676e-05,
+      "loss": 2.9727,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.462990533224767e-05,
+      "loss": 2.9112,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.462428857518155e-05,
+      "loss": 2.7942,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.461866923610743e-05,
+      "loss": 2.9305,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.461304731576464e-05,
+      "loss": 2.7279,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.460742281489289e-05,
+      "loss": 2.8178,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.460179573423219e-05,
+      "loss": 2.8496,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.459616607452291e-05,
+      "loss": 2.7226,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4590533836505754e-05,
+      "loss": 2.6094,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.458489902092178e-05,
+      "loss": 2.7987,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.457926162851235e-05,
+      "loss": 2.8275,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45736216600192e-05,
+      "loss": 2.9981,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45679791161844e-05,
+      "loss": 3.2024,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.456233399775033e-05,
+      "loss": 2.8265,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.455668630545975e-05,
+      "loss": 2.7685,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.455103604005574e-05,
+      "loss": 2.7132,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4545383202281695e-05,
+      "loss": 2.5893,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.453972779288139e-05,
+      "loss": 2.8711,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.453406981259891e-05,
+      "loss": 2.809,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.452840926217869e-05,
+      "loss": 3.0222,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45227461423655e-05,
+      "loss": 3.0353,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4517080453904444e-05,
+      "loss": 2.8875,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.451141219754098e-05,
+      "loss": 2.8494,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4505741374020885e-05,
+      "loss": 2.8058,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450006798409028e-05,
+      "loss": 2.7531,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449439202849562e-05,
+      "loss": 2.789,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448871350798371e-05,
+      "loss": 2.8187,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448303242330168e-05,
+      "loss": 2.5067,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4477348775197004e-05,
+      "loss": 2.6852,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4471662564417497e-05,
+      "loss": 2.6071,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446597379171129e-05,
+      "loss": 2.9022,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446028245782689e-05,
+      "loss": 2.8588,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.445458856351309e-05,
+      "loss": 2.989,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.444889210951906e-05,
+      "loss": 2.6594,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4443193096594305e-05,
+      "loss": 2.8531,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.443749152548864e-05,
+      "loss": 2.7102,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.443178739695224e-05,
+      "loss": 2.9075,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4426080711735606e-05,
+      "loss": 2.8848,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4420371470589584e-05,
+      "loss": 2.7723,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4414659674265346e-05,
+      "loss": 2.8478,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4408945323514406e-05,
+      "loss": 2.7601,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4403228419088604e-05,
+      "loss": 2.6046,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.439750896174014e-05,
+      "loss": 2.8319,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.439178695222152e-05,
+      "loss": 2.9674,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4386062391285624e-05,
+      "loss": 2.7796,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.438033527968561e-05,
+      "loss": 2.6986,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4374605618175034e-05,
+      "loss": 2.6985,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.436887340750775e-05,
+      "loss": 2.5806,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4363138648437954e-05,
+      "loss": 2.8784,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.435740134172018e-05,
+      "loss": 2.8374,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43516614881093e-05,
+      "loss": 2.701,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434591908836052e-05,
+      "loss": 2.7012,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4340174143229383e-05,
+      "loss": 2.9281,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.433442665347175e-05,
+      "loss": 2.8686,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4328676619843846e-05,
+      "loss": 2.7874,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43229240431022e-05,
+      "loss": 2.685,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4317168924003694e-05,
+      "loss": 2.8924,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4311411263305546e-05,
+      "loss": 2.772,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43056510617653e-05,
+      "loss": 2.7862,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4299888320140845e-05,
+      "loss": 2.8316,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.429412303919038e-05,
+      "loss": 2.7856,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.428835521967247e-05,
+      "loss": 3.0113,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4282584862346e-05,
+      "loss": 2.662,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427681196797017e-05,
+      "loss": 2.7329,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427103653730455e-05,
+      "loss": 3.0092,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.426525857110902e-05,
+      "loss": 2.9914,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.425947807014379e-05,
+      "loss": 2.7202,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.425369503516943e-05,
+      "loss": 2.7234,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.424790946694681e-05,
+      "loss": 2.9346,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.424212136623716e-05,
+      "loss": 2.9135,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.423633073380202e-05,
+      "loss": 2.6759,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.423053757040329e-05,
+      "loss": 2.8477,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.422474187680317e-05,
+      "loss": 2.8737,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4218943653764236e-05,
+      "loss": 2.6044,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.421314290204934e-05,
+      "loss": 2.6522,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.420733962242174e-05,
+      "loss": 3.1418,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4201533815644946e-05,
+      "loss": 2.6718,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4195725482482865e-05,
+      "loss": 2.6837,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4189914623699694e-05,
+      "loss": 2.8023,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.418410124005999e-05,
+      "loss": 2.8515,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.417828533232864e-05,
+      "loss": 3.0546,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4172466901270837e-05,
+      "loss": 2.8812,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.416664594765214e-05,
+      "loss": 2.8995,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.416082247223841e-05,
+      "loss": 2.7968,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4154996475795864e-05,
+      "loss": 2.725,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.414916795909103e-05,
+      "loss": 2.8632,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4143336922890785e-05,
+      "loss": 2.7476,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.413750336796234e-05,
+      "loss": 2.6829,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.413166729507321e-05,
+      "loss": 2.8259,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.412582870499126e-05,
+      "loss": 2.68,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4119987598484704e-05,
+      "loss": 2.6546,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.411414397632204e-05,
+      "loss": 2.675,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.410829783927216e-05,
+      "loss": 2.8779,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4102449188104226e-05,
+      "loss": 2.8369,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.409659802358776e-05,
+      "loss": 2.6957,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.409074434649262e-05,
+      "loss": 2.494,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.408488815758899e-05,
+      "loss": 2.6491,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.407902945764736e-05,
+      "loss": 2.7142,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4073168247438594e-05,
+      "loss": 3.1302,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.406730452773386e-05,
+      "loss": 3.0014,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4061438299304644e-05,
+      "loss": 2.9092,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.405556956292279e-05,
+      "loss": 2.6699,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.404969831936046e-05,
+      "loss": 2.9881,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4043824569390135e-05,
+      "loss": 2.6011,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4037948313784655e-05,
+      "loss": 3.0388,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4032069553317154e-05,
+      "loss": 2.7822,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.402618828876113e-05,
+      "loss": 2.9776,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4020304520890365e-05,
+      "loss": 2.736,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4014418250479026e-05,
+      "loss": 3.1772,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400852947830157e-05,
+      "loss": 2.9263,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4002638205132794e-05,
+      "loss": 2.7566,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.399674443174783e-05,
+      "loss": 2.7789,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3990848158922135e-05,
+      "loss": 2.7511,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.398494938743148e-05,
+      "loss": 2.8283,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3979048118052e-05,
+      "loss": 2.8229,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397314435156012e-05,
+      "loss": 2.892,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396723808873261e-05,
+      "loss": 3.0366,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3961329330346585e-05,
+      "loss": 2.4538,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395541807717946e-05,
+      "loss": 2.9172,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.394950433000899e-05,
+      "loss": 2.845,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3943588089613265e-05,
+      "loss": 2.9275,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39376693567707e-05,
+      "loss": 2.651,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.393174813226002e-05,
+      "loss": 2.8281,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39258244168603e-05,
+      "loss": 2.7679,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391989821135094e-05,
+      "loss": 2.593,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391396951651167e-05,
+      "loss": 2.8644,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390803833312251e-05,
+      "loss": 2.5961,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390210466196388e-05,
+      "loss": 2.7998,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389616850381645e-05,
+      "loss": 2.6677,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389022985946128e-05,
+      "loss": 2.735,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.388428872967971e-05,
+      "loss": 2.7994,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387834511525343e-05,
+      "loss": 2.6903,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387239901696446e-05,
+      "loss": 2.6663,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.386645043559514e-05,
+      "loss": 2.7221,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3860499371928124e-05,
+      "loss": 2.8346,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385454582674642e-05,
+      "loss": 2.7408,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3848589800833353e-05,
+      "loss": 2.7763,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.384263129497255e-05,
+      "loss": 2.9106,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3836670309947995e-05,
+      "loss": 2.8871,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383070684654399e-05,
+      "loss": 2.6968,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.382474090554516e-05,
+      "loss": 2.4839,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.381877248773645e-05,
+      "loss": 2.7959,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.381280159390314e-05,
+      "loss": 2.8914,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3806828224830834e-05,
+      "loss": 2.7587,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3800852381305445e-05,
+      "loss": 2.8654,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.379487406411326e-05,
+      "loss": 2.8859,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.378889327404083e-05,
+      "loss": 2.7839,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.378291001187507e-05,
+      "loss": 2.6749,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3776924278403206e-05,
+      "loss": 2.6796,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37709360744128e-05,
+      "loss": 2.6623,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3764945400691736e-05,
+      "loss": 2.9752,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3758952258028205e-05,
+      "loss": 2.8634,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.375295664721075e-05,
+      "loss": 3.0746,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3746958569028214e-05,
+      "loss": 2.8948,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.374095802426978e-05,
+      "loss": 2.9437,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.373495501372497e-05,
+      "loss": 2.9527,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.372894953818358e-05,
+      "loss": 2.7345,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3722941598435784e-05,
+      "loss": 2.9209,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.371693119527206e-05,
+      "loss": 2.8116,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3710918329483204e-05,
+      "loss": 2.7021,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.370490300186034e-05,
+      "loss": 2.9279,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 2.827476739883423,
+      "eval_runtime": 1.3594,
+      "eval_samples_per_second": 73.56,
+      "eval_steps_per_second": 12.505,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6084461683442176e-05,
+      "loss": 2.5561,
+      "step": 2001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6061874804424e-05,
+      "loss": 2.9095,
+      "step": 2002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6039296287481695e-05,
+      "loss": 2.875,
+      "step": 2003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6016726153738638e-05,
+      "loss": 2.9714,
+      "step": 2004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.599416442431034e-05,
+      "loss": 2.955,
+      "step": 2005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5971611120304453e-05,
+      "loss": 2.707,
+      "step": 2006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.594906626282076e-05,
+      "loss": 2.7539,
