diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.3547409651910428,
+  "global_step": 900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0509721492380452e-08,
+      "loss": 1.1226,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1019442984760905e-08,
+      "loss": 1.035,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.152916447714136e-08,
+      "loss": 1.0924,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.203888596952181e-08,
+      "loss": 1.0726,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2548607461902266e-08,
+      "loss": 1.1055,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.305832895428272e-08,
+      "loss": 1.1069,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.356805044666317e-08,
+      "loss": 1.1214,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.407777193904362e-08,
+      "loss": 1.0352,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.458749343142408e-08,
+      "loss": 1.0834,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0509721492380453e-07,
+      "loss": 1.0943,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1560693641618497e-07,
+      "loss": 1.0665,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2611665790856543e-07,
+      "loss": 1.038,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3662637940094587e-07,
+      "loss": 1.0735,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4713610089332634e-07,
+      "loss": 1.1459,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5764582238570677e-07,
+      "loss": 1.0382,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6815554387808724e-07,
+      "loss": 1.0474,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.786652653704677e-07,
+      "loss": 1.0968,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8917498686284816e-07,
+      "loss": 1.0508,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996847083552286e-07,
+      "loss": 1.0399,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1019442984760907e-07,
+      "loss": 1.0031,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2070415133998953e-07,
+      "loss": 0.9566,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3121387283236994e-07,
+      "loss": 1.0053,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.417235943247504e-07,
+      "loss": 1.0701,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5223331581713087e-07,
+      "loss": 1.019,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.627430373095113e-07,
+      "loss": 1.0403,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7325275880189174e-07,
+      "loss": 0.9973,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.837624802942722e-07,
+      "loss": 0.9612,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9427220178665267e-07,
+      "loss": 0.9842,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0478192327903313e-07,
+      "loss": 1.01,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1529164477141355e-07,
+      "loss": 0.9919,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2580136626379406e-07,
+      "loss": 0.9786,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3631108775617447e-07,
+      "loss": 0.9469,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.468208092485549e-07,
+      "loss": 0.9505,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.573305307409354e-07,
+      "loss": 0.9357,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678402522333158e-07,
+      "loss": 1.0086,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7834997372569633e-07,
+      "loss": 1.0187,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8885969521807674e-07,
+      "loss": 0.9877,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.993694167104572e-07,
+      "loss": 0.8856,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0987913820283767e-07,
+      "loss": 0.9854,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2038885969521813e-07,
+      "loss": 1.0107,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3089858118759854e-07,
+      "loss": 0.933,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4140830267997906e-07,
+      "loss": 0.9874,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5191802417235947e-07,
+      "loss": 0.9812,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.624277456647399e-07,
+      "loss": 0.9467,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.729374671571204e-07,
+      "loss": 0.9949,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.834471886495008e-07,
+      "loss": 1.005,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.939569101418812e-07,
+      "loss": 0.935,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.044666316342617e-07,
+      "loss": 0.9481,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.149763531266421e-07,
+      "loss": 0.9681,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.254860746190226e-07,
+      "loss": 0.9508,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.359957961114031e-07,
+      "loss": 0.8342,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.465055176037835e-07,
+      "loss": 0.9349,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.57015239096164e-07,
+      "loss": 0.9312,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.675249605885444e-07,
+      "loss": 0.8591,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.780346820809249e-07,
+      "loss": 0.9763,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.885444035733053e-07,
+      "loss": 0.9114,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.990541250656859e-07,
+      "loss": 0.9064,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.095638465580663e-07,
+      "loss": 0.9438,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.200735680504468e-07,
+      "loss": 0.865,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.305832895428271e-07,
+      "loss": 0.8593,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.410930110352076e-07,
+      "loss": 0.9031,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.516027325275881e-07,
+      "loss": 0.9431,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.621124540199684e-07,
+      "loss": 0.8947,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.726221755123489e-07,
+      "loss": 0.9237,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.831318970047295e-07,
+      "loss": 0.9067,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.936416184971098e-07,
+      "loss": 0.905,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.041513399894903e-07,
+      "loss": 0.8822,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.146610614818708e-07,
+      "loss": 0.9116,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.251707829742513e-07,
+      "loss": 0.8923,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.356805044666316e-07,
+      "loss": 1.0088,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.461902259590121e-07,
+      "loss": 0.9253,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.566999474513927e-07,
+      "loss": 0.875,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.672096689437731e-07,
+      "loss": 0.9378,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.777193904361535e-07,
+      "loss": 0.9202,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.88229111928534e-07,
+      "loss": 0.9185,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.987388334209144e-07,
+      "loss": 0.9362,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.092485549132949e-07,
+      "loss": 0.8857,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.197582764056753e-07,
+      "loss": 0.8658,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.302679978980557e-07,
+      "loss": 0.8919,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.407777193904363e-07,
+      "loss": 0.9018,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.512874408828168e-07,
+      "loss": 0.8743,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.617971623751971e-07,
+      "loss": 0.9393,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.723068838675776e-07,
+      "loss": 0.8945,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.828166053599581e-07,
+      "loss": 0.9197,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.933263268523384e-07,
+      "loss": 0.9709,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.038360483447189e-07,
+      "loss": 0.8581,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.143457698370995e-07,
+      "loss": 0.8924,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.248554913294798e-07,
+      "loss": 0.8823,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.353652128218603e-07,
+      "loss": 0.8701,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.458749343142408e-07,
+      "loss": 0.8422,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.563846558066212e-07,
+      "loss": 0.8954,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.668943772990016e-07,
+      "loss": 0.9193,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.774040987913822e-07,
+      "loss": 0.8889,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879138202837624e-07,
+      "loss": 0.8538,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98423541776143e-07,
+      "loss": 0.8416,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0089332632685235e-06,
+      "loss": 0.9169,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0194429847609039e-06,
+      "loss": 0.8734,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0299527062532843e-06,
+      "loss": 0.9196,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.040462427745665e-06,
+      "loss": 0.9213,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0509721492380451e-06,
+      "loss": 0.8665,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0614818707304257e-06,
+      "loss": 0.867,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0719915922228062e-06,
+      "loss": 0.8649,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0825013137151866e-06,
+      "loss": 0.8447,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.093011035207567e-06,
+      "loss": 0.865,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1035207566999476e-06,
+      "loss": 0.9243,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.114030478192328e-06,
+      "loss": 0.916,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1245401996847084e-06,
+      "loss": 0.8486,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1350499211770888e-06,
+      "loss": 0.9284,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1455596426694692e-06,
+      "loss": 0.8814,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1560693641618499e-06,
+      "loss": 0.9023,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1665790856542303e-06,
+      "loss": 0.9036,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1770888071466107e-06,
+      "loss": 0.839,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.187598528638991e-06,
+      "loss": 0.83,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1981082501313717e-06,
+      "loss": 0.8136,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2086179716237521e-06,
+      "loss": 0.8475,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2191276931161325e-06,
