diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,41454 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9994569840896337,
+  "eval_steps": 500,
+  "global_step": 6904,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 6.161,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 6.1682,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 5.9098,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 6.2003,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 6.1261,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 5.9273,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4038461538461542e-05,
+      "loss": 5.7911,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4038461538461542e-05,
+      "loss": 5.6916,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.884615384615385e-05,
+      "loss": 5.4199,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.884615384615385e-05,
+      "loss": 5.2784,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.365384615384615e-05,
+      "loss": 5.4758,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 5.4711,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3269230769230766e-05,
+      "loss": 5.4178,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3269230769230766e-05,
+      "loss": 5.1831,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 5.1222,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.288461538461539e-05,
+      "loss": 4.9677,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-05,
+      "loss": 5.014,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-05,
+      "loss": 4.8392,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.73076923076923e-05,
+      "loss": 4.8018,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.211538461538461e-05,
+      "loss": 4.8535,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 4.5942,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.173076923076923e-05,
+      "loss": 4.5563,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846153e-05,
+      "loss": 4.6241,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.134615384615384e-05,
+      "loss": 4.6345,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 4.5706,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010096153846153847,
+      "loss": 4.5761,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010576923076923077,
+      "loss": 4.5912,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011057692307692308,
+      "loss": 4.4882,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001153846153846154,
+      "loss": 4.5608,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001201923076923077,
+      "loss": 4.6495,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000125,
+      "loss": 4.3678,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012980769230769233,
+      "loss": 4.5512,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001346153846153846,
+      "loss": 4.4474,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013942307692307694,
+      "loss": 4.2791,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014423076923076922,
+      "loss": 4.4434,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014903846153846155,
+      "loss": 4.3315,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 4.3086,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015865384615384616,
+      "loss": 4.4628,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 4.4949,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001682692307692308,
+      "loss": 4.5125,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017307692307692307,
+      "loss": 4.56,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001778846153846154,
+      "loss": 4.2873,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018269230769230767,
+      "loss": 4.2763,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001875,
+      "loss": 4.4362,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 4.2658,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001971153846153846,
+      "loss": 4.3142,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020192307692307694,
+      "loss": 4.2805,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020673076923076922,
+      "loss": 4.2614,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021153846153846155,
+      "loss": 4.3094,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021634615384615385,
+      "loss": 4.2918,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022115384615384616,
+      "loss": 4.2968,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022596153846153846,
+      "loss": 4.2043,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002307692307692308,
+      "loss": 4.2329,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00023557692307692307,
+      "loss": 4.3264,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002403846153846154,
+      "loss": 4.3367,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002451923076923077,
+      "loss": 4.1987,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025,
+      "loss": 4.1895,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002548076923076923,
+      "loss": 4.1371,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025961538461538467,
+      "loss": 4.2447,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002644230769230769,
+      "loss": 4.1681,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002692307692307692,
+      "loss": 4.1076,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002740384615384616,
+      "loss": 4.0651,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002788461538461539,
+      "loss": 4.0728,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028365384615384613,
+      "loss": 3.9783,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028846153846153843,
+      "loss": 4.1753,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002932692307692308,
+      "loss": 4.1236,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002980769230769231,
+      "loss": 4.1174,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030288461538461535,
+      "loss": 3.8908,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003076923076923077,
+      "loss": 4.1196,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003125,
+      "loss": 4.1691,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003173076923076923,
+      "loss": 4.2177,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032211538461538467,
+      "loss": 4.1797,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003269230769230769,
+      "loss": 4.1321,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003317307692307692,
+      "loss": 3.9375,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003365384615384616,
+      "loss": 3.9576,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003413461538461539,
+      "loss": 3.9569,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00034615384615384613,
+      "loss": 4.0711,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035096153846153844,
+      "loss": 4.0995,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003557692307692308,
+      "loss": 3.912,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003605769230769231,
+      "loss": 3.8976,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036538461538461535,
+      "loss": 4.0411,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003701923076923077,
+      "loss": 4.0273,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000375,
+      "loss": 3.9225,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003798076923076923,
+      "loss": 3.9683,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00038461538461538467,
+      "loss": 3.7877,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003894230769230769,
+      "loss": 3.9174,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003942307692307692,
+      "loss": 3.9229,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003990384615384616,
+      "loss": 3.9332,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004038461538461539,
+      "loss": 3.9367,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00040865384615384613,
+      "loss": 3.9915,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00041346153846153844,
+      "loss": 3.8203,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004182692307692308,
+      "loss": 4.0365,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004230769230769231,
+      "loss": 3.8866,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00042788461538461535,
+      "loss": 3.8819,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004326923076923077,
+      "loss": 3.9308,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004375,
+      "loss": 3.8048,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004423076923076923,
+      "loss": 3.9687,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00044711538461538467,
+      "loss": 3.8629,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004519230769230769,
+      "loss": 3.9706,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004567307692307692,
+      "loss": 3.881,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004615384615384616,
+      "loss": 4.0128,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004663461538461539,
+      "loss": 3.8763,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00047115384615384613,
+      "loss": 3.7673,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00047596153846153844,
+      "loss": 3.8364,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004807692307692308,
+      "loss": 3.8304,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004855769230769231,
+      "loss": 3.837,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004903846153846154,
+      "loss": 3.9548,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004951923076923077,
+      "loss": 3.9055,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 3.898,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005048076923076923,
+      "loss": 3.9116,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005096153846153846,
+      "loss": 3.7062,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005144230769230769,
+      "loss": 3.8398,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005192307692307693,
+      "loss": 4.0592,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005240384615384615,
+      "loss": 3.7921,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005288461538461538,
+      "loss": 3.7489,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005336538461538461,
+      "loss": 3.9802,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005384615384615384,
+      "loss": 3.7342,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005432692307692307,
+      "loss": 3.9971,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005480769230769232,
+      "loss": 3.907,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005528846153846155,
+      "loss": 3.7726,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005576923076923078,
+      "loss": 3.7401,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005625000000000001,
+      "loss": 3.7839,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005673076923076923,
+      "loss": 3.9549,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005721153846153846,
+      "loss": 3.8995,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005769230769230769,
+      "loss": 3.8765,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005817307692307693,
+      "loss": 3.8703,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005865384615384616,
+      "loss": 3.7604,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005913461538461539,
+      "loss": 3.8104,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005961538461538462,
+      "loss": 3.8254,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006009615384615385,
+      "loss": 3.8615,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006057692307692307,
+      "loss": 3.8292,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006105769230769231,
+      "loss": 3.8351,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006153846153846154,
+      "loss": 3.7863,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006201923076923077,
+      "loss": 3.7492,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000625,
+      "loss": 3.7104,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006298076923076923,
+      "loss": 3.9247,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006346153846153846,
+      "loss": 3.8404,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006394230769230769,
+      "loss": 3.7626,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006442307692307693,
+      "loss": 3.6955,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006490384615384615,
+      "loss": 3.722,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006538461538461538,
+      "loss": 3.7595,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006586538461538461,
+      "loss": 3.9546,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006634615384615384,
+      "loss": 3.9042,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006682692307692307,
+      "loss": 3.7585,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006730769230769232,
+      "loss": 3.6973,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006778846153846155,
+      "loss": 3.8009,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006826923076923078,
+      "loss": 3.7775,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006875,
+      "loss": 3.7912,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006923076923076923,
+      "loss": 3.6862,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006971153846153846,
+      "loss": 3.8582,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007019230769230769,
+      "loss": 3.8413,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007067307692307693,
+      "loss": 3.7278,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007115384615384616,
+      "loss": 3.7517,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007163461538461539,
+      "loss": 3.702,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007211538461538462,
+      "loss": 3.7415,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007259615384615385,
+      "loss": 3.9373,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007307692307692307,
+      "loss": 3.7964,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007355769230769231,
+      "loss": 3.7393,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007403846153846154,
+      "loss": 3.7706,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007451923076923077,
+      "loss": 3.6625,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00075,
+      "loss": 3.8063,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007548076923076923,
+      "loss": 3.6056,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007596153846153846,
+      "loss": 3.8231,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007644230769230769,
+      "loss": 3.7259,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007692307692307693,
+      "loss": 3.826,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007740384615384615,
+      "loss": 3.7214,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007788461538461538,
+      "loss": 3.7655,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007836538461538461,
+      "loss": 3.8019,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007884615384615384,
+      "loss": 3.925,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007932692307692307,
+      "loss": 3.8384,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007980769230769232,
+      "loss": 3.688,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008028846153846155,
+      "loss": 3.7548,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008076923076923078,
+      "loss": 3.5522,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008125000000000001,
+      "loss": 3.7104,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008173076923076923,
+      "loss": 3.7556,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008221153846153846,
+      "loss": 3.7124,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008269230769230769,
+      "loss": 3.6851,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008317307692307693,
+      "loss": 3.777,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008365384615384616,
+      "loss": 3.6594,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008413461538461539,
+      "loss": 3.7439,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008461538461538462,
+      "loss": 3.8061,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008509615384615385,
+      "loss": 3.7258,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008557692307692307,
+      "loss": 3.6793,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008605769230769231,
+      "loss": 3.6215,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008653846153846154,
+      "loss": 3.7049,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008701923076923077,
+      "loss": 3.8242,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000875,
+      "loss": 3.778,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008798076923076923,
+      "loss": 3.6382,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008846153846153846,
+      "loss": 3.7708,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008894230769230769,
+      "loss": 3.6522,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008942307692307693,
+      "loss": 3.7645,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008990384615384615,
+      "loss": 3.7917,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009038461538461538,
+      "loss": 3.6763,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009086538461538461,
+      "loss": 3.7301,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009134615384615384,
+      "loss": 3.5736,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009182692307692308,
+      "loss": 3.6986,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009230769230769232,
+      "loss": 3.6325,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009278846153846155,
+      "loss": 3.8108,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009326923076923078,
+      "loss": 3.8029,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009375,
+      "loss": 3.7307,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009423076923076923,
+      "loss": 3.7974,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009471153846153846,
+      "loss": 3.8639,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009519230769230769,
+      "loss": 3.6149,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009567307692307693,
+      "loss": 3.7643,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009615384615384616,
+      "loss": 3.6839,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009663461538461539,
+      "loss": 3.7104,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009711538461538462,
+      "loss": 3.7614,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009759615384615385,
+      "loss": 3.6576,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009807692307692308,
+      "loss": 3.7488,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009855769230769232,
+      "loss": 3.7574,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009903846153846154,
+      "loss": 3.8157,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009951923076923078,
+      "loss": 3.6628,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001,
+      "loss": 3.6348,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000999999944968816,
+      "loss": 3.6941,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999997798752765,
+      "loss": 3.8112,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999995047194174,
+      "loss": 3.8529,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999991195012996,
+      "loss": 3.5949,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999986242210078,
+      "loss": 3.6972,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000999998018878651,
+      "loss": 3.6609,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999973034743625,
+      "loss": 3.6864,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999964780082996,
+      "loss": 3.8264,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999955424806442,
+      "loss": 3.6914,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999944968916022,
+      "loss": 3.7121,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999933412414039,
+      "loss": 3.7541,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999920755303033,
+      "loss": 3.8026,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999906997585793,
+      "loss": 3.5693,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999892139265346,
+      "loss": 3.6715,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999876180344966,
+      "loss": 3.7445,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999859120828163,
+      "loss": 3.796,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999984096071869,
+      "loss": 3.657,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999982170002055,
+      "loss": 3.6771,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999801338737978,
+      "loss": 3.8769,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999779876875459,
+      "loss": 3.6235,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999757314437716,
+      "loss": 3.6209,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999733651429716,
+      "loss": 3.6345,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999708887856666,
+      "loss": 3.7145,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999968302372402,
+      "loss": 3.8341,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999656059037471,
+      "loss": 3.5662,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999627993802953,
+      "loss": 3.8158,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999598828026644,
+      "loss": 3.6846,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999568561714965,
+      "loss": 3.6518,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999537194874577,
+      "loss": 3.7659,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999504727512384,
+      "loss": 3.6396,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999471159635538,
+      "loss": 3.6343,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999436491251424,
+      "loss": 3.758,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999400722367673,
+      "loss": 3.6296,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999363852992158,
+      "loss": 3.7105,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999325883132998,
+      "loss": 3.7128,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999928681279855,
+      "loss": 3.7588,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999246641997413,
+      "loss": 3.7156,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999920537073843,
+      "loss": 3.7028,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999162999030687,
+      "loss": 3.5924,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999911952688351,
+      "loss": 3.7117,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999074954306468,
+      "loss": 3.7178,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999029281309376,
+      "loss": 3.7393,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009998982507902281,
+      "loss": 3.7616,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009998934634095485,
+      "loss": 3.7342,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009998885659899524,
+      "loss": 3.6621,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998835585325178,
+      "loss": 3.5621,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999878441038347,
+      "loss": 3.7805,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998732135085666,
+      "loss": 3.6375,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998678759443273,
+      "loss": 3.8603,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998624283468036,
+      "loss": 3.7728,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999856870717195,
+      "loss": 3.6819,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998512030567252,
+      "loss": 3.625,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998454253666413,
+      "loss": 3.7778,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998395376482152,
+      "loss": 3.6591,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998335399027433,
+      "loss": 3.6708,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998274321315452,
+      "loss": 3.6223,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998212143359658,
+      "loss": 3.6819,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998148865173737,
+      "loss": 3.7105,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999808448677162,
+      "loss": 3.5446,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998019008167476,
+      "loss": 3.611,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999795242937572,
+      "loss": 3.7123,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997884750411004,
+      "loss": 3.6479,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997815971288229,
+      "loss": 3.664,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997746092022535,
+      "loss": 3.5848,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997675112629304,
+      "loss": 3.6461,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999760303312416,
+      "loss": 3.5358,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997529853522967,
+      "loss": 3.7093,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997455573841837,
+      "loss": 3.7852,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997380194097121,
+      "loss": 3.7664,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997303714305412,
+      "loss": 3.6975,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997226134483542,
+      "loss": 3.5344,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999714745464859,
+      "loss": 3.7226,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997067674817877,
+      "loss": 3.704,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996986795008961,
+      "loss": 3.8532,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999690481523965,
+      "loss": 3.7976,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996821735527987,
+      "loss": 3.7236,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996737555892257,
+      "loss": 3.6443,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996652276350996,
+      "loss": 3.6849,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996565896922974,
+      "loss": 3.6088,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996478417627203,
+      "loss": 3.5341,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009996389838482942,
+      "loss": 3.7067,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999630015950969,
+      "loss": 3.7872,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009996209380727184,
+      "loss": 3.7297,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009996117502155408,
+      "loss": 3.655,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999602452381459,
+      "loss": 3.6808,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995930445725193,
+      "loss": 3.5957,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995835267907927,
+      "loss": 3.57,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995738990383743,
+      "loss": 3.6477,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995641613173834,
+      "loss": 3.689,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995543136299637,
+      "loss": 3.6932,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995443559782827,
+      "loss": 3.6288,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995342883645324,
+      "loss": 3.7541,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999524110790929,
+      "loss": 3.7432,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995138232597126,
+      "loss": 3.7528,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999503425773148,
+      "loss": 3.7018,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994929183335236,
+      "loss": 3.6876,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994823009431529,
+      "loss": 3.7658,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994715736043726,
+      "loss": 3.4898,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994607363195442,
+      "loss": 3.6119,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994497890910533,
+      "loss": 3.6885,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994387319213094,
+      "loss": 3.7195,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999427564812747,
+      "loss": 3.6904,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994162877678236,
+      "loss": 3.7741,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994049007890221,
+      "loss": 3.709,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993934038788487,
+      "loss": 3.679,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993817970398342,
+      "loss": 3.7931,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993700802745338,
+      "loss": 3.7332,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993582535855264,
+      "loss": 3.6608,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993463169754154,
+      "loss": 3.5741,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993342704468285,
+      "loss": 3.7506,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999322114002417,
+      "loss": 3.6312,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993098476448576,
+      "loss": 3.5923,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009992974713768498,
+      "loss": 3.5332,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999284985201118,
+      "loss": 3.634,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009992723891204106,
+      "loss": 3.5849,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000999259683137501,
+      "loss": 3.6371,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009992468672551853,
+      "loss": 3.6672,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009992339414762851,
+      "loss": 3.6747,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009992209058036455,
+      "loss": 3.6194,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009992077602401356,
+      "loss": 3.6531,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00099919450478865,
+      "loss": 3.6961,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991811394521055,
+      "loss": 3.7749,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991676642334447,
+      "loss": 3.8673,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991540791356342,
+      "loss": 3.7134,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991403841616636,
+      "loss": 3.7489,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991265793145479,
+      "loss": 3.834,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991126645973259,
+      "loss": 3.827,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990986400130607,
+      "loss": 3.7506,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000999084505564839,
+      "loss": 3.6117,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990702612557728,
+      "loss": 3.7151,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990559070889968,
+      "loss": 3.8299,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990414430676715,
+      "loss": 3.7023,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990268691949804,
+      "loss": 3.7124,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990121854741316,
+      "loss": 3.6357,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989973919083575,
+      "loss": 3.7341,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989824885009142,
+      "loss": 3.7055,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989674752550826,
+      "loss": 3.6428,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989523521741674,
+      "loss": 3.7944,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989371192614977,
+      "loss": 3.6954,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989217765204266,
+      "loss": 3.7625,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000998906323954331,
+      "loss": 3.6765,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988907615666128,
+      "loss": 3.6974,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988750893606976,
+      "loss": 3.6291,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988593073400353,
+      "loss": 3.6986,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988434155080998,
+      "loss": 3.7308,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988274138683895,
+      "loss": 3.6717,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988113024244263,
+      "loss": 3.6114,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009987950811797572,
+      "loss": 3.7942,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009987787501379527,
+      "loss": 3.7083,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009987623093026074,
+      "loss": 3.7808,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009987457586773409,
+      "loss": 3.5259,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998729098265796,
+      "loss": 3.6476,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009987123280716403,
+      "loss": 3.7561,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998695448098565,
+      "loss": 3.7506,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009986784583502861,
+      "loss": 3.616,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009986613588305435,
+      "loss": 3.629,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998644149543101,
+      "loss": 3.9004,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009986268304917468,
+      "loss": 4.0177,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009986094016802935,
+      "loss": 3.8866,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985918631125773,
+      "loss": 4.063,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985742147924593,
+      "loss": 4.1941,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985564567238237,
+      "loss": 3.9636,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00099853858891058,
+      "loss": 4.0097,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985206113566614,
+      "loss": 3.8793,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985025240660248,
+      "loss": 3.9372,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984843270426519,
+      "loss": 3.8242,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984660202905483,
+      "loss": 3.9139,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984476038137435,
+      "loss": 3.8506,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984290776162922,
+      "loss": 3.9393,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984104417022715,
+      "loss": 3.9364,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983916960757842,
+      "loss": 3.7773,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983728407409564,
+      "loss": 3.8694,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983538757019391,
+      "loss": 3.8089,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983348009629063,
+      "loss": 4.0018,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983156165280574,
+      "loss": 3.7097,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998296322401615,
+      "loss": 3.8345,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009982769185878264,
+      "loss": 3.773,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998257405090963,
+      "loss": 3.8152,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009982377819153197,
+      "loss": 3.8483,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009982180490652165,
+      "loss": 3.784,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009981982065449969,
+      "loss": 3.7975,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009981782543590288,
+      "loss": 3.7096,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009981581925117044,
+      "loss": 3.8472,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998138021007439,