+      "step": 2007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5926529872951142e-05,
+      "loss": 2.7266,
+      "step": 2008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.590400197177954e-05,
+      "loss": 2.8762,
+      "step": 2009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.588148258038197e-05,
+      "loss": 2.7436,
+      "step": 2010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5858971719826477e-05,
+      "loss": 2.8887,
+      "step": 2011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.583646941117313e-05,
+      "loss": 2.8921,
+      "step": 2012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5813975675473996e-05,
+      "loss": 2.7331,
+      "step": 2013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.579149053377313e-05,
+      "loss": 2.9938,
+      "step": 2014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5769014007106528e-05,
+      "loss": 3.1149,
+      "step": 2015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.574654611650214e-05,
+      "loss": 2.92,
+      "step": 2016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.572408688297984e-05,
+      "loss": 2.8857,
+      "step": 2017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5701636327551382e-05,
+      "loss": 2.8479,
+      "step": 2018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.567919447122043e-05,
+      "loss": 2.8883,
+      "step": 2019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5656761334982488e-05,
+      "loss": 2.8902,
+      "step": 2020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5634336939824907e-05,
+      "loss": 2.9305,
+      "step": 2021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5611921306726872e-05,
+      "loss": 2.8719,
+      "step": 2022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.558951445665935e-05,
+      "loss": 2.7734,
+      "step": 2023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.55671164105851e-05,
+      "loss": 2.9862,
+      "step": 2024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5544727189458652e-05,
+      "loss": 2.9412,
+      "step": 2025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5522346814226278e-05,
+      "loss": 2.8944,
+      "step": 2026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5499975305825957e-05,
+      "loss": 2.9142,
+      "step": 2027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5477612685187403e-05,
+      "loss": 3.0047,
+      "step": 2028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.545525897323199e-05,
+      "loss": 2.8495,
+      "step": 2029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5432914190872757e-05,
+      "loss": 2.8696,
+      "step": 2030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.541057835901441e-05,
+      "loss": 2.9301,
+      "step": 2031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5388251498553265e-05,
+      "loss": 2.8553,
+      "step": 2032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5365933630377246e-05,
+      "loss": 2.8989,
+      "step": 2033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.534362477536586e-05,
+      "loss": 2.8498,
+      "step": 2034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5321324954390207e-05,
+      "loss": 2.8857,
+      "step": 2035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.52990341883129e-05,
+      "loss": 2.9694,
+      "step": 2036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.52767524979881e-05,
+      "loss": 2.8461,
+      "step": 2037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5254479904261487e-05,
+      "loss": 2.8749,
+      "step": 2038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5232216427970208e-05,
+      "loss": 2.9075,
+      "step": 2039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5209962089942884e-05,
+      "loss": 3.0062,
+      "step": 2040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.51877169109996e-05,
+      "loss": 2.9603,
+      "step": 2041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5165480911951867e-05,
+      "loss": 2.8281,
+      "step": 2042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.51432541136026e-05,
+      "loss": 2.9713,
+      "step": 2043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.512103653674612e-05,
+      "loss": 2.9926,
+      "step": 2044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5098828202168103e-05,
+      "loss": 2.8648,
+      "step": 2045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5076629130645598e-05,
+      "loss": 2.741,
+      "step": 2046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5054439342946974e-05,
+      "loss": 3.0197,
+      "step": 2047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5032258859831915e-05,
+      "loss": 2.7776,
+      "step": 2048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.501008770205141e-05,
+      "loss": 2.8135,
+      "step": 2049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4987925890347721e-05,
+      "loss": 2.7974,
+      "step": 2050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4965773445454349e-05,
+      "loss": 2.8464,
+      "step": 2051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4943630388096055e-05,
+      "loss": 2.7397,
+      "step": 2052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4921496738988811e-05,
+      "loss": 2.8935,
+      "step": 2053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4899372518839778e-05,
+      "loss": 2.8744,
+      "step": 2054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4877257748347302e-05,
+      "loss": 2.9179,
+      "step": 2055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4855152448200901e-05,
+      "loss": 2.8878,
+      "step": 2056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4833056639081205e-05,
+      "loss": 2.9468,
+      "step": 2057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.481097034165998e-05,
+      "loss": 2.9015,
+      "step": 2058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4788893576600099e-05,
+      "loss": 2.9941,
+      "step": 2059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4766826364555513e-05,
+      "loss": 2.8453,
+      "step": 2060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.474476872617123e-05,
+      "loss": 3.0022,
+      "step": 2061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4722720682083308e-05,
+      "loss": 2.8543,
+      "step": 2062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4700682252918824e-05,
+      "loss": 2.8437,
+      "step": 2063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.467865345929586e-05,
+      "loss": 2.8291,
+      "step": 2064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4656634321823493e-05,
+      "loss": 3.009,
+      "step": 2065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.463462486110175e-05,
+      "loss": 2.8785,
+      "step": 2066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4612625097721618e-05,
+      "loss": 2.7481,
+      "step": 2067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4590635052265009e-05,
+      "loss": 2.844,
+      "step": 2068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.456865474530475e-05,
+      "loss": 2.9429,
+      "step": 2069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4546684197404525e-05,
+      "loss": 2.905,
+      "step": 2070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4524723429118927e-05,
+      "loss": 2.8942,
+      "step": 2071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4502772460993385e-05,
+      "loss": 2.9746,
+      "step": 2072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4480831313564164e-05,
+      "loss": 2.8491,
+      "step": 2073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4458900007358317e-05,
+      "loss": 2.9227,
+      "step": 2074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4436978562893721e-05,
+      "loss": 2.8403,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4415067000679027e-05,
+      "loss": 2.8856,
+      "step": 2076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4393165341213605e-05,
+      "loss": 2.979,
+      "step": 2077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4371273604987596e-05,
+      "loss": 2.8605,
+      "step": 2078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.434939181248184e-05,
+      "loss": 2.8144,
+      "step": 2079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4327519984167886e-05,
+      "loss": 2.7847,
+      "step": 2080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4305658140507954e-05,
+      "loss": 2.9887,
+      "step": 2081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4283806301954927e-05,
+      "loss": 2.8478,
+      "step": 2082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4261964488952311e-05,
+      "loss": 2.856,
+      "step": 2083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4240132721934255e-05,
+      "loss": 2.8707,
+      "step": 2084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4218311021325495e-05,
+      "loss": 2.9052,
+      "step": 2085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4196499407541359e-05,
+      "loss": 2.8124,
+      "step": 2086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4174697900987726e-05,
+      "loss": 2.8079,
+      "step": 2087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4152906522061048e-05,
+      "loss": 2.9324,
+      "step": 2088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4131125291148256e-05,
+      "loss": 2.956,
+      "step": 2089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.410935422862683e-05,
+      "loss": 2.9399,
+      "step": 2090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4087593354864699e-05,
+      "loss": 2.9021,
+      "step": 2091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4065842690220293e-05,
+      "loss": 2.7797,
+      "step": 2092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4044102255042475e-05,
+      "loss": 2.8815,
+      "step": 2093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4022372069670547e-05,
+      "loss": 2.9354,
+      "step": 2094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4000652154434205e-05,
+      "loss": 2.9515,
+      "step": 2095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.397894252965355e-05,
+      "loss": 2.9134,
+      "step": 2096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3957243215639049e-05,
+      "loss": 2.8267,
+      "step": 2097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3935554232691533e-05,
+      "loss": 2.8528,
+      "step": 2098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3913875601102161e-05,
+      "loss": 2.9268,
+      "step": 2099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3892207341152416e-05,
+      "loss": 2.8621,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.387054947311405e-05,
+      "loss": 2.8718,
+      "step": 2101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.384890201724912e-05,
+      "loss": 2.8735,
+      "step": 2102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.382726499380994e-05,
+      "loss": 2.8243,
+      "step": 2103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3805638423039055e-05,
+      "loss": 3.0961,
+      "step": 2104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3784022325169232e-05,
+      "loss": 2.8567,
+      "step": 2105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.376241672042346e-05,
+      "loss": 3.0248,
+      "step": 2106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3740821629014874e-05,
+      "loss": 2.9685,
+      "step": 2107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.371923707114679e-05,
+      "loss": 2.7854,
+      "step": 2108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3697663067012678e-05,
+      "loss": 3.0109,
+      "step": 2109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3676099636796132e-05,
+      "loss": 2.7401,
+      "step": 2110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3654546800670848e-05,
+      "loss": 2.8947,
+      "step": 2111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3633004578800612e-05,
+      "loss": 2.7469,
+      "step": 2112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3611472991339294e-05,
+      "loss": 2.9693,
+      "step": 2113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3589952058430778e-05,
+      "loss": 2.8203,
+      "step": 2114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3568441800209014e-05,
+      "loss": 2.8807,
+      "step": 2115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3546942236797961e-05,
+      "loss": 2.9635,
+      "step": 2116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3525453388311554e-05,
+      "loss": 2.8924,
+      "step": 2117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.350397527485374e-05,
+      "loss": 3.1362,
+      "step": 2118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.348250791651837e-05,
+      "loss": 2.9281,
+      "step": 2119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3461051333389274e-05,
+      "loss": 2.8207,
+      "step": 2120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.343960554554019e-05,
+      "loss": 2.8106,
+      "step": 2121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3418170573034758e-05,
+      "loss": 2.8099,
+      "step": 2122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3396746435926493e-05,
+      "loss": 2.9131,
+      "step": 2123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3375333154258787e-05,
+      "loss": 2.9742,
+      "step": 2124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.335393074806486e-05,
+      "loss": 3.0005,
+      "step": 2125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3332539237367753e-05,
+      "loss": 2.9072,
+      "step": 2126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3311158642180332e-05,
+      "loss": 3.0594,
+      "step": 2127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.328978898250525e-05,
+      "loss": 2.8639,
+      "step": 2128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3268430278334915e-05,
+      "loss": 2.8487,
+      "step": 2129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3247082549651495e-05,
+      "loss": 2.9104,
+      "step": 2130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3225745816426905e-05,
+      "loss": 2.8196,
+      "step": 2131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3204420098622728e-05,
+      "loss": 2.9795,
+      "step": 2132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3183105416190284e-05,
+      "loss": 2.7966,
+      "step": 2133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3161801789070555e-05,
+      "loss": 2.8508,
+      "step": 2134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3140509237194176e-05,
+      "loss": 2.8787,
+      "step": 2135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3119227780481442e-05,
+      "loss": 2.9546,
+      "step": 2136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.309795743884222e-05,
+      "loss": 2.9807,
+      "step": 2137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.307669823217603e-05,
+      "loss": 2.8689,
+      "step": 2138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3055450180371948e-05,
+      "loss": 2.8805,
+      "step": 2139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3034213303308626e-05,
+      "loss": 2.7788,
+      "step": 2140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3012987620854239e-05,
+      "loss": 2.823,
+      "step": 2141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2991773152866515e-05,
+      "loss": 2.7614,
+      "step": 2142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2970569919192691e-05,
+      "loss": 2.7125,
+      "step": 2143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.294937793966946e-05,
+      "loss": 2.9344,
+      "step": 2144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2928197234123018e-05,
+      "loss": 2.8638,
+      "step": 2145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2907027822369005e-05,
+      "loss": 2.8825,
+      "step": 2146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.28858697242125e-05,
+      "loss": 2.8795,
+      "step": 2147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.286472295944799e-05,
+      "loss": 2.8853,
+      "step": 2148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2843587547859361e-05,