+      "loss": 0.8561,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.229637414608513e-06,
+      "loss": 0.8307,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2401471361008936e-06,
+      "loss": 0.8672,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.250656857593274e-06,
+      "loss": 0.9235,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2611665790856542e-06,
+      "loss": 0.9005,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2716763005780348e-06,
+      "loss": 0.9471,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2821860220704152e-06,
+      "loss": 0.8485,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2926957435627956e-06,
+      "loss": 0.8306,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3032054650551763e-06,
+      "loss": 0.8953,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3137151865475567e-06,
+      "loss": 0.8887,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3242249080399369e-06,
+      "loss": 0.9285,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3347346295323175e-06,
+      "loss": 0.8876,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3452443510246979e-06,
+      "loss": 0.8815,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3557540725170785e-06,
+      "loss": 0.8883,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.366263794009459e-06,
+      "loss": 0.8416,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3767735155018393e-06,
+      "loss": 0.8636,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3872832369942195e-06,
+      "loss": 0.9279,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3977929584866002e-06,
+      "loss": 0.8635,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4083026799789806e-06,
+      "loss": 0.8704,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4188124014713612e-06,
+      "loss": 0.9254,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4293221229637416e-06,
+      "loss": 0.8566,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.439831844456122e-06,
+      "loss": 0.8802,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4503415659485026e-06,
+      "loss": 0.8882,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4608512874408828e-06,
+      "loss": 0.8894,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4713610089332632e-06,
+      "loss": 0.8016,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4818707304256439e-06,
+      "loss": 0.843,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4923804519180243e-06,
+      "loss": 0.8225,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.502890173410405e-06,
+      "loss": 0.9416,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5133998949027853e-06,
+      "loss": 0.8928,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5239096163951655e-06,
+      "loss": 0.8778,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5344193378875461e-06,
+      "loss": 0.8778,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5449290593799265e-06,
+      "loss": 0.8203,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.555438780872307e-06,
+      "loss": 0.8298,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5659485023646876e-06,
+      "loss": 0.8581,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.576458223857068e-06,
+      "loss": 0.8292,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5869679453494482e-06,
+      "loss": 0.8326,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5974776668418288e-06,
+      "loss": 0.8218,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6079873883342092e-06,
+      "loss": 0.845,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6184971098265898e-06,
+      "loss": 0.8576,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6290068313189703e-06,
+      "loss": 0.87,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6395165528113507e-06,
+      "loss": 0.9236,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6500262743037309e-06,
+      "loss": 0.8897,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6605359957961115e-06,
+      "loss": 0.8678,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.671045717288492e-06,
+      "loss": 0.8475,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6815554387808725e-06,
+      "loss": 0.8253,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.692065160273253e-06,
+      "loss": 0.8887,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7025748817656336e-06,
+      "loss": 0.9127,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7130846032580138e-06,
+      "loss": 0.8863,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7235943247503942e-06,
+      "loss": 0.8783,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7341040462427746e-06,
+      "loss": 0.8765,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7446137677351552e-06,
+      "loss": 0.8933,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7551234892275356e-06,
+      "loss": 0.8049,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7656332107199162e-06,
+      "loss": 0.8465,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7761429322122964e-06,
+      "loss": 0.8307,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7866526537046768e-06,
+      "loss": 0.8415,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7971623751970575e-06,
+      "loss": 0.9014,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8076720966894379e-06,
+      "loss": 0.8472,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.8447,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.828691539674199e-06,
+      "loss": 0.8928,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8392012611665791e-06,
+      "loss": 0.9114,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8497109826589595e-06,
+      "loss": 0.8536,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8602207041513401e-06,
+      "loss": 0.8554,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8707304256437206e-06,
+      "loss": 0.7809,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8812401471361012e-06,
+      "loss": 0.8485,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8917498686284816e-06,
+      "loss": 0.8215,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.902259590120862e-06,
+      "loss": 0.8399,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9127693116132424e-06,
+      "loss": 0.8658,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.923279033105623e-06,
+      "loss": 0.8194,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9337887545980032e-06,
+      "loss": 0.9043,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9442984760903836e-06,
+      "loss": 0.806,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9548081975827645e-06,
+      "loss": 0.8179,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.965317919075145e-06,
+      "loss": 0.8461,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975827640567525e-06,
+      "loss": 0.8699,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9863373620599057e-06,
+      "loss": 0.9114,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996847083552286e-06,
+      "loss": 0.846,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0073568050446665e-06,
+      "loss": 0.8631,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.017866526537047e-06,
+      "loss": 0.8162,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0283762480294274e-06,
+      "loss": 0.8563,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0388859695218078e-06,
+      "loss": 0.8067,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.049395691014188e-06,
+      "loss": 0.8742,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0599054125065686e-06,
+      "loss": 0.8952,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0704151339989494e-06,
+      "loss": 0.8416,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.08092485549133e-06,
+      "loss": 0.8678,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0914345769837102e-06,
+      "loss": 0.8226,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1019442984760902e-06,
+      "loss": 0.7962,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.112454019968471e-06,
+      "loss": 0.8959,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1229637414608515e-06,
+      "loss": 0.8077,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.133473462953232e-06,
+      "loss": 0.8309,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1439831844456123e-06,
+      "loss": 0.825,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.154492905937993e-06,
+      "loss": 0.8421,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.165002627430373e-06,
+      "loss": 0.8885,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1755123489227535e-06,
+      "loss": 0.8628,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186022070415134e-06,
+      "loss": 0.8114,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1965317919075148e-06,
+      "loss": 0.8257,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.207041513399895e-06,
+      "loss": 0.8555,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2175512348922756e-06,
+      "loss": 0.8733,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.228060956384656e-06,
+      "loss": 0.8273,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2385706778770364e-06,
+      "loss": 0.8197,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.249080399369417e-06,
+      "loss": 0.7955,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2595901208617972e-06,
+      "loss": 0.806,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2700998423541777e-06,
+      "loss": 0.8325,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2806095638465585e-06,
+      "loss": 0.8283,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2911192853389385e-06,
+      "loss": 0.7938,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.301629006831319e-06,
+      "loss": 0.8119,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3121387283236997e-06,
+      "loss": 0.8418,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.32264844981608e-06,
+      "loss": 0.8606,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3331581713084605e-06,
+      "loss": 0.8541,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.343667892800841e-06,
+      "loss": 0.8402,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3541776142932214e-06,
+      "loss": 0.8204,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3646873357856018e-06,
+      "loss": 0.8525,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.375197057277982e-06,
+      "loss": 0.8229,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3857067787703626e-06,
+      "loss": 0.7948,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3962165002627434e-06,
+      "loss": 0.8312,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.406726221755124e-06,
+      "loss": 0.8345,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4172359432475043e-06,
+      "loss": 0.816,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4277456647398847e-06,
+      "loss": 0.7643,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.438255386232265e-06,
+      "loss": 0.8797,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4487651077246455e-06,