+      "loss": 3.6477,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000998117739850674,
+      "loss": 3.8306,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980973490458728,
+      "loss": 3.6405,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980768485975244,
+      "loss": 3.7399,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980562385101413,
+      "loss": 3.7544,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980355187882606,
+      "loss": 3.754,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980146894364428,
+      "loss": 3.7543,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997993750459273,
+      "loss": 3.836,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009979727018613607,
+      "loss": 3.6722,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997951543647339,
+      "loss": 3.829,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009979302758218653,
+      "loss": 3.8155,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009979088983896214,
+      "loss": 3.7308,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978874113553127,
+      "loss": 3.6827,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978658147236695,
+      "loss": 3.7619,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978441084994452,
+      "loss": 3.71,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978222926874181,
+      "loss": 3.8948,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978003672923904,
+      "loss": 3.7231,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009977783323191884,
+      "loss": 3.8278,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009977561877726627,
+      "loss": 3.6713,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009977339336576876,
+      "loss": 3.7997,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997711569979162,
+      "loss": 3.6536,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009976890967420089,
+      "loss": 3.7295,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009976665139511745,
+      "loss": 3.6741,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009976438216116306,
+      "loss": 3.5868,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997621019728372,
+      "loss": 3.8072,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975981083064176,
+      "loss": 3.7895,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975750873508116,
+      "loss": 3.7789,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975519568666206,
+      "loss": 3.7323,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997528716858937,
+      "loss": 3.756,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975053673328758,
+      "loss": 3.7363,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009974819082935772,
+      "loss": 3.7528,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009974583397462052,
+      "loss": 3.7039,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009974346616959476,
+      "loss": 3.6738,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009974108741480166,
+      "loss": 3.6975,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009973869771076483,
+      "loss": 3.6814,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009973629705801034,
+      "loss": 3.7683,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009973388545706657,
+      "loss": 3.7217,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009973146290846444,
+      "loss": 3.6759,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972902941273716,
+      "loss": 3.6048,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972658497042044,
+      "loss": 3.8603,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972412958205235,
+      "loss": 3.8097,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972166324817339,
+      "loss": 3.7029,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009971918596932642,
+      "loss": 3.662,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009971669774605678,
+      "loss": 3.5865,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009971419857891222,
+      "loss": 3.7332,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009971168846844283,
+      "loss": 3.6936,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009970916741520114,
+      "loss": 3.8466,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009970663541974214,
+      "loss": 3.7223,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009970409248262315,
+      "loss": 3.615,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009970153860440394,
+      "loss": 3.774,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009969897378564667,
+      "loss": 3.561,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009969639802691593,
+      "loss": 3.6403,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009969381132877873,
+      "loss": 3.8436,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009969121369180444,
+      "loss": 3.6757,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968860511656485,
+      "loss": 3.6232,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968598560363422,
+      "loss": 3.7082,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968335515358915,
+      "loss": 3.6463,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968071376700864,
+      "loss": 3.7536,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009967806144447415,
+      "loss": 3.8328,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009967539818656953,
+      "loss": 3.7297,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00099672723993881,
+      "loss": 3.7524,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009967003886699722,
+      "loss": 3.5577,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000996673428065093,
+      "loss": 3.7276,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009966463581301062,
+      "loss": 3.5709,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009966191788709714,
+      "loss": 3.6987,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009965918902936713,
+      "loss": 3.6349,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009965644924042124,
+      "loss": 3.5932,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000996536985208626,
+      "loss": 3.716,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009965093687129669,
+      "loss": 3.8616,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009964816429233142,
+      "loss": 3.5844,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009964538078457712,
+      "loss": 3.7315,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009964258634864648,
+      "loss": 3.7489,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009963978098515467,
+      "loss": 3.6614,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009963696469471918,
+      "loss": 3.6779,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009963413747795995,
+      "loss": 3.7368,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009963129933549932,
+      "loss": 3.5822,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009962845026796206,
+      "loss": 3.7772,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009962559027597532,
+      "loss": 3.7422,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000996227193601686,
+      "loss": 3.7444,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000996198375211739,
+      "loss": 3.6498,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009961694475962562,
+      "loss": 3.6996,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009961404107616048,
+      "loss": 3.6961,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009961112647141764,
+      "loss": 3.6028,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009960820094603872,
+      "loss": 3.6861,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009960526450066766,
+      "loss": 3.7575,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009960231713595088,
+      "loss": 3.7073,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009959935885253714,
+      "loss": 3.6649,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009959638965107767,
+      "loss": 3.6906,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009959340953222603,
+      "loss": 3.6002,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000995904184966382,
+      "loss": 3.6651,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009958741654497263,
+      "loss": 3.6374,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000995844036778901,
+      "loss": 3.6679,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000995813798960538,
+      "loss": 3.6765,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009957834520012937,
+      "loss": 3.6081,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009957529959078479,
+      "loss": 3.572,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009957224306869053,
+      "loss": 3.6963,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009956917563451934,
+      "loss": 3.7229,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009956609728894647,
+      "loss": 3.6748,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009956300803264954,
+      "loss": 3.7678,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009955990786630858,
+      "loss": 3.726,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00099556796790606,
+      "loss": 3.6513,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000995536748062266,
+      "loss": 3.5193,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009955054191385763,
+      "loss": 3.5712,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009954739811418875,
+      "loss": 3.585,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009954424340791195,
+      "loss": 3.6559,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009954107779572166,
+      "loss": 3.6704,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009953790127831472,
+      "loss": 3.6832,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009953471385639036,
+      "loss": 3.7979,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000995315155306502,
+      "loss": 3.6549,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009952830630179826,
+      "loss": 3.5736,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00099525086170541,
+      "loss": 3.5251,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009952185513758724,
+      "loss": 3.6032,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009951861320364821,
+      "loss": 3.5613,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009951536036943754,
+      "loss": 3.6615,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009951209663567125,
+      "loss": 3.6612,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009950882200306778,
+      "loss": 3.5431,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009950553647234796,
+      "loss": 3.6175,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00099502240044235,
+      "loss": 3.6892,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009949893271945455,
+      "loss": 3.6055,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994956144987346,
+      "loss": 3.7621,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994922853828056,
+      "loss": 3.6264,
+      "step": 517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009948894537240038,
+      "loss": 3.6447,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009948559446825412,
+      "loss": 3.6252,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009948223267110447,
+      "loss": 3.5332,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009947885998169142,
+      "loss": 3.742,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994754764007574,
+      "loss": 3.6361,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009947208192904722,
+      "loss": 3.7173,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994686765673081,
+      "loss": 3.6875,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009946526031628959,
+      "loss": 3.6247,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009946183317674376,
+      "loss": 3.6669,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009945839514942495,
+      "loss": 3.7014,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009945494623509001,
+      "loss": 3.6868,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009945148643449807,
+      "loss": 3.7157,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009944801574841078,
+      "loss": 3.8294,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994445341775921,
+      "loss": 3.733,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009944104172280837,
+      "loss": 3.7909,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009943753838482844,
+      "loss": 3.6931,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994340241644234,
+      "loss": 3.6485,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009943049906236687,
+      "loss": 3.7484,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009942696307943483,
+      "loss": 3.6667,
+      "step": 536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009942341621640557,
+      "loss": 3.7458,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000994198584740599,
+      "loss": 3.5692,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009941628985318093,
+      "loss": 3.5952,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009941271035455424,
+      "loss": 3.7567,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009940911997896773,
+      "loss": 3.8045,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009940551872721175,
+      "loss": 3.6537,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009940190660007902,
+      "loss": 3.6709,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009939828359836467,
+      "loss": 3.7905,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009939464972286618,
+      "loss": 3.5699,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993910049743835,
+      "loss": 3.7881,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993873493537189,
+      "loss": 3.6279,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009938368286167706,
+      "loss": 3.5849,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993800054990651,
+      "loss": 3.6961,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009937631726669245,
+      "loss": 3.7376,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009937261816537106,
+      "loss": 3.7351,
+      "step": 551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009936890819591511,
+      "loss": 3.6581,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009936518735914133,
+      "loss": 3.6157,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993614556558687,
+      "loss": 3.6744,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993577130869187,
+      "loss": 3.5829,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009935395965311516,
+      "loss": 3.6365,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009935019535528432,
+      "loss": 3.6775,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009934642019425472,
+      "loss": 3.7224,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009934263417085746,
+      "loss": 3.7002,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009933883728592589,
+      "loss": 3.7048,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993350295402958,
+      "loss": 3.4936,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009933121093480537,
+      "loss": 3.529,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009932738147029516,
+      "loss": 3.665,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009932354114760818,
+      "loss": 3.7,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009931968996758972,
+      "loss": 3.6423,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009931582793108753,
+      "loss": 3.5859,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009931195503895178,
+      "loss": 3.7281,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009930807129203494,
+      "loss": 3.6349,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009930417669119194,
+      "loss": 3.6389,
+      "step": 569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000993002712372801,
+      "loss": 3.6347,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009929635493115907,
+      "loss": 3.6897,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009929242777369095,
+      "loss": 3.75,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009928848976574018,
+      "loss": 3.6016,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009928454090817364,
+      "loss": 3.6417,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009928058120186058,
+      "loss": 3.5364,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009927661064767258,
+      "loss": 3.7325,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000992726292464837,
+      "loss": 3.6492,
+      "step": 577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009926863699917033,
+      "loss": 3.6142,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009926463390661127,
+      "loss": 3.7066,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000992606199696877,
+      "loss": 3.4474,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009925659518928314,
+      "loss": 3.7053,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009925255956628363,
+      "loss": 3.6484,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009924851310157744,
+      "loss": 3.5367,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009924445579605533,
+      "loss": 3.6106,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000992403876506104,
+      "loss": 3.5737,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009923630866613816,
+      "loss": 3.7044,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009923221884353649,
+      "loss": 3.5919,
+      "step": 587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009922811818370567,
+      "loss": 3.6284,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009922400668754834,
+      "loss": 3.6501,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009921988435596953,
+      "loss": 3.526,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000992157511898767,
+      "loss": 3.737,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009921160719017967,
+      "loss": 3.5667,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009920745235779058,
+      "loss": 3.7342,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009920328669362408,
+      "loss": 3.6082,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000991991101985971,
+      "loss": 3.5576,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009919492287362897,
+      "loss": 3.6667,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009919072471964146,
+      "loss": 3.5459,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009918651573755865,
+      "loss": 3.7746,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000991822959283071,
+      "loss": 3.7523,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009917806529281565,
+      "loss": 3.6561,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009917382383201555,
+      "loss": 3.7697,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009916957154684051,
+      "loss": 3.649,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009916530843822654,
+      "loss": 3.606,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009916103450711203,
+      "loss": 3.6625,
+      "step": 604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009915674975443777,
+      "loss": 3.6538,
+      "step": 605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00099152454181147,
+      "loss": 3.6074,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009914814778818524,
+      "loss": 3.5671,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009914383057650043,
+      "loss": 3.7064,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991395025470429,
+      "loss": 3.5435,
+      "step": 609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009913516370076536,
+      "loss": 3.644,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991308140386229,
+      "loss": 3.5077,
+      "step": 611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00099126453561573,
+      "loss": 3.5789,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991220822705755,
+      "loss": 3.6847,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991177001665926,
+      "loss": 3.5346,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009911330725058895,
+      "loss": 3.6991,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009910890352353154,
+      "loss": 3.5078,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009910448898638971,
+      "loss": 3.6742,
+      "step": 617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991000636401352,
+      "loss": 3.5215,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009909562748574218,
+      "loss": 3.6943,
+      "step": 619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009909118052418716,
+      "loss": 3.5329,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009908672275644897,
+      "loss": 3.6148,
+      "step": 621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990822541835089,
+      "loss": 3.6845,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009907777480635063,
+      "loss": 3.6781,
+      "step": 623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009907328462596015,
+      "loss": 3.6402,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009906878364332585,
+      "loss": 3.7434,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009906427185943853,
+      "loss": 3.6494,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009905974927529135,
+      "loss": 3.4317,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990552158918798,
+      "loss": 3.5232,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009905067171020185,
+      "loss": 3.5147,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009904611673125774,
+      "loss": 3.5913,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009904155095605013,
+      "loss": 3.6906,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009903697438558411,
+      "loss": 3.6337,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009903238702086705,
+      "loss": 3.5287,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009902778886290878,
+      "loss": 3.5244,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990231799127214,
+      "loss": 3.744,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009901856017131953,
+      "loss": 3.5939,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009901392963972007,
+      "loss": 3.6093,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990092883189423,
+      "loss": 3.568,
+      "step": 638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009900463621000787,
+      "loss": 3.6674,
+      "step": 639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009899997331394089,
+      "loss": 3.6506,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989952996317677,
+      "loss": 3.7021,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009899061516451715,
+      "loss": 3.5932,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009898591991322036,
+      "loss": 3.5443,
+      "step": 643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009898121387891092,
+      "loss": 3.6242,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009897649706262473,
+      "loss": 3.6105,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009897176946540005,
+      "loss": 3.6434,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989670310882776,
+      "loss": 3.6683,
+      "step": 647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009896228193230032,
+      "loss": 3.6425,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989575219985137,
+      "loss": 3.7033,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989527512879655,
+      "loss": 3.5329,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009894796980170584,
+      "loss": 3.6693,
+      "step": 651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009894317754078727,
+      "loss": 3.6757,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009893837450626471,
+      "loss": 3.5769,
+      "step": 653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009893356069919537,
+      "loss": 3.5945,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009892873612063894,
+      "loss": 3.5734,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009892390077165737,
+      "loss": 3.4647,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989190546533151,
+      "loss": 3.5565,
+      "step": 657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009891419776667885,
+      "loss": 3.5983,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009890933011281774,
+      "loss": 3.6734,
+      "step": 659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009890445169280328,
+      "loss": 3.5085,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009889956250770933,
+      "loss": 3.5911,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009889466255861208,
+      "loss": 3.6021,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009888975184659018,
+      "loss": 3.4324,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009888483037272455,
+      "loss": 3.705,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000988798981380986,
+      "loss": 3.5961,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009887495514379796,
+      "loss": 3.5768,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009887000139091073,
+      "loss": 3.6342,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009886503688052739,
+      "loss": 3.5585,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000988600616137407,
+      "loss": 3.5758,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000988550755916459,
+      "loss": 3.5678,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009885007881534045,
+      "loss": 3.5479,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009884507128592435,
+      "loss": 3.6072,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009884005300449983,
+      "loss": 3.6328,
+      "step": 673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009883502397217156,
+      "loss": 3.5589,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009882998419004653,
+      "loss": 3.5256,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009882493365923415,
+      "loss": 3.5706,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009881987238084615,
+      "loss": 3.4686,
+      "step": 677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009881480035599667,
+      "loss": 3.5833,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009880971758580216,
+      "loss": 3.5522,
+      "step": 679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009880462407138144,
+      "loss": 3.6313,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009879951981385578,
+      "loss": 3.4857,
+      "step": 681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000987944048143487,
+      "loss": 3.7416,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009878927907398616,
+      "loss": 3.517,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009878414259389646,
+      "loss": 3.4619,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009877899537521027,
+      "loss": 3.6346,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009877383741906062,
+      "loss": 3.6255,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000987686687265829,
+      "loss": 3.5642,
+      "step": 687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009876348929891486,
+      "loss": 3.474,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009875829913719664,
+      "loss": 3.5055,
+      "step": 689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009875309824257071,
+      "loss": 3.5864,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000987478866161819,
+      "loss": 3.4751,
+      "step": 691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009874266425917745,
+      "loss": 3.6781,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009873743117270691,
+      "loss": 3.6705,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009873218735792222,
+      "loss": 3.5399,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009872693281597767,
+      "loss": 3.4929,
+      "step": 695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000987216675480299,
+      "loss": 3.5202,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009871639155523795,
+      "loss": 3.7003,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000987111048387632,
+      "loss": 3.5224,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009870580739976935,
+      "loss": 3.4661,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009870049923942253,
+      "loss": 3.5702,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000986951803588912,
+      "loss": 3.6086,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009868985075934616,
+      "loss": 3.6399,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009868451044196059,
+      "loss": 3.6272,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009867915940791,
+      "loss": 3.6246,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009867379765837235,
+      "loss": 3.4941,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009866842519452785,
+      "loss": 3.6187,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009866304201755912,
+      "loss": 3.6571,
+      "step": 707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009865764812865112,
+      "loss": 3.467,
+      "step": 708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009865224352899118,
+      "loss": 3.5198,
+      "step": 709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009864682821976902,
+      "loss": 3.5924,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009864140220217665,
+      "loss": 3.6005,
+      "step": 711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009863596547740846,
+      "loss": 3.5977,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009863051804666124,
+      "loss": 3.4406,
+      "step": 713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009862505991113408,
+      "loss": 3.5955,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009861959107202846,
+      "loss": 3.7426,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009861411153054822,
+      "loss": 3.5773,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009860862128789954,
+      "loss": 3.5234,
+      "step": 717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009860312034529093,
+      "loss": 3.561,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009859760870393332,
+      "loss": 3.615,
+      "step": 719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985920863650399,
+      "loss": 3.5786,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009858655332982632,
+      "loss": 3.5686,
+      "step": 721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009858100959951057,
+      "loss": 3.6361,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985754551753129,
+      "loss": 3.6035,
+      "step": 723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00098569890058456,
+      "loss": 3.4891,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985643142501649,
+      "loss": 3.5108,
+      "step": 725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009855872775166696,
+      "loss": 3.5662,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985531305641919,
+      "loss": 3.5337,
+      "step": 727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009854752268897181,
+      "loss": 3.5486,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009854190412724112,
+      "loss": 3.5992,
+      "step": 729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009853627488023663,
+      "loss": 3.5737,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009853063494919745,
+      "loss": 3.6321,
+      "step": 731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985249843353651,
+      "loss": 3.6426,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009851932303998336,
+      "loss": 3.61,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009851365106429849,
+      "loss": 3.6078,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00098507968409559,
+      "loss": 3.5538,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985022750770158,
+      "loss": 3.4704,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000984965710679221,
+      "loss": 3.5393,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000984908563835335,
+      "loss": 3.6422,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00098485131025108,
+      "loss": 3.4752,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000984793949939058,
+      "loss": 3.53,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009847364829118962,
+      "loss": 3.5809,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009846789091822441,
+      "loss": 3.6176,
+      "step": 742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009846212287627754,
+      "loss": 3.5731,
+      "step": 743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009845634416661866,
+      "loss": 3.6697,
+      "step": 744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009845055479051985,
+      "loss": 3.6432,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009844475474925548,
+      "loss": 3.4783,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009843894404410224,
+      "loss": 3.5522,
+      "step": 747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009843312267633928,
+      "loss": 3.6019,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00098427290647248,
+      "loss": 3.5303,
+      "step": 749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009842144795811215,
+      "loss": 3.6123,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009841559461021786,
+      "loss": 3.5729,
+      "step": 751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000984097306048536,
+      "loss": 3.4976,
+      "step": 752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000984038559433102,
+      "loss": 3.5393,
+      "step": 753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009839797062688083,
+      "loss": 3.6539,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009839207465686093,
+      "loss": 3.5819,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000983861680345484,
+      "loss": 3.5266,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009838025076124345,
+      "loss": 3.6981,
+      "step": 757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009837432283824855,
+      "loss": 3.6821,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009836838426686863,
+      "loss": 3.4628,
+      "step": 759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009836243504841092,
+      "loss": 3.6034,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009835647518418498,
+      "loss": 3.6074,
+      "step": 761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009835050467550273,
+      "loss": 3.6137,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000983445235236784,
+      "loss": 3.4411,
+      "step": 763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009833853173002861,
+      "loss": 3.5851,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000983325292958723,
+      "loss": 3.4749,
+      "step": 765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009832651622253077,
+      "loss": 3.5695,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009832049251132762,
+      "loss": 3.6013,
+      "step": 767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009831445816358884,
+      "loss": 3.5717,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009830841318064273,
+      "loss": 3.6076,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009830235756381992,
+      "loss": 3.499,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009829629131445341,
+      "loss": 3.4922,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009829021443387856,
+      "loss": 3.6222,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009828412692343303,
+      "loss": 3.3502,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000982780287844568,
+      "loss": 3.351,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009827192001829227,
+      "loss": 3.5911,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000982658006262841,