+      "loss": 2.869,
+      "step": 2149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2822463509219874e-05,
+      "loss": 2.8893,
+      "step": 2150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.280135086329215e-05,
+      "loss": 2.9045,
+      "step": 2151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2780249629828162e-05,
+      "loss": 3.0404,
+      "step": 2152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.27591598285692e-05,
+      "loss": 2.7655,
+      "step": 2153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2738081479245863e-05,
+      "loss": 2.8406,
+      "step": 2154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2717014601578017e-05,
+      "loss": 2.9295,
+      "step": 2155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2695959215274816e-05,
+      "loss": 2.9663,
+      "step": 2156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2674915340034669e-05,
+      "loss": 2.9116,
+      "step": 2157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.265388299554518e-05,
+      "loss": 2.7031,
+      "step": 2158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2632862201483204e-05,
+      "loss": 2.7214,
+      "step": 2159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.261185297751477e-05,
+      "loss": 2.945,
+      "step": 2160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.25908553432951e-05,
+      "loss": 2.8524,
+      "step": 2161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2569869318468542e-05,
+      "loss": 2.8709,
+      "step": 2162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2548894922668612e-05,
+      "loss": 2.9877,
+      "step": 2163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2527932175517934e-05,
+      "loss": 2.8508,
+      "step": 2164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2506981096628237e-05,
+      "loss": 2.9388,
+      "step": 2165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2486041705600335e-05,
+      "loss": 2.8851,
+      "step": 2166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.246511402202411e-05,
+      "loss": 2.8306,
+      "step": 2167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2444198065478474e-05,
+      "loss": 2.8156,
+      "step": 2168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2423293855531384e-05,
+      "loss": 2.9053,
+      "step": 2169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2402401411739806e-05,
+      "loss": 3.0263,
+      "step": 2170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2381520753649694e-05,
+      "loss": 2.8923,
+      "step": 2171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2360651900795995e-05,
+      "loss": 2.8178,
+      "step": 2172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2339794872702562e-05,
+      "loss": 2.841,
+      "step": 2173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2318949688882247e-05,
+      "loss": 2.8855,
+      "step": 2174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.229811636883677e-05,
+      "loss": 2.9196,
+      "step": 2175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2277294932056782e-05,
+      "loss": 2.7278,
+      "step": 2176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2256485398021808e-05,
+      "loss": 2.8811,
+      "step": 2177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.223568778620024e-05,
+      "loss": 2.8608,
+      "step": 2178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.221490211604932e-05,
+      "loss": 2.8541,
+      "step": 2179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2194128407015093e-05,
+      "loss": 2.8228,
+      "step": 2180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2173366678532444e-05,
+      "loss": 2.7869,
+      "step": 2181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2152616950025037e-05,
+      "loss": 2.9077,
+      "step": 2182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2131879240905309e-05,
+      "loss": 2.956,
+      "step": 2183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2111153570574454e-05,
+      "loss": 2.9574,
+      "step": 2184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2090439958422415e-05,
+      "loss": 2.7951,
+      "step": 2185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2069738423827817e-05,
+      "loss": 3.0131,
+      "step": 2186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2049048986158023e-05,
+      "loss": 2.8445,
+      "step": 2187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2028371664769069e-05,
+      "loss": 2.8382,
+      "step": 2188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2007706479005651e-05,
+      "loss": 2.847,
+      "step": 2189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1987053448201118e-05,
+      "loss": 2.8652,
+      "step": 2190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.196641259167743e-05,
+      "loss": 2.7963,
+      "step": 2191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1945783928745186e-05,
+      "loss": 2.8637,
+      "step": 2192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1925167478703544e-05,
+      "loss": 2.8801,
+      "step": 2193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1904563260840259e-05,
+      "loss": 2.9815,
+      "step": 2194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1883971294431642e-05,
+      "loss": 2.8868,
+      "step": 2195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1863391598742535e-05,
+      "loss": 2.8857,
+      "step": 2196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1842824193026308e-05,
+      "loss": 2.9094,
+      "step": 2197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1822269096524812e-05,
+      "loss": 2.8555,
+      "step": 2198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1801726328468405e-05,
+      "loss": 2.8582,
+      "step": 2199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1781195908075904e-05,
+      "loss": 2.9729,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1760677854554569e-05,
+      "loss": 2.8234,
+      "step": 2201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1740172187100101e-05,
+      "loss": 2.8885,
+      "step": 2202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1719678924896616e-05,
+      "loss": 2.8788,
+      "step": 2203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1699198087116589e-05,
+      "loss": 2.8799,
+      "step": 2204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1678729692920911e-05,
+      "loss": 2.7425,
+      "step": 2205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1658273761458819e-05,
+      "loss": 2.9197,
+      "step": 2206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1637830311867886e-05,
+      "loss": 2.8818,
+      "step": 2207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1617399363274022e-05,
+      "loss": 2.7187,
+      "step": 2208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1596980934791413e-05,
+      "loss": 2.8619,
+      "step": 2209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1576575045522566e-05,
+      "loss": 2.7934,
+      "step": 2210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.155618171455822e-05,
+      "loss": 2.8264,
+      "step": 2211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1535800960977397e-05,
+      "loss": 2.7591,
+      "step": 2212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1515432803847337e-05,
+      "loss": 2.9177,
+      "step": 2213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1495077262223503e-05,
+      "loss": 2.8239,
+      "step": 2214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1474734355149558e-05,
+      "loss": 2.9357,
+      "step": 2215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1454404101657317e-05,
+      "loss": 2.8378,
+      "step": 2216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1434086520766792e-05,
+      "loss": 2.7312,
+      "step": 2217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1413781631486122e-05,
+      "loss": 2.8687,
+      "step": 2218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.139348945281158e-05,
+      "loss": 2.8116,
+      "step": 2219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1373210003727535e-05,
+      "loss": 2.8897,
+      "step": 2220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.135294330320647e-05,
+      "loss": 2.8922,
+      "step": 2221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1332689370208905e-05,
+      "loss": 2.8321,
+      "step": 2222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1312448223683444e-05,
+      "loss": 2.7708,
+      "step": 2223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1292219882566727e-05,
+      "loss": 2.77,
+      "step": 2224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1272004365783408e-05,
+      "loss": 2.8925,
+      "step": 2225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.125180169224613e-05,
+      "loss": 2.8388,
+      "step": 2226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1231611880855545e-05,
+      "loss": 2.9976,
+      "step": 2227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.121143495050026e-05,
+      "loss": 2.7072,
+      "step": 2228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.119127092005682e-05,
+      "loss": 3.0128,
+      "step": 2229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1171119808389721e-05,
+      "loss": 2.8828,
+      "step": 2230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1150981634351362e-05,
+      "loss": 2.8628,
+      "step": 2231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1130856416782045e-05,
+      "loss": 2.8084,
+      "step": 2232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1110744174509952e-05,
+      "loss": 2.8735,
+      "step": 2233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1090644926351102e-05,
+      "loss": 2.8878,
+      "step": 2234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1070558691109385e-05,
+      "loss": 2.7629,
+      "step": 2235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1050485487576506e-05,
+      "loss": 2.7802,
+      "step": 2236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1030425334531982e-05,
+      "loss": 2.8205,
+      "step": 2237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1010378250743117e-05,
+      "loss": 2.7462,
+      "step": 2238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0990344254964999e-05,
+      "loss": 2.8447,
+      "step": 2239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0970323365940444e-05,
+      "loss": 2.8913,
+      "step": 2240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0950315602400027e-05,
+      "loss": 2.8424,
+      "step": 2241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0930320983062057e-05,
+      "loss": 2.8733,
+      "step": 2242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.09103395266325e-05,
+      "loss": 2.7048,
+      "step": 2243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0890371251805059e-05,
+      "loss": 2.8917,
+      "step": 2244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0870416177261072e-05,
+      "loss": 2.8007,
+      "step": 2245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0850474321669552e-05,
+      "loss": 2.8481,
+      "step": 2246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0830545703687109e-05,
+      "loss": 2.8765,
+      "step": 2247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0810630341958004e-05,
+      "loss": 2.8442,
+      "step": 2248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0790728255114074e-05,
+      "loss": 2.8408,
+      "step": 2249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0770839461774753e-05,
+      "loss": 2.8774,
+      "step": 2250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0750963980547027e-05,
+      "loss": 2.8717,
+      "step": 2251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0731101830025441e-05,
+      "loss": 2.7995,
+      "step": 2252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0711253028792039e-05,
+      "loss": 2.8324,
+      "step": 2253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0691417595416407e-05,
+      "loss": 2.9077,
+      "step": 2254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.067159554845561e-05,
+      "loss": 3.0149,
+      "step": 2255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0651786906454193e-05,
+      "loss": 2.8544,
+      "step": 2256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.063199168794417e-05,
+      "loss": 2.7706,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.061220991144497e-05,
+      "loss": 2.7437,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0592441595463477e-05,
+      "loss": 2.9893,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.057268675849395e-05,
+      "loss": 2.94,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0552945419018065e-05,
+      "loss": 2.9912,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0533217595504858e-05,
+      "loss": 2.8175,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0513503306410726e-05,
+      "loss": 2.9742,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0493802570179412e-05,
+      "loss": 2.7633,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0474115405241944e-05,
+      "loss": 2.8938,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0454441830016687e-05,
+      "loss": 2.8038,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0434781862909287e-05,
+      "loss": 2.8253,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.041513552231265e-05,
+      "loss": 3.038,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0395502826606942e-05,
+      "loss": 2.9512,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0375883794159565e-05,
+      "loss": 2.8576,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0356278443325118e-05,
+      "loss": 2.9123,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0336686792445425e-05,
+      "loss": 2.867,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0317108859849477e-05,
+      "loss": 3.0427,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0297544663853446e-05,
+      "loss": 2.7808,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0277994222760645e-05,
+      "loss": 2.8698,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.02584575548615e-05,
+      "loss": 2.7553,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0238934678433593e-05,
+      "loss": 2.8405,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0219425611741554e-05,
+      "loss": 2.8603,
+      "step": 2278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0199930373037131e-05,
+      "loss": 3.0028,
+      "step": 2279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0180448980559124e-05,
+      "loss": 2.7881,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0160981452533374e-05,
+      "loss": 2.8985,
+      "step": 2281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0141527807172766e-05,
+      "loss": 3.02,
+      "step": 2282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0122088062677165e-05,
+      "loss": 2.9063,
+      "step": 2283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0102662237233464e-05,
+      "loss": 2.9892,
+      "step": 2284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0083250349015525e-05,
+      "loss": 2.871,
+      "step": 2285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.006385241618416e-05,
+      "loss": 2.9004,
+      "step": 2286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.004446845688714e-05,
+      "loss": 2.8046,
+      "step": 2287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.002509848925916e-05,
+      "loss": 2.7761,
+      "step": 2288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0005742531421805e-05,
+      "loss": 2.9844,
+      "step": 2289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.986400601483581e-06,
+      "loss": 2.8416,
+      "step": 2290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.967072717539851e-06,
+      "loss": 2.8851,
+      "step": 2291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.947758897672854e-06,
+      "loss": 2.9523,
+      "step": 2292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.928459159951665e-06,
+      "loss": 2.8292,