+      "loss": 0.8267,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.459274829217026e-06,
+      "loss": 0.834,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4697845507094063e-06,
+      "loss": 0.8035,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.480294272201787e-06,
+      "loss": 0.8127,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.490803993694167e-06,
+      "loss": 0.8404,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.501313715186548e-06,
+      "loss": 0.8017,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5118234366789284e-06,
+      "loss": 0.783,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5223331581713084e-06,
+      "loss": 0.8523,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5328428796636888e-06,
+      "loss": 0.8818,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5433526011560696e-06,
+      "loss": 0.8113,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.55386232264845e-06,
+      "loss": 0.8489,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5643720441408304e-06,
+      "loss": 0.8477,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.574881765633211e-06,
+      "loss": 0.8228,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5853914871255913e-06,
+      "loss": 0.8161,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.595901208617972e-06,
+      "loss": 0.8977,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6064109301103525e-06,
+      "loss": 0.797,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.616920651602733e-06,
+      "loss": 0.7946,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6274303730951133e-06,
+      "loss": 0.8672,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6379400945874937e-06,
+      "loss": 0.8719,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6484498160798737e-06,
+      "loss": 0.81,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.658959537572254e-06,
+      "loss": 0.7862,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.669469259064635e-06,
+      "loss": 0.8419,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6799789805570154e-06,
+      "loss": 0.8263,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6904887020493958e-06,
+      "loss": 0.9026,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.700998423541776e-06,
+      "loss": 0.8,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.711508145034157e-06,
+      "loss": 0.7938,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7220178665265374e-06,
+      "loss": 0.857,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.732527588018918e-06,
+      "loss": 0.7823,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7430373095112983e-06,
+      "loss": 0.8051,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7535470310036787e-06,
+      "loss": 0.7843,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7640567524960595e-06,
+      "loss": 0.8237,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.774566473988439e-06,
+      "loss": 0.8013,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.78507619548082e-06,
+      "loss": 0.8332,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7955859169732003e-06,
+      "loss": 0.863,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8060956384655807e-06,
+      "loss": 0.789,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.816605359957961e-06,
+      "loss": 0.7895,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8271150814503416e-06,
+      "loss": 0.8474,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8376248029427224e-06,
+      "loss": 0.9381,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.848134524435103e-06,
+      "loss": 0.7906,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.858644245927483e-06,
+      "loss": 0.8295,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8691539674198636e-06,
+      "loss": 0.8607,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.879663688912244e-06,
+      "loss": 0.8294,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.890173410404625e-06,
+      "loss": 0.8384,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9006831318970053e-06,
+      "loss": 0.8402,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9111928533893853e-06,
+      "loss": 0.8587,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9217025748817657e-06,
+      "loss": 0.7734,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.932212296374146e-06,
+      "loss": 0.7796,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9427220178665265e-06,
+      "loss": 0.8167,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9532317393589073e-06,
+      "loss": 0.8551,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9637414608512877e-06,
+      "loss": 0.8125,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.974251182343668e-06,
+      "loss": 0.7928,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9847609038360486e-06,
+      "loss": 0.7864,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995270625328429e-06,
+      "loss": 0.8908,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.00578034682081e-06,
+      "loss": 0.8098,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0162900683131902e-06,
+      "loss": 0.784,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0267997898055706e-06,
+      "loss": 0.8152,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0373095112979506e-06,
+      "loss": 0.8079,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.047819232790331e-06,
+      "loss": 0.8098,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0583289542827114e-06,
+      "loss": 0.8008,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0688386757750923e-06,
+      "loss": 0.8185,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0793483972674727e-06,
+      "loss": 0.8014,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.089858118759853e-06,
+      "loss": 0.7557,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1003678402522335e-06,
+      "loss": 0.7783,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.110877561744614e-06,
+      "loss": 0.8218,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1213872832369948e-06,
+      "loss": 0.8198,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.131897004729375e-06,
+      "loss": 0.8062,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1424067262217556e-06,
+      "loss": 0.8037,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.152916447714136e-06,
+      "loss": 0.8203,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.163426169206516e-06,
+      "loss": 0.8182,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1739358906988964e-06,
+      "loss": 0.8488,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1844456121912772e-06,
+      "loss": 0.81,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1949553336836576e-06,
+      "loss": 0.8073,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.205465055176038e-06,
+      "loss": 0.8169,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2159747766684185e-06,
+      "loss": 0.8016,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.226484498160799e-06,
+      "loss": 0.8066,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2369942196531797e-06,
+      "loss": 0.8052,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.24750394114556e-06,
+      "loss": 0.7699,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2580136626379405e-06,
+      "loss": 0.8497,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.268523384130321e-06,
+      "loss": 0.7892,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2790331056227013e-06,
+      "loss": 0.8514,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.289542827115082e-06,
+      "loss": 0.7749,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3000525486074617e-06,
+      "loss": 0.7904,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3105622700998426e-06,
+      "loss": 0.7986,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.321071991592223e-06,
+      "loss": 0.779,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3315817130846034e-06,
+      "loss": 0.8299,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.342091434576984e-06,
+      "loss": 0.7993,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3526011560693642e-06,
+      "loss": 0.8412,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.363110877561745e-06,
+      "loss": 0.8111,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3736205990541255e-06,
+      "loss": 0.8285,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.384130320546506e-06,
+      "loss": 0.8332,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3946400420388863e-06,
+      "loss": 0.8837,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.405149763531267e-06,
+      "loss": 0.8275,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4156594850236475e-06,
+      "loss": 0.8397,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4261692065160275e-06,
+      "loss": 0.835,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.436678928008408e-06,
+      "loss": 0.8153,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4471886495007883e-06,
+      "loss": 0.8656,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4576983709931688e-06,
+      "loss": 0.8272,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.468208092485549e-06,
+      "loss": 0.8442,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.47871781397793e-06,
+      "loss": 0.8567,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4892275354703104e-06,
+      "loss": 0.7913,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.499737256962691e-06,
+      "loss": 0.8499,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5102469784550712e-06,
+      "loss": 0.7344,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5207566999474516e-06,
+      "loss": 0.8471,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5312664214398325e-06,
+      "loss": 0.8772,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.541776142932213e-06,
+      "loss": 0.9182,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.552285864424593e-06,
+      "loss": 0.7932,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5627955859169733e-06,
+      "loss": 0.8078,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5733053074093537e-06,
+      "loss": 0.7449,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.583815028901734e-06,
+      "loss": 0.8372,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.594324750394115e-06,
+      "loss": 0.8061,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6048344718864954e-06,
+      "loss": 0.7962,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6153441933788758e-06,
+      "loss": 0.7833,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.625853914871256e-06,
+      "loss": 0.8089,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.8276,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6468733578560174e-06,
+      "loss": 0.7673,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.657383079348398e-06,
+      "loss": 0.85,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6678928008407782e-06,
+      "loss": 0.8004,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6784025223331582e-06,