+      "loss": 3.5716,
+      "step": 776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009825967060977932,
+      "loss": 3.5324,
+      "step": 777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000982535299701273,
+      "loss": 3.6293,
+      "step": 778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009824737870867976,
+      "loss": 3.5073,
+      "step": 779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009824121682679072,
+      "loss": 3.5074,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000982350443258166,
+      "loss": 3.5036,
+      "step": 781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000982288612071161,
+      "loss": 3.5905,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009822266747205024,
+      "loss": 3.5023,
+      "step": 783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009821646312198249,
+      "loss": 3.5743,
+      "step": 784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009821024815827853,
+      "loss": 3.5681,
+      "step": 785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009820402258230642,
+      "loss": 3.6316,
+      "step": 786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000981977863954366,
+      "loss": 3.5863,
+      "step": 787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009819153959904178,
+      "loss": 3.6011,
+      "step": 788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009818528219449705,
+      "loss": 3.5443,
+      "step": 789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000981790141831798,
+      "loss": 3.5701,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009817273556646982,
+      "loss": 3.5753,
+      "step": 791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009816644634574913,
+      "loss": 3.549,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009816014652240218,
+      "loss": 3.5605,
+      "step": 793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000981538360978157,
+      "loss": 3.601,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009814751507337877,
+      "loss": 3.5687,
+      "step": 795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009814118345048285,
+      "loss": 3.6279,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009813484123052162,
+      "loss": 3.5233,
+      "step": 797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009812848841489119,
+      "loss": 3.4506,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009812212500498996,
+      "loss": 3.6081,
+      "step": 799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000981157510022187,
+      "loss": 3.6848,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009810936640798045,
+      "loss": 3.4994,
+      "step": 801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009810297122368067,
+      "loss": 3.6364,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009809656545072703,
+      "loss": 3.4953,
+      "step": 803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009809014909052966,
+      "loss": 3.5484,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009808372214450093,
+      "loss": 3.5691,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009807728461405556,
+      "loss": 3.5722,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009807083650061063,
+      "loss": 3.5811,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009806437780558552,
+      "loss": 3.6131,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009805790853040196,
+      "loss": 3.5346,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009805142867648398,
+      "loss": 3.6232,
+      "step": 810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009804493824525796,
+      "loss": 3.5237,
+      "step": 811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009803843723815262,
+      "loss": 3.6583,
+      "step": 812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009803192565659897,
+      "loss": 3.5203,
+      "step": 813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009802540350203038,
+      "loss": 3.5106,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009801887077588254,
+      "loss": 3.6142,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009801232747959348,
+      "loss": 3.4449,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000980057736146035,
+      "loss": 3.4883,
+      "step": 817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009799920918235532,
+      "loss": 3.4582,
+      "step": 818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000979926341842939,
+      "loss": 3.638,
+      "step": 819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009798604862186655,
+      "loss": 3.4349,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009797945249652295,
+      "loss": 3.5413,
+      "step": 821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009797284580971503,
+      "loss": 3.5257,
+      "step": 822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009796622856289713,
+      "loss": 3.4987,
+      "step": 823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009795960075752585,
+      "loss": 3.6352,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000979529623950601,
+      "loss": 3.5039,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009794631347696122,
+      "loss": 3.3518,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009793965400469273,
+      "loss": 3.6784,
+      "step": 827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009793298397972057,
+      "loss": 3.5734,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009792630340351301,
+      "loss": 3.5829,
+      "step": 829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009791961227754056,
+      "loss": 3.5014,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009791291060327612,
+      "loss": 3.5623,
+      "step": 831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000979061983821949,
+      "loss": 3.6146,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009789947561577445,
+      "loss": 3.5619,
+      "step": 833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009789274230549456,
+      "loss": 3.4901,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009788599845283745,
+      "loss": 3.5296,
+      "step": 835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009787924405928758,
+      "loss": 3.4983,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009787247912633177,
+      "loss": 3.4982,
+      "step": 837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009786570365545916,
+      "loss": 3.5671,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009785891764816116,
+      "loss": 3.4727,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009785212110593158,
+      "loss": 3.4206,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009784531403026649,
+      "loss": 3.5059,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000978384964226643,
+      "loss": 3.6528,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009783166828462573,
+      "loss": 3.56,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009782482961765383,
+      "loss": 3.4675,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000978179804232539,
+      "loss": 3.5227,
+      "step": 845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009781112070293373,
+      "loss": 3.5736,
+      "step": 846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009780425045820323,
+      "loss": 3.658,
+      "step": 847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009779736969057473,
+      "loss": 3.6315,
+      "step": 848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009779047840156288,
+      "loss": 3.5286,
+      "step": 849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977835765926846,
+      "loss": 3.5504,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009777666426545916,
+      "loss": 3.5619,
+      "step": 851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977697414214081,
+      "loss": 3.4639,
+      "step": 852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009776280806205536,
+      "loss": 3.585,
+      "step": 853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977558641889271,
+      "loss": 3.6519,
+      "step": 854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009774890980355187,
+      "loss": 3.4163,
+      "step": 855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977419449074605,
+      "loss": 3.5057,
+      "step": 856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009773496950218612,
+      "loss": 3.5909,
+      "step": 857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977279835892642,
+      "loss": 3.6187,
+      "step": 858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977209871702325,
+      "loss": 3.5689,
+      "step": 859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009771398024663112,
+      "loss": 3.5449,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009770696282000244,
+      "loss": 3.5003,
+      "step": 861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009769993489189119,
+      "loss": 3.5458,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009769289646384437,
+      "loss": 3.4318,
+      "step": 863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009768584753741135,
+      "loss": 3.5175,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009767878811414372,
+      "loss": 3.5169,
+      "step": 865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009767171819559549,
+      "loss": 3.5585,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009766463778332285,
+      "loss": 3.4212,
+      "step": 867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009765754687888445,
+      "loss": 3.5118,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009765044548384113,
+      "loss": 3.5994,
+      "step": 869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000976433335997561,
+      "loss": 3.4767,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009763621122819484,
+      "loss": 3.5178,
+      "step": 871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009762907837072519,
+      "loss": 3.5422,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009762193502891725,
+      "loss": 3.4347,
+      "step": 873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009761478120434345,
+      "loss": 3.4119,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009760761689857852,
+      "loss": 3.6649,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009760044211319952,
+      "loss": 3.6256,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009759325684978576,
+      "loss": 3.4753,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009758606110991892,
+      "loss": 3.3769,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009757885489518296,
+      "loss": 3.5747,
+      "step": 879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009757163820716416,
+      "loss": 3.5021,
+      "step": 880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009756441104745106,
+      "loss": 3.6369,
+      "step": 881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009755717341763455,
+      "loss": 3.5371,
+      "step": 882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009754992531930782,
+      "loss": 3.5898,
+      "step": 883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009754266675406634,
+      "loss": 3.59,
+      "step": 884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009753539772350791,
+      "loss": 3.4418,
+      "step": 885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009752811822923262,
+      "loss": 3.5543,
+      "step": 886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009752082827284288,
+      "loss": 3.6001,
+      "step": 887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009751352785594337,
+      "loss": 3.503,
+      "step": 888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009750621698014111,
+      "loss": 3.4875,
+      "step": 889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009749889564704537,
+      "loss": 3.5998,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009749156385826781,
+      "loss": 3.4877,
+      "step": 891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009748422161542228,
+      "loss": 3.4938,
+      "step": 892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009747686892012504,
+      "loss": 3.4152,
+      "step": 893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009746950577399458,
+      "loss": 3.5186,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000974621321786517,
+      "loss": 3.5221,
+      "step": 895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009745474813571953,
+      "loss": 3.5841,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009744735364682345,
+      "loss": 3.7238,
+      "step": 897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000974399487135912,
+      "loss": 3.6349,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009743253333765279,
+      "loss": 3.5319,
+      "step": 899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009742510752064051,
+      "loss": 3.4815,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009741767126418898,
+      "loss": 3.452,
+      "step": 901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009741022456993509,
+      "loss": 3.6086,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009740276743951806,
+      "loss": 3.43,
+      "step": 903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009739529987457936,
+      "loss": 3.5281,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009738782187676281,
+      "loss": 3.3784,
+      "step": 905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000973803334477145,
+      "loss": 3.445,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009737283458908281,
+      "loss": 3.6006,
+      "step": 907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009736532530251842,
+      "loss": 3.525,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009735780558967434,
+      "loss": 3.5538,
+      "step": 909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000973502754522058,
+      "loss": 3.4658,
+      "step": 910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009734273489177041,
+      "loss": 3.6239,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009733518391002803,
+      "loss": 3.3687,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009732762250864078,
+      "loss": 3.5682,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009732005068927314,
+      "loss": 3.526,
+      "step": 914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009731246845359185,
+      "loss": 3.6061,
+      "step": 915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009730487580326594,
+      "loss": 3.4868,
+      "step": 916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009729727273996675,
+      "loss": 3.4171,
+      "step": 917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009728965926536793,
+      "loss": 3.4232,
+      "step": 918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009728203538114533,
+      "loss": 3.288,
+      "step": 919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000972744010889772,
+      "loss": 3.5914,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009726675639054403,
+      "loss": 3.5839,
+      "step": 921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009725910128752861,
+      "loss": 3.5542,
+      "step": 922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00097251435781616,
+      "loss": 3.4124,
+      "step": 923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009724375987449358,
+      "loss": 3.383,
+      "step": 924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009723607356785102,
+      "loss": 3.4318,
+      "step": 925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009722837686338024,
+      "loss": 3.5786,
+      "step": 926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009722066976277548,
+      "loss": 3.371,
+      "step": 927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009721295226773329,
+      "loss": 3.5054,
+      "step": 928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009720522437995245,
+      "loss": 3.5444,
+      "step": 929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009719748610113408,
+      "loss": 3.4402,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009718973743298154,
+      "loss": 3.4786,
+      "step": 931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009718197837720053,
+      "loss": 3.5061,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009717420893549903,
+      "loss": 3.627,
+      "step": 933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009716642910958721,
+      "loss": 3.6156,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009715863890117769,
+      "loss": 3.5177,
+      "step": 935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009715083831198521,
+      "loss": 3.5053,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009714302734372694,
+      "loss": 3.4878,
+      "step": 937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009713520599812222,
+      "loss": 3.4091,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009712737427689276,
+      "loss": 3.497,
+      "step": 939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009711953218176249,
+      "loss": 3.4804,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009711167971445764,
+      "loss": 3.4217,
+      "step": 941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009710381687670675,
+      "loss": 3.5062,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009709594367024062,
+      "loss": 3.5981,
+      "step": 943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009708806009679235,
+      "loss": 3.6186,
+      "step": 944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009708016615809729,
+      "loss": 3.4613,
+      "step": 945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009707226185589311,
+      "loss": 3.5123,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009706434719191972,
+      "loss": 3.5741,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009705642216791935,
+      "loss": 3.4768,
+      "step": 948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009704848678563648,
+      "loss": 3.5631,
+      "step": 949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000970405410468179,
+      "loss": 3.5149,
+      "step": 950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009703258495321266,
+      "loss": 3.5053,
+      "step": 951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009702461850657208,
+      "loss": 3.606,
+      "step": 952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009701664170864979,
+      "loss": 3.5679,
+      "step": 953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009700865456120167,
+      "loss": 3.5625,
+      "step": 954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009700065706598588,
+      "loss": 3.5396,
+      "step": 955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009699264922476289,
+      "loss": 3.5066,
+      "step": 956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 3.4495,
+      "step": 957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009697660251134846,
+      "loss": 3.5757,
+      "step": 958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009696856364268928,
+      "loss": 3.4091,
+      "step": 959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009696051443508745,
+      "loss": 3.3879,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009695245489031478,
+      "loss": 3.5407,
+      "step": 961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009694438501014539,
+      "loss": 3.5848,
+      "step": 962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009693630479635566,
+      "loss": 3.4625,
+      "step": 963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009692821425072427,
+      "loss": 3.4761,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009692011337503211,
+      "loss": 3.5363,
+      "step": 965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009691200217106237,
+      "loss": 3.4911,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009690388064060059,
+      "loss": 3.5524,
+      "step": 967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009689574878543446,
+      "loss": 3.5366,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009688760660735403,
+      "loss": 3.5017,
+      "step": 969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009687945410815158,
+      "loss": 3.4439,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000968712912896217,
+      "loss": 3.6677,
+      "step": 971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009686311815356121,
+      "loss": 3.5207,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009685493470176922,
+      "loss": 3.6109,
+      "step": 973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009684674093604713,
+      "loss": 3.4524,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009683853685819856,
+      "loss": 3.4903,
+      "step": 975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009683032247002945,
+      "loss": 3.5888,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009682209777334798,
+      "loss": 3.5445,
+      "step": 977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009681386276996462,
+      "loss": 3.5005,
+      "step": 978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000968056174616921,
+      "loss": 3.4943,
+      "step": 979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009679736185034539,
+      "loss": 3.5839,
+      "step": 980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009678909593774179,
+      "loss": 3.5211,
+      "step": 981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000967808197257008,
+      "loss": 3.3511,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009677253321604425,
+      "loss": 3.5873,
+      "step": 983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009676423641059617,
+      "loss": 3.4395,
+      "step": 984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009675592931118293,
+      "loss": 3.6192,
+      "step": 985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009674761191963311,
+      "loss": 3.5369,
+      "step": 986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009673928423777756,
+      "loss": 3.5729,
+      "step": 987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009673094626744943,
+      "loss": 3.5083,
+      "step": 988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000967225980104841,
+      "loss": 3.5692,
+      "step": 989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009671423946871924,
+      "loss": 3.5435,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009670587064399476,
+      "loss": 3.4407,
+      "step": 991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009669749153815285,
+      "loss": 3.4946,
+      "step": 992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009668910215303797,
+      "loss": 3.4737,
+      "step": 993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000966807024904968,
+      "loss": 3.4256,
+      "step": 994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009667229255237835,
+      "loss": 3.4798,
+      "step": 995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009666387234053385,
+      "loss": 3.4783,
+      "step": 996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009665544185681677,
+      "loss": 3.4432,
+      "step": 997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009664700110308287,
+      "loss": 3.4962,
+      "step": 998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000966385500811902,
+      "loss": 3.3048,
+      "step": 999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009663008879299902,
+      "loss": 3.5994,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009662161724037187,
+      "loss": 3.5069,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009661313542517354,
+      "loss": 3.5633,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009660464334927112,
+      "loss": 3.5597,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009659614101453389,
+      "loss": 3.4308,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009658762842283342,
+      "loss": 3.5643,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009657910557604358,
+      "loss": 3.45,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009657057247604042,
+      "loss": 3.5006,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009656202912470232,
+      "loss": 3.5166,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009655347552390986,
+      "loss": 3.5072,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009654491167554591,
+      "loss": 3.4849,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009653633758149558,
+      "loss": 3.4952,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009652775324364624,
+      "loss": 3.4285,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009651915866388752,
+      "loss": 3.617,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009651055384411128,
+      "loss": 3.4667,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009650193878621169,
+      "loss": 3.423,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000964933134920851,
+      "loss": 3.5756,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009648467796363017,
+      "loss": 3.5222,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009647603220274781,
+      "loss": 3.5495,
+      "step": 1018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009646737621134112,
+      "loss": 3.4985,
+      "step": 1019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009645870999131554,
+      "loss": 3.5168,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009645003354457871,
+      "loss": 3.4323,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009644134687304052,
+      "loss": 3.4265,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009643264997861312,
+      "loss": 3.584,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009642394286321091,
+      "loss": 3.4916,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009641522552875055,
+      "loss": 3.3122,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009640649797715095,
+      "loss": 3.5789,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009639776021033326,
+      "loss": 3.4873,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009638901223022085,
+      "loss": 3.5818,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009638025403873939,
+      "loss": 3.401,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009637148563781677,
+      "loss": 3.6326,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009636270702938313,
+      "loss": 3.641,
+      "step": 1031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009635391821537087,
+      "loss": 3.5108,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000963451191977146,
+      "loss": 3.4928,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009633630997835124,
+      "loss": 3.4518,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009632749055921987,
+      "loss": 3.5628,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009631866094226189,
+      "loss": 3.4975,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009630982112942092,
+      "loss": 3.557,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009630097112264283,
+      "loss": 3.4622,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009629211092387569,
+      "loss": 3.6069,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009628324053506988,
+      "loss": 3.43,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009627435995817798,
+      "loss": 3.4051,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009626546919515482,
+      "loss": 3.4932,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000962565682479575,
+      "loss": 3.3411,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009624765711854531,
+      "loss": 3.6247,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009623873580887981,
+      "loss": 3.4938,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009622980432092483,
+      "loss": 3.5675,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000962208626566464,
+      "loss": 3.4404,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009621191081801277,
+      "loss": 3.4603,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009620294880699449,
+      "loss": 3.5026,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009619397662556434,
+      "loss": 3.4933,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009618499427569729,
+      "loss": 3.423,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009617600175937057,
+      "loss": 3.5696,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009616699907856368,
+      "loss": 3.4475,
+      "step": 1053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009615798623525831,
+      "loss": 3.3406,
+      "step": 1054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009614896323143844,
+      "loss": 3.5829,
+      "step": 1055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009613993006909022,
+      "loss": 3.5474,
+      "step": 1056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000961308867502021,
+      "loss": 3.4575,
+      "step": 1057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009612183327676472,
+      "loss": 3.3231,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009611276965077098,
+      "loss": 3.631,
+      "step": 1059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009610369587421603,
+      "loss": 3.4946,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009609461194909719,
+      "loss": 3.4722,
+      "step": 1061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009608551787741408,
+      "loss": 3.3482,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009607641366116855,
+      "loss": 3.4333,
+      "step": 1063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009606729930236463,
+      "loss": 3.3754,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009605817480300863,
+      "loss": 3.3337,
+      "step": 1065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009604904016510907,
+      "loss": 3.4779,
+      "step": 1066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009603989539067673,
+      "loss": 3.5222,
+      "step": 1067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009603074048172458,
+      "loss": 3.6245,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009602157544026784,
+      "loss": 3.5671,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009601240026832398,
+      "loss": 3.457,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009600321496791268,
+      "loss": 3.5599,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009599401954105583,
+      "loss": 3.5857,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000959848139897776,
+      "loss": 3.3973,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009597559831610434,
+      "loss": 3.495,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009596637252206466,
+      "loss": 3.508,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009595713660968937,
+      "loss": 3.4039,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009594789058101153,
+      "loss": 3.543,
+      "step": 1077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009593863443806643,
+      "loss": 3.4228,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009592936818289158,
+      "loss": 3.4424,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009592009181752667,
+      "loss": 3.5694,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000959108053440137,
+      "loss": 3.4883,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009590150876439685,
+      "loss": 3.4609,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000958922020807225,
+      "loss": 3.5559,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000958828852950393,
+      "loss": 3.4133,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009587355840939812,
+      "loss": 3.4119,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00095864221425852,
+      "loss": 3.5892,
+      "step": 1086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009585487434645627,
+      "loss": 3.477,
+      "step": 1087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009584551717326846,
+      "loss": 3.4365,
+      "step": 1088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009583614990834828,
+      "loss": 3.5156,
+      "step": 1089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009582677255375773,
+      "loss": 3.5673,
+      "step": 1090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009581738511156099,
+      "loss": 3.5105,
+      "step": 1091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009580798758382445,
+      "loss": 3.4694,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009579857997261676,
+      "loss": 3.5583,
+      "step": 1093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009578916228000875,
+      "loss": 3.5915,
+      "step": 1094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009577973450807352,
+      "loss": 3.4807,
+      "step": 1095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009577029665888631,
+      "loss": 3.5021,
+      "step": 1096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009576084873452465,
+      "loss": 3.487,
+      "step": 1097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009575139073706827,
+      "loss": 3.5604,
+      "step": 1098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009574192266859909,
+      "loss": 3.4013,
+      "step": 1099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009573244453120127,
+      "loss": 3.3565,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009572295632696119,
+      "loss": 3.5115,
+      "step": 1101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009571345805796744,
+      "loss": 3.3657,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009570394972631081,
+      "loss": 3.5658,
+      "step": 1103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009569443133408433,
+      "loss": 3.4801,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009568490288338324,
+      "loss": 3.4802,
+      "step": 1105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009567536437630498,
+      "loss": 3.5203,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000956658158149492,
+      "loss": 3.3927,
+      "step": 1107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009565625720141779,
+      "loss": 3.5339,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009564668853781482,
+      "loss": 3.4401,
+      "step": 1109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009563710982624661,
+      "loss": 3.3405,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009562752106882167,
+      "loss": 3.5112,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009561792226765072,
+      "loss": 3.4346,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009560831342484667,
+      "loss": 3.542,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000955986945425247,
+      "loss": 3.4283,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009558906562280214,
+      "loss": 3.4636,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009557942666779855,
+      "loss": 3.4943,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009556977767963573,
+      "loss": 3.3695,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009556011866043763,
+      "loss": 3.298,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009555044961233047,
+      "loss": 3.4682,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009554077053744262,
+      "loss": 3.6107,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000955310814379047,
+      "loss": 3.4354,
+      "step": 1121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000955213823158495,
+      "loss": 3.4328,
+      "step": 1122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009551167317341208,
+      "loss": 3.4753,
+      "step": 1123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000955019540127296,
+      "loss": 3.3754,
+      "step": 1124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009549222483594154,
+      "loss": 3.5392,
+      "step": 1125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000954824856451895,
+      "loss": 3.3951,
+      "step": 1126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009547273644261733,