+      "step": 2293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.909173522432164e-06,
+      "loss": 2.8427,
+      "step": 2294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.889902003157073e-06,
+      "loss": 2.7954,
+      "step": 2295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.870644620155877e-06,
+      "loss": 2.8794,
+      "step": 2296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.851401391444853e-06,
+      "loss": 2.9071,
+      "step": 2297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.832172335027034e-06,
+      "loss": 2.81,
+      "step": 2298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.81295746889219e-06,
+      "loss": 2.8052,
+      "step": 2299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.793756811016824e-06,
+      "loss": 2.9173,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.774570379364123e-06,
+      "loss": 2.8691,
+      "step": 2301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.75539819188399e-06,
+      "loss": 2.9394,
+      "step": 2302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.736240266512992e-06,
+      "loss": 2.9292,
+      "step": 2303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.717096621174354e-06,
+      "loss": 2.8595,
+      "step": 2304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.697967273777939e-06,
+      "loss": 2.8051,
+      "step": 2305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.678852242220249e-06,
+      "loss": 2.7951,
+      "step": 2306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.659751544384355e-06,
+      "loss": 2.943,
+      "step": 2307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.640665198139956e-06,
+      "loss": 2.8633,
+      "step": 2308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.621593221343308e-06,
+      "loss": 2.7292,
+      "step": 2309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.60253563183724e-06,
+      "loss": 2.852,
+      "step": 2310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.58349244745108e-06,
+      "loss": 2.8262,
+      "step": 2311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.564463686000727e-06,
+      "loss": 2.8771,
+      "step": 2312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.545449365288567e-06,
+      "loss": 3.0363,
+      "step": 2313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.52644950310346e-06,
+      "loss": 2.951,
+      "step": 2314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.507464117220762e-06,
+      "loss": 2.893,
+      "step": 2315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.48849322540228e-06,
+      "loss": 2.8273,
+      "step": 2316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.46953684539626e-06,
+      "loss": 2.9894,
+      "step": 2317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.450594994937372e-06,
+      "loss": 2.8104,
+      "step": 2318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.431667691746685e-06,
+      "loss": 2.7629,
+      "step": 2319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.412754953531663e-06,
+      "loss": 2.7911,
+      "step": 2320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.393856797986153e-06,
+      "loss": 2.9343,
+      "step": 2321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.374973242790346e-06,
+      "loss": 3.0493,
+      "step": 2322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.356104305610786e-06,
+      "loss": 3.116,
+      "step": 2323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.337250004100337e-06,
+      "loss": 2.883,
+      "step": 2324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.318410355898153e-06,
+      "loss": 2.9603,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.299585378629703e-06,
+      "loss": 2.9324,
+      "step": 2326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.280775089906726e-06,
+      "loss": 2.881,
+      "step": 2327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.261979507327204e-06,
+      "loss": 2.8256,
+      "step": 2328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.243198648475373e-06,
+      "loss": 2.7481,
+      "step": 2329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.224432530921698e-06,
+      "loss": 2.7813,
+      "step": 2330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.205681172222854e-06,
+      "loss": 2.9316,
+      "step": 2331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.186944589921687e-06,
+      "loss": 2.9574,
+      "step": 2332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.168222801547236e-06,
+      "loss": 2.6641,
+      "step": 2333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.149515824614701e-06,
+      "loss": 2.8854,
+      "step": 2334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.130823676625421e-06,
+      "loss": 2.8471,
+      "step": 2335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.112146375066871e-06,
+      "loss": 3.0796,
+      "step": 2336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.093483937412608e-06,
+      "loss": 2.7595,
+      "step": 2337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.074836381122312e-06,
+      "loss": 2.892,
+      "step": 2338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.056203723641732e-06,
+      "loss": 2.9215,
+      "step": 2339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.037585982402677e-06,
+      "loss": 2.9295,
+      "step": 2340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.018983174823e-06,
+      "loss": 2.7754,
+      "step": 2341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.000395318306598e-06,
+      "loss": 2.8985,
+      "step": 2342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.981822430243353e-06,
+      "loss": 2.8717,
+      "step": 2343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.963264528009149e-06,
+      "loss": 2.8557,
+      "step": 2344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.944721628965868e-06,
+      "loss": 2.8117,
+      "step": 2345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.926193750461346e-06,
+      "loss": 2.8976,
+      "step": 2346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.907680909829366e-06,
+      "loss": 2.9686,
+      "step": 2347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.889183124389644e-06,
+      "loss": 2.793,
+      "step": 2348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.870700411447816e-06,
+      "loss": 2.7771,
+      "step": 2349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.852232788295396e-06,
+      "loss": 2.7553,
+      "step": 2350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.8337802722098e-06,
+      "loss": 2.7789,
+      "step": 2351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.815342880454311e-06,
+      "loss": 2.8173,
+      "step": 2352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.796920630278051e-06,
+      "loss": 2.793,
+      "step": 2353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.778513538916e-06,
+      "loss": 3.022,
+      "step": 2354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.760121623588912e-06,
+      "loss": 2.7686,
+      "step": 2355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.741744901503388e-06,
+      "loss": 2.8284,
+      "step": 2356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.723383389851788e-06,
+      "loss": 2.8944,
+      "step": 2357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.705037105812259e-06,
+      "loss": 2.9449,
+      "step": 2358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.686706066548686e-06,
+      "loss": 2.9533,
+      "step": 2359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.66839028921071e-06,
+      "loss": 2.8747,
+      "step": 2360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.650089790933674e-06,
+      "loss": 2.7906,
+      "step": 2361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.63180458883863e-06,
+      "loss": 2.8167,
+      "step": 2362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.61353470003233e-06,
+      "loss": 3.0015,
+      "step": 2363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.595280141607198e-06,
+      "loss": 2.8545,
+      "step": 2364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.577040930641308e-06,
+      "loss": 2.828,
+      "step": 2365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.558817084198387e-06,
+      "loss": 2.8722,
+      "step": 2366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.540608619327781e-06,
+      "loss": 2.7584,
+      "step": 2367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.522415553064433e-06,
+      "loss": 2.9249,
+      "step": 2368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.504237902428902e-06,
+      "loss": 2.7009,
+      "step": 2369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.48607568442731e-06,
+      "loss": 2.802,
+      "step": 2370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.46792891605135e-06,
+      "loss": 2.7911,
+      "step": 2371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.44979761427826e-06,
+      "loss": 2.9434,
+      "step": 2372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.431681796070809e-06,
+      "loss": 2.7824,
+      "step": 2373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.413581478377261e-06,
+      "loss": 2.8451,
+      "step": 2374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.395496678131401e-06,
+      "loss": 2.9685,
+      "step": 2375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.377427412252495e-06,
+      "loss": 2.84,
+      "step": 2376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.359373697645276e-06,
+      "loss": 2.7653,
+      "step": 2377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.341335551199902e-06,
+      "loss": 2.7476,
+      "step": 2378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.323312989792012e-06,
+      "loss": 2.865,
+      "step": 2379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.305306030282617e-06,
+      "loss": 2.9375,
+      "step": 2380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.287314689518161e-06,
+      "loss": 2.9395,
+      "step": 2381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.269338984330474e-06,
+      "loss": 2.8775,
+      "step": 2382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.251378931536749e-06,
+      "loss": 2.8315,
+      "step": 2383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.23343454793954e-06,
+      "loss": 2.6765,
+      "step": 2384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.215505850326747e-06,
+      "loss": 2.8353,
+      "step": 2385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.197592855471575e-06,
+      "loss": 2.9233,
+      "step": 2386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.179695580132563e-06,
+      "loss": 2.9171,
+      "step": 2387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.161814041053526e-06,
+      "loss": 2.8952,
+      "step": 2388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.143948254963568e-06,
+      "loss": 2.7238,
+      "step": 2389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.126098238577048e-06,
+      "loss": 3.0894,
+      "step": 2390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.108264008593588e-06,
+      "loss": 2.9038,
+      "step": 2391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.090445581698006e-06,
+      "loss": 2.9138,
+      "step": 2392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.072642974560369e-06,
+      "loss": 2.8821,
+      "step": 2393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.054856203835934e-06,
+      "loss": 2.8913,
+      "step": 2394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.037085286165128e-06,
+      "loss": 2.766,
+      "step": 2395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.019330238173567e-06,
+      "loss": 2.8754,
+      "step": 2396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.00159107647201e-06,
+      "loss": 2.9049,
+      "step": 2397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.983867817656365e-06,
+      "loss": 2.954,
+      "step": 2398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.966160478307632e-06,
+      "loss": 2.9743,
+      "step": 2399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.948469074991954e-06,
+      "loss": 2.8024,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.93079362426054e-06,
+      "loss": 2.8056,
+      "step": 2401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.91313414264969e-06,
+      "loss": 2.7608,
+      "step": 2402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.895490646680758e-06,
+      "loss": 2.9297,
+      "step": 2403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.877863152860133e-06,
+      "loss": 2.9054,
+      "step": 2404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.860251677679247e-06,
+      "loss": 2.9372,
+      "step": 2405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.842656237614543e-06,
+      "loss": 3.0439,
+      "step": 2406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.825076849127458e-06,
+      "loss": 2.9003,
+      "step": 2407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.807513528664414e-06,
+      "loss": 2.8052,
+      "step": 2408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.78996629265681e-06,
+      "loss": 2.9056,
+      "step": 2409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.772435157520972e-06,
+      "loss": 2.7973,
+      "step": 2410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.754920139658195e-06,
+      "loss": 3.0256,
+      "step": 2411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.73742125545466e-06,
+      "loss": 2.8423,
+      "step": 2412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.719938521281487e-06,
+      "loss": 2.9581,
+      "step": 2413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.702471953494667e-06,
+      "loss": 2.812,
+      "step": 2414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.685021568435074e-06,
+      "loss": 2.7616,
+      "step": 2415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.667587382428454e-06,
+      "loss": 3.0359,
+      "step": 2416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.650169411785362e-06,
+      "loss": 2.7896,
+      "step": 2417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.63276767280122e-06,
+      "loss": 2.8915,
+      "step": 2418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.6153821817562425e-06,
+      "loss": 2.8447,
+      "step": 2419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.598012954915456e-06,
+      "loss": 2.8562,
+      "step": 2420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.580660008528673e-06,
+      "loss": 2.9372,
+      "step": 2421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.563323358830448e-06,
+      "loss": 2.8603,
+      "step": 2422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.546003022040121e-06,
+      "loss": 2.8884,
+      "step": 2423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.528699014361756e-06,
+      "loss": 2.8088,
+      "step": 2424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5114113519841405e-06,
+      "loss": 2.8706,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494140051080775e-06,
+      "loss": 2.8231,
+      "step": 2426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.476885127809857e-06,
+      "loss": 3.0574,
+      "step": 2427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.459646598314246e-06,
+      "loss": 2.6782,
+      "step": 2428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4424244787214656e-06,
+      "loss": 2.9559,
+      "step": 2429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.425218785143703e-06,
+      "loss": 3.0178,
+      "step": 2430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.408029533677768e-06,
+      "loss": 2.9536,
+      "step": 2431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.390856740405092e-06,
+      "loss": 2.9444,
+      "step": 2432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.373700421391702e-06,
+      "loss": 2.8442,
+      "step": 2433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3565605926882306e-06,
+      "loss": 2.924,
+      "step": 2434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.339437270329847e-06,
+      "loss": 2.8849,