+      "loss": 0.8385,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6889122438255386e-06,
+      "loss": 0.8413,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.699421965317919e-06,
+      "loss": 0.8346,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7099316868103e-06,
+      "loss": 0.7425,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7204414083026803e-06,
+      "loss": 0.8421,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7309511297950607e-06,
+      "loss": 0.8623,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.741460851287441e-06,
+      "loss": 0.7881,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7519705727798215e-06,
+      "loss": 0.8611,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7624802942722024e-06,
+      "loss": 0.8196,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7729900157645828e-06,
+      "loss": 0.8406,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.783499737256963e-06,
+      "loss": 0.7493,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7940094587493436e-06,
+      "loss": 0.7704,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.804519180241724e-06,
+      "loss": 0.7866,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.815028901734104e-06,
+      "loss": 0.8068,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.825538623226485e-06,
+      "loss": 0.7198,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.836048344718865e-06,
+      "loss": 0.8212,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.846558066211246e-06,
+      "loss": 0.777,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.857067787703626e-06,
+      "loss": 0.8647,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8675775091960065e-06,
+      "loss": 0.7956,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.878087230688387e-06,
+      "loss": 0.8092,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.888596952180767e-06,
+      "loss": 0.7958,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.899106673673148e-06,
+      "loss": 0.85,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.909616395165529e-06,
+      "loss": 0.7967,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.920126116657909e-06,
+      "loss": 0.7974,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.93063583815029e-06,
+      "loss": 0.7822,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.941145559642669e-06,
+      "loss": 0.8445,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.95165528113505e-06,
+      "loss": 0.8093,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.96216500262743e-06,
+      "loss": 0.8476,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9726747241198114e-06,
+      "loss": 0.8214,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.983184445612192e-06,
+      "loss": 0.8533,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.993694167104572e-06,
+      "loss": 0.7723,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.004203888596953e-06,
+      "loss": 0.807,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.014713610089333e-06,
+      "loss": 0.8492,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.0252233315817135e-06,
+      "loss": 0.8389,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.035733053074094e-06,
+      "loss": 0.7838,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.046242774566474e-06,
+      "loss": 0.8013,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.056752496058855e-06,
+      "loss": 0.8072,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.067262217551235e-06,
+      "loss": 0.8006,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.0777719390436155e-06,
+      "loss": 0.8104,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.088281660535996e-06,
+      "loss": 0.8065,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.098791382028376e-06,
+      "loss": 0.7914,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.109301103520757e-06,
+      "loss": 0.8036,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.119810825013137e-06,
+      "loss": 0.8041,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.130320546505518e-06,
+      "loss": 0.7746,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.140830267997899e-06,
+      "loss": 0.765,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.151339989490279e-06,
+      "loss": 0.7855,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.16184971098266e-06,
+      "loss": 0.7624,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.17235943247504e-06,
+      "loss": 0.7905,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1828691539674205e-06,
+      "loss": 0.7837,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.193378875459801e-06,
+      "loss": 0.8263,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2038885969521805e-06,
+      "loss": 0.7906,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.214398318444562e-06,
+      "loss": 0.829,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.224908039936942e-06,
+      "loss": 0.8397,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2354177614293225e-06,
+      "loss": 0.8143,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.245927482921703e-06,
+      "loss": 0.8318,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.256437204414083e-06,
+      "loss": 0.7841,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.266946925906464e-06,
+      "loss": 0.8352,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.277456647398844e-06,
+      "loss": 0.8099,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.287966368891225e-06,
+      "loss": 0.7887,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.298476090383605e-06,
+      "loss": 0.8013,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.308985811875986e-06,
+      "loss": 0.7538,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.319495533368367e-06,
+      "loss": 0.8405,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.330005254860746e-06,
+      "loss": 0.7894,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.340514976353127e-06,
+      "loss": 0.7704,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.351024697845507e-06,
+      "loss": 0.781,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3615344193378875e-06,
+      "loss": 0.7339,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.372044140830268e-06,
+      "loss": 0.8059,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.382553862322649e-06,
+      "loss": 0.7805,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3930635838150296e-06,
+      "loss": 0.8449,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.40357330530741e-06,
+      "loss": 0.8233,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.41408302679979e-06,
+      "loss": 0.8101,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.424592748292171e-06,
+      "loss": 0.7501,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.435102469784551e-06,
+      "loss": 0.8055,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.445612191276932e-06,
+      "loss": 0.8196,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.456121912769312e-06,
+      "loss": 0.8087,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4666316342616924e-06,
+      "loss": 0.8175,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.477141355754073e-06,
+      "loss": 0.7983,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.487651077246453e-06,
+      "loss": 0.7686,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.498160798738834e-06,
+      "loss": 0.8567,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.508670520231214e-06,
+      "loss": 0.8355,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5191802417235945e-06,
+      "loss": 0.7395,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.529689963215975e-06,
+      "loss": 0.7831,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.540199684708355e-06,
+      "loss": 0.7781,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5507094062007366e-06,
+      "loss": 0.7756,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561219127693117e-06,
+      "loss": 0.7922,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.571728849185497e-06,
+      "loss": 0.743,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.582238570677877e-06,
+      "loss": 0.7949,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.592748292170257e-06,
+      "loss": 0.766,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.603258013662638e-06,
+      "loss": 0.7464,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.613767735155019e-06,
+      "loss": 0.824,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6242774566473994e-06,
+      "loss": 0.7683,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.63478717813978e-06,
+      "loss": 0.8004,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.64529689963216e-06,
+      "loss": 0.8034,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655806621124541e-06,
+      "loss": 0.7548,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.666316342616921e-06,
+      "loss": 0.8016,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6768260641093015e-06,
+      "loss": 0.7472,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.687335785601682e-06,
+      "loss": 0.7707,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697845507094062e-06,
+      "loss": 0.8231,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708355228586443e-06,
+      "loss": 0.851,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718864950078823e-06,
+      "loss": 0.7595,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7293746715712036e-06,
+      "loss": 0.795,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.739884393063584e-06,
+      "loss": 0.8017,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.750394114555964e-06,
+      "loss": 0.7624,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.760903836048345e-06,
+      "loss": 0.7707,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.771413557540725e-06,
+      "loss": 0.8561,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7819232790331065e-06,
+      "loss": 0.7829,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.792433000525487e-06,
+      "loss": 0.8147,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.802942722017867e-06,
+      "loss": 0.7943,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.813452443510248e-06,
+      "loss": 0.8013,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.823962165002628e-06,
+      "loss": 0.7784,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8344718864950085e-06,
+      "loss": 0.7541,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.844981607987388e-06,
+      "loss": 0.7895,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.855491329479769e-06,
+      "loss": 0.8256,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.86600105097215e-06,
+      "loss": 0.7822,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.87651077246453e-06,
+      "loss": 0.8249,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8870204939569106e-06,
+      "loss": 0.8216,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897530215449291e-06,
+      "loss": 0.7614,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908039936941671e-06,