+      "loss": 3.4474,
+      "step": 1127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009546297723037108,
+      "loss": 3.585,
+      "step": 1128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009545320801059899,
+      "loss": 3.4031,
+      "step": 1129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000954434287854515,
+      "loss": 3.4119,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009543363955708124,
+      "loss": 3.4096,
+      "step": 1131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009542384032764311,
+      "loss": 3.3013,
+      "step": 1132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009541403109929412,
+      "loss": 3.3416,
+      "step": 1133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009540421187419354,
+      "loss": 3.3955,
+      "step": 1134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009539438265450282,
+      "loss": 3.4547,
+      "step": 1135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009538454344238563,
+      "loss": 3.5079,
+      "step": 1136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000953746942400078,
+      "loss": 3.4254,
+      "step": 1137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000953648350495374,
+      "loss": 3.3355,
+      "step": 1138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009535496587314467,
+      "loss": 3.5066,
+      "step": 1139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009534508671300208,
+      "loss": 3.5104,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009533519757128426,
+      "loss": 3.5015,
+      "step": 1141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009532529845016804,
+      "loss": 3.5309,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009531538935183251,
+      "loss": 3.5318,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009530547027845885,
+      "loss": 3.5455,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009529554123223053,
+      "loss": 3.5305,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009528560221533318,
+      "loss": 3.4937,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009527565322995459,
+      "loss": 3.5614,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009526569427828482,
+      "loss": 3.3454,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009525572536251606,
+      "loss": 3.6448,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009524574648484272,
+      "loss": 3.6655,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009523575764746138,
+      "loss": 3.4615,
+      "step": 1151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009522575885257085,
+      "loss": 3.4586,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009521575010237213,
+      "loss": 3.4774,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009520573139906835,
+      "loss": 3.3822,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000951957027448649,
+      "loss": 3.3402,
+      "step": 1155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009518566414196933,
+      "loss": 3.5972,
+      "step": 1156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009517561559259138,
+      "loss": 3.4594,
+      "step": 1157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009516555709894299,
+      "loss": 3.3494,
+      "step": 1158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009515548866323829,
+      "loss": 3.4584,
+      "step": 1159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009514541028769359,
+      "loss": 3.4955,
+      "step": 1160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009513532197452737,
+      "loss": 3.442,
+      "step": 1161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009512522372596033,
+      "loss": 3.4369,
+      "step": 1162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009511511554421536,
+      "loss": 3.3733,
+      "step": 1163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000951049974315175,
+      "loss": 3.5298,
+      "step": 1164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009509486939009399,
+      "loss": 3.406,
+      "step": 1165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009508473142217429,
+      "loss": 3.4008,
+      "step": 1166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009507458352999001,
+      "loss": 3.5708,
+      "step": 1167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009506442571577494,
+      "loss": 3.401,
+      "step": 1168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009505425798176506,
+      "loss": 3.5604,
+      "step": 1169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009504408033019856,
+      "loss": 3.4011,
+      "step": 1170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009503389276331578,
+      "loss": 3.4842,
+      "step": 1171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009502369528335925,
+      "loss": 3.4707,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009501348789257373,
+      "loss": 3.426,
+      "step": 1173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009500327059320606,
+      "loss": 3.4284,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009499304338750535,
+      "loss": 3.4284,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009498280627772286,
+      "loss": 3.3262,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009497255926611204,
+      "loss": 3.4692,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009496230235492848,
+      "loss": 3.6114,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009495203554643002,
+      "loss": 3.506,
+      "step": 1179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009494175884287659,
+      "loss": 3.4448,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000949314722465304,
+      "loss": 3.4518,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009492117575965576,
+      "loss": 3.452,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009491086938451916,
+      "loss": 3.4753,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009490055312338932,
+      "loss": 3.4168,
+      "step": 1184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009489022697853709,
+      "loss": 3.4701,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000948798909522355,
+      "loss": 3.3463,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000948695450467598,
+      "loss": 3.4065,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009485918926438735,
+      "loss": 3.648,
+      "step": 1188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009484882360739771,
+      "loss": 3.337,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009483844807807265,
+      "loss": 3.5844,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009482806267869604,
+      "loss": 3.5108,
+      "step": 1191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00094817667411554,
+      "loss": 3.5523,
+      "step": 1192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009480726227893474,
+      "loss": 3.4023,
+      "step": 1193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009479684728312873,
+      "loss": 3.4591,
+      "step": 1194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009478642242642856,
+      "loss": 3.4727,
+      "step": 1195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00094775987711129,
+      "loss": 3.5113,
+      "step": 1196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009476554313952697,
+      "loss": 3.527,
+      "step": 1197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009475508871392157,
+      "loss": 3.4104,
+      "step": 1198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009474462443661414,
+      "loss": 3.5622,
+      "step": 1199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009473415030990805,
+      "loss": 3.4857,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009472366633610896,
+      "loss": 3.3896,
+      "step": 1201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009471317251752464,
+      "loss": 3.4251,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009470266885646504,
+      "loss": 3.5522,
+      "step": 1203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009469215535524227,
+      "loss": 3.3946,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009468163201617061,
+      "loss": 3.3417,
+      "step": 1205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009467109884156653,
+      "loss": 3.5147,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009466055583374862,
+      "loss": 3.3857,
+      "step": 1207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009465000299503766,
+      "loss": 3.4718,
+      "step": 1208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009463944032775659,
+      "loss": 3.3237,
+      "step": 1209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009462886783423052,
+      "loss": 3.5157,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000946182855167867,
+      "loss": 3.4439,
+      "step": 1211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000946076933777546,
+      "loss": 3.3567,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009459709141946578,
+      "loss": 3.3312,
+      "step": 1213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00094586479644254,
+      "loss": 3.3966,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009457585805445517,
+      "loss": 3.3479,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009456522665240737,
+      "loss": 3.5239,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009455458544045084,
+      "loss": 3.4368,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009454393442092795,
+      "loss": 3.4658,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009453327359618329,
+      "loss": 3.4498,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009452260296856354,
+      "loss": 3.5219,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009451192254041758,
+      "loss": 3.3854,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009450123231409645,
+      "loss": 3.5392,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000944905322919533,
+      "loss": 3.4415,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000944798224763435,
+      "loss": 3.4649,
+      "step": 1224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009446910286962454,
+      "loss": 3.4086,
+      "step": 1225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009445837347415607,
+      "loss": 3.5799,
+      "step": 1226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009444763429229988,
+      "loss": 3.4461,
+      "step": 1227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009443688532641994,
+      "loss": 3.4643,
+      "step": 1228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009442612657888236,
+      "loss": 3.4334,
+      "step": 1229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009441535805205541,
+      "loss": 3.4699,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009440457974830952,
+      "loss": 3.4442,
+      "step": 1231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009439379167001725,
+      "loss": 3.4856,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009438299381955332,
+      "loss": 3.5447,
+      "step": 1233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009437218619929461,
+      "loss": 3.5375,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009436136881162014,
+      "loss": 3.4859,
+      "step": 1235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009435054165891108,
+      "loss": 3.5139,
+      "step": 1236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009433970474355078,
+      "loss": 3.431,
+      "step": 1237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009432885806792468,
+      "loss": 3.5373,
+      "step": 1238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009431800163442042,
+      "loss": 3.5694,
+      "step": 1239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009430713544542775,
+      "loss": 3.5188,
+      "step": 1240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009429625950333863,
+      "loss": 3.4209,
+      "step": 1241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009428537381054707,
+      "loss": 3.5914,
+      "step": 1242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009427447836944931,
+      "loss": 3.4998,
+      "step": 1243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009426357318244371,
+      "loss": 3.375,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009425265825193077,
+      "loss": 3.5494,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009424173358031313,
+      "loss": 3.4496,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009423079916999557,
+      "loss": 3.4364,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009421985502338503,
+      "loss": 3.3708,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000942089011428906,
+      "loss": 3.4589,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000941979375309235,
+      "loss": 3.5386,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009418696418989707,
+      "loss": 3.3466,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009417598112222684,
+      "loss": 3.4504,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009416498833033043,
+      "loss": 3.4788,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009415398581662766,
+      "loss": 3.5419,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009414297358354041,
+      "loss": 3.5052,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009413195163349278,
+      "loss": 3.5594,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009412091996891096,
+      "loss": 3.4582,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009410987859222328,
+      "loss": 3.5344,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009409882750586027,
+      "loss": 3.4305,
+      "step": 1259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009408776671225448,
+      "loss": 3.3858,
+      "step": 1260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009407669621384071,
+      "loss": 3.4205,
+      "step": 1261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009406561601305583,
+      "loss": 3.3421,
+      "step": 1262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009405452611233888,
+      "loss": 3.4428,
+      "step": 1263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009404342651413102,
+      "loss": 3.4048,
+      "step": 1264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009403231722087552,
+      "loss": 3.4089,
+      "step": 1265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009402119823501786,
+      "loss": 3.533,
+      "step": 1266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009401006955900555,
+      "loss": 3.4595,
+      "step": 1267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009399893119528832,
+      "loss": 3.5276,
+      "step": 1268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009398778314631801,
+      "loss": 3.4688,
+      "step": 1269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009397662541454854,
+      "loss": 3.4219,
+      "step": 1270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009396545800243603,
+      "loss": 3.515,
+      "step": 1271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000939542809124387,
+      "loss": 3.5304,
+      "step": 1272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009394309414701691,
+      "loss": 3.4831,
+      "step": 1273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009393189770863313,
+      "loss": 3.3917,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009392069159975199,
+      "loss": 3.384,
+      "step": 1275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000939094758228402,
+      "loss": 3.4211,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009389825038036665,
+      "loss": 3.3417,
+      "step": 1277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009388701527480237,
+      "loss": 3.3621,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009387577050862043,
+      "loss": 3.521,
+      "step": 1279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000938645160842961,
+      "loss": 3.4802,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009385325200430678,
+      "loss": 3.3666,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009384197827113195,
+      "loss": 3.4192,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009383069488725323,
+      "loss": 3.373,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009381940185515439,
+      "loss": 3.4617,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009380809917732131,
+      "loss": 3.4004,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009379678685624196,
+      "loss": 3.3708,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000937854648944065,
+      "loss": 3.3608,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009377413329430715,
+      "loss": 3.2966,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009376279205843827,
+      "loss": 3.3409,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009375144118929637,
+      "loss": 3.532,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009374008068938003,
+      "loss": 3.3268,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009372871056118998,
+      "loss": 3.4235,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009371733080722911,
+      "loss": 3.2505,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009370594143000233,
+      "loss": 3.3253,
+      "step": 1294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009369454243201676,
+      "loss": 3.4299,
+      "step": 1295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009368313381578157,
+      "loss": 3.4081,
+      "step": 1296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009367171558380812,
+      "loss": 3.4095,
+      "step": 1297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009366028773860979,
+      "loss": 3.4344,
+      "step": 1298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009364885028270219,
+      "loss": 3.4723,
+      "step": 1299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009363740321860294,
+      "loss": 3.5192,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009362594654883185,
+      "loss": 3.4867,
+      "step": 1301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000936144802759108,
+      "loss": 3.5584,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009360300440236383,
+      "loss": 3.4309,
+      "step": 1303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009359151893071701,
+      "loss": 3.4978,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009358002386349862,
+      "loss": 3.2717,
+      "step": 1305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009356851920323901,
+      "loss": 3.3344,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009355700495247061,
+      "loss": 3.5188,
+      "step": 1307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009354548111372802,
+      "loss": 3.4985,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000935339476895479,
+      "loss": 3.4764,
+      "step": 1309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009352240468246907,
+      "loss": 3.29,
+      "step": 1310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000935108520950324,
+      "loss": 3.3709,
+      "step": 1311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009349928992978093,
+      "loss": 3.4172,
+      "step": 1312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009348771818925974,
+      "loss": 3.3744,
+      "step": 1313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009347613687601611,
+      "loss": 3.4147,
+      "step": 1314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009346454599259931,
+      "loss": 3.421,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009345294554156084,
+      "loss": 3.4694,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000934413355254542,
+      "loss": 3.4997,
+      "step": 1317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009342971594683506,
+      "loss": 3.3398,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009341808680826118,
+      "loss": 3.4477,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009340644811229242,
+      "loss": 3.5037,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009339479986149075,
+      "loss": 3.5472,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009338314205842022,
+      "loss": 3.469,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009337147470564702,
+      "loss": 3.3621,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009335979780573941,
+      "loss": 3.3787,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009334811136126778,
+      "loss": 3.4651,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009333641537480459,
+      "loss": 3.4207,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000933247098489244,
+      "loss": 3.3333,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009331299478620394,
+      "loss": 3.427,
+      "step": 1328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009330127018922195,
+      "loss": 3.5332,
+      "step": 1329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009328953606055928,
+      "loss": 3.4831,
+      "step": 1330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009327779240279894,
+      "loss": 3.3182,
+      "step": 1331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009326603921852599,
+      "loss": 3.4366,
+      "step": 1332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009325427651032758,
+      "loss": 3.4651,
+      "step": 1333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009324250428079301,
+      "loss": 3.4332,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000932307225325136,
+      "loss": 3.3028,
+      "step": 1335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009321893126808283,
+      "loss": 3.4168,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009320713049009624,
+      "loss": 3.4024,
+      "step": 1337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009319532020115146,
+      "loss": 3.2594,
+      "step": 1338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009318350040384824,
+      "loss": 3.3419,
+      "step": 1339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009317167110078842,
+      "loss": 3.3499,
+      "step": 1340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000931598322945759,
+      "loss": 3.4765,
+      "step": 1341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000931479839878167,
+      "loss": 3.4306,
+      "step": 1342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009313612618311896,
+      "loss": 3.5119,
+      "step": 1343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009312425888309281,
+      "loss": 3.4875,
+      "step": 1344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000931123820903506,
+      "loss": 3.5151,
+      "step": 1345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009310049580750666,
+      "loss": 3.4533,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009308860003717749,
+      "loss": 3.4276,
+      "step": 1347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009307669478198162,
+      "loss": 3.444,
+      "step": 1348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009306478004453969,
+      "loss": 3.4755,
+      "step": 1349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009305285582747444,
+      "loss": 3.4241,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009304092213341068,
+      "loss": 3.3772,
+      "step": 1351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009302897896497532,
+      "loss": 3.3596,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009301702632479734,
+      "loss": 3.5062,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009300506421550781,
+      "loss": 3.3361,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009299309263973987,
+      "loss": 3.5178,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000929811116001288,
+      "loss": 3.346,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009296912109931188,
+      "loss": 3.4259,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009295712113992856,
+      "loss": 3.5577,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000929451117246203,
+      "loss": 3.2834,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009293309285603067,
+      "loss": 3.4307,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009292106453680531,
+      "loss": 3.2813,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009290902676959197,
+      "loss": 3.2926,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009289697955704047,
+      "loss": 3.3531,
+      "step": 1363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009288492290180266,
+      "loss": 3.4795,
+      "step": 1364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009287285680653254,
+      "loss": 3.399,
+      "step": 1365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009286078127388617,
+      "loss": 3.3518,
+      "step": 1366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009284869630652164,
+      "loss": 3.4574,
+      "step": 1367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009283660190709914,
+      "loss": 3.5092,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00092824498078281,
+      "loss": 3.5326,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009281238482273153,
+      "loss": 3.3758,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009280026214311716,
+      "loss": 3.402,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000927881300421064,
+      "loss": 3.3841,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009277598852236984,
+      "loss": 3.4685,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000927638375865801,
+      "loss": 3.3931,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009275167723741193,
+      "loss": 3.4629,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009273950747754209,
+      "loss": 3.5442,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009272732830964948,
+      "loss": 3.3796,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009271513973641501,
+      "loss": 3.4243,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000927029417605217,
+      "loss": 3.333,
+      "step": 1379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009269073438465462,
+      "loss": 3.3831,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009267851761150093,
+      "loss": 3.371,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000926662914437498,
+      "loss": 3.3735,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009265405588409257,
+      "loss": 3.5322,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009264181093522257,
+      "loss": 3.4954,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009262955659983521,
+      "loss": 3.3605,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009261729288062795,
+      "loss": 3.4737,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009260501978030038,
+      "loss": 3.3139,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009259273730155409,
+      "loss": 3.4509,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009258044544709274,
+      "loss": 3.4176,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009256814421962212,
+      "loss": 3.4684,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009255583362184998,
+      "loss": 3.4119,
+      "step": 1391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009254351365648623,
+      "loss": 3.4409,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009253118432624279,
+      "loss": 3.4422,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009251884563383362,
+      "loss": 3.4088,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009250649758197481,
+      "loss": 3.4316,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009249414017338444,
+      "loss": 3.5577,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009248177341078272,
+      "loss": 3.3473,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009246939729689185,
+      "loss": 3.3608,
+      "step": 1398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009245701183443612,
+      "loss": 3.4354,
+      "step": 1399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009244461702614189,
+      "loss": 3.4208,
+      "step": 1400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009243221287473755,
+      "loss": 3.3903,
+      "step": 1401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009241979938295358,
+      "loss": 3.5084,
+      "step": 1402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009240737655352247,
+      "loss": 3.4489,
+      "step": 1403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000923949443891788,
+      "loss": 3.4788,
+      "step": 1404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009238250289265921,
+      "loss": 3.4446,
+      "step": 1405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009237005206670238,
+      "loss": 3.2476,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009235759191404904,
+      "loss": 3.4619,
+      "step": 1407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009234512243744197,
+      "loss": 3.3443,
+      "step": 1408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009233264363962601,
+      "loss": 3.458,
+      "step": 1409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009232015552334807,
+      "loss": 3.4374,
+      "step": 1410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009230765809135708,
+      "loss": 3.4712,
+      "step": 1411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009229515134640405,
+      "loss": 3.4655,
+      "step": 1412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009228263529124198,
+      "loss": 3.3235,
+      "step": 1413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009227010992862603,
+      "loss": 3.4861,
+      "step": 1414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009225757526131328,
+      "loss": 3.4776,
+      "step": 1415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009224503129206294,
+      "loss": 3.4416,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009223247802363628,
+      "loss": 3.414,
+      "step": 1417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009221991545879653,
+      "loss": 3.3897,
+      "step": 1418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009220734360030907,
+      "loss": 3.4098,
+      "step": 1419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009219476245094125,
+      "loss": 3.3968,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009218217201346251,
+      "loss": 3.4156,
+      "step": 1421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009216957229064429,
+      "loss": 3.4565,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009215696328526012,
+      "loss": 3.3153,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009214434500008555,
+      "loss": 3.4571,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009213171743789819,
+      "loss": 3.3284,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009211908060147764,
+      "loss": 3.4729,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009210643449360563,
+      "loss": 3.4779,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009209377911706584,
+      "loss": 3.3601,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009208111447464406,
+      "loss": 3.5096,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009206844056912809,
+      "loss": 3.4168,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009205575740330774,
+      "loss": 3.5034,
+      "step": 1431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009204306497997492,
+      "loss": 3.4343,
+      "step": 1432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009203036330192354,
+      "loss": 3.3579,
+      "step": 1433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009201765237194954,
+      "loss": 3.3178,
+      "step": 1434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009200493219285091,
+      "loss": 3.5167,
+      "step": 1435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000919922027674277,
+      "loss": 3.3586,
+      "step": 1436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009197946409848195,
+      "loss": 3.3971,
+      "step": 1437
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009196671618881775,
+      "loss": 3.4838,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009195395904124127,
+      "loss": 3.3883,
+      "step": 1439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009194119265856063,
+      "loss": 3.3505,
+      "step": 1440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009192841704358603,
+      "loss": 3.4158,
+      "step": 1441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009191563219912973,
+      "loss": 3.3792,
+      "step": 1442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009190283812800596,
+      "loss": 3.3501,
+      "step": 1443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009189003483303103,
+      "loss": 3.4569,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009187722231702326,
+      "loss": 3.3888,
+      "step": 1445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009186440058280298,
+      "loss": 3.3838,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009185156963319261,
+      "loss": 3.4199,
+      "step": 1447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009183872947101652,
+      "loss": 3.4749,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009182588009910118,
+      "loss": 3.3924,
+      "step": 1449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009181302152027503,
+      "loss": 3.4622,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009180015373736856,
+      "loss": 3.3308,
+      "step": 1451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009178727675321432,
+      "loss": 3.4488,
+      "step": 1452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009177439057064682,
+      "loss": 3.3675,
+      "step": 1453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009176149519250264,
+      "loss": 3.437,
+      "step": 1454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009174859062162038,
+      "loss": 3.4502,
+      "step": 1455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009173567686084063,
+      "loss": 3.4229,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009172275391300604,
+      "loss": 3.5385,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009170982178096128,
+      "loss": 3.4238,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009169688046755302,
+      "loss": 3.3664,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009168392997562999,
+      "loss": 3.5207,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009167097030804288,
+      "loss": 3.3588,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009165800146764445,
+      "loss": 3.5378,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009164502345728945,
+      "loss": 3.3924,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009163203627983468,
+      "loss": 3.472,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009161903993813892,
+      "loss": 3.4022,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009160603443506299,
+      "loss": 3.404,
+      "step": 1466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009159301977346975,
+      "loss": 3.314,
+      "step": 1467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009157999595622399,
+      "loss": 3.2916,
+      "step": 1468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009156696298619264,
+      "loss": 3.3927,
+      "step": 1469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009155392086624454,
+      "loss": 3.3874,
+      "step": 1470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009154086959925061,
+      "loss": 3.5102,
+      "step": 1471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009152780918808372,
+      "loss": 3.4116,
+      "step": 1472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009151473963561883,
+      "loss": 3.5719,
+      "step": 1473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009150166094473283,
+      "loss": 3.4166,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000914885731183047,
+      "loss": 3.4619,
+      "step": 1475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009147547615921536,