+      "step": 2435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3223304703363135e-06,
+      "loss": 2.8317,
+      "step": 2436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3052402087119195e-06,
+      "loss": 2.7201,
+      "step": 2437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2881665014454846e-06,
+      "loss": 2.9706,
+      "step": 2438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.271109364510345e-06,
+      "loss": 2.9302,
+      "step": 2439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.254068813864315e-06,
+      "loss": 2.8115,
+      "step": 2440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2370448654497185e-06,
+      "loss": 2.9235,
+      "step": 2441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.22003753519333e-06,
+      "loss": 2.887,
+      "step": 2442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.203046839006383e-06,
+      "loss": 2.7604,
+      "step": 2443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.186072792784548e-06,
+      "loss": 2.8275,
+      "step": 2444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.169115412407926e-06,
+      "loss": 2.6798,
+      "step": 2445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.152174713741014e-06,
+      "loss": 2.8493,
+      "step": 2446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.135250712632693e-06,
+      "loss": 2.8738,
+      "step": 2447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.118343424916249e-06,
+      "loss": 2.7124,
+      "step": 2448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.101452866409316e-06,
+      "loss": 2.8791,
+      "step": 2449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.084579052913884e-06,
+      "loss": 2.822,
+      "step": 2450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.067722000216271e-06,
+      "loss": 2.9093,
+      "step": 2451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.050881724087125e-06,
+      "loss": 2.9447,
+      "step": 2452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.034058240281377e-06,
+      "loss": 2.8167,
+      "step": 2453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0172515645382655e-06,
+      "loss": 2.937,
+      "step": 2454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.000461712581302e-06,
+      "loss": 2.8283,
+      "step": 2455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9836887001182575e-06,
+      "loss": 2.8178,
+      "step": 2456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.966932542841156e-06,
+      "loss": 2.8533,
+      "step": 2457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.950193256426227e-06,
+      "loss": 2.9091,
+      "step": 2458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.933470856533944e-06,
+      "loss": 3.0768,
+      "step": 2459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9167653588089685e-06,
+      "loss": 2.8239,
+      "step": 2460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9000767788801575e-06,
+      "loss": 2.8526,
+      "step": 2461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.88340513236054e-06,
+      "loss": 2.8931,
+      "step": 2462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.866750434847286e-06,
+      "loss": 2.8146,
+      "step": 2463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8501127019217346e-06,
+      "loss": 2.8345,
+      "step": 2464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.833491949149329e-06,
+      "loss": 2.8496,
+      "step": 2465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8168881920796425e-06,
+      "loss": 3.0632,
+      "step": 2466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.800301446246346e-06,
+      "loss": 2.7952,
+      "step": 2467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.7837317271671945e-06,
+      "loss": 2.7625,
+      "step": 2468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.767179050344008e-06,
+      "loss": 2.9106,
+      "step": 2469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.750643431262679e-06,
+      "loss": 2.8671,
+      "step": 2470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.734124885393111e-06,
+      "loss": 2.7338,
+      "step": 2471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.7176234281892624e-06,
+      "loss": 2.8307,
+      "step": 2472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.7011390750891e-06,
+      "loss": 2.785,
+      "step": 2473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.6846718415145754e-06,
+      "loss": 2.8346,
+      "step": 2474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.668221742871647e-06,
+      "loss": 2.8168,
+      "step": 2475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.651788794550209e-06,
+      "loss": 2.8889,
+      "step": 2476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.63537301192414e-06,
+      "loss": 2.8807,
+      "step": 2477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.618974410351247e-06,
+      "loss": 2.9146,
+      "step": 2478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.602593005173274e-06,
+      "loss": 2.7171,
+      "step": 2479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.586228811715853e-06,
+      "loss": 2.835,
+      "step": 2480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.569881845288536e-06,
+      "loss": 2.8659,
+      "step": 2481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.553552121184759e-06,
+      "loss": 2.8949,
+      "step": 2482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.537239654681804e-06,
+      "loss": 2.8137,
+      "step": 2483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.520944461040829e-06,
+      "loss": 2.8673,
+      "step": 2484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.504666555506825e-06,
+      "loss": 2.8487,
+      "step": 2485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4884059533086125e-06,
+      "loss": 2.855,
+      "step": 2486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4721626696588185e-06,
+      "loss": 3.024,
+      "step": 2487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.455936719753882e-06,
+      "loss": 2.8465,
+      "step": 2488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.439728118773994e-06,
+      "loss": 2.8115,
+      "step": 2489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4235368818831435e-06,
+      "loss": 2.8499,
+      "step": 2490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.407363024229069e-06,
+      "loss": 2.8255,
+      "step": 2491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3912065609432415e-06,
+      "loss": 2.8856,
+      "step": 2492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.375067507140864e-06,
+      "loss": 2.8392,
+      "step": 2493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.358945877920861e-06,
+      "loss": 2.9186,
+      "step": 2494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.34284168836583e-06,
+      "loss": 2.8818,
+      "step": 2495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3267549535420855e-06,
+      "loss": 2.8125,
+      "step": 2496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.310685688499576e-06,
+      "loss": 2.8353,
+      "step": 2497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.294633908271938e-06,
+      "loss": 2.8927,
+      "step": 2498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.278599627876433e-06,
+      "loss": 2.7562,
+      "step": 2499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.262582862313968e-06,
+      "loss": 2.9507,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.24658362656903e-06,
+      "loss": 2.8167,
+      "step": 2501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.230601935609737e-06,
+      "loss": 2.6208,
+      "step": 2502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.214637804387785e-06,
+      "loss": 2.8704,
+      "step": 2503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.198691247838437e-06,
+      "loss": 2.8939,
+      "step": 2504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.1827622808805194e-06,
+      "loss": 2.7895,
+      "step": 2505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.166850918416406e-06,
+      "loss": 3.0149,
+      "step": 2506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.150957175331979e-06,
+      "loss": 2.8745,
+      "step": 2507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.135081066496662e-06,
+      "loss": 2.9412,
+      "step": 2508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.11922260676337e-06,
+      "loss": 2.7951,
+      "step": 2509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.103381810968512e-06,
+      "loss": 3.1414,
+      "step": 2510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.087558693931958e-06,
+      "loss": 2.8724,
+      "step": 2511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.071753270457064e-06,
+      "loss": 3.0466,
+      "step": 2512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.055965555330606e-06,
+      "loss": 2.759,
+      "step": 2513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.040195563322798e-06,
+      "loss": 2.8331,
+      "step": 2514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.024443309187283e-06,
+      "loss": 2.9937,
+      "step": 2515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.008708807661109e-06,
+      "loss": 2.8217,
+      "step": 2516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.992992073464706e-06,
+      "loss": 2.9669,
+      "step": 2517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.977293121301891e-06,
+      "loss": 2.857,
+      "step": 2518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.961611965859848e-06,
+      "loss": 2.8132,
+      "step": 2519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.945948621809091e-06,
+      "loss": 3.0184,
+      "step": 2520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.930303103803489e-06,
+      "loss": 3.0046,
+      "step": 2521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.914675426480232e-06,
+      "loss": 2.881,
+      "step": 2522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.899065604459814e-06,
+      "loss": 2.7689,
+      "step": 2523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.88347365234603e-06,
+      "loss": 2.7619,
+      "step": 2524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.867899584725944e-06,
+      "loss": 2.8044,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.852343416169906e-06,
+      "loss": 2.8638,
+      "step": 2526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.836805161231507e-06,
+      "loss": 2.952,
+      "step": 2527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.821284834447585e-06,
+      "loss": 2.9526,
+      "step": 2528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.805782450338204e-06,
+      "loss": 2.9153,
+      "step": 2529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.790298023406651e-06,
+      "loss": 2.9041,
+      "step": 2530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.774831568139394e-06,
+      "loss": 2.8723,
+      "step": 2531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.759383099006094e-06,
+      "loss": 2.905,
+      "step": 2532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.74395263045959e-06,
+      "loss": 2.8739,
+      "step": 2533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7285401769358845e-06,
+      "loss": 2.7912,
+      "step": 2534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.713145752854118e-06,
+      "loss": 2.7187,
+      "step": 2535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.697769372616566e-06,
+      "loss": 2.9975,
+      "step": 2536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.68241105060863e-06,
+      "loss": 2.8755,
+      "step": 2537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.667070801198798e-06,
+      "loss": 2.7839,
+      "step": 2538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.651748638738669e-06,
+      "loss": 2.7758,
+      "step": 2539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.63644457756291e-06,
+      "loss": 2.8013,
+      "step": 2540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6211586319892625e-06,
+      "loss": 2.8308,
+      "step": 2541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6058908163185205e-06,
+      "loss": 2.7252,
+      "step": 2542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5906411448344965e-06,
+      "loss": 2.7297,
+      "step": 2543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5754096318040485e-06,
+      "loss": 2.886,
+      "step": 2544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.56019629147704e-06,
+      "loss": 2.866,
+      "step": 2545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.545001138086345e-06,
+      "loss": 2.8624,
+      "step": 2546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.529824185847793e-06,
+      "loss": 2.8282,
+      "step": 2547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.51466544896021e-06,
+      "loss": 2.898,
+      "step": 2548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.499524941605386e-06,
+      "loss": 2.9047,
+      "step": 2549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.484402677948022e-06,
+      "loss": 2.8554,
+      "step": 2550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.469298672135789e-06,
+      "loss": 2.8777,
+      "step": 2551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.454212938299255e-06,
+      "loss": 2.9583,
+      "step": 2552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.439145490551905e-06,
+      "loss": 2.9001,
+      "step": 2553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4240963429901095e-06,
+      "loss": 2.8113,
+      "step": 2554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.409065509693126e-06,
+      "loss": 2.8529,
+      "step": 2555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3940530047230594e-06,
+      "loss": 2.9421,
+      "step": 2556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.379058842124884e-06,
+      "loss": 2.9824,
+      "step": 2557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.364083035926417e-06,
+      "loss": 2.8012,
+      "step": 2558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3491256001382915e-06,
+      "loss": 2.9638,
+      "step": 2559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.334186548753961e-06,
+      "loss": 2.9604,
+      "step": 2560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.31926589574967e-06,
+      "loss": 2.7859,
+      "step": 2561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.30436365508446e-06,
+      "loss": 2.8665,
+      "step": 2562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.289479840700149e-06,
+      "loss": 2.7662,
+      "step": 2563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2746144665213e-06,
+      "loss": 2.7897,
+      "step": 2564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.259767546455241e-06,
+      "loss": 2.9153,
+      "step": 2565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2449390943920315e-06,
+      "loss": 2.9,
+      "step": 2566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2301291242044554e-06,
+      "loss": 2.8879,
+      "step": 2567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.215337649747986e-06,
+      "loss": 2.9107,
+      "step": 2568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.20056468486082e-06,
+      "loss": 2.9445,
+      "step": 2569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.185810243363826e-06,
+      "loss": 2.8211,
+      "step": 2570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.171074339060536e-06,
+      "loss": 2.7081,
+      "step": 2571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.156356985737154e-06,
+      "loss": 3.0078,
+      "step": 2572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.141658197162527e-06,
+      "loss": 2.8632,
+      "step": 2573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.126977987088108e-06,
+      "loss": 2.7864,
+      "step": 2574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.112316369248002e-06,
+      "loss": 2.8319,
+      "step": 2575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.097673357358907e-06,
+      "loss": 2.9457,
+      "step": 2576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0830489651201075e-06,