+      "loss": 0.7757,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.918549658434052e-06,
+      "loss": 0.7565,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.929059379926432e-06,
+      "loss": 0.8064,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.939569101418813e-06,
+      "loss": 0.7756,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.950078822911194e-06,
+      "loss": 0.798,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.960588544403574e-06,
+      "loss": 0.7709,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971098265895954e-06,
+      "loss": 0.7985,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.981607987388334e-06,
+      "loss": 0.7588,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.992117708880715e-06,
+      "loss": 0.7575,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.002627430373096e-06,
+      "loss": 0.7779,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.0131371518654755e-06,
+      "loss": 0.7804,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.023646873357857e-06,
+      "loss": 0.7687,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.034156594850237e-06,
+      "loss": 0.8063,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.044666316342617e-06,
+      "loss": 0.8207,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.055176037834998e-06,
+      "loss": 0.7946,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.0656857593273776e-06,
+      "loss": 0.8119,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.076195480819759e-06,
+      "loss": 0.8291,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.086705202312139e-06,
+      "loss": 0.7851,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.09721492380452e-06,
+      "loss": 0.8316,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1077246452969e-06,
+      "loss": 0.7544,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1182343667892805e-06,
+      "loss": 0.7991,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.128744088281661e-06,
+      "loss": 0.7846,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.139253809774042e-06,
+      "loss": 0.8036,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.149763531266422e-06,
+      "loss": 0.8046,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.160273252758803e-06,
+      "loss": 0.8217,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1707829742511825e-06,
+      "loss": 0.792,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.181292695743563e-06,
+      "loss": 0.7791,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.191802417235944e-06,
+      "loss": 0.8335,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.202312138728324e-06,
+      "loss": 0.8251,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.212821860220705e-06,
+      "loss": 0.7816,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2233315817130846e-06,
+      "loss": 0.7564,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.233841303205466e-06,
+      "loss": 0.7769,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.244351024697845e-06,
+      "loss": 0.7552,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.254860746190227e-06,
+      "loss": 0.7614,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.265370467682607e-06,
+      "loss": 0.8261,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2758801891749875e-06,
+      "loss": 0.8149,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.286389910667368e-06,
+      "loss": 0.7631,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2968996321597474e-06,
+      "loss": 0.7762,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.307409353652129e-06,
+      "loss": 0.7666,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.317919075144508e-06,
+      "loss": 0.8196,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.3284287966368895e-06,
+      "loss": 0.7519,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.33893851812927e-06,
+      "loss": 0.7955,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.34944823962165e-06,
+      "loss": 0.7994,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.359957961114031e-06,
+      "loss": 0.8109,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.370467682606412e-06,
+      "loss": 0.8304,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.3809774040987916e-06,
+      "loss": 0.7815,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.391487125591173e-06,
+      "loss": 0.7933,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.401996847083552e-06,
+      "loss": 0.7836,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.412506568575934e-06,
+      "loss": 0.7597,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.423016290068314e-06,
+      "loss": 0.8288,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.433526011560694e-06,
+      "loss": 0.7912,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.444035733053075e-06,
+      "loss": 0.7629,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.7808,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.465055176037836e-06,
+      "loss": 0.8425,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.475564897530215e-06,
+      "loss": 0.7491,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4860746190225965e-06,
+      "loss": 0.7645,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.496584340514977e-06,
+      "loss": 0.7341,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.507094062007357e-06,
+      "loss": 0.8188,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.517603783499738e-06,
+      "loss": 0.7907,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.528113504992119e-06,
+      "loss": 0.7354,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.538623226484499e-06,
+      "loss": 0.7496,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.549132947976878e-06,
+      "loss": 0.7818,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.559642669469259e-06,
+      "loss": 0.7998,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.57015239096164e-06,
+      "loss": 0.7906,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.58066211245402e-06,
+      "loss": 0.747,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.591171833946401e-06,
+      "loss": 0.7924,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.601681555438782e-06,
+      "loss": 0.7703,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6121912769311615e-06,
+      "loss": 0.7799,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.622700998423543e-06,
+      "loss": 0.7895,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.633210719915922e-06,
+      "loss": 0.7431,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6437204414083035e-06,
+      "loss": 0.8335,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.654230162900683e-06,
+      "loss": 0.8167,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.664739884393064e-06,
+      "loss": 0.8001,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.675249605885445e-06,
+      "loss": 0.8377,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.685759327377824e-06,
+      "loss": 0.7522,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.696269048870206e-06,
+      "loss": 0.7745,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.706778770362585e-06,
+      "loss": 0.7422,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.717288491854966e-06,
+      "loss": 0.75,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.727798213347347e-06,
+      "loss": 0.8685,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.738307934839727e-06,
+      "loss": 0.7686,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.748817656332108e-06,
+      "loss": 0.7747,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.759327377824488e-06,
+      "loss": 0.7944,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7698370993168685e-06,
+      "loss": 0.7937,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.78034682080925e-06,
+      "loss": 0.7406,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.790856542301629e-06,
+      "loss": 0.7916,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8013662637940106e-06,
+      "loss": 0.8029,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.81187598528639e-06,
+      "loss": 0.8132,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8223857067787705e-06,
+      "loss": 0.8217,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.832895428271152e-06,
+      "loss": 0.7689,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.843405149763531e-06,
+      "loss": 0.84,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.853914871255913e-06,
+      "loss": 0.7647,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.864424592748292e-06,
+      "loss": 0.8234,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8749343142406734e-06,
+      "loss": 0.7762,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.885444035733053e-06,
+      "loss": 0.7357,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.895953757225434e-06,
+      "loss": 0.7967,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.906463478717815e-06,
+      "loss": 0.8429,
+      "step": 562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.916973200210195e-06,
+      "loss": 0.7876,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9274829217025755e-06,
+      "loss": 0.8194,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.937992643194955e-06,
+      "loss": 0.7162,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.948502364687336e-06,
+      "loss": 0.8076,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.959012086179716e-06,
+      "loss": 0.7697,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.969521807672097e-06,
+      "loss": 0.8177,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9800315291644775e-06,
+      "loss": 0.798,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.990541250656858e-06,
+      "loss": 0.7574,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.001050972149238e-06,
+      "loss": 0.8,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.01156069364162e-06,
+      "loss": 0.7675,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.022070415133999e-06,
+      "loss": 0.8142,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0325801366263804e-06,
+      "loss": 0.7923,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.04308985811876e-06,
+      "loss": 0.7586,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.053599579611141e-06,
+      "loss": 0.8027,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.064109301103521e-06,
+      "loss": 0.7852,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.074619022595901e-06,
+      "loss": 0.7369,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0851287440882825e-06,
+      "loss": 0.7629,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.095638465580662e-06,
+      "loss": 0.752,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.106148187073043e-06,
+      "loss": 0.729,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.116657908565423e-06,
+      "loss": 0.7968,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.127167630057804e-06,
+      "loss": 0.7879,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1376773515501846e-06,