+      "loss": 3.4612,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009146237007034781,
+      "loss": 3.5615,
+      "step": 1477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009144925485458701,
+      "loss": 3.4768,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009143613051481994,
+      "loss": 3.3689,
+      "step": 1479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009142299705393559,
+      "loss": 3.4848,
+      "step": 1480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009140985447482497,
+      "loss": 3.479,
+      "step": 1481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009139670278038108,
+      "loss": 3.27,
+      "step": 1482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009138354197349892,
+      "loss": 3.5017,
+      "step": 1483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009137037205707553,
+      "loss": 3.4301,
+      "step": 1484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009135719303400993,
+      "loss": 3.51,
+      "step": 1485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009134400490720313,
+      "loss": 3.493,
+      "step": 1486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000913308076795582,
+      "loss": 3.3665,
+      "step": 1487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009131760135398015,
+      "loss": 3.2569,
+      "step": 1488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009130438593337602,
+      "loss": 3.4917,
+      "step": 1489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009129116142065486,
+      "loss": 3.3882,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009127792781872769,
+      "loss": 3.4547,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009126468513050758,
+      "loss": 3.5479,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009125143335890957,
+      "loss": 3.326,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000912381725068507,
+      "loss": 3.3763,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009122490257724999,
+      "loss": 3.4186,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000912116235730285,
+      "loss": 3.3406,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009119833549710928,
+      "loss": 3.4783,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009118503835241734,
+      "loss": 3.3811,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009117173214187971,
+      "loss": 3.4911,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009115841686842544,
+      "loss": 3.4621,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009114509253498553,
+      "loss": 3.4161,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009113175914449301,
+      "loss": 3.4388,
+      "step": 1502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009111841669988288,
+      "loss": 3.4246,
+      "step": 1503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009110506520409213,
+      "loss": 3.3159,
+      "step": 1504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009109170466005979,
+      "loss": 3.5645,
+      "step": 1505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009107833507072679,
+      "loss": 3.3976,
+      "step": 1506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009106495643903616,
+      "loss": 3.3902,
+      "step": 1507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009105156876793286,
+      "loss": 3.2418,
+      "step": 1508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009103817206036382,
+      "loss": 3.4133,
+      "step": 1509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009102476631927801,
+      "loss": 3.3553,
+      "step": 1510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009101135154762635,
+      "loss": 3.4636,
+      "step": 1511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009099792774836178,
+      "loss": 3.3586,
+      "step": 1512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009098449492443921,
+      "loss": 3.3726,
+      "step": 1513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009097105307881552,
+      "loss": 3.3504,
+      "step": 1514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009095760221444959,
+      "loss": 3.3367,
+      "step": 1515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000909441423343023,
+      "loss": 3.4277,
+      "step": 1516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009093067344133652,
+      "loss": 3.4782,
+      "step": 1517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009091719553851706,
+      "loss": 3.3788,
+      "step": 1518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009090370862881074,
+      "loss": 3.4783,
+      "step": 1519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009089021271518639,
+      "loss": 3.2593,
+      "step": 1520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009087670780061475,
+      "loss": 3.4685,
+      "step": 1521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009086319388806863,
+      "loss": 3.448,
+      "step": 1522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009084967098052275,
+      "loss": 3.3183,
+      "step": 1523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009083613908095385,
+      "loss": 3.4089,
+      "step": 1524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009082259819234062,
+      "loss": 3.3306,
+      "step": 1525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009080904831766375,
+      "loss": 3.485,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009079548945990592,
+      "loss": 3.442,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009078192162205174,
+      "loss": 3.4291,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009076834480708786,
+      "loss": 3.4991,
+      "step": 1529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009075475901800285,
+      "loss": 3.3433,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009074116425778729,
+      "loss": 3.4155,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000907275605294337,
+      "loss": 3.4234,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009071394783593663,
+      "loss": 3.4906,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009070032618029255,
+      "loss": 3.4373,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009068669556549993,
+      "loss": 3.3876,
+      "step": 1535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000906730559945592,
+      "loss": 3.3738,
+      "step": 1536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009065940747047277,
+      "loss": 3.4782,
+      "step": 1537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009064574999624502,
+      "loss": 3.4552,
+      "step": 1538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000906320835748823,
+      "loss": 3.5532,
+      "step": 1539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009061840820939291,
+      "loss": 3.4128,
+      "step": 1540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009060472390278717,
+      "loss": 3.4418,
+      "step": 1541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000905910306580773,
+      "loss": 3.4463,
+      "step": 1542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009057732847827754,
+      "loss": 3.5065,
+      "step": 1543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009056361736640407,
+      "loss": 3.3871,
+      "step": 1544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009054989732547506,
+      "loss": 3.3726,
+      "step": 1545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009053616835851062,
+      "loss": 3.3427,
+      "step": 1546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009052243046853283,
+      "loss": 3.3802,
+      "step": 1547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009050868365856575,
+      "loss": 3.4166,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009049492793163538,
+      "loss": 3.3535,
+      "step": 1549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000904811632907697,
+      "loss": 3.3599,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009046738973899866,
+      "loss": 3.5453,
+      "step": 1551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009045360727935414,
+      "loss": 3.3058,
+      "step": 1552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009043981591487002,
+      "loss": 3.3802,
+      "step": 1553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009042601564858213,
+      "loss": 3.5476,
+      "step": 1554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009041220648352819,
+      "loss": 3.4963,
+      "step": 1555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009039838842274801,
+      "loss": 3.3665,
+      "step": 1556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009038456146928325,
+      "loss": 3.3512,
+      "step": 1557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009037072562617757,
+      "loss": 3.4731,
+      "step": 1558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009035688089647659,
+      "loss": 3.386,
+      "step": 1559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009034302728322787,
+      "loss": 3.2112,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009032916478948092,
+      "loss": 3.4905,
+      "step": 1561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009031529341828724,
+      "loss": 3.3176,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009030141317270026,
+      "loss": 3.3519,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009028752405577535,
+      "loss": 3.4219,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009027362607056986,
+      "loss": 3.3804,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009025971922014307,
+      "loss": 3.3938,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009024580350755623,
+      "loss": 3.3151,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009023187893587255,
+      "loss": 3.5547,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009021794550815713,
+      "loss": 3.4192,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000902040032274771,
+      "loss": 3.392,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009019005209690147,
+      "loss": 3.4099,
+      "step": 1571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009017609211950127,
+      "loss": 3.2996,
+      "step": 1572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000901621232983494,
+      "loss": 3.4289,
+      "step": 1573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009014814563652077,
+      "loss": 3.3132,
+      "step": 1574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009013415913709217,
+      "loss": 3.3513,
+      "step": 1575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000901201638031424,
+      "loss": 3.5028,
+      "step": 1576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009010615963775219,
+      "loss": 3.3635,
+      "step": 1577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009009214664400419,
+      "loss": 3.3513,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009007812482498301,
+      "loss": 3.4214,
+      "step": 1579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000900640941837752,
+      "loss": 3.3444,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009005005472346923,
+      "loss": 3.3484,
+      "step": 1581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009003600644715557,
+      "loss": 3.5227,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009002194935792656,
+      "loss": 3.3583,
+      "step": 1583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009000788345887654,
+      "loss": 3.2395,
+      "step": 1584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008999380875310175,
+      "loss": 3.434,
+      "step": 1585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008997972524370037,
+      "loss": 3.3589,
+      "step": 1586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008996563293377254,
+      "loss": 3.3679,
+      "step": 1587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008995153182642032,
+      "loss": 3.3344,
+      "step": 1588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008993742192474771,
+      "loss": 3.3964,
+      "step": 1589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008992330323186068,
+      "loss": 3.4193,
+      "step": 1590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008990917575086708,
+      "loss": 3.4544,
+      "step": 1591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000898950394848767,
+      "loss": 3.4655,
+      "step": 1592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008988089443700131,
+      "loss": 3.4745,
+      "step": 1593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008986674061035456,
+      "loss": 3.4333,
+      "step": 1594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008985257800805209,
+      "loss": 3.3412,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008983840663321141,
+      "loss": 3.3424,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00089824226488952,
+      "loss": 3.32,
+      "step": 1597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008981003757839527,
+      "loss": 3.4338,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008979583990466454,
+      "loss": 3.2909,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008978163347088507,
+      "loss": 3.3318,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008976741828018402,
+      "loss": 3.3918,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008975319433569055,
+      "loss": 3.3622,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008973896164053569,
+      "loss": 3.4138,
+      "step": 1603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008972472019785243,
+      "loss": 3.3027,
+      "step": 1604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000897104700107756,
+      "loss": 3.4346,
+      "step": 1605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008969621108244208,
+      "loss": 3.4899,
+      "step": 1606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008968194341599056,
+      "loss": 3.3582,
+      "step": 1607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008966766701456176,
+      "loss": 3.2857,
+      "step": 1608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008965338188129824,
+      "loss": 3.4243,
+      "step": 1609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008963908801934451,
+      "loss": 3.3265,
+      "step": 1610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00089624785431847,
+      "loss": 3.3795,
+      "step": 1611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008961047412195409,
+      "loss": 3.3041,
+      "step": 1612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008959615409281603,
+      "loss": 3.3915,
+      "step": 1613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008958182534758501,
+      "loss": 3.4096,
+      "step": 1614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008956748788941514,
+      "loss": 3.441,
+      "step": 1615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008955314172146247,
+      "loss": 3.4681,
+      "step": 1616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008953878684688492,
+      "loss": 3.4491,
+      "step": 1617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008952442326884239,
+      "loss": 3.3392,
+      "step": 1618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008951005099049663,
+      "loss": 3.4295,
+      "step": 1619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008949567001501134,
+      "loss": 3.3247,
+      "step": 1620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008948128034555211,
+      "loss": 3.3867,
+      "step": 1621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000894668819852865,
+      "loss": 3.3997,
+      "step": 1622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008945247493738391,
+      "loss": 3.3472,
+      "step": 1623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008943805920501571,
+      "loss": 3.3823,
+      "step": 1624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008942363479135516,
+      "loss": 3.453,
+      "step": 1625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000894092016995774,
+      "loss": 3.4222,
+      "step": 1626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008939475993285956,
+      "loss": 3.2948,
+      "step": 1627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008938030949438059,
+      "loss": 3.3695,
+      "step": 1628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008936585038732142,
+      "loss": 3.2836,
+      "step": 1629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008935138261486482,
+      "loss": 3.4491,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008933690618019552,
+      "loss": 3.3722,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008932242108650015,
+      "loss": 3.4862,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008930792733696725,
+      "loss": 3.3029,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008929342493478721,
+      "loss": 3.3983,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008927891388315241,
+      "loss": 3.296,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008926439418525707,
+      "loss": 3.3845,
+      "step": 1636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008924986584429732,
+      "loss": 3.4169,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008923532886347125,
+      "loss": 3.4701,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008922078324597878,
+      "loss": 3.3467,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008920622899502178,
+      "loss": 3.4706,
+      "step": 1640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008919166611380396,
+      "loss": 3.3655,
+      "step": 1641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008917709460553101,
+      "loss": 3.3584,
+      "step": 1642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008916251447341046,
+      "loss": 3.516,
+      "step": 1643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008914792572065178,
+      "loss": 3.4284,
+      "step": 1644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008913332835046628,
+      "loss": 3.3717,
+      "step": 1645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008911872236606723,
+      "loss": 3.3326,
+      "step": 1646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008910410777066975,
+      "loss": 3.276,
+      "step": 1647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008908948456749089,
+      "loss": 3.4644,
+      "step": 1648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008907485275974958,
+      "loss": 3.4554,
+      "step": 1649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008906021235066664,
+      "loss": 3.4571,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008904556334346477,
+      "loss": 3.3694,
+      "step": 1651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008903090574136858,
+      "loss": 3.3984,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008901623954760459,
+      "loss": 3.3359,
+      "step": 1653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008900156476540119,
+      "loss": 3.2889,
+      "step": 1654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008898688139798865,
+      "loss": 3.3656,
+      "step": 1655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008897218944859915,
+      "loss": 3.2288,
+      "step": 1656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008895748892046674,
+      "loss": 3.3871,
+      "step": 1657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008894277981682739,
+      "loss": 3.3536,
+      "step": 1658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008892806214091893,
+      "loss": 3.5329,
+      "step": 1659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008891333589598107,
+      "loss": 3.4646,
+      "step": 1660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008889860108525543,
+      "loss": 3.3202,
+      "step": 1661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008888385771198552,
+      "loss": 3.4292,
+      "step": 1662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008886910577941672,
+      "loss": 3.3812,
+      "step": 1663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008885434529079625,
+      "loss": 3.3298,
+      "step": 1664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008883957624937332,
+      "loss": 3.3088,
+      "step": 1665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008882479865839893,
+      "loss": 3.4496,
+      "step": 1666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008881001252112599,
+      "loss": 3.2832,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008879521784080932,
+      "loss": 3.4611,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008878041462070556,
+      "loss": 3.3892,
+      "step": 1669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008876560286407329,
+      "loss": 3.3901,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008875078257417294,
+      "loss": 3.4996,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008873595375426681,
+      "loss": 3.4846,
+      "step": 1672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008872111640761911,
+      "loss": 3.3232,
+      "step": 1673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000887062705374959,
+      "loss": 3.4312,
+      "step": 1674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008869141614716511,
+      "loss": 3.3542,
+      "step": 1675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008867655323989656,
+      "loss": 3.3957,
+      "step": 1676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008866168181896197,
+      "loss": 3.4165,
+      "step": 1677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008864680188763488,
+      "loss": 3.4353,
+      "step": 1678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008863191344919074,
+      "loss": 3.4652,
+      "step": 1679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008861701650690686,
+      "loss": 3.424,
+      "step": 1680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008860211106406244,
+      "loss": 3.3843,
+      "step": 1681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008858719712393851,
+      "loss": 3.3123,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008857227468981804,
+      "loss": 3.4461,
+      "step": 1683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008855734376498576,
+      "loss": 3.4767,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008854240435272841,
+      "loss": 3.3446,
+      "step": 1685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008852745645633448,
+      "loss": 3.4318,
+      "step": 1686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008851250007909439,
+      "loss": 3.2335,
+      "step": 1687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008849753522430037,
+      "loss": 3.3963,
+      "step": 1688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008848256189524662,
+      "loss": 3.4311,
+      "step": 1689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008846758009522908,
+      "loss": 3.4011,
+      "step": 1690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008845258982754566,
+      "loss": 3.4417,
+      "step": 1691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008843759109549606,
+      "loss": 3.4263,
+      "step": 1692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008842258390238188,
+      "loss": 3.3227,
+      "step": 1693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008840756825150657,
+      "loss": 3.4103,
+      "step": 1694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008839254414617546,
+      "loss": 3.331,
+      "step": 1695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008837751158969571,
+      "loss": 3.417,
+      "step": 1696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008836247058537638,
+      "loss": 3.3434,
+      "step": 1697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008834742113652834,
+      "loss": 3.3571,
+      "step": 1698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008833236324646437,
+      "loss": 3.3045,
+      "step": 1699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008831729691849906,
+      "loss": 3.4677,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000883022221559489,
+      "loss": 3.288,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008828713896213222,
+      "loss": 3.4299,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008827204734036919,
+      "loss": 3.3239,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008825694729398186,
+      "loss": 3.5415,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000882418388262941,
+      "loss": 3.4223,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008822672194063171,
+      "loss": 3.378,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008821159664032224,
+      "loss": 3.4145,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008819646292869515,
+      "loss": 3.3313,
+      "step": 1708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008818132080908176,
+      "loss": 3.4405,
+      "step": 1709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008816617028481523,
+      "loss": 3.3754,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008815101135923055,
+      "loss": 3.4556,
+      "step": 1711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000881358440356646,
+      "loss": 3.376,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008812066831745602,
+      "loss": 3.5598,
+      "step": 1713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008810548420794543,
+      "loss": 3.3608,
+      "step": 1714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000880902917104752,
+      "loss": 3.3658,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008807509082838957,
+      "loss": 3.3063,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008805988156503464,
+      "loss": 3.3708,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008804466392375833,
+      "loss": 3.3033,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008802943790791045,
+      "loss": 3.3869,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008801420352084259,
+      "loss": 3.4691,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008799896076590823,
+      "loss": 3.4929,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008798370964646268,
+      "loss": 3.3101,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008796845016586308,
+      "loss": 3.3632,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008795318232746842,
+      "loss": 3.4584,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008793790613463954,
+      "loss": 3.519,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008792262159073911,
+      "loss": 3.4006,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008790732869913161,
+      "loss": 3.3995,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008789202746318341,
+      "loss": 3.3605,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008787671788626268,
+      "loss": 3.3151,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008786139997173944,
+      "loss": 3.3407,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008784607372298552,
+      "loss": 3.4603,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008783073914337465,
+      "loss": 3.3278,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008781539623628232,
+      "loss": 3.3586,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008780004500508588,
+      "loss": 3.3956,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008778468545316454,
+      "loss": 3.4038,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008776931758389928,
+      "loss": 3.3723,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008775394140067299,
+      "loss": 3.352,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008773855690687031,
+      "loss": 3.5563,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008772316410587779,
+      "loss": 3.359,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008770776300108373,
+      "loss": 3.4804,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000876923535958783,
+      "loss": 3.3439,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008767693589365352,
+      "loss": 3.4438,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008766150989780317,
+      "loss": 3.3398,
+      "step": 1743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008764607561172292,
+      "loss": 3.3372,
+      "step": 1744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008763063303881022,
+      "loss": 3.4553,
+      "step": 1745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008761518218246436,
+      "loss": 3.4436,
+      "step": 1746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008759972304608648,
+      "loss": 3.4692,
+      "step": 1747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008758425563307951,
+      "loss": 3.2915,
+      "step": 1748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008756877994684818,
+      "loss": 3.4361,
+      "step": 1749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008755329599079911,
+      "loss": 3.2118,
+      "step": 1750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008753780376834069,
+      "loss": 3.6013,
+      "step": 1751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008752230328288313,
+      "loss": 3.4064,
+      "step": 1752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008750679453783847,
+      "loss": 3.3342,
+      "step": 1753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008749127753662059,
+      "loss": 3.2794,
+      "step": 1754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008747575228264515,
+      "loss": 3.39,
+      "step": 1755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008746021877932965,
+      "loss": 3.4412,
+      "step": 1756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008744467703009339,
+      "loss": 3.4493,
+      "step": 1757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008742912703835749,
+      "loss": 3.307,
+      "step": 1758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000874135688075449,
+      "loss": 3.3551,
+      "step": 1759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008739800234108036,
+      "loss": 3.4491,
+      "step": 1760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008738242764239046,
+      "loss": 3.4067,
+      "step": 1761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008736684471490352,
+      "loss": 3.2745,
+      "step": 1762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000873512535620498,
+      "loss": 3.3658,
+      "step": 1763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008733565418726125,
+      "loss": 3.4044,
+      "step": 1764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008732004659397169,
+      "loss": 3.3933,
+      "step": 1765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008730443078561674,
+      "loss": 3.3054,
+      "step": 1766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008728880676563383,
+      "loss": 3.4028,
+      "step": 1767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008727317453746218,
+      "loss": 3.4266,
+      "step": 1768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008725753410454284,
+      "loss": 3.4602,
+      "step": 1769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008724188547031866,
+      "loss": 3.2936,
+      "step": 1770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008722622863823427,
+      "loss": 3.3524,
+      "step": 1771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008721056361173616,
+      "loss": 3.3057,
+      "step": 1772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008719489039427256,
+      "loss": 3.3193,
+      "step": 1773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008717920898929355,
+      "loss": 3.3594,
+      "step": 1774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008716351940025097,
+      "loss": 3.3979,
+      "step": 1775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008714782163059852,
+      "loss": 3.3464,
+      "step": 1776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008713211568379166,
+      "loss": 3.3244,
+      "step": 1777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008711640156328764,
+      "loss": 3.2849,
+      "step": 1778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008710067927254555,
+      "loss": 3.3964,
+      "step": 1779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008708494881502622,
+      "loss": 3.3725,
+      "step": 1780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008706921019419236,
+      "loss": 3.4193,
+      "step": 1781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008705346341350839,
+      "loss": 3.5048,
+      "step": 1782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008703770847644058,
+      "loss": 3.3834,
+      "step": 1783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008702194538645698,
+      "loss": 3.3296,
+      "step": 1784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008700617414702745,
+      "loss": 3.5018,
+      "step": 1785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000869903947616236,
+      "loss": 3.3917,
+      "step": 1786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008697460723371891,
+      "loss": 3.4325,
+      "step": 1787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008695881156678856,
+      "loss": 3.2443,
+      "step": 1788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008694300776430958,
+      "loss": 3.4001,
+      "step": 1789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008692719582976078,
+      "loss": 3.5383,
+      "step": 1790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008691137576662275,
+      "loss": 3.4524,
+      "step": 1791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008689554757837792,
+      "loss": 3.3572,
+      "step": 1792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008687971126851041,
+      "loss": 3.4432,
+      "step": 1793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000868638668405062,
+      "loss": 3.5079,
+      "step": 1794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008684801429785305,
+      "loss": 3.3333,
+      "step": 1795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008683215364404051,
+      "loss": 3.291,
+      "step": 1796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008681628488255986,
+      "loss": 3.4844,
+      "step": 1797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008680040801690425,
+      "loss": 3.4243,
+      "step": 1798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008678452305056854,
+      "loss": 3.236,
+      "step": 1799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008676862998704944,
+      "loss": 3.2824,
+      "step": 1800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008675272882984536,
+      "loss": 3.4751,
+      "step": 1801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008673681958245657,
+      "loss": 3.3383,
+      "step": 1802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008672090224838508,
+      "loss": 3.3328,
+      "step": 1803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000867049768311347,
+      "loss": 3.4494,
+      "step": 1804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008668904333421097,
+      "loss": 3.3132,
+      "step": 1805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000866731017611213,
+      "loss": 3.292,
+      "step": 1806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000866571521153748,
+      "loss": 3.278,
+      "step": 1807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008664119440048236,
+      "loss": 3.3565,
+      "step": 1808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000866252286199567,
+      "loss": 3.3865,
+      "step": 1809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008660925477731226,
+      "loss": 3.2417,
+      "step": 1810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008659327287606531,
+      "loss": 3.4384,
+      "step": 1811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008657728291973383,
+      "loss": 3.5091,
+      "step": 1812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008656128491183762,
+      "loss": 3.323,
+      "step": 1813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008654527885589823,