+      "loss": 2.7857,
+      "step": 2577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.068443206213486e-06,
+      "loss": 2.9109,
+      "step": 2578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.053856094303467e-06,
+      "loss": 2.8667,
+      "step": 2579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.039287643037058e-06,
+      "loss": 2.8759,
+      "step": 2580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.02473786604378e-06,
+      "loss": 2.8398,
+      "step": 2581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.010206776935705e-06,
+      "loss": 2.9464,
+      "step": 2582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.995694389307412e-06,
+      "loss": 2.9544,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981200716735992e-06,
+      "loss": 3.0624,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.966725772781022e-06,
+      "loss": 2.8769,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9522695709845466e-06,
+      "loss": 2.9129,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.937832124871092e-06,
+      "loss": 2.9796,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.92341344794763e-06,
+      "loss": 2.8726,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9090135537035755e-06,
+      "loss": 2.8063,
+      "step": 2589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.894632455610773e-06,
+      "loss": 2.7427,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8802701671234826e-06,
+      "loss": 2.8348,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.865926701678353e-06,
+      "loss": 2.9688,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.851602072694439e-06,
+      "loss": 2.9121,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.837296293573168e-06,
+      "loss": 2.7691,
+      "step": 2594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.823009377698332e-06,
+      "loss": 2.8642,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.808741338436082e-06,
+      "loss": 2.8864,
+      "step": 2596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.794492189134892e-06,
+      "loss": 2.8057,
+      "step": 2597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780261943125586e-06,
+      "loss": 2.9223,
+      "step": 2598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.766050613721276e-06,
+      "loss": 2.8324,
+      "step": 2599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7518582142174e-06,
+      "loss": 2.8208,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7376847578916776e-06,
+      "loss": 2.7604,
+      "step": 2601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.723530258004106e-06,
+      "loss": 2.815,
+      "step": 2602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.709394727796953e-06,
+      "loss": 2.9323,
+      "step": 2603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.695278180494725e-06,
+      "loss": 2.8119,
+      "step": 2604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.681180629304182e-06,
+      "loss": 2.87,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.667102087414307e-06,
+      "loss": 2.9878,
+      "step": 2606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.653042567996305e-06,
+      "loss": 2.9581,
+      "step": 2607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.639002084203575e-06,
+      "loss": 2.8871,
+      "step": 2608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6249806491717175e-06,
+      "loss": 3.017,
+      "step": 2609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610978276018496e-06,
+      "loss": 2.8463,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5969949778438575e-06,
+      "loss": 2.6449,
+      "step": 2611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.583030767729899e-06,
+      "loss": 2.8081,
+      "step": 2612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.569085658740848e-06,
+      "loss": 2.7938,
+      "step": 2613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.55515966392309e-06,
+      "loss": 2.8405,
+      "step": 2614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.541252796305087e-06,
+      "loss": 2.8232,
+      "step": 2615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.527365068897443e-06,
+      "loss": 2.8335,
+      "step": 2616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.513496494692834e-06,
+      "loss": 2.7136,
+      "step": 2617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.499647086666029e-06,
+      "loss": 2.8938,
+      "step": 2618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4858168577738586e-06,
+      "loss": 2.9398,
+      "step": 2619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.472005820955216e-06,
+      "loss": 2.9115,
+      "step": 2620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.458213989131041e-06,
+      "loss": 2.8326,
+      "step": 2621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.444441375204289e-06,
+      "loss": 2.7317,
+      "step": 2622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.430687992059959e-06,
+      "loss": 2.8838,
+      "step": 2623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.416953852565045e-06,
+      "loss": 2.8852,
+      "step": 2624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.40323896956854e-06,
+      "loss": 2.8155,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.389543355901423e-06,
+      "loss": 2.9085,
+      "step": 2626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.375867024376653e-06,
+      "loss": 2.8759,
+      "step": 2627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.362209987789128e-06,
+      "loss": 2.9316,
+      "step": 2628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.348572258915717e-06,
+      "loss": 2.6159,
+      "step": 2629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.334953850515214e-06,
+      "loss": 3.0065,
+      "step": 2630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.321354775328345e-06,
+      "loss": 2.8054,
+      "step": 2631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.307775046077739e-06,
+      "loss": 2.8513,
+      "step": 2632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.294214675467928e-06,
+      "loss": 2.9036,
+      "step": 2633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.280673676185351e-06,
+      "loss": 2.7477,
+      "step": 2634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.267152060898291e-06,
+      "loss": 2.8,
+      "step": 2635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.253649842256924e-06,
+      "loss": 2.906,
+      "step": 2636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.240167032893266e-06,
+      "loss": 2.87,
+      "step": 2637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.226703645421185e-06,
+      "loss": 2.8813,
+      "step": 2638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.213259692436367e-06,
+      "loss": 2.9296,
+      "step": 2639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.199835186516332e-06,
+      "loss": 2.8701,
+      "step": 2640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.186430140220379e-06,
+      "loss": 2.8627,
+      "step": 2641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1730445660896286e-06,
+      "loss": 2.7852,
+      "step": 2642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.159678476646972e-06,
+      "loss": 2.8237,
+      "step": 2643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.146331884397073e-06,
+      "loss": 3.0268,
+      "step": 2644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.133004801826362e-06,
+      "loss": 2.8711,
+      "step": 2645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.119697241402998e-06,
+      "loss": 2.9021,
+      "step": 2646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.106409215576895e-06,
+      "loss": 2.8882,
+      "step": 2647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.093140736779691e-06,
+      "loss": 2.8109,
+      "step": 2648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.079891817424713e-06,
+      "loss": 2.9575,
+      "step": 2649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.06666246990702e-06,
+      "loss": 2.748,
+      "step": 2650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0534527066033445e-06,
+      "loss": 2.9601,
+      "step": 2651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.040262539872106e-06,
+      "loss": 2.7988,
+      "step": 2652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.027091982053369e-06,
+      "loss": 2.7798,
+      "step": 2653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.013941045468875e-06,
+      "loss": 2.9611,
+      "step": 2654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.000809742422004e-06,
+      "loss": 2.8617,
+      "step": 2655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.987698085197761e-06,
+      "loss": 3.1099,
+      "step": 2656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.974606086062782e-06,
+      "loss": 2.9096,
+      "step": 2657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9615337572653064e-06,
+      "loss": 2.8605,
+      "step": 2658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.94848111103516e-06,
+      "loss": 2.9583,
+      "step": 2659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.935448159583774e-06,
+      "loss": 2.6665,
+      "step": 2660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.922434915104136e-06,
+      "loss": 2.8698,
+      "step": 2661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9094413897708155e-06,
+      "loss": 2.8919,
+      "step": 2662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.896467595739925e-06,
+      "loss": 2.8849,
+      "step": 2663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.883513545149104e-06,
+      "loss": 2.6926,
+      "step": 2664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.870579250117545e-06,
+      "loss": 2.936,
+      "step": 2665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.857664722745938e-06,
+      "loss": 3.0362,
+      "step": 2666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.844769975116488e-06,
+      "loss": 2.8382,
+      "step": 2667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.831895019292897e-06,
+      "loss": 3.1252,
+      "step": 2668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.819039867320351e-06,
+      "loss": 2.9346,
+      "step": 2669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8062045312255074e-06,
+      "loss": 2.8826,
+      "step": 2670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.793389023016475e-06,
+      "loss": 2.8377,
+      "step": 2671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7805933546828264e-06,
+      "loss": 2.9818,
+      "step": 2672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7678175381955672e-06,
+      "loss": 2.8366,
+      "step": 2673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7550615855071277e-06,
+      "loss": 2.8071,
+      "step": 2674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.742325508551364e-06,
+      "loss": 2.6971,
+      "step": 2675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.729609319243532e-06,
+      "loss": 2.8526,
+      "step": 2676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7169130294802707e-06,
+      "loss": 2.891,
+      "step": 2677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.704236651139617e-06,
+      "loss": 2.8703,
+      "step": 2678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6915801960809765e-06,
+      "loss": 3.0021,
+      "step": 2679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6789436761451133e-06,
+      "loss": 2.7637,
+      "step": 2680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.666327103154149e-06,
+      "loss": 2.8743,
+      "step": 2681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.653730488911522e-06,
+      "loss": 2.8012,
+      "step": 2682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6411538452020277e-06,
+      "loss": 2.8854,
+      "step": 2683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6285971837917517e-06,
+      "loss": 2.9325,
+      "step": 2684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6160605164280963e-06,
+      "loss": 2.8606,
+      "step": 2685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6035438548397666e-06,
+      "loss": 2.8581,
+      "step": 2686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5910472107367395e-06,
+      "loss": 3.0375,
+      "step": 2687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.578570595810274e-06,
+      "loss": 2.8048,
+      "step": 2688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.566114021732875e-06,
+      "loss": 2.7386,
+      "step": 2689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.553677500158312e-06,
+      "loss": 2.9632,
+      "step": 2690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5412610427215914e-06,
+      "loss": 2.907,
+      "step": 2691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5288646610389498e-06,
+      "loss": 2.8461,
+      "step": 2692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5164883667078375e-06,
+      "loss": 2.7286,
+      "step": 2693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5041321713069202e-06,
+      "loss": 2.993,
+      "step": 2694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491796086396043e-06,
+      "loss": 3.0817,
+      "step": 2695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479480123516257e-06,
+      "loss": 2.9267,
+      "step": 2696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4671842941897765e-06,
+      "loss": 2.9563,
+      "step": 2697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4549086099199794e-06,
+      "loss": 2.9004,
+      "step": 2698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4426530821914078e-06,
+      "loss": 2.9166,
+      "step": 2699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.430417722469728e-06,
+      "loss": 2.9933,
+      "step": 2700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4182025422017622e-06,
+      "loss": 2.7474,
+      "step": 2701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.406007552815421e-06,
+      "loss": 2.8888,
+      "step": 2702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3938327657197567e-06,
+      "loss": 2.8623,
+      "step": 2703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3816781923049046e-06,
+      "loss": 2.8028,
+      "step": 2704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.369543843942094e-06,
+      "loss": 2.8243,
+      "step": 2705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3574297319836414e-06,
+      "loss": 2.8949,
+      "step": 2706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3453358677629054e-06,
+      "loss": 2.9395,
+      "step": 2707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.333262262594328e-06,
+      "loss": 2.8821,
+      "step": 2708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.321208927773384e-06,
+      "loss": 2.885,
+      "step": 2709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3091758745765922e-06,
+      "loss": 2.9708,
+      "step": 2710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2971631142614894e-06,
+      "loss": 2.9501,
+      "step": 2711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2851706580666356e-06,
+      "loss": 2.7875,
+      "step": 2712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2731985172115815e-06,
+      "loss": 2.9755,
+      "step": 2713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.261246702896889e-06,
+      "loss": 2.8083,
+      "step": 2714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2493152263040894e-06,
+      "loss": 2.8479,
+      "step": 2715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2374040985957004e-06,
+      "loss": 2.8804,
+      "step": 2716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2255133309151862e-06,
+      "loss": 2.8522,
+      "step": 2717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2136429343869735e-06,
+      "loss": 2.8193,
+      "step": 2718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2017929201164367e-06,
+      "loss": 2.8344,
+      "step": 2719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.189963299189863e-06,
+      "loss": 2.7804,