+      "loss": 0.7143,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.148187073042565e-06,
+      "loss": 0.8175,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.158696794534945e-06,
+      "loss": 0.7768,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.169206516027327e-06,
+      "loss": 0.8143,
+      "step": 587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.179716237519706e-06,
+      "loss": 0.7779,
+      "step": 588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1902259590120875e-06,
+      "loss": 0.813,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.200735680504467e-06,
+      "loss": 0.806,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.2112454019968474e-06,
+      "loss": 0.8452,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.221755123489228e-06,
+      "loss": 0.776,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.232264844981608e-06,
+      "loss": 0.7981,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.2427745664739895e-06,
+      "loss": 0.7873,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.253284287966369e-06,
+      "loss": 0.7843,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.26379400945875e-06,
+      "loss": 0.7289,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.27430373095113e-06,
+      "loss": 0.7581,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.284813452443511e-06,
+      "loss": 0.7745,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.295323173935891e-06,
+      "loss": 0.768,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.305832895428272e-06,
+      "loss": 0.7967,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.316342616920652e-06,
+      "loss": 0.7992,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.326852338413032e-06,
+      "loss": 0.7745,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.337362059905413e-06,
+      "loss": 0.7966,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.347871781397793e-06,
+      "loss": 0.7472,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.358381502890174e-06,
+      "loss": 0.7194,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3688912243825544e-06,
+      "loss": 0.7344,
+      "step": 606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.379400945874935e-06,
+      "loss": 0.7493,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.389910667367315e-06,
+      "loss": 0.8069,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.400420388859696e-06,
+      "loss": 0.8284,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.410930110352076e-06,
+      "loss": 0.7601,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.421439831844457e-06,
+      "loss": 0.766,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.431949553336837e-06,
+      "loss": 0.7906,
+      "step": 612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.442459274829218e-06,
+      "loss": 0.728,
+      "step": 613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.452968996321598e-06,
+      "loss": 0.7995,
+      "step": 614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.463478717813978e-06,
+      "loss": 0.7608,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.473988439306359e-06,
+      "loss": 0.7801,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.484498160798739e-06,
+      "loss": 0.7507,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.49500788229112e-06,
+      "loss": 0.8051,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.5055176037835e-06,
+      "loss": 0.791,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.516027325275881e-06,
+      "loss": 0.7625,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.526537046768261e-06,
+      "loss": 0.7028,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.537046768260642e-06,
+      "loss": 0.7875,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.547556489753022e-06,
+      "loss": 0.775,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.558066211245403e-06,
+      "loss": 0.7493,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.568575932737783e-06,
+      "loss": 0.8062,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.579085654230164e-06,
+      "loss": 0.7098,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.589595375722544e-06,
+      "loss": 0.7582,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6001050972149235e-06,
+      "loss": 0.7779,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.610614818707305e-06,
+      "loss": 0.7466,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.621124540199685e-06,
+      "loss": 0.7856,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6316342616920656e-06,
+      "loss": 0.7639,
+      "step": 631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.642143983184446e-06,
+      "loss": 0.7871,
+      "step": 632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.652653704676827e-06,
+      "loss": 0.7422,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.663163426169207e-06,
+      "loss": 0.7893,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.673673147661588e-06,
+      "loss": 0.7736,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.684182869153968e-06,
+      "loss": 0.8063,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.694692590646349e-06,
+      "loss": 0.769,
+      "step": 637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7052023121387284e-06,
+      "loss": 0.7561,
+      "step": 638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.715712033631109e-06,
+      "loss": 0.7822,
+      "step": 639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.72622175512349e-06,
+      "loss": 0.7915,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.73673147661587e-06,
+      "loss": 0.7615,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.747241198108251e-06,
+      "loss": 0.737,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7577509196006305e-06,
+      "loss": 0.7573,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.768260641093012e-06,
+      "loss": 0.7806,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.778770362585392e-06,
+      "loss": 0.7488,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7892800840777726e-06,
+      "loss": 0.7691,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.799789805570153e-06,
+      "loss": 0.7952,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.810299527062534e-06,
+      "loss": 0.8166,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.820809248554914e-06,
+      "loss": 0.7445,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.831318970047295e-06,
+      "loss": 0.7769,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.841828691539675e-06,
+      "loss": 0.794,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.852338413032055e-06,
+      "loss": 0.7516,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.8628481345244354e-06,
+      "loss": 0.735,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.873357856016816e-06,
+      "loss": 0.7748,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.883867577509197e-06,
+      "loss": 0.7853,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.894377299001577e-06,
+      "loss": 0.7373,
+      "step": 656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.904887020493958e-06,
+      "loss": 0.7545,
+      "step": 657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9153967419863375e-06,
+      "loss": 0.7999,
+      "step": 658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.925906463478719e-06,
+      "loss": 0.8137,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.936416184971098e-06,
+      "loss": 0.7768,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.94692590646348e-06,
+      "loss": 0.7407,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.95743562795586e-06,
+      "loss": 0.7633,
+      "step": 662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9679453494482396e-06,
+      "loss": 0.7699,
+      "step": 663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.978455070940621e-06,
+      "loss": 0.7418,
+      "step": 664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.988964792433e-06,
+      "loss": 0.7519,
+      "step": 665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.999474513925382e-06,
+      "loss": 0.7914,
+      "step": 666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.009984235417762e-06,
+      "loss": 0.8432,
+      "step": 667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0204939569101425e-06,
+      "loss": 0.7886,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.031003678402523e-06,
+      "loss": 0.7764,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.041513399894903e-06,
+      "loss": 0.7937,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.052023121387284e-06,
+      "loss": 0.7701,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.062532842879665e-06,
+      "loss": 0.806,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0730425643720445e-06,
+      "loss": 0.6984,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.083552285864426e-06,
+      "loss": 0.7063,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.094062007356805e-06,
+      "loss": 0.7426,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.104571728849186e-06,
+      "loss": 0.7533,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.115081450341567e-06,
+      "loss": 0.7619,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.1255911718339466e-06,
+      "loss": 0.7628,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.136100893326328e-06,
+      "loss": 0.7579,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.146610614818707e-06,
+      "loss": 0.7938,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.157120336311089e-06,
+      "loss": 0.7901,
+      "step": 681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.167630057803468e-06,
+      "loss": 0.8017,
+      "step": 682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.1781397792958495e-06,
+      "loss": 0.7116,
+      "step": 683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.18864950078823e-06,
+      "loss": 0.7804,
+      "step": 684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.19915922228061e-06,
+      "loss": 0.7738,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.209668943772991e-06,
+      "loss": 0.8055,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.220178665265372e-06,
+      "loss": 0.7584,
+      "step": 687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2306883867577515e-06,
+      "loss": 0.8173,
+      "step": 688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.241198108250131e-06,
+      "loss": 0.8321,
+      "step": 689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.251707829742512e-06,
+      "loss": 0.8037,
+      "step": 690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.262217551234893e-06,
+      "loss": 0.7945,
+      "step": 691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.7635,
+      "step": 692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.283236994219654e-06,
+      "loss": 0.797,
+      "step": 693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.293746715712035e-06,
+      "loss": 0.7834,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.304256437204414e-06,