+      "loss": 3.395,
+      "step": 1814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008652926475543898,
+      "loss": 3.2598,
+      "step": 1815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00086513242613985,
+      "loss": 3.5224,
+      "step": 1816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008649721243506313,
+      "loss": 3.4678,
+      "step": 1817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008648117422220203,
+      "loss": 3.3788,
+      "step": 1818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008646512797893209,
+      "loss": 3.3157,
+      "step": 1819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008644907370878549,
+      "loss": 3.3316,
+      "step": 1820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008643301141529619,
+      "loss": 3.5024,
+      "step": 1821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008641694110199987,
+      "loss": 3.4628,
+      "step": 1822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008640086277243402,
+      "loss": 3.403,
+      "step": 1823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008638477643013787,
+      "loss": 3.4095,
+      "step": 1824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008636868207865244,
+      "loss": 3.3465,
+      "step": 1825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008635257972152046,
+      "loss": 3.3619,
+      "step": 1826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008633646936228649,
+      "loss": 3.289,
+      "step": 1827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008632035100449682,
+      "loss": 3.1592,
+      "step": 1828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008630422465169947,
+      "loss": 3.439,
+      "step": 1829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008628809030744427,
+      "loss": 3.4329,
+      "step": 1830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008627194797528277,
+      "loss": 3.3465,
+      "step": 1831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008625579765876832,
+      "loss": 3.4276,
+      "step": 1832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00086239639361456,
+      "loss": 3.4684,
+      "step": 1833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008622347308690263,
+      "loss": 3.3125,
+      "step": 1834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008620729883866685,
+      "loss": 3.5461,
+      "step": 1835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008619111662030896,
+      "loss": 3.3789,
+      "step": 1836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008617492643539109,
+      "loss": 3.4293,
+      "step": 1837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008615872828747711,
+      "loss": 3.2324,
+      "step": 1838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008614252218013263,
+      "loss": 3.4566,
+      "step": 1839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00086126308116925,
+      "loss": 3.2924,
+      "step": 1840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008611008610142335,
+      "loss": 3.1984,
+      "step": 1841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008609385613719854,
+      "loss": 3.3983,
+      "step": 1842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008607761822782317,
+      "loss": 3.458,
+      "step": 1843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008606137237687165,
+      "loss": 3.295,
+      "step": 1844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008604511858792005,
+      "loss": 3.2751,
+      "step": 1845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008602885686454627,
+      "loss": 3.2126,
+      "step": 1846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008601258721032988,
+      "loss": 3.4474,
+      "step": 1847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008599630962885226,
+      "loss": 3.4366,
+      "step": 1848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008598002412369647,
+      "loss": 3.335,
+      "step": 1849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008596373069844741,
+      "loss": 3.3877,
+      "step": 1850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008594742935669164,
+      "loss": 3.3503,
+      "step": 1851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008593112010201747,
+      "loss": 3.3764,
+      "step": 1852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008591480293801499,
+      "loss": 3.3217,
+      "step": 1853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00085898477868276,
+      "loss": 3.3967,
+      "step": 1854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008588214489639407,
+      "loss": 3.3666,
+      "step": 1855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008586580402596447,
+      "loss": 3.4239,
+      "step": 1856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008584945526058425,
+      "loss": 3.3969,
+      "step": 1857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008583309860385216,
+      "loss": 3.3483,
+      "step": 1858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008581673405936872,
+      "loss": 3.3936,
+      "step": 1859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008580036163073615,
+      "loss": 3.4937,
+      "step": 1860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008578398132155845,
+      "loss": 3.2547,
+      "step": 1861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008576759313544131,
+      "loss": 3.3604,
+      "step": 1862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008575119707599219,
+      "loss": 3.2823,
+      "step": 1863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008573479314682026,
+      "loss": 3.4998,
+      "step": 1864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008571838135153644,
+      "loss": 3.1938,
+      "step": 1865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008570196169375337,
+      "loss": 3.282,
+      "step": 1866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008568553417708542,
+      "loss": 3.3254,
+      "step": 1867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000856690988051487,
+      "loss": 3.3979,
+      "step": 1868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008565265558156101,
+      "loss": 3.2725,
+      "step": 1869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008563620450994196,
+      "loss": 3.313,
+      "step": 1870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008561974559391279,
+      "loss": 3.2979,
+      "step": 1871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008560327883709656,
+      "loss": 3.4767,
+      "step": 1872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008558680424311796,
+      "loss": 3.3036,
+      "step": 1873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008557032181560351,
+      "loss": 3.4406,
+      "step": 1874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008555383155818135,
+      "loss": 3.4227,
+      "step": 1875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008553733347448143,
+      "loss": 3.3977,
+      "step": 1876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008552082756813537,
+      "loss": 3.3401,
+      "step": 1877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008550431384277652,
+      "loss": 3.2511,
+      "step": 1878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008548779230204,
+      "loss": 3.4007,
+      "step": 1879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008547126294956255,
+      "loss": 3.3354,
+      "step": 1880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008545472578898274,
+      "loss": 3.4504,
+      "step": 1881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008543818082394077,
+      "loss": 3.3961,
+      "step": 1882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008542162805807862,
+      "loss": 3.2174,
+      "step": 1883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008540506749503996,
+      "loss": 3.2711,
+      "step": 1884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008538849913847017,
+      "loss": 3.345,
+      "step": 1885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008537192299201637,
+      "loss": 3.2601,
+      "step": 1886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008535533905932737,
+      "loss": 3.4235,
+      "step": 1887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008533874734405371,
+      "loss": 3.4112,
+      "step": 1888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008532214784984762,
+      "loss": 3.3361,
+      "step": 1889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008530554058036308,
+      "loss": 3.3305,
+      "step": 1890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008528892553925576,
+      "loss": 3.4485,
+      "step": 1891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008527230273018303,
+      "loss": 3.3027,
+      "step": 1892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008525567215680398,
+      "loss": 3.2698,
+      "step": 1893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008523903382277941,
+      "loss": 3.4226,
+      "step": 1894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008522238773177184,
+      "loss": 3.3046,
+      "step": 1895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008520573388744548,
+      "loss": 3.3348,
+      "step": 1896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008518907229346626,
+      "loss": 3.3518,
+      "step": 1897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000851724029535018,
+      "loss": 3.4021,
+      "step": 1898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008515572587122143,
+      "loss": 3.3515,
+      "step": 1899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000851390410502962,
+      "loss": 3.3212,
+      "step": 1900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008512234849439886,
+      "loss": 3.4275,
+      "step": 1901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008510564820720383,
+      "loss": 3.2734,
+      "step": 1902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008508894019238726,
+      "loss": 3.5278,
+      "step": 1903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008507222445362701,
+      "loss": 3.5169,
+      "step": 1904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008505550099460264,
+      "loss": 3.468,
+      "step": 1905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008503876981899536,
+      "loss": 3.3539,
+      "step": 1906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008502203093048815,
+      "loss": 3.2993,
+      "step": 1907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008500528433276562,
+      "loss": 3.3819,
+      "step": 1908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008498853002951414,
+      "loss": 3.3609,
+      "step": 1909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008497176802442175,
+      "loss": 3.4236,
+      "step": 1910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008495499832117815,
+      "loss": 3.3728,
+      "step": 1911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000849382209234748,
+      "loss": 3.2567,
+      "step": 1912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008492143583500479,
+      "loss": 3.4208,
+      "step": 1913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008490464305946295,
+      "loss": 3.2458,
+      "step": 1914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008488784260054579,
+      "loss": 3.4283,
+      "step": 1915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000848710344619515,
+      "loss": 3.3848,
+      "step": 1916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008485421864737996,
+      "loss": 3.4151,
+      "step": 1917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008483739516053274,
+      "loss": 3.2856,
+      "step": 1918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008482056400511315,
+      "loss": 3.3593,
+      "step": 1919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000848037251848261,
+      "loss": 3.3285,
+      "step": 1920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008478687870337824,
+      "loss": 3.4107,
+      "step": 1921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008477002456447791,
+      "loss": 3.4317,
+      "step": 1922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008475316277183509,
+      "loss": 3.3094,
+      "step": 1923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008473629332916153,
+      "loss": 3.3118,
+      "step": 1924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008471941624017057,
+      "loss": 3.5174,
+      "step": 1925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008470253150857729,
+      "loss": 3.3851,
+      "step": 1926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008468563913809843,
+      "loss": 3.489,
+      "step": 1927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008466873913245243,
+      "loss": 3.4461,
+      "step": 1928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008465183149535939,
+      "loss": 3.2474,
+      "step": 1929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008463491623054109,
+      "loss": 3.3681,
+      "step": 1930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008461799334172104,
+      "loss": 3.3447,
+      "step": 1931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008460106283262432,
+      "loss": 3.3607,
+      "step": 1932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008458412470697781,
+      "loss": 3.5179,
+      "step": 1933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008456717896851,
+      "loss": 3.3276,
+      "step": 1934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008455022562095104,
+      "loss": 3.3412,
+      "step": 1935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008453326466803281,
+      "loss": 3.4028,
+      "step": 1936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008451629611348881,
+      "loss": 3.4494,
+      "step": 1937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008449931996105427,
+      "loss": 3.3164,
+      "step": 1938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008448233621446603,
+      "loss": 3.3388,
+      "step": 1939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008446534487746265,
+      "loss": 3.3252,
+      "step": 1940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008444834595378434,
+      "loss": 3.2484,
+      "step": 1941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008443133944717298,
+      "loss": 3.5031,
+      "step": 1942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008441432536137213,
+      "loss": 3.4053,
+      "step": 1943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008439730370012699,
+      "loss": 3.3728,
+      "step": 1944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008438027446718449,
+      "loss": 3.339,
+      "step": 1945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008436323766629314,
+      "loss": 3.2442,
+      "step": 1946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008434619330120319,
+      "loss": 3.3336,
+      "step": 1947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008432914137566651,
+      "loss": 3.2944,
+      "step": 1948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008431208189343669,
+      "loss": 3.2119,
+      "step": 1949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008429501485826889,
+      "loss": 3.3018,
+      "step": 1950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008427794027392002,
+      "loss": 3.3234,
+      "step": 1951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008426085814414859,
+      "loss": 3.2862,
+      "step": 1952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008424376847271482,
+      "loss": 3.3701,
+      "step": 1953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008422667126338057,
+      "loss": 3.3421,
+      "step": 1954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008420956651990937,
+      "loss": 3.3834,
+      "step": 1955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008419245424606638,
+      "loss": 3.3293,
+      "step": 1956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008417533444561843,
+      "loss": 3.3661,
+      "step": 1957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008415820712233402,
+      "loss": 3.2568,
+      "step": 1958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008414107227998329,
+      "loss": 3.364,
+      "step": 1959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008412392992233806,
+      "loss": 3.3625,
+      "step": 1960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008410678005317177,
+      "loss": 3.3368,
+      "step": 1961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008408962267625954,
+      "loss": 3.4848,
+      "step": 1962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008407245779537812,
+      "loss": 3.3013,
+      "step": 1963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008405528541430594,
+      "loss": 3.3423,
+      "step": 1964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008403810553682306,
+      "loss": 3.2173,
+      "step": 1965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008402091816671121,
+      "loss": 3.4899,
+      "step": 1966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008400372330775372,
+      "loss": 3.4638,
+      "step": 1967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008398652096373565,
+      "loss": 3.3639,
+      "step": 1968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008396931113844362,
+      "loss": 3.4823,
+      "step": 1969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008395209383566596,
+      "loss": 3.3618,
+      "step": 1970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000839348690591926,
+      "loss": 3.397,
+      "step": 1971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000839176368128152,
+      "loss": 3.2737,
+      "step": 1972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008390039710032692,
+      "loss": 3.3572,
+      "step": 1973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008388314992552271,
+      "loss": 3.2798,
+      "step": 1974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008386589529219908,
+      "loss": 3.3979,
+      "step": 1975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008384863320415419,
+      "loss": 3.4746,
+      "step": 1976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008383136366518787,
+      "loss": 3.4024,
+      "step": 1977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008381408667910157,
+      "loss": 3.4492,
+      "step": 1978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008379680224969837,
+      "loss": 3.3682,
+      "step": 1979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008377951038078302,
+      "loss": 3.2377,
+      "step": 1980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008376221107616186,
+      "loss": 3.3355,
+      "step": 1981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008374490433964292,
+      "loss": 3.4091,
+      "step": 1982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008372759017503584,
+      "loss": 3.3204,
+      "step": 1983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008371026858615188,
+      "loss": 3.3658,
+      "step": 1984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008369293957680396,
+      "loss": 3.2496,
+      "step": 1985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008367560315080663,
+      "loss": 3.3147,
+      "step": 1986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008365825931197603,
+      "loss": 3.3209,
+      "step": 1987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008364090806413003,
+      "loss": 3.3813,
+      "step": 1988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008362354941108802,
+      "loss": 3.2619,
+      "step": 1989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008360618335667108,
+      "loss": 3.4472,
+      "step": 1990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008358880990470192,
+      "loss": 3.2791,
+      "step": 1991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008357142905900485,
+      "loss": 3.5078,
+      "step": 1992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008355404082340585,
+      "loss": 3.2592,
+      "step": 1993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008353664520173248,
+      "loss": 3.3373,
+      "step": 1994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008351924219781393,
+      "loss": 3.544,
+      "step": 1995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008350183181548106,
+      "loss": 3.3556,
+      "step": 1996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008348441405856633,
+      "loss": 3.3404,
+      "step": 1997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008346698893090379,
+      "loss": 3.3787,
+      "step": 1998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008344955643632917,
+      "loss": 3.2588,
+      "step": 1999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008343211657867978,
+      "loss": 3.3134,
+      "step": 2000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008341466936179455,
+      "loss": 3.3613,
+      "step": 2001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008339721478951408,
+      "loss": 3.3776,
+      "step": 2002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008337975286568052,
+      "loss": 3.3187,
+      "step": 2003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008336228359413769,
+      "loss": 3.2765,
+      "step": 2004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00083344806978731,
+      "loss": 3.4083,
+      "step": 2005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008332732302330748,
+      "loss": 3.3402,
+      "step": 2006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000833098317317158,
+      "loss": 3.4671,
+      "step": 2007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008329233310780622,
+      "loss": 3.1967,
+      "step": 2008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000832748271554306,
+      "loss": 3.2754,
+      "step": 2009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008325731387844247,
+      "loss": 3.3834,
+      "step": 2010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008323979328069688,
+      "loss": 3.4381,
+      "step": 2011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008322226536605062,
+      "loss": 3.3755,
+      "step": 2012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008320473013836196,
+      "loss": 3.3647,
+      "step": 2013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008318718760149085,
+      "loss": 3.2551,
+      "step": 2014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008316963775929885,
+      "loss": 3.3266,
+      "step": 2015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008315208061564909,
+      "loss": 3.3726,
+      "step": 2016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008313451617440637,
+      "loss": 3.3969,
+      "step": 2017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008311694443943702,
+      "loss": 3.3384,
+      "step": 2018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008309936541460904,
+      "loss": 3.4492,
+      "step": 2019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008308177910379199,
+      "loss": 3.4375,
+      "step": 2020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008306418551085706,
+      "loss": 3.3988,
+      "step": 2021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008304658463967704,
+      "loss": 3.3608,
+      "step": 2022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008302897649412632,
+      "loss": 3.2904,
+      "step": 2023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008301136107808086,
+      "loss": 3.4792,
+      "step": 2024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008299373839541828,
+      "loss": 3.3866,
+      "step": 2025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008297610845001775,
+      "loss": 3.5166,
+      "step": 2026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008295847124576007,
+      "loss": 3.3389,
+      "step": 2027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008294082678652761,
+      "loss": 3.3418,
+      "step": 2028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008292317507620437,
+      "loss": 3.308,
+      "step": 2029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008290551611867593,
+      "loss": 3.399,
+      "step": 2030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008288784991782945,
+      "loss": 3.3189,
+      "step": 2031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008287017647755368,
+      "loss": 3.2981,
+      "step": 2032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008285249580173901,
+      "loss": 3.3544,
+      "step": 2033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000828348078942774,
+      "loss": 3.28,
+      "step": 2034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008281711275906237,
+      "loss": 3.2638,
+      "step": 2035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008279941039998906,
+      "loss": 3.3358,
+      "step": 2036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008278170082095422,
+      "loss": 3.3871,
+      "step": 2037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008276398402585614,
+      "loss": 3.4146,
+      "step": 2038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008274626001859475,
+      "loss": 3.3006,
+      "step": 2039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008272852880307154,
+      "loss": 3.41,
+      "step": 2040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008271079038318955,
+      "loss": 3.3944,
+      "step": 2041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008269304476285349,
+      "loss": 3.2374,
+      "step": 2042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008267529194596959,
+      "loss": 3.3536,
+      "step": 2043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008265753193644568,
+      "loss": 3.3419,
+      "step": 2044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000826397647381912,
+      "loss": 3.328,
+      "step": 2045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008262199035511713,
+      "loss": 3.4855,
+      "step": 2046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008260420879113605,
+      "loss": 3.412,
+      "step": 2047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008258642005016213,
+      "loss": 3.3747,
+      "step": 2048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008256862413611113,
+      "loss": 3.3767,
+      "step": 2049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008255082105290033,
+      "loss": 3.4244,
+      "step": 2050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008253301080444864,
+      "loss": 3.3287,
+      "step": 2051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008251519339467657,
+      "loss": 3.3744,
+      "step": 2052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008249736882750614,
+      "loss": 3.2847,
+      "step": 2053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008247953710686098,
+      "loss": 3.3765,
+      "step": 2054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000824616982366663,
+      "loss": 3.3201,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008244385222084889,
+      "loss": 3.3447,
+      "step": 2056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008242599906333707,
+      "loss": 3.4563,
+      "step": 2057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008240813876806079,
+      "loss": 3.38,
+      "step": 2058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008239027133895153,
+      "loss": 3.4187,
+      "step": 2059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008237239677994234,
+      "loss": 3.2578,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008235451509496788,
+      "loss": 3.3162,
+      "step": 2061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008233662628796435,
+      "loss": 3.5046,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008231873036286947,
+      "loss": 3.327,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008230082732362264,
+      "loss": 3.3191,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008228291717416471,
+      "loss": 3.3624,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000822649999184382,
+      "loss": 3.245,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008224707556038711,
+      "loss": 3.3217,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00082229144103957,
+      "loss": 3.3457,
+      "step": 2068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000822112055530951,
+      "loss": 3.3784,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008219325991175009,
+      "loss": 3.293,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008217530718387223,
+      "loss": 3.3497,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000821573473734134,
+      "loss": 3.4371,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008213938048432696,
+      "loss": 3.2978,
+      "step": 2073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000821214065205679,
+      "loss": 3.2809,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008210342548609273,
+      "loss": 3.2783,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000820854373848595,
+      "loss": 3.2221,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008206744222082785,
+      "loss": 3.3784,
+      "step": 2077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008204943999795896,
+      "loss": 3.4696,
+      "step": 2078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008203143072021555,
+      "loss": 3.3715,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008201341439156194,
+      "loss": 3.3256,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008199539101596393,
+      "loss": 3.386,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008197736059738894,
+      "loss": 3.1622,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000819593231398059,
+      "loss": 3.2572,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008194127864718529,
+      "loss": 3.4193,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008192322712349917,
+      "loss": 3.3013,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008190516857272113,
+      "loss": 3.3038,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008188710299882628,
+      "loss": 3.3601,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008186903040579131,
+      "loss": 3.4277,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008185095079759444,
+      "loss": 3.166,
+      "step": 2089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008183286417821548,
+      "loss": 3.3043,
+      "step": 2090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008181477055163567,
+      "loss": 3.3465,
+      "step": 2091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008179666992183791,
+      "loss": 3.3567,
+      "step": 2092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008177856229280659,
+      "loss": 3.198,
+      "step": 2093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008176044766852766,
+      "loss": 3.3882,
+      "step": 2094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008174232605298856,
+      "loss": 3.3909,
+      "step": 2095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008172419745017834,
+      "loss": 3.2309,
+      "step": 2096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008170606186408755,
+      "loss": 3.2033,
+      "step": 2097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008168791929870824,
+      "loss": 3.373,
+      "step": 2098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008166976975803411,
+      "loss": 3.3108,
+      "step": 2099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008165161324606026,
+      "loss": 3.3767,
+      "step": 2100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008163344976678342,
+      "loss": 3.291,
+      "step": 2101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008161527932420181,
+      "loss": 3.2666,
+      "step": 2102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000815971019223152,
+      "loss": 3.33,
+      "step": 2103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008157891756512488,
+      "loss": 3.4123,
+      "step": 2104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008156072625663369,
+      "loss": 3.3861,
+      "step": 2105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008154252800084595,
+      "loss": 3.3821,
+      "step": 2106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008152432280176758,
+      "loss": 3.3206,
+      "step": 2107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008150611066340599,
+      "loss": 3.4567,
+      "step": 2108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008148789158977012,
+      "loss": 3.2937,
+      "step": 2109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008146966558487043,
+      "loss": 3.3507,
+      "step": 2110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008145143265271894,
+      "loss": 3.2633,
+      "step": 2111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008143319279732914,
+      "loss": 3.2777,
+      "step": 2112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008141494602271609,
+      "loss": 3.3884,
+      "step": 2113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008139669233289634,
+      "loss": 3.2869,
+      "step": 2114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00081378431731888,
+      "loss": 3.3211,
+      "step": 2115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008136016422371065,
+      "loss": 3.3305,
+      "step": 2116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008134188981238546,
+      "loss": 3.3877,
+      "step": 2117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008132360850193506,
+      "loss": 3.2927,
+      "step": 2118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008130532029638361,
+      "loss": 3.3749,
+      "step": 2119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008128702519975679,
+      "loss": 3.1715,
+      "step": 2120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008126872321608184,
+      "loss": 3.2586,
+      "step": 2121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008125041434938746,
+      "loss": 3.412,
+      "step": 2122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008123209860370388,
+      "loss": 3.3046,
+      "step": 2123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008121377598306283,
+      "loss": 3.3415,
+      "step": 2124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008119544649149761,
+      "loss": 3.2916,
+      "step": 2125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008117711013304298,
+      "loss": 3.4712,
+      "step": 2126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008115876691173522,
+      "loss": 3.3304,
+      "step": 2127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008114041683161213,
+      "loss": 3.2743,
+      "step": 2128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008112205989671302,
+      "loss": 3.3253,
+      "step": 2129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008110369611107869,
+      "loss": 3.3865,
+      "step": 2130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008108532547875149,
+      "loss": 3.2456,
+      "step": 2131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008106694800377523,
+      "loss": 3.1915,
+      "step": 2132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008104856369019524,
+      "loss": 3.2667,
+      "step": 2133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008103017254205839,
+      "loss": 3.2965,
+      "step": 2134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00081011774563413,
+      "loss": 3.3934,
+      "step": 2135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008099336975830896,
+      "loss": 3.317,
+      "step": 2136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008097495813079756,
+      "loss": 3.3609,
+      "step": 2137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000809565396849317,
+      "loss": 3.2618,