+      "step": 2720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1781540826744754e-06,
+      "loss": 2.8532,
+      "step": 2721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1663652816184027e-06,
+      "loss": 2.8064,
+      "step": 2722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1545969070506747e-06,
+      "loss": 3.0441,
+      "step": 2723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1428489699812185e-06,
+      "loss": 2.7098,
+      "step": 2724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1311214814008226e-06,
+      "loss": 2.8701,
+      "step": 2725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.119414452281158e-06,
+      "loss": 2.9668,
+      "step": 2726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1077278935747578e-06,
+      "loss": 2.8268,
+      "step": 2727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.096061816214993e-06,
+      "loss": 2.7426,
+      "step": 2728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0844162311160856e-06,
+      "loss": 2.7378,
+      "step": 2729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0727911491730764e-06,
+      "loss": 2.9177,
+      "step": 2730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0611865812618267e-06,
+      "loss": 2.9037,
+      "step": 2731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.049602538239002e-06,
+      "loss": 2.8679,
+      "step": 2732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.038039030942083e-06,
+      "loss": 2.7918,
+      "step": 2733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0264960701893137e-06,
+      "loss": 2.8846,
+      "step": 2734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.01497366677973e-06,
+      "loss": 3.0358,
+      "step": 2735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0034718314931376e-06,
+      "loss": 2.8916,
+      "step": 2736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.991990575090095e-06,
+      "loss": 2.7949,
+      "step": 2737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9805299083118993e-06,
+      "loss": 2.9441,
+      "step": 2738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9690898418806013e-06,
+      "loss": 2.7636,
+      "step": 2739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.95767038649897e-06,
+      "loss": 2.7842,
+      "step": 2740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9462715528504937e-06,
+      "loss": 2.9159,
+      "step": 2741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9348933515993742e-06,
+      "loss": 2.822,
+      "step": 2742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9235357933904887e-06,
+      "loss": 2.8339,
+      "step": 2743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9121988888494297e-06,
+      "loss": 2.7715,
+      "step": 2744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9008826485824455e-06,
+      "loss": 2.7307,
+      "step": 2745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8895870831764675e-06,
+      "loss": 2.8201,
+      "step": 2746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8783122031990794e-06,
+      "loss": 2.9387,
+      "step": 2747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8670580191985097e-06,
+      "loss": 2.9221,
+      "step": 2748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.855824541703628e-06,
+      "loss": 2.9487,
+      "step": 2749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8446117812239213e-06,
+      "loss": 2.7838,
+      "step": 2750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.833419748249511e-06,
+      "loss": 2.9237,
+      "step": 2751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8222484532511166e-06,
+      "loss": 2.851,
+      "step": 2752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.811097906680063e-06,
+      "loss": 2.8791,
+      "step": 2753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7999681189682574e-06,
+      "loss": 2.8781,
+      "step": 2754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.788859100528196e-06,
+      "loss": 2.7557,
+      "step": 2755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7777708617529263e-06,
+      "loss": 3.0072,
+      "step": 2756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7667034130160666e-06,
+      "loss": 2.928,
+      "step": 2757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7556567646717907e-06,
+      "loss": 2.8089,
+      "step": 2758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.744630927054803e-06,
+      "loss": 2.9014,
+      "step": 2759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7336259104803504e-06,
+      "loss": 2.962,
+      "step": 2760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7226417252441768e-06,
+      "loss": 2.7461,
+      "step": 2761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7116783816225595e-06,
+      "loss": 2.8586,
+      "step": 2762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.700735889872269e-06,
+      "loss": 2.8897,
+      "step": 2763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6898142602305748e-06,
+      "loss": 2.9038,
+      "step": 2764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6789135029152173e-06,
+      "loss": 2.8104,
+      "step": 2765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.668033628124422e-06,
+      "loss": 2.8478,
+      "step": 2766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6571746460368695e-06,
+      "loss": 3.0549,
+      "step": 2767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6463365668116857e-06,
+      "loss": 2.7293,
+      "step": 2768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6355194005884597e-06,
+      "loss": 2.8216,
+      "step": 2769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.624723157487208e-06,
+      "loss": 2.9176,
+      "step": 2770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6139478476083697e-06,
+      "loss": 2.8354,
+      "step": 2771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.603193481032801e-06,
+      "loss": 2.9651,
+      "step": 2772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5924600678217696e-06,
+      "loss": 2.8313,
+      "step": 2773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5817476180169275e-06,
+      "loss": 3.0902,
+      "step": 2774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5710561416403267e-06,
+      "loss": 3.0483,
+      "step": 2775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.560385648694394e-06,
+      "loss": 2.9698,
+      "step": 2776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5497361491619214e-06,
+      "loss": 2.8939,
+      "step": 2777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5391076530060674e-06,
+      "loss": 2.9264,
+      "step": 2778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.528500170170339e-06,
+      "loss": 2.8897,
+      "step": 2779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.517913710578573e-06,
+      "loss": 2.7828,
+      "step": 2780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5073482841349526e-06,
+      "loss": 2.8739,
+      "step": 2781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.496803900723979e-06,
+      "loss": 2.727,
+      "step": 2782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4862805702104642e-06,
+      "loss": 2.845,
+      "step": 2783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.475778302439524e-06,
+      "loss": 2.9388,
+      "step": 2784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.465297107236572e-06,
+      "loss": 2.8443,
+      "step": 2785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4548369944073004e-06,
+      "loss": 2.9481,
+      "step": 2786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.444397973737686e-06,
+      "loss": 2.9039,
+      "step": 2787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.43398005499397e-06,
+      "loss": 2.796,
+      "step": 2788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4235832479226504e-06,
+      "loss": 2.8702,
+      "step": 2789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4132075622504775e-06,
+      "loss": 2.9974,
+      "step": 2790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4028530076844414e-06,
+      "loss": 3.0094,
+      "step": 2791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3925195939117516e-06,
+      "loss": 2.776,
+      "step": 2792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3822073305998534e-06,
+      "loss": 2.9103,
+      "step": 2793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.371916227396398e-06,
+      "loss": 2.8028,
+      "step": 2794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3616462939292427e-06,
+      "loss": 3.0544,
+      "step": 2795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.351397539806438e-06,
+      "loss": 2.8176,
+      "step": 2796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.341169974616228e-06,
+      "loss": 2.9229,
+      "step": 2797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.330963607927014e-06,
+      "loss": 2.8702,
+      "step": 2798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.320778449287381e-06,
+      "loss": 2.8059,
+      "step": 2799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.310614508226078e-06,
+      "loss": 2.7564,
+      "step": 2800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3004717942519794e-06,
+      "loss": 2.8647,
+      "step": 2801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.290350316854123e-06,
+      "loss": 2.6485,
+      "step": 2802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2802500855016727e-06,
+      "loss": 2.8402,
+      "step": 2803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.270171109643918e-06,
+      "loss": 2.791,
+      "step": 2804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2601133987102505e-06,
+      "loss": 2.9546,
+      "step": 2805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2500769621101786e-06,
+      "loss": 2.956,
+      "step": 2806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.24006180923331e-06,
+      "loss": 2.8918,
+      "step": 2807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.23006794944933e-06,
+      "loss": 2.88,
+      "step": 2808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2200953921080153e-06,
+      "loss": 2.8292,
+      "step": 2809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2101441465391975e-06,
+      "loss": 2.8915,
+      "step": 2810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2002142220527823e-06,
+      "loss": 2.8749,
+      "step": 2811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.190305627938724e-06,
+      "loss": 2.876,
+      "step": 2812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1804183734670277e-06,
+      "loss": 2.9485,
+      "step": 2813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.170552467887721e-06,
+      "loss": 2.8715,
+      "step": 2814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.160707920430874e-06,
+      "loss": 2.8444,
+      "step": 2815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.150884740306558e-06,
+      "loss": 2.7653,
+      "step": 2816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1410829367048717e-06,
+      "loss": 2.9073,
+      "step": 2817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1313025187958995e-06,
+      "loss": 2.7274,
+      "step": 2818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1215434957297276e-06,
+      "loss": 2.7487,
+      "step": 2819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1118058766364245e-06,
+      "loss": 2.8312,
+      "step": 2820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1020896706260367e-06,
+      "loss": 2.9037,
+      "step": 2821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0923948867885732e-06,
+      "loss": 2.82,
+      "step": 2822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.082721534193999e-06,
+      "loss": 2.8257,
+      "step": 2823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0730696218922373e-06,
+      "loss": 2.7371,
+      "step": 2824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.063439158913147e-06,
+      "loss": 2.7925,
+      "step": 2825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0538301542665202e-06,
+      "loss": 2.8649,
+      "step": 2826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0442426169420826e-06,
+      "loss": 2.7718,
+      "step": 2827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0346765559094567e-06,
+      "loss": 2.829,
+      "step": 2828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.025131980118192e-06,
+      "loss": 2.7513,
+      "step": 2829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0156088984977285e-06,
+      "loss": 2.8226,
+      "step": 2830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.006107319957398e-06,
+      "loss": 2.9319,
+      "step": 2831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9966272533864183e-06,
+      "loss": 2.8025,
+      "step": 2832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.98716870765388e-06,
+      "loss": 2.727,
+      "step": 2833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.977731691608739e-06,
+      "loss": 2.9308,
+      "step": 2834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9683162140798045e-06,
+      "loss": 2.9318,
+      "step": 2835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9589222838757415e-06,
+      "loss": 2.938,
+      "step": 2836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.949549909785056e-06,
+      "loss": 2.807,
+      "step": 2837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9401991005760842e-06,
+      "loss": 2.8642,
+      "step": 2838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9308698649969954e-06,
+      "loss": 3.0066,
+      "step": 2839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.921562211775768e-06,
+      "loss": 2.8282,
+      "step": 2840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.912276149620182e-06,
+      "loss": 2.8787,
+      "step": 2841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9030116872178316e-06,
+      "loss": 2.7224,
+      "step": 2842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8937688332361008e-06,
+      "loss": 2.8434,
+      "step": 2843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8845475963221503e-06,
+      "loss": 2.818,
+      "step": 2844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8753479851029265e-06,
+      "loss": 2.7493,
+      "step": 2845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8661700081851335e-06,
+      "loss": 2.7258,
+      "step": 2846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8570136741552419e-06,
+      "loss": 2.8435,
+      "step": 2847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8478789915794769e-06,
+      "loss": 2.9438,
+      "step": 2848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8387659690038e-06,
+      "loss": 2.8978,
+      "step": 2849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8296746149539135e-06,
+      "loss": 2.9833,
+      "step": 2850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8206049379352525e-06,
+      "loss": 2.9016,
+      "step": 2851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.81155694643296e-06,
+      "loss": 2.8846,
+      "step": 2852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8025306489118931e-06,
+      "loss": 2.7741,
+      "step": 2853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7935260538166243e-06,
+      "loss": 2.946,
+      "step": 2854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7845431695714148e-06,
+      "loss": 2.8811,
+      "step": 2855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7755820045802145e-06,
+      "loss": 2.9491,
+      "step": 2856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7666425672266534e-06,
+      "loss": 2.9312,
+      "step": 2857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7577248658740387e-06,
+      "loss": 2.8942,
+      "step": 2858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7488289088653326e-06,
+      "loss": 2.884,
+      "step": 2859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7399547045231613e-06,
+      "loss": 2.9245,
+      "step": 2860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.731102261149803e-06,
+      "loss": 2.8007,
+      "step": 2861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7222715870271717e-06,