+      "loss": 0.7692,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.314766158696796e-06,
+      "loss": 0.782,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.325275880189175e-06,
+      "loss": 0.8031,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3357856016815565e-06,
+      "loss": 0.7787,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.346295323173936e-06,
+      "loss": 0.7711,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3568050446663165e-06,
+      "loss": 0.7156,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.367314766158698e-06,
+      "loss": 0.8187,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.377824487651077e-06,
+      "loss": 0.7703,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3883342091434585e-06,
+      "loss": 0.7594,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.398843930635838e-06,
+      "loss": 0.771,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.409353652128219e-06,
+      "loss": 0.7844,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4198633736206e-06,
+      "loss": 0.7794,
+      "step": 706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.43037309511298e-06,
+      "loss": 0.7641,
+      "step": 707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.440882816605361e-06,
+      "loss": 0.7565,
+      "step": 708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.451392538097742e-06,
+      "loss": 0.7471,
+      "step": 709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.461902259590121e-06,
+      "loss": 0.7408,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.472411981082503e-06,
+      "loss": 0.7169,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.482921702574882e-06,
+      "loss": 0.7622,
+      "step": 712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.493431424067263e-06,
+      "loss": 0.7655,
+      "step": 713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.503941145559643e-06,
+      "loss": 0.7409,
+      "step": 714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5144508670520235e-06,
+      "loss": 0.8388,
+      "step": 715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.524960588544405e-06,
+      "loss": 0.7318,
+      "step": 716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.535470310036784e-06,
+      "loss": 0.7832,
+      "step": 717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5459800315291655e-06,
+      "loss": 0.8179,
+      "step": 718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.556489753021545e-06,
+      "loss": 0.7869,
+      "step": 719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.566999474513926e-06,
+      "loss": 0.7611,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.577509196006306e-06,
+      "loss": 0.747,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.588018917498687e-06,
+      "loss": 0.7801,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.598528638991068e-06,
+      "loss": 0.8057,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.609038360483448e-06,
+      "loss": 0.7504,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.619548081975828e-06,
+      "loss": 0.8344,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.630057803468209e-06,
+      "loss": 0.7311,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.64056752496059e-06,
+      "loss": 0.7417,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.65107724645297e-06,
+      "loss": 0.8558,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.66158696794535e-06,
+      "loss": 0.7519,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.67209668943773e-06,
+      "loss": 0.7799,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.682606410930111e-06,
+      "loss": 0.7661,
+      "step": 731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.693116132422491e-06,
+      "loss": 0.7743,
+      "step": 732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.703625853914872e-06,
+      "loss": 0.7495,
+      "step": 733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.714135575407252e-06,
+      "loss": 0.8155,
+      "step": 734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.724645296899633e-06,
+      "loss": 0.7618,
+      "step": 735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.735155018392013e-06,
+      "loss": 0.7464,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.745664739884393e-06,
+      "loss": 0.7517,
+      "step": 737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.756174461376774e-06,
+      "loss": 0.7446,
+      "step": 738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.766684182869154e-06,
+      "loss": 0.7448,
+      "step": 739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.777193904361535e-06,
+      "loss": 0.7553,
+      "step": 740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.787703625853915e-06,
+      "loss": 0.7546,
+      "step": 741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.798213347346295e-06,
+      "loss": 0.774,
+      "step": 742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.808723068838676e-06,
+      "loss": 0.7909,
+      "step": 743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.819232790331058e-06,
+      "loss": 0.7567,
+      "step": 744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.829742511823437e-06,
+      "loss": 0.8226,
+      "step": 745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.840252233315819e-06,
+      "loss": 0.7328,
+      "step": 746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.850761954808197e-06,
+      "loss": 0.7556,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.86127167630058e-06,
+      "loss": 0.7505,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.87178139779296e-06,
+      "loss": 0.7099,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.882291119285339e-06,
+      "loss": 0.7661,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.89280084077772e-06,
+      "loss": 0.7586,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9033105622701e-06,
+      "loss": 0.7893,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.913820283762482e-06,
+      "loss": 0.742,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.92433000525486e-06,
+      "loss": 0.7481,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.934839726747242e-06,
+      "loss": 0.7572,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.945349448239623e-06,
+      "loss": 0.6884,
+      "step": 756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.955859169732003e-06,
+      "loss": 0.7334,
+      "step": 757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.966368891224384e-06,
+      "loss": 0.7577,
+      "step": 758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.976878612716764e-06,
+      "loss": 0.7246,
+      "step": 759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.987388334209145e-06,
+      "loss": 0.7789,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.997898055701525e-06,
+      "loss": 0.7203,
+      "step": 761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.008407777193905e-06,
+      "loss": 0.7356,
+      "step": 762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.018917498686286e-06,
+      "loss": 0.798,
+      "step": 763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.029427220178666e-06,
+      "loss": 0.7849,
+      "step": 764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.039936941671047e-06,
+      "loss": 0.845,
+      "step": 765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.050446663163427e-06,
+      "loss": 0.8198,
+      "step": 766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.060956384655807e-06,
+      "loss": 0.6849,
+      "step": 767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.071466106148188e-06,
+      "loss": 0.809,
+      "step": 768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.081975827640568e-06,
+      "loss": 0.7842,
+      "step": 769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.092485549132949e-06,
+      "loss": 0.7655,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.102995270625329e-06,
+      "loss": 0.7263,
+      "step": 771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.11350499211771e-06,
+      "loss": 0.7766,
+      "step": 772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.12401471361009e-06,
+      "loss": 0.7554,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13452443510247e-06,
+      "loss": 0.7954,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14503415659485e-06,
+      "loss": 0.7477,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.155543878087231e-06,
+      "loss": 0.823,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.166053599579611e-06,
+      "loss": 0.8016,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.176563321071992e-06,
+      "loss": 0.7548,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.187073042564372e-06,
+      "loss": 0.7682,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.197582764056753e-06,
+      "loss": 0.7819,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.208092485549133e-06,
+      "loss": 0.6974,
+      "step": 781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.218602207041514e-06,
+      "loss": 0.808,
+      "step": 782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.229111928533896e-06,
+      "loss": 0.7267,
+      "step": 783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.239621650026274e-06,
+      "loss": 0.7472,
+      "step": 784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.250131371518656e-06,
+      "loss": 0.7971,
+      "step": 785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.260641093011035e-06,
+      "loss": 0.7911,
+      "step": 786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.271150814503416e-06,
+      "loss": 0.7287,
+      "step": 787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.281660535995798e-06,
+      "loss": 0.7724,
+      "step": 788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292170257488176e-06,
+      "loss": 0.6968,
+      "step": 789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302679978980559e-06,
+      "loss": 0.7586,
+      "step": 790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.313189700472937e-06,
+      "loss": 0.7669,
+      "step": 791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32369942196532e-06,
+      "loss": 0.7405,
+      "step": 792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.334209143457698e-06,
+      "loss": 0.7313,
+      "step": 793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34471886495008e-06,
+      "loss": 0.7788,
+      "step": 794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35522858644246e-06,
+      "loss": 0.7164,
+      "step": 795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.365738307934841e-06,
+      "loss": 0.7731,
+      "step": 796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.376248029427221e-06,
+      "loss": 0.7352,
+      "step": 797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.386757750919602e-06,
+      "loss": 0.7368,
+      "step": 798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.397267472411982e-06,
+      "loss": 0.7649,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.407777193904361e-06,