+      "step": 2138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008093811442476573,
+      "loss": 3.2941,
+      "step": 2139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008091968235435551,
+      "loss": 3.4445,
+      "step": 2140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008090124347775836,
+      "loss": 3.182,
+      "step": 2141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008088279779903318,
+      "loss": 3.4259,
+      "step": 2142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008086434532224028,
+      "loss": 3.3834,
+      "step": 2143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008084588605144154,
+      "loss": 3.3397,
+      "step": 2144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008082741999070029,
+      "loss": 3.2279,
+      "step": 2145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008080894714408135,
+      "loss": 3.2908,
+      "step": 2146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008079046751565107,
+      "loss": 3.4242,
+      "step": 2147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008077198110947725,
+      "loss": 3.3703,
+      "step": 2148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008075348792962924,
+      "loss": 3.381,
+      "step": 2149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008073498798017785,
+      "loss": 3.254,
+      "step": 2150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008071648126519532,
+      "loss": 3.4739,
+      "step": 2151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008069796778875548,
+      "loss": 3.3563,
+      "step": 2152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008067944755493359,
+      "loss": 3.2915,
+      "step": 2153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008066092056780643,
+      "loss": 3.2568,
+      "step": 2154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008064238683145223,
+      "loss": 3.3293,
+      "step": 2155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008062384634995072,
+      "loss": 3.2936,
+      "step": 2156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008060529912738315,
+      "loss": 3.2912,
+      "step": 2157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008058674516783217,
+      "loss": 3.3809,
+      "step": 2158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008056818447538202,
+      "loss": 3.3233,
+      "step": 2159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008054961705411834,
+      "loss": 3.3607,
+      "step": 2160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008053104290812827,
+      "loss": 3.3175,
+      "step": 2161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008051246204150046,
+      "loss": 3.3384,
+      "step": 2162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008049387445832499,
+      "loss": 3.3329,
+      "step": 2163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008047528016269349,
+      "loss": 3.2286,
+      "step": 2164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008045667915869899,
+      "loss": 3.4024,
+      "step": 2165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008043807145043603,
+      "loss": 3.2363,
+      "step": 2166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008041945704200064,
+      "loss": 3.3358,
+      "step": 2167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008040083593749034,
+      "loss": 3.3777,
+      "step": 2168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008038220814100403,
+      "loss": 3.4601,
+      "step": 2169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008036357365664219,
+      "loss": 3.2846,
+      "step": 2170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008034493248850674,
+      "loss": 3.2696,
+      "step": 2171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008032628464070105,
+      "loss": 3.3914,
+      "step": 2172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008030763011732995,
+      "loss": 3.1885,
+      "step": 2173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008028896892249979,
+      "loss": 3.3052,
+      "step": 2174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008027030106031835,
+      "loss": 3.2597,
+      "step": 2175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008025162653489491,
+      "loss": 3.3002,
+      "step": 2176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008023294535034016,
+      "loss": 3.2688,
+      "step": 2177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008021425751076629,
+      "loss": 3.312,
+      "step": 2178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00080195563020287,
+      "loss": 3.3452,
+      "step": 2179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008017686188301737,
+      "loss": 3.4421,
+      "step": 2180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008015815410307398,
+      "loss": 3.3151,
+      "step": 2181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008013943968457491,
+      "loss": 3.3593,
+      "step": 2182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008012071863163964,
+      "loss": 3.3991,
+      "step": 2183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008010199094838914,
+      "loss": 3.4062,
+      "step": 2184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008008325663894586,
+      "loss": 3.3301,
+      "step": 2185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008006451570743365,
+      "loss": 3.2703,
+      "step": 2186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008004576815797786,
+      "loss": 3.386,
+      "step": 2187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008002701399470531,
+      "loss": 3.364,
+      "step": 2188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008000825322174423,
+      "loss": 3.3026,
+      "step": 2189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007998948584322434,
+      "loss": 3.3486,
+      "step": 2190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007997071186327682,
+      "loss": 3.2615,
+      "step": 2191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007995193128603426,
+      "loss": 3.3764,
+      "step": 2192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007993314411563075,
+      "loss": 3.3229,
+      "step": 2193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000799143503562018,
+      "loss": 3.3121,
+      "step": 2194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000798955500118844,
+      "loss": 3.1774,
+      "step": 2195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007987674308681694,
+      "loss": 3.3891,
+      "step": 2196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007985792958513931,
+      "loss": 3.3273,
+      "step": 2197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007983910951099283,
+      "loss": 3.2368,
+      "step": 2198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007982028286852024,
+      "loss": 3.3407,
+      "step": 2199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007980144966186578,
+      "loss": 3.2784,
+      "step": 2200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007978260989517508,
+      "loss": 3.1925,
+      "step": 2201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007976376357259526,
+      "loss": 3.1958,
+      "step": 2202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007974491069827484,
+      "loss": 3.3584,
+      "step": 2203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007972605127636383,
+      "loss": 3.3274,
+      "step": 2204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007970718531101365,
+      "loss": 3.3039,
+      "step": 2205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007968831280637713,
+      "loss": 3.3252,
+      "step": 2206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000796694337666086,
+      "loss": 3.2968,
+      "step": 2207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007965054819586381,
+      "loss": 3.36,
+      "step": 2208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007963165609829995,
+      "loss": 3.3232,
+      "step": 2209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007961275747807562,
+      "loss": 3.4068,
+      "step": 2210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007959385233935086,
+      "loss": 3.3558,
+      "step": 2211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007957494068628718,
+      "loss": 3.4937,
+      "step": 2212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007955602252304752,
+      "loss": 3.187,
+      "step": 2213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007953709785379621,
+      "loss": 3.4378,
+      "step": 2214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007951816668269904,
+      "loss": 3.3885,
+      "step": 2215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007949922901392322,
+      "loss": 3.2606,
+      "step": 2216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007948028485163743,
+      "loss": 3.4462,
+      "step": 2217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007946133420001171,
+      "loss": 3.2161,
+      "step": 2218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000794423770632176,
+      "loss": 3.2039,
+      "step": 2219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007942341344542801,
+      "loss": 3.4182,
+      "step": 2220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007940444335081732,
+      "loss": 3.2639,
+      "step": 2221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007938546678356131,
+      "loss": 3.4901,
+      "step": 2222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000793664837478372,
+      "loss": 3.2681,
+      "step": 2223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007934749424782361,
+      "loss": 3.2701,
+      "step": 2224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007932849828770061,
+      "loss": 3.347,
+      "step": 2225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007930949587164968,
+      "loss": 3.3469,
+      "step": 2226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007929048700385371,
+      "loss": 3.3277,
+      "step": 2227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007927147168849704,
+      "loss": 3.4065,
+      "step": 2228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007925244992976537,
+      "loss": 3.4147,
+      "step": 2229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007923342173184594,
+      "loss": 3.2844,
+      "step": 2230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007921438709892724,
+      "loss": 3.2478,
+      "step": 2231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007919534603519934,
+      "loss": 3.3189,
+      "step": 2232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007917629854485357,
+      "loss": 3.1947,
+      "step": 2233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007915724463208282,
+      "loss": 3.4606,
+      "step": 2234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000791381843010813,
+      "loss": 3.4158,
+      "step": 2235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007911911755604466,
+      "loss": 3.365,
+      "step": 2236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007910004440116996,
+      "loss": 3.3101,
+      "step": 2237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007908096484065569,
+      "loss": 3.3731,
+      "step": 2238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007906187887870172,
+      "loss": 3.4101,
+      "step": 2239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007904278651950933,
+      "loss": 3.3002,
+      "step": 2240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007902368776728125,
+      "loss": 3.4901,
+      "step": 2241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007900458262622156,
+      "loss": 3.2595,
+      "step": 2242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007898547110053581,
+      "loss": 3.2746,
+      "step": 2243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007896635319443087,
+      "loss": 3.2214,
+      "step": 2244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000789472289121151,
+      "loss": 3.3826,
+      "step": 2245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007892809825779821,
+      "loss": 3.2913,
+      "step": 2246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007890896123569135,
+      "loss": 3.2191,
+      "step": 2247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007888981785000704,
+      "loss": 3.3184,
+      "step": 2248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000788706681049592,
+      "loss": 3.2472,
+      "step": 2249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007885151200476319,
+      "loss": 3.3683,
+      "step": 2250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007883234955363572,
+      "loss": 3.3096,
+      "step": 2251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007881318075579492,
+      "loss": 3.3168,
+      "step": 2252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007879400561546033,
+      "loss": 3.4886,
+      "step": 2253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007877482413685286,
+      "loss": 3.2121,
+      "step": 2254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007875563632419484,
+      "loss": 3.336,
+      "step": 2255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007873644218170996,
+      "loss": 3.3045,
+      "step": 2256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007871724171362336,
+      "loss": 3.3418,
+      "step": 2257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000786980349241615,
+      "loss": 3.166,
+      "step": 2258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007867882181755231,
+      "loss": 3.3934,
+      "step": 2259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007865960239802503,
+      "loss": 3.4106,
+      "step": 2260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007864037666981036,
+      "loss": 3.3376,
+      "step": 2261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007862114463714034,
+      "loss": 3.2463,
+      "step": 2262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007860190630424842,
+      "loss": 3.2966,
+      "step": 2263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007858266167536944,
+      "loss": 3.286,
+      "step": 2264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007856341075473961,
+      "loss": 3.3309,
+      "step": 2265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007854415354659654,
+      "loss": 3.3617,
+      "step": 2266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007852489005517922,
+      "loss": 3.3939,
+      "step": 2267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007850562028472801,
+      "loss": 3.4517,
+      "step": 2268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007848634423948469,
+      "loss": 3.3077,
+      "step": 2269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007846706192369235,
+      "loss": 3.3066,
+      "step": 2270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007844777334159554,
+      "loss": 3.3159,
+      "step": 2271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007842847849744013,
+      "loss": 3.3739,
+      "step": 2272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007840917739547342,
+      "loss": 3.2778,
+      "step": 2273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007838987003994405,
+      "loss": 3.4168,
+      "step": 2274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007837055643510202,
+      "loss": 3.3508,
+      "step": 2275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007835123658519878,
+      "loss": 3.3684,
+      "step": 2276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007833191049448706,
+      "loss": 3.3352,
+      "step": 2277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007831257816722104,
+      "loss": 3.294,
+      "step": 2278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007829323960765625,
+      "loss": 3.2529,
+      "step": 2279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007827389482004954,
+      "loss": 3.3472,
+      "step": 2280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007825454380865922,
+      "loss": 3.3397,
+      "step": 2281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007823518657774492,
+      "loss": 3.2644,
+      "step": 2282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007821582313156763,
+      "loss": 3.406,
+      "step": 2283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007819645347438973,
+      "loss": 3.283,
+      "step": 2284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007817707761047497,
+      "loss": 3.4015,
+      "step": 2285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007815769554408845,
+      "loss": 3.3053,
+      "step": 2286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007813830727949663,
+      "loss": 3.3879,
+      "step": 2287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007811891282096737,
+      "loss": 3.2801,
+      "step": 2288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007809951217276985,
+      "loss": 3.2409,
+      "step": 2289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007808010533917464,
+      "loss": 3.3723,
+      "step": 2290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007806069232445368,
+      "loss": 3.3152,
+      "step": 2291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007804127313288023,
+      "loss": 3.4493,
+      "step": 2292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007802184776872894,
+      "loss": 3.2221,
+      "step": 2293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007800241623627582,
+      "loss": 3.2853,
+      "step": 2294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007798297853979823,
+      "loss": 3.2532,
+      "step": 2295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007796353468357489,
+      "loss": 3.2091,
+      "step": 2296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007794408467188586,
+      "loss": 3.3899,
+      "step": 2297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007792462850901258,
+      "loss": 3.3981,
+      "step": 2298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007790516619923783,
+      "loss": 3.2673,
+      "step": 2299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007788569774684575,
+      "loss": 3.3157,
+      "step": 2300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007786622315612182,
+      "loss": 3.3404,
+      "step": 2301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007784674243135289,
+      "loss": 3.4063,
+      "step": 2302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007782725557682713,
+      "loss": 3.3032,
+      "step": 2303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007780776259683411,
+      "loss": 3.3349,
+      "step": 2304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000777882634956647,
+      "loss": 3.1847,
+      "step": 2305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007776875827761113,
+      "loss": 3.3071,
+      "step": 2306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007774924694696698,
+      "loss": 3.5699,
+      "step": 2307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007772972950802718,
+      "loss": 3.2079,
+      "step": 2308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007771020596508799,
+      "loss": 3.3501,
+      "step": 2309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007769067632244707,
+      "loss": 3.2777,
+      "step": 2310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007767114058440332,
+      "loss": 3.4477,
+      "step": 2311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007765159875525706,
+      "loss": 3.2204,
+      "step": 2312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007763205083930995,
+      "loss": 3.264,
+      "step": 2313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007761249684086493,
+      "loss": 3.3216,
+      "step": 2314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007759293676422636,
+      "loss": 3.3178,
+      "step": 2315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007757337061369989,
+      "loss": 3.2545,
+      "step": 2316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007755379839359248,
+      "loss": 3.3306,
+      "step": 2317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007753422010821248,
+      "loss": 3.3021,
+      "step": 2318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007751463576186957,
+      "loss": 3.2546,
+      "step": 2319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007749504535887472,
+      "loss": 3.3511,
+      "step": 2320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000774754489035403,
+      "loss": 3.1737,
+      "step": 2321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007745584640017995,
+      "loss": 3.267,
+      "step": 2322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007743623785310866,
+      "loss": 3.2648,
+      "step": 2323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007741662326664278,
+      "loss": 3.3597,
+      "step": 2324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007739700264509992,
+      "loss": 3.3107,
+      "step": 2325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007737737599279913,
+      "loss": 3.3417,
+      "step": 2326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007735774331406066,
+      "loss": 3.3739,
+      "step": 2327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007733810461320619,
+      "loss": 3.3832,
+      "step": 2328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007731845989455866,
+      "loss": 3.3474,
+      "step": 2329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007729880916244236,
+      "loss": 3.3403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007727915242118292,
+      "loss": 3.3854,
+      "step": 2331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007725948967510726,
+      "loss": 3.2718,
+      "step": 2332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007723982092854365,
+      "loss": 3.3024,
+      "step": 2333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007722014618582166,
+      "loss": 3.3894,
+      "step": 2334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007720046545127218,
+      "loss": 3.336,
+      "step": 2335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007718077872922742,
+      "loss": 3.3541,
+      "step": 2336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007716108602402094,
+      "loss": 3.2757,
+      "step": 2337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007714138733998758,
+      "loss": 3.3654,
+      "step": 2338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007712168268146351,
+      "loss": 3.2871,
+      "step": 2339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007710197205278619,
+      "loss": 3.2296,
+      "step": 2340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007708225545829446,
+      "loss": 3.2833,
+      "step": 2341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007706253290232838,
+      "loss": 3.3375,
+      "step": 2342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007704280438922943,
+      "loss": 3.2294,
+      "step": 2343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000770230699233403,
+      "loss": 3.3513,
+      "step": 2344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007700332950900504,
+      "loss": 3.2919,
+      "step": 2345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00076983583150569,
+      "loss": 3.2693,
+      "step": 2346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007696383085237886,
+      "loss": 3.2395,
+      "step": 2347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007694407261878258,
+      "loss": 3.2252,
+      "step": 2348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007692430845412945,
+      "loss": 3.3333,
+      "step": 2349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007690453836277002,
+      "loss": 3.3296,
+      "step": 2350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007688476234905622,
+      "loss": 3.3576,
+      "step": 2351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000768649804173412,
+      "loss": 3.3986,
+      "step": 2352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007684519257197946,
+      "loss": 3.2361,
+      "step": 2353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007682539881732681,
+      "loss": 3.4595,
+      "step": 2354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007680559915774033,
+      "loss": 3.2328,
+      "step": 2355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007678579359757842,
+      "loss": 3.2864,
+      "step": 2356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007676598214120078,
+      "loss": 3.3988,
+      "step": 2357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007674616479296841,
+      "loss": 3.2249,
+      "step": 2358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007672634155724356,
+      "loss": 3.3609,
+      "step": 2359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007670651243838986,
+      "loss": 3.3224,
+      "step": 2360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007668667744077216,
+      "loss": 3.2143,
+      "step": 2361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007666683656875664,
+      "loss": 3.2994,
+      "step": 2362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007664698982671078,
+      "loss": 3.3185,
+      "step": 2363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007662713721900331,
+      "loss": 3.1976,
+      "step": 2364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007660727875000431,
+      "loss": 3.2574,
+      "step": 2365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000765874144240851,
+      "loss": 3.2942,
+      "step": 2366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007656754424561834,
+      "loss": 3.3466,
+      "step": 2367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000765476682189779,
+      "loss": 3.2743,
+      "step": 2368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007652778634853903,
+      "loss": 3.3277,
+      "step": 2369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000765078986386782,
+      "loss": 3.2839,
+      "step": 2370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007648800509377318,
+      "loss": 3.1942,
+      "step": 2371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007646810571820304,
+      "loss": 3.2388,
+      "step": 2372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007644820051634812,
+      "loss": 3.426,
+      "step": 2373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007642828949259007,
+      "loss": 3.3784,
+      "step": 2374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007640837265131177,
+      "loss": 3.3261,
+      "step": 2375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007638844999689743,
+      "loss": 3.307,
+      "step": 2376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007636852153373251,
+      "loss": 3.2962,
+      "step": 2377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007634858726620373,
+      "loss": 3.2425,
+      "step": 2378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007632864719869917,
+      "loss": 3.3222,
+      "step": 2379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000763087013356081,
+      "loss": 3.185,
+      "step": 2380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007628874968132111,
+      "loss": 3.3223,
+      "step": 2381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007626879224023002,
+      "loss": 3.2877,
+      "step": 2382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00076248829016728,
+      "loss": 3.3245,
+      "step": 2383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007622886001520944,
+      "loss": 3.2963,
+      "step": 2384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007620888524006999,
+      "loss": 3.1869,
+      "step": 2385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007618890469570661,
+      "loss": 3.2662,
+      "step": 2386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007616891838651749,
+      "loss": 3.2476,
+      "step": 2387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007614892631690216,
+      "loss": 3.228,
+      "step": 2388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007612892849126132,
+      "loss": 3.2759,
+      "step": 2389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007610892491399701,
+      "loss": 3.3322,
+      "step": 2390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007608891558951249,
+      "loss": 3.2079,
+      "step": 2391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007606890052221232,
+      "loss": 3.232,
+      "step": 2392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007604887971650233,
+      "loss": 3.3401,
+      "step": 2393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007602885317678957,
+      "loss": 3.1664,
+      "step": 2394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007600882090748238,
+      "loss": 3.2771,
+      "step": 2395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007598878291299036,
+      "loss": 3.3346,
+      "step": 2396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007596873919772438,
+      "loss": 3.3885,
+      "step": 2397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007594868976609657,
+      "loss": 3.2252,
+      "step": 2398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007592863462252024,
+      "loss": 3.1511,
+      "step": 2399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007590857377141009,
+      "loss": 3.3808,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007588850721718199,
+      "loss": 3.3095,
+      "step": 2401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007586843496425309,
+      "loss": 3.3894,
+      "step": 2402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007584835701704176,
+      "loss": 3.3223,
+      "step": 2403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007582827337996768,
+      "loss": 3.3342,
+      "step": 2404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007580818405745176,
+      "loss": 3.3465,
+      "step": 2405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007578808905391615,
+      "loss": 3.3457,
+      "step": 2406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007576798837378425,
+      "loss": 3.3582,
+      "step": 2407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007574788202148071,
+      "loss": 3.2792,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007572777000143145,
+      "loss": 3.3602,
+      "step": 2409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007570765231806362,
+      "loss": 3.254,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007568752897580562,
+      "loss": 3.4319,
+      "step": 2411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007566739997908709,
+      "loss": 3.3807,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007564726533233892,
+      "loss": 3.2483,
+      "step": 2413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007562712503999327,
+      "loss": 3.4165,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007560697910648347,
+      "loss": 3.2272,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007558682753624419,
+      "loss": 3.3466,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007556667033371123,
+      "loss": 3.3784,
+      "step": 2417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007554650750332175,
+      "loss": 3.165,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007552633904951405,
+      "loss": 3.3042,
+      "step": 2419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007550616497672771,
+      "loss": 3.3435,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007548598528940353,
+      "loss": 3.2307,
+      "step": 2421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000754657999919836,
+      "loss": 3.3242,
+      "step": 2422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007544560908891116,
+      "loss": 3.255,
+      "step": 2423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007542541258463075,
+      "loss": 3.1781,
+      "step": 2424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007540521048358814,
+      "loss": 3.2379,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007538500279023026,
+      "loss": 3.366,
+      "step": 2426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007536478950900537,
+      "loss": 3.3713,
+      "step": 2427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007534457064436289,
+      "loss": 3.3321,
+      "step": 2428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007532434620075349,
+      "loss": 3.2991,
+      "step": 2429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007530411618262906,
+      "loss": 3.3414,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007528388059444278,
+      "loss": 3.4424,
+      "step": 2431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007526363944064895,
+      "loss": 3.3561,
+      "step": 2432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007524339272570316,
+      "loss": 3.4258,
+      "step": 2433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007522314045406223,
+      "loss": 3.1827,
+      "step": 2434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007520288263018418,
+      "loss": 3.2712,
+      "step": 2435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007518261925852823,
+      "loss": 3.2915,
+      "step": 2436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000751623503435549,
+      "loss": 3.3503,
+      "step": 2437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007514207588972583,
+      "loss": 3.2194,
+      "step": 2438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007512179590150396,
+      "loss": 3.2236,
+      "step": 2439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000751015103833534,
+      "loss": 3.2936,
+      "step": 2440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000750812193397395,
+      "loss": 3.2713,
+      "step": 2441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007506092277512884,
+      "loss": 3.28,
+      "step": 2442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007504062069398918,
+      "loss": 3.2698,
+      "step": 2443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007502031310078949,
+      "loss": 3.378,
+      "step": 2444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00075,
+      "loss": 3.3048,
+      "step": 2445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007497968139609213,
+      "loss": 3.4182,
+      "step": 2446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007495935729353849,
+      "loss": 3.3861,
+      "step": 2447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007493902769681292,
+      "loss": 3.325,
+      "step": 2448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007491869261039046,
+      "loss": 3.3921,
+      "step": 2449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007489835203874738,
+      "loss": 3.3489,
+      "step": 2450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007487800598636116,
+      "loss": 3.1312,
+      "step": 2451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007485765445771042,
+      "loss": 3.307,
+      "step": 2452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007483729745727508,
+      "loss": 3.3065,
+      "step": 2453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007481693498953621,
+      "loss": 3.4184,
+      "step": 2454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007479656705897607,
+      "loss": 3.3657,
+      "step": 2455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007477619367007815,
+      "loss": 3.3791,
+      "step": 2456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007475581482732716,
+      "loss": 3.3281,
+      "step": 2457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00074735430535209,
+      "loss": 3.4043,
+      "step": 2458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000747150407982107,
+      "loss": 3.3448,
+      "step": 2459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007469464562082059,
+      "loss": 3.3861,
+      "step": 2460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007467424500752813,