+      "loss": 2.7854,
+      "step": 2862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7134626904168228e-06,
+      "loss": 2.8426,
+      "step": 2863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7046755795599222e-06,
+      "loss": 2.8691,
+      "step": 2864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6959102626772694e-06,
+      "loss": 2.8861,
+      "step": 2865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.687166747969271e-06,
+      "loss": 2.942,
+      "step": 2866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6784450436159282e-06,
+      "loss": 2.9529,
+      "step": 2867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6697451577768557e-06,
+      "loss": 2.8582,
+      "step": 2868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661067098591232e-06,
+      "loss": 2.737,
+      "step": 2869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6524108741778372e-06,
+      "loss": 2.9332,
+      "step": 2870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6437764926350074e-06,
+      "loss": 2.8545,
+      "step": 2871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6351639620406506e-06,
+      "loss": 2.7507,
+      "step": 2872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.626573290452238e-06,
+      "loss": 2.7969,
+      "step": 2873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6180044859067793e-06,
+      "loss": 2.9482,
+      "step": 2874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6094575564208348e-06,
+      "loss": 2.9738,
+      "step": 2875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.600932509990502e-06,
+      "loss": 2.8291,
+      "step": 2876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5924293545913876e-06,
+      "loss": 2.929,
+      "step": 2877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.583948098178642e-06,
+      "loss": 2.8427,
+      "step": 2878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5754887486869097e-06,
+      "loss": 2.9497,
+      "step": 2879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.567051314030349e-06,
+      "loss": 2.9459,
+      "step": 2880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5586358021026204e-06,
+      "loss": 2.8897,
+      "step": 2881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5502422207768597e-06,
+      "loss": 2.9311,
+      "step": 2882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5418705779056992e-06,
+      "loss": 3.0635,
+      "step": 2883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5335208813212375e-06,
+      "loss": 2.739,
+      "step": 2884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5251931388350516e-06,
+      "loss": 2.9569,
+      "step": 2885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5168873582381644e-06,
+      "loss": 2.817,
+      "step": 2886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.508603547301063e-06,
+      "loss": 3.0103,
+      "step": 2887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.500341713773687e-06,
+      "loss": 2.8666,
+      "step": 2888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4921018653853953e-06,
+      "loss": 2.8361,
+      "step": 2889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4838840098449963e-06,
+      "loss": 3.0003,
+      "step": 2890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4756881548407153e-06,
+      "loss": 3.0305,
+      "step": 2891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4675143080401965e-06,
+      "loss": 2.9352,
+      "step": 2892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4593624770904924e-06,
+      "loss": 2.9673,
+      "step": 2893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4512326696180662e-06,
+      "loss": 2.9013,
+      "step": 2894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4431248932287616e-06,
+      "loss": 2.8263,
+      "step": 2895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4350391555078251e-06,
+      "loss": 2.8856,
+      "step": 2896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4269754640198807e-06,
+      "loss": 2.8266,
+      "step": 2897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4189338263089241e-06,
+      "loss": 2.9295,
+      "step": 2898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4109142498983263e-06,
+      "loss": 2.7419,
+      "step": 2899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4029167422908107e-06,
+      "loss": 2.9524,
+      "step": 2900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.394941310968456e-06,
+      "loss": 2.8492,
+      "step": 2901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3869879633926913e-06,
+      "loss": 3.0581,
+      "step": 2902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3790567070042754e-06,
+      "loss": 2.9282,
+      "step": 2903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3711475492233117e-06,
+      "loss": 2.7634,
+      "step": 2904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3632604974492257e-06,
+      "loss": 2.8397,
+      "step": 2905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3553955590607599e-06,
+      "loss": 2.7876,
+      "step": 2906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.347552741415964e-06,
+      "loss": 2.8102,
+      "step": 2907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3397320518521994e-06,
+      "loss": 2.6528,
+      "step": 2908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3319334976861248e-06,
+      "loss": 2.9018,
+      "step": 2909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.324157086213687e-06,
+      "loss": 2.6609,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3164028247101222e-06,
+      "loss": 2.7542,
+      "step": 2911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3086707204299414e-06,
+      "loss": 2.8282,
+      "step": 2912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3009607806069192e-06,
+      "loss": 2.9218,
+      "step": 2913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2932730124541049e-06,
+      "loss": 2.7919,
+      "step": 2914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.285607423163801e-06,
+      "loss": 2.8642,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2779640199075627e-06,
+      "loss": 2.9521,
+      "step": 2916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2703428098361836e-06,
+      "loss": 2.8633,
+      "step": 2917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.262743800079702e-06,
+      "loss": 2.9045,
+      "step": 2918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2551669977473813e-06,
+      "loss": 2.8079,
+      "step": 2919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2476124099277037e-06,
+      "loss": 2.8846,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.240080043688377e-06,
+      "loss": 2.8991,
+      "step": 2921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2325699060763195e-06,
+      "loss": 2.7793,
+      "step": 2922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.225082004117653e-06,
+      "loss": 2.8436,
+      "step": 2923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.217616344817693e-06,
+      "loss": 2.7127,
+      "step": 2924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2101729351609499e-06,
+      "loss": 2.8851,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2027517821111112e-06,
+      "loss": 2.8838,
+      "step": 2926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1953528926110485e-06,
+      "loss": 2.8392,
+      "step": 2927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1879762735828081e-06,
+      "loss": 2.844,
+      "step": 2928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.180621931927592e-06,
+      "loss": 2.7431,
+      "step": 2929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1732898745257709e-06,
+      "loss": 2.9194,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1659801082368526e-06,
+      "loss": 2.8165,
+      "step": 2931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1586926398995057e-06,
+      "loss": 2.8925,
+      "step": 2932
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1514274763315292e-06,
+      "loss": 2.9467,
+      "step": 2933
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.144184624329861e-06,
+      "loss": 2.8752,
+      "step": 2934
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1369640906705558e-06,
+      "loss": 2.7855,
+      "step": 2935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1297658821088019e-06,
+      "loss": 2.7959,
+      "step": 2936
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1225900053788902e-06,
+      "loss": 2.7816,
+      "step": 2937
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1154364671942198e-06,
+      "loss": 2.842,
+      "step": 2938
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1083052742472927e-06,
+      "loss": 2.8565,
+      "step": 2939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1011964332097114e-06,
+      "loss": 2.7393,
+      "step": 2940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0941099507321585e-06,
+      "loss": 2.8378,
+      "step": 2941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0870458334444056e-06,
+      "loss": 2.7078,
+      "step": 2942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0800040879553025e-06,
+      "loss": 2.918,
+      "step": 2943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0729847208527517e-06,
+      "loss": 2.6711,
+      "step": 2944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0659877387037415e-06,
+      "loss": 2.9084,
+      "step": 2945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0590131480543054e-06,
+      "loss": 2.7004,
+      "step": 2946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0520609554295346e-06,
+      "loss": 2.9429,
+      "step": 2947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0451311673335629e-06,
+      "loss": 2.7329,
+      "step": 2948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0382237902495596e-06,
+      "loss": 2.8692,
+      "step": 2949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0313388306397303e-06,
+      "loss": 2.7771,
+      "step": 2950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0244762949453147e-06,
+      "loss": 3.044,
+      "step": 2951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0176361895865683e-06,
+      "loss": 2.8087,
+      "step": 2952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.010818520962753e-06,
+      "loss": 2.816,
+      "step": 2953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0040232954521557e-06,
+      "loss": 2.6872,
+      "step": 2954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.9725051941206e-07,
+      "loss": 2.9794,
+      "step": 2955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.9050019917874e-07,
+      "loss": 2.7165,
+      "step": 2956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.83772341067471e-07,
+      "loss": 2.8668,
+      "step": 2957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.770669513725128e-07,
+      "loss": 2.8682,
+      "step": 2958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.703840363670975e-07,
+      "loss": 3.0101,
+      "step": 2959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.6372360230344e-07,
+      "loss": 2.9713,
+      "step": 2960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.570856554127205e-07,
+      "loss": 2.9232,
+      "step": 2961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.504702019050737e-07,
+      "loss": 2.792,
+      "step": 2962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.43877247969599e-07,
+      "loss": 2.8986,
+      "step": 2963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.373067997743429e-07,
+      "loss": 2.8635,
+      "step": 2964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.307588634663022e-07,
+      "loss": 2.8837,
+      "step": 2965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.242334451714075e-07,
+      "loss": 2.9795,
+      "step": 2966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.177305509945166e-07,
+      "loss": 2.8286,
+      "step": 2967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.112501870194273e-07,
+      "loss": 2.6828,
+      "step": 2968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.047923593088536e-07,
+      "loss": 2.9126,
+      "step": 2969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.983570739044211e-07,
+      "loss": 2.7636,
+      "step": 2970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.91944336826675e-07,
+      "loss": 2.799,
+      "step": 2971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.855541540750578e-07,
+      "loss": 3.0232,
+      "step": 2972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.791865316279152e-07,
+      "loss": 2.7541,
+      "step": 2973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.728414754424874e-07,
+      "loss": 2.9554,
+      "step": 2974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.665189914548955e-07,
+      "loss": 2.9384,
+      "step": 2975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.602190855801523e-07,
+      "loss": 2.8562,
+      "step": 2976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.539417637121461e-07,
+      "loss": 3.0199,
+      "step": 2977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.476870317236291e-07,
+      "loss": 2.8572,
+      "step": 2978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.414548954662316e-07,
+      "loss": 2.9349,
+      "step": 2979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.352453607704286e-07,
+      "loss": 2.8086,
+      "step": 2980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.290584334455675e-07,
+      "loss": 2.9076,
+      "step": 2981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.228941192798323e-07,
+      "loss": 2.9103,
+      "step": 2982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.167524240402541e-07,
+      "loss": 2.9223,
+      "step": 2983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.106333534727146e-07,
+      "loss": 2.9259,
+      "step": 2984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.045369133019093e-07,
+      "loss": 2.8218,
+      "step": 2985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.98463109231376e-07,
+      "loss": 2.9647,
+      "step": 2986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.924119469434665e-07,
+      "loss": 2.8277,
+      "step": 2987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.86383432099358e-07,
+      "loss": 2.9032,
+      "step": 2988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.803775703390359e-07,
+      "loss": 2.9221,
+      "step": 2989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.74394367281292e-07,
+      "loss": 2.9624,
+      "step": 2990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.684338285237236e-07,
+      "loss": 2.9366,
+      "step": 2991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.624959596427145e-07,
+      "loss": 2.8924,
+      "step": 2992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.565807661934543e-07,
+      "loss": 2.8395,
+      "step": 2993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.506882537099025e-07,
+      "loss": 2.8759,
+      "step": 2994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.448184277048159e-07,
+      "loss": 2.8467,
+      "step": 2995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.389712936697129e-07,
+      "loss": 2.8125,
+      "step": 2996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.331468570748956e-07,
+      "loss": 2.6775,
+      "step": 2997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.273451233694134e-07,
+      "loss": 2.8979,
+      "step": 2998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.21566097981094e-07,
+      "loss": 2.9253,
+      "step": 2999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.15809786316507e-07,
+      "loss": 2.8341,
+      "step": 3000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 2.8934926986694336,
+      "eval_runtime": 1.3141,
+      "eval_samples_per_second": 76.096,
+      "eval_steps_per_second": 5.327,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3248,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 7.4060100796416e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}