+      "loss": 0.7943,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.418286915396743e-06,
+      "loss": 0.7244,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.428796636889123e-06,
+      "loss": 0.7227,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.439306358381504e-06,
+      "loss": 0.7827,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.449816079873884e-06,
+      "loss": 0.7631,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.460325801366265e-06,
+      "loss": 0.8117,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.470835522858645e-06,
+      "loss": 0.7492,
+      "step": 806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.481345244351026e-06,
+      "loss": 0.7419,
+      "step": 807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.491854965843406e-06,
+      "loss": 0.763,
+      "step": 808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.502364687335786e-06,
+      "loss": 0.7746,
+      "step": 809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.512874408828167e-06,
+      "loss": 0.7402,
+      "step": 810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.523384130320547e-06,
+      "loss": 0.7607,
+      "step": 811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.533893851812928e-06,
+      "loss": 0.6862,
+      "step": 812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.544403573305308e-06,
+      "loss": 0.7702,
+      "step": 813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.554913294797688e-06,
+      "loss": 0.7228,
+      "step": 814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.565423016290069e-06,
+      "loss": 0.7032,
+      "step": 815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.57593273778245e-06,
+      "loss": 0.7233,
+      "step": 816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.58644245927483e-06,
+      "loss": 0.7629,
+      "step": 817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.59695218076721e-06,
+      "loss": 0.6951,
+      "step": 818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.60746190225959e-06,
+      "loss": 0.8245,
+      "step": 819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.617971623751973e-06,
+      "loss": 0.7504,
+      "step": 820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.628481345244351e-06,
+      "loss": 0.7719,
+      "step": 821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.638991066736733e-06,
+      "loss": 0.7961,
+      "step": 822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.649500788229112e-06,
+      "loss": 0.7055,
+      "step": 823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.660010509721492e-06,
+      "loss": 0.7163,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.670520231213873e-06,
+      "loss": 0.7388,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.681029952706253e-06,
+      "loss": 0.758,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.691539674198635e-06,
+      "loss": 0.7891,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.702049395691014e-06,
+      "loss": 0.6774,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.712559117183396e-06,
+      "loss": 0.7652,
+      "step": 829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.723068838675775e-06,
+      "loss": 0.7522,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.733578560168157e-06,
+      "loss": 0.774,
+      "step": 831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.744088281660536e-06,
+      "loss": 0.7496,
+      "step": 832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.754598003152918e-06,
+      "loss": 0.7677,
+      "step": 833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.765107724645298e-06,
+      "loss": 0.7399,
+      "step": 834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.775617446137677e-06,
+      "loss": 0.7568,
+      "step": 835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.786127167630059e-06,
+      "loss": 0.7512,
+      "step": 836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.796636889122438e-06,
+      "loss": 0.756,
+      "step": 837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.80714661061482e-06,
+      "loss": 0.7868,
+      "step": 838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.8176563321072e-06,
+      "loss": 0.7159,
+      "step": 839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.82816605359958e-06,
+      "loss": 0.7721,
+      "step": 840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.838675775091961e-06,
+      "loss": 0.7449,
+      "step": 841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.849185496584342e-06,
+      "loss": 0.737,
+      "step": 842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.859695218076722e-06,
+      "loss": 0.7306,
+      "step": 843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.870204939569102e-06,
+      "loss": 0.7732,
+      "step": 844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.880714661061483e-06,
+      "loss": 0.7752,
+      "step": 845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.891224382553863e-06,
+      "loss": 0.7554,
+      "step": 846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.901734104046244e-06,
+      "loss": 0.8126,
+      "step": 847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.912243825538624e-06,
+      "loss": 0.7632,
+      "step": 848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.922753547031004e-06,
+      "loss": 0.7316,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.933263268523385e-06,
+      "loss": 0.8087,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.943772990015765e-06,
+      "loss": 0.7079,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.954282711508146e-06,
+      "loss": 0.7437,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.964792433000526e-06,
+      "loss": 0.7766,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.975302154492907e-06,
+      "loss": 0.767,
+      "step": 854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.985811875985287e-06,
+      "loss": 0.7668,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.996321597477667e-06,
+      "loss": 0.7701,
+      "step": 856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.006831318970048e-06,
+      "loss": 0.7061,
+      "step": 857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.017341040462428e-06,
+      "loss": 0.6904,
+      "step": 858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.02785076195481e-06,
+      "loss": 0.7267,
+      "step": 859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.038360483447189e-06,
+      "loss": 0.7632,
+      "step": 860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.04887020493957e-06,
+      "loss": 0.7453,
+      "step": 861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.05937992643195e-06,
+      "loss": 0.7961,
+      "step": 862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.06988964792433e-06,
+      "loss": 0.7276,
+      "step": 863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.08039936941671e-06,
+      "loss": 0.7405,
+      "step": 864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.7306,
+      "step": 865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.101418812401473e-06,
+      "loss": 0.7752,
+      "step": 866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.111928533893852e-06,
+      "loss": 0.742,
+      "step": 867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.122438255386234e-06,
+      "loss": 0.7216,
+      "step": 868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.132947976878613e-06,
+      "loss": 0.7545,
+      "step": 869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.143457698370995e-06,
+      "loss": 0.7734,
+      "step": 870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.153967419863374e-06,
+      "loss": 0.74,
+      "step": 871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.164477141355754e-06,
+      "loss": 0.7086,
+      "step": 872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.174986862848136e-06,
+      "loss": 0.7173,
+      "step": 873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.185496584340515e-06,
+      "loss": 0.7585,
+      "step": 874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.196006305832897e-06,
+      "loss": 0.7284,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.206516027325276e-06,
+      "loss": 0.7226,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.217025748817658e-06,
+      "loss": 0.751,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.227535470310038e-06,
+      "loss": 0.7323,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.238045191802418e-06,
+      "loss": 0.7772,
+      "step": 879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.248554913294799e-06,
+      "loss": 0.738,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.25906463478718e-06,
+      "loss": 0.7128,
+      "step": 881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.26957435627956e-06,
+      "loss": 0.7456,
+      "step": 882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.28008407777194e-06,
+      "loss": 0.7583,
+      "step": 883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.29059379926432e-06,
+      "loss": 0.7162,
+      "step": 884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.301103520756701e-06,
+      "loss": 0.716,
+      "step": 885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.311613242249081e-06,
+      "loss": 0.734,
+      "step": 886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.322122963741462e-06,
+      "loss": 0.7537,
+      "step": 887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.332632685233842e-06,
+      "loss": 0.7442,
+      "step": 888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.343142406726223e-06,
+      "loss": 0.7305,
+      "step": 889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.353652128218603e-06,
+      "loss": 0.7629,
+      "step": 890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.364161849710983e-06,
+      "loss": 0.6891,
+      "step": 891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.374671571203364e-06,
+      "loss": 0.6873,
+      "step": 892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.385181292695744e-06,
+      "loss": 0.7411,
+      "step": 893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.395691014188125e-06,
+      "loss": 0.7575,
+      "step": 894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.406200735680505e-06,
+      "loss": 0.7546,
+      "step": 895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.416710457172885e-06,
+      "loss": 0.7051,
+      "step": 896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.427220178665266e-06,
+      "loss": 0.736,
+      "step": 897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.437729900157646e-06,
+      "loss": 0.7112,
+      "step": 898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.448239621650027e-06,
+      "loss": 0.7521,
+      "step": 899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.458749343142407e-06,
+      "loss": 0.7765,
+      "step": 900
+    }
+  ],
+  "max_steps": 63425,
+  "num_train_epochs": 25,
+  "total_flos": 2.3846875513749504e+18,
+  "trial_name": null,
+  "trial_params": null
+}