+      "loss": 3.1929,
+      "step": 2461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007465383896282403,
+      "loss": 3.2409,
+      "step": 2462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007463342749120013,
+      "loss": 3.2858,
+      "step": 2463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007461301059714953,
+      "loss": 3.3522,
+      "step": 2464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007459258828516645,
+      "loss": 3.2691,
+      "step": 2465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000745721605597464,
+      "loss": 3.377,
+      "step": 2466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007455172742538599,
+      "loss": 3.3386,
+      "step": 2467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007453128888658307,
+      "loss": 3.3165,
+      "step": 2468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007451084494783667,
+      "loss": 3.207,
+      "step": 2469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007449039561364701,
+      "loss": 3.395,
+      "step": 2470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007446994088851548,
+      "loss": 3.2555,
+      "step": 2471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000744494807769447,
+      "loss": 3.3541,
+      "step": 2472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007442901528343841,
+      "loss": 3.3464,
+      "step": 2473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007440854441250159,
+      "loss": 3.4178,
+      "step": 2474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007438806816864039,
+      "loss": 3.418,
+      "step": 2475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007436758655636212,
+      "loss": 3.2953,
+      "step": 2476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007434709958017531,
+      "loss": 3.2259,
+      "step": 2477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007432660724458963,
+      "loss": 3.1261,
+      "step": 2478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007430610955411597,
+      "loss": 3.3843,
+      "step": 2479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007428560651326637,
+      "loss": 3.269,
+      "step": 2480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007426509812655407,
+      "loss": 3.3477,
+      "step": 2481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007424458439849342,
+      "loss": 3.2765,
+      "step": 2482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007422406533360007,
+      "loss": 3.4115,
+      "step": 2483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007420354093639073,
+      "loss": 3.3335,
+      "step": 2484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007418301121138335,
+      "loss": 3.2718,
+      "step": 2485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007416247616309701,
+      "loss": 3.2432,
+      "step": 2486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007414193579605198,
+      "loss": 3.3722,
+      "step": 2487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007412139011476973,
+      "loss": 3.1986,
+      "step": 2488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007410083912377286,
+      "loss": 3.3292,
+      "step": 2489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007408028282758515,
+      "loss": 3.2684,
+      "step": 2490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007405972123073153,
+      "loss": 3.3755,
+      "step": 2491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007403915433773815,
+      "loss": 3.2879,
+      "step": 2492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007401858215313228,
+      "loss": 3.3021,
+      "step": 2493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007399800468144234,
+      "loss": 3.3069,
+      "step": 2494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007397742192719798,
+      "loss": 3.305,
+      "step": 2495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007395683389492995,
+      "loss": 3.2709,
+      "step": 2496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007393624058917019,
+      "loss": 3.3354,
+      "step": 2497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007391564201445181,
+      "loss": 3.2793,
+      "step": 2498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007389503817530904,
+      "loss": 3.2333,
+      "step": 2499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007387442907627732,
+      "loss": 3.2478,
+      "step": 2500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007385381472189321,
+      "loss": 3.2216,
+      "step": 2501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007383319511669443,
+      "loss": 3.3405,
+      "step": 2502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007381257026521988,
+      "loss": 3.2993,
+      "step": 2503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007379194017200958,
+      "loss": 3.2292,
+      "step": 2504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007377130484160476,
+      "loss": 3.3309,
+      "step": 2505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007375066427854774,
+      "loss": 3.4123,
+      "step": 2506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007373001848738202,
+      "loss": 3.2293,
+      "step": 2507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007370936747265226,
+      "loss": 3.2575,
+      "step": 2508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007368871123890425,
+      "loss": 3.2899,
+      "step": 2509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007366804979068493,
+      "loss": 3.2131,
+      "step": 2510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007364738313254243,
+      "loss": 3.3758,
+      "step": 2511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007362671126902594,
+      "loss": 3.3223,
+      "step": 2512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007360603420468589,
+      "loss": 3.206,
+      "step": 2513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000735853519440738,
+      "loss": 3.3188,
+      "step": 2514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007356466449174235,
+      "loss": 3.3324,
+      "step": 2515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007354397185224535,
+      "loss": 3.2082,
+      "step": 2516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007352327403013779,
+      "loss": 3.2798,
+      "step": 2517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007350257102997573,
+      "loss": 3.2976,
+      "step": 2518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007348186285631646,
+      "loss": 3.187,
+      "step": 2519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000734611495137183,
+      "loss": 3.2669,
+      "step": 2520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007344043100674083,
+      "loss": 3.2487,
+      "step": 2521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007341970733994466,
+      "loss": 3.3945,
+      "step": 2522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007339897851789162,
+      "loss": 3.2723,
+      "step": 2523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007337824454514462,
+      "loss": 3.2243,
+      "step": 2524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007335750542626771,
+      "loss": 3.3568,
+      "step": 2525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000733367611658261,
+      "loss": 3.179,
+      "step": 2526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007331601176838612,
+      "loss": 3.2023,
+      "step": 2527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007329525723851519,
+      "loss": 3.3097,
+      "step": 2528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007327449758078193,
+      "loss": 3.4292,
+      "step": 2529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007325373279975602,
+      "loss": 3.2049,
+      "step": 2530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007323296290000835,
+      "loss": 3.362,
+      "step": 2531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007321218788611086,
+      "loss": 3.2865,
+      "step": 2532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007319140776263664,
+      "loss": 3.3078,
+      "step": 2533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007317062253415993,
+      "loss": 3.2789,
+      "step": 2534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007314983220525604,
+      "loss": 3.3022,
+      "step": 2535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007312903678050146,
+      "loss": 3.2268,
+      "step": 2536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007310823626447378,
+      "loss": 3.2673,
+      "step": 2537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000730874306617517,
+      "loss": 3.2883,
+      "step": 2538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007306661997691504,
+      "loss": 3.2493,
+      "step": 2539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007304580421454476,
+      "loss": 3.2458,
+      "step": 2540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007302498337922293,
+      "loss": 3.2342,
+      "step": 2541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007300415747553271,
+      "loss": 3.3653,
+      "step": 2542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000729833265080584,
+      "loss": 3.2753,
+      "step": 2543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007296249048138543,
+      "loss": 3.1613,
+      "step": 2544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007294164940010031,
+      "loss": 3.2398,
+      "step": 2545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007292080326879067,
+      "loss": 3.3029,
+      "step": 2546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007289995209204529,
+      "loss": 3.3486,
+      "step": 2547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007287909587445398,
+      "loss": 3.2953,
+      "step": 2548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007285823462060776,
+      "loss": 3.3083,
+      "step": 2549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007283736833509867,
+      "loss": 3.314,
+      "step": 2550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007281649702251993,
+      "loss": 3.1833,
+      "step": 2551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007279562068746581,
+      "loss": 3.3149,
+      "step": 2552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007277473933453169,
+      "loss": 3.2666,
+      "step": 2553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007275385296831413,
+      "loss": 3.3772,
+      "step": 2554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007273296159341069,
+      "loss": 3.4231,
+      "step": 2555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007271206521442007,
+      "loss": 3.3128,
+      "step": 2556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007269116383594211,
+      "loss": 3.3952,
+      "step": 2557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000726702574625777,
+      "loss": 3.2771,
+      "step": 2558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007264934609892888,
+      "loss": 3.214,
+      "step": 2559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007262842974959873,
+      "loss": 3.2437,
+      "step": 2560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007260750841919144,
+      "loss": 3.2546,
+      "step": 2561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007258658211231235,
+      "loss": 3.2326,
+      "step": 2562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007256565083356785,
+      "loss": 3.3274,
+      "step": 2563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007254471458756543,
+      "loss": 3.2817,
+      "step": 2564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007252377337891364,
+      "loss": 3.301,
+      "step": 2565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007250282721222222,
+      "loss": 3.1949,
+      "step": 2566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000724818760921019,
+      "loss": 3.2885,
+      "step": 2567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007246092002316456,
+      "loss": 3.2824,
+      "step": 2568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007243995901002312,
+      "loss": 3.3139,
+      "step": 2569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007241899305729162,
+      "loss": 3.1892,
+      "step": 2570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007239802216958522,
+      "loss": 3.3741,
+      "step": 2571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007237704635152011,
+      "loss": 3.3037,
+      "step": 2572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007235606560771359,
+      "loss": 3.3483,
+      "step": 2573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007233507994278403,
+      "loss": 3.2818,
+      "step": 2574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007231408936135091,
+      "loss": 3.1891,
+      "step": 2575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007229309386803476,
+      "loss": 3.3037,
+      "step": 2576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007227209346745723,
+      "loss": 3.3566,
+      "step": 2577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00072251088164241,
+      "loss": 3.2835,
+      "step": 2578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007223007796300986,
+      "loss": 3.2505,
+      "step": 2579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007220906286838868,
+      "loss": 3.2368,
+      "step": 2580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007218804288500342,
+      "loss": 3.3388,
+      "step": 2581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000721670180174811,
+      "loss": 3.2866,
+      "step": 2582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007214598827044979,
+      "loss": 3.3073,
+      "step": 2583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007212495364853866,
+      "loss": 3.3432,
+      "step": 2584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007210391415637796,
+      "loss": 3.2473,
+      "step": 2585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007208286979859899,
+      "loss": 3.2712,
+      "step": 2586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007206182057983415,
+      "loss": 3.2972,
+      "step": 2587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007204076650471689,
+      "loss": 3.2631,
+      "step": 2588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007201970757788173,
+      "loss": 3.2795,
+      "step": 2589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007199864380396424,
+      "loss": 3.3941,
+      "step": 2590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007197757518760112,
+      "loss": 3.3492,
+      "step": 2591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007195650173343007,
+      "loss": 3.2969,
+      "step": 2592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007193542344608987,
+      "loss": 3.367,
+      "step": 2593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007191434033022038,
+      "loss": 3.2465,
+      "step": 2594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007189325239046253,
+      "loss": 3.2793,
+      "step": 2595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007187215963145827,
+      "loss": 3.3054,
+      "step": 2596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007185106205785067,
+      "loss": 3.3007,
+      "step": 2597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007182995967428379,
+      "loss": 3.3126,
+      "step": 2598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007180885248540283,
+      "loss": 3.2236,
+      "step": 2599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007178774049585397,
+      "loss": 3.0986,
+      "step": 2600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000717666237102845,
+      "loss": 3.2269,
+      "step": 2601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007174550213334273,
+      "loss": 3.2539,
+      "step": 2602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007172437576967805,
+      "loss": 3.4594,
+      "step": 2603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007170324462394091,
+      "loss": 3.2076,
+      "step": 2604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007168210870078277,
+      "loss": 3.4187,
+      "step": 2605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007166096800485618,
+      "loss": 3.2836,
+      "step": 2606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007163982254081475,
+      "loss": 3.2759,
+      "step": 2607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007161867231331309,
+      "loss": 3.2833,
+      "step": 2608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000715975173270069,
+      "loss": 3.3048,
+      "step": 2609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007157635758655294,
+      "loss": 3.3058,
+      "step": 2610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007155519309660896,
+      "loss": 3.2224,
+      "step": 2611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007153402386183378,
+      "loss": 3.2768,
+      "step": 2612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000715128498868873,
+      "loss": 3.3039,
+      "step": 2613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007149167117643043,
+      "loss": 3.3904,
+      "step": 2614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007147048773512513,
+      "loss": 3.2719,
+      "step": 2615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007144929956763437,
+      "loss": 3.4837,
+      "step": 2616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007142810667862223,
+      "loss": 3.1921,
+      "step": 2617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007140690907275376,
+      "loss": 3.3359,
+      "step": 2618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007138570675469511,
+      "loss": 3.1794,
+      "step": 2619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007136449972911339,
+      "loss": 3.3222,
+      "step": 2620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007134328800067684,
+      "loss": 3.3043,
+      "step": 2621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007132207157405462,
+      "loss": 3.3152,
+      "step": 2622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007130085045391706,
+      "loss": 3.3887,
+      "step": 2623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007127962464493542,
+      "loss": 3.2894,
+      "step": 2624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007125839415178203,
+      "loss": 3.1888,
+      "step": 2625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007123715897913025,
+      "loss": 3.3921,
+      "step": 2626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007121591913165446,
+      "loss": 3.3132,
+      "step": 2627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007119467461403009,
+      "loss": 3.2523,
+      "step": 2628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007117342543093357,
+      "loss": 3.4282,
+      "step": 2629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007115217158704237,
+      "loss": 3.3204,
+      "step": 2630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007113091308703497,
+      "loss": 3.2632,
+      "step": 2631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007110964993559094,
+      "loss": 3.2652,
+      "step": 2632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007108838213739078,
+      "loss": 3.239,
+      "step": 2633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000710671096971161,
+      "loss": 3.4205,
+      "step": 2634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007104583261944943,
+      "loss": 3.2946,
+      "step": 2635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007102455090907445,
+      "loss": 3.3062,
+      "step": 2636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007100326457067575,
+      "loss": 3.3325,
+      "step": 2637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007098197360893898,
+      "loss": 3.3041,
+      "step": 2638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007096067802855082,
+      "loss": 3.3117,
+      "step": 2639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007093937783419896,
+      "loss": 3.356,
+      "step": 2640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007091807303057207,
+      "loss": 3.1935,
+      "step": 2641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007089676362235991,
+      "loss": 3.1475,
+      "step": 2642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007087544961425316,
+      "loss": 3.2449,
+      "step": 2643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007085413101094358,
+      "loss": 3.3046,
+      "step": 2644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007083280781712394,
+      "loss": 3.3148,
+      "step": 2645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007081148003748799,
+      "loss": 3.275,
+      "step": 2646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007079014767673047,
+      "loss": 3.3395,
+      "step": 2647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007076881073954721,
+      "loss": 3.3111,
+      "step": 2648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007074746923063497,
+      "loss": 3.3318,
+      "step": 2649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007072612315469155,
+      "loss": 3.2956,
+      "step": 2650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007070477251641575,
+      "loss": 3.2945,
+      "step": 2651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007068341732050737,
+      "loss": 3.3266,
+      "step": 2652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007066205757166722,
+      "loss": 3.3074,
+      "step": 2653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000706406932745971,
+      "loss": 3.332,
+      "step": 2654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007061932443399985,
+      "loss": 3.3611,
+      "step": 2655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007059795105457925,
+      "loss": 3.1384,
+      "step": 2656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007057657314104011,
+      "loss": 3.1879,
+      "step": 2657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007055519069808827,
+      "loss": 3.2258,
+      "step": 2658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000705338037304305,
+      "loss": 3.3975,
+      "step": 2659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000705124122427746,
+      "loss": 3.1883,
+      "step": 2660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007049101623982937,
+      "loss": 3.3707,
+      "step": 2661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007046961572630462,
+      "loss": 3.3252,
+      "step": 2662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007044821070691113,
+      "loss": 3.4178,
+      "step": 2663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007042680118636065,
+      "loss": 3.3224,
+      "step": 2664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007040538716936597,
+      "loss": 3.3855,
+      "step": 2665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007038396866064083,
+      "loss": 3.4197,
+      "step": 2666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007036254566489997,
+      "loss": 3.2673,
+      "step": 2667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007034111818685913,
+      "loss": 3.2082,
+      "step": 2668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007031968623123502,
+      "loss": 3.4433,
+      "step": 2669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007029824980274535,
+      "loss": 3.4111,
+      "step": 2670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007027680890610881,
+      "loss": 3.3303,
+      "step": 2671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007025536354604508,
+      "loss": 3.1658,
+      "step": 2672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000702339137272748,
+      "loss": 3.2231,
+      "step": 2673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007021245945451961,
+      "loss": 3.2974,
+      "step": 2674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007019100073250212,
+      "loss": 3.2037,
+      "step": 2675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007016953756594594,
+      "loss": 3.3163,
+      "step": 2676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007014806995957562,
+      "loss": 3.2365,
+      "step": 2677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007012659791811674,
+      "loss": 3.3193,
+      "step": 2678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000701051214462958,
+      "loss": 3.2385,
+      "step": 2679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007008364054884032,
+      "loss": 3.3654,
+      "step": 2680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007006215523047879,
+      "loss": 3.3026,
+      "step": 2681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007004066549594063,
+      "loss": 3.2484,
+      "step": 2682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007001917134995627,
+      "loss": 3.1977,
+      "step": 2683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000699976727972571,
+      "loss": 3.392,
+      "step": 2684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006997616984257552,
+      "loss": 3.2801,
+      "step": 2685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006995466249064482,
+      "loss": 3.3344,
+      "step": 2686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006993315074619931,
+      "loss": 3.3117,
+      "step": 2687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006991163461397425,
+      "loss": 3.1371,
+      "step": 2688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006989011409870591,
+      "loss": 3.3533,
+      "step": 2689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006986858920513145,
+      "loss": 3.2598,
+      "step": 2690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006984705993798904,
+      "loss": 3.3226,
+      "step": 2691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006982552630201782,
+      "loss": 3.2958,
+      "step": 2692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006980398830195785,
+      "loss": 3.2262,
+      "step": 2693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006978244594255019,
+      "loss": 3.1237,
+      "step": 2694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006976089922853685,
+      "loss": 3.332,
+      "step": 2695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006973934816466078,
+      "loss": 3.2218,
+      "step": 2696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006971779275566593,
+      "loss": 3.217,
+      "step": 2697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006969623300629716,
+      "loss": 3.2229,
+      "step": 2698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000696746689213003,
+      "loss": 3.2771,
+      "step": 2699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006965310050542215,
+      "loss": 3.2674,
+      "step": 2700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006963152776341043,
+      "loss": 3.3478,
+      "step": 2701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006960995070001387,
+      "loss": 3.3323,
+      "step": 2702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006958836931998209,
+      "loss": 3.282,
+      "step": 2703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000695667836280657,
+      "loss": 3.1812,
+      "step": 2704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006954519362901621,
+      "loss": 3.1915,
+      "step": 2705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006952359932758616,
+      "loss": 3.2424,
+      "step": 2706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006950200072852896,
+      "loss": 3.2786,
+      "step": 2707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00069480397836599,
+      "loss": 3.3386,
+      "step": 2708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006945879065655164,
+      "loss": 3.3075,
+      "step": 2709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006943717919314311,
+      "loss": 3.2141,
+      "step": 2710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006941556345113066,
+      "loss": 3.2514,
+      "step": 2711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006939394343527243,
+      "loss": 3.2487,
+      "step": 2712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006937231915032751,
+      "loss": 3.2856,
+      "step": 2713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006935069060105599,
+      "loss": 3.1706,
+      "step": 2714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000693290577922188,
+      "loss": 3.3809,
+      "step": 2715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006930742072857787,
+      "loss": 3.4066,
+      "step": 2716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006928577941489606,
+      "loss": 3.2659,
+      "step": 2717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006926413385593715,
+      "loss": 3.4152,
+      "step": 2718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006924248405646589,
+      "loss": 3.2901,
+      "step": 2719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006922083002124789,
+      "loss": 3.3123,
+      "step": 2720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006919917175504977,
+      "loss": 3.3301,
+      "step": 2721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006917750926263903,
+      "loss": 3.324,
+      "step": 2722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006915584254878416,
+      "loss": 3.2039,
+      "step": 2723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000691341716182545,
+      "loss": 3.309,
+      "step": 2724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006911249647582037,
+      "loss": 3.5544,
+      "step": 2725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006909081712625299,
+      "loss": 3.2108,
+      "step": 2726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006906913357432454,
+      "loss": 3.1977,
+      "step": 2727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006904744582480812,
+      "loss": 3.1041,
+      "step": 2728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006902575388247772,
+      "loss": 3.3276,
+      "step": 2729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006900405775210828,
+      "loss": 3.4057,
+      "step": 2730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006898235743847563,
+      "loss": 3.2728,
+      "step": 2731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006896065294635657,
+      "loss": 3.26,
+      "step": 2732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000689389442805288,
+      "loss": 3.2951,
+      "step": 2733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006891723144577094,
+      "loss": 3.2983,
+      "step": 2734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006889551444686249,
+      "loss": 3.195,
+      "step": 2735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006887379328858392,
+      "loss": 3.2447,
+      "step": 2736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006885206797571659,
+      "loss": 3.166,
+      "step": 2737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006883033851304279,
+      "loss": 3.2477,
+      "step": 2738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006880860490534568,
+      "loss": 3.2551,
+      "step": 2739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006878686715740942,
+      "loss": 3.1548,
+      "step": 2740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006876512527401897,
+      "loss": 3.2234,
+      "step": 2741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006874337925996028,
+      "loss": 3.1728,
+      "step": 2742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000687216291200202,
+      "loss": 3.1858,
+      "step": 2743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006869987485898643,
+      "loss": 3.2024,
+      "step": 2744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006867811648164768,
+      "loss": 3.2454,
+      "step": 2745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006865635399279347,
+      "loss": 3.2286,
+      "step": 2746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006863458739721428,
+      "loss": 3.2516,
+      "step": 2747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006861281669970143,
+      "loss": 3.294,
+      "step": 2748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006859104190504725,
+      "loss": 3.4409,
+      "step": 2749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006856926301804488,
+      "loss": 3.3722,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000685474800434884,
+      "loss": 3.3042,
+      "step": 2751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006852569298617277,
+      "loss": 3.3402,
+      "step": 2752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006850390185089385,
+      "loss": 3.2451,
+      "step": 2753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006848210664244846,
+      "loss": 3.3127,
+      "step": 2754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006846030736563422,
+      "loss": 3.2795,
+      "step": 2755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000684385040252497,
+      "loss": 3.2501,
+      "step": 2756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006841669662609437,
+      "loss": 3.2865,
+      "step": 2757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006839488517296855,
+      "loss": 3.3314,
+      "step": 2758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006837306967067349,
+      "loss": 3.2642,
+      "step": 2759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006835125012401133,
+      "loss": 3.3216,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006832942653778509,
+      "loss": 3.1626,
+      "step": 2761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006830759891679866,
+      "loss": 3.173,
+      "step": 2762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006828576726585686,
+      "loss": 3.4248,
+      "step": 2763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000682639315897654,
+      "loss": 3.2822,
+      "step": 2764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006824209189333081,
+      "loss": 3.2014,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006822024818136056,
+      "loss": 3.5576,
+      "step": 2766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006819840045866301,
+      "loss": 3.4001,
+      "step": 2767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006817654873004737,
+      "loss": 3.2767,
+      "step": 2768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006815469300032373,
+      "loss": 3.2805,
+      "step": 2769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000681328332743031,
+      "loss": 3.3265,
+      "step": 2770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006811096955679734,
+      "loss": 3.4123,
+      "step": 2771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006808910185261918,
+      "loss": 3.2895,
+      "step": 2772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006806723016658228,
+      "loss": 3.3089,
+      "step": 2773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006804535450350111,
+      "loss": 3.1922,
+      "step": 2774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006802347486819104,
+      "loss": 3.2314,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006800159126546833,
+      "loss": 3.273,
+      "step": 2776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000679797037001501,
+      "loss": 3.3195,
+      "step": 2777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006795781217705435,
+      "loss": 3.2962,
+      "step": 2778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006793591670099994,
+      "loss": 3.305,
+      "step": 2779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006791401727680662,
+      "loss": 3.3173,
+      "step": 2780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006789211390929496,
+      "loss": 3.2163,
+      "step": 2781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006787020660328646,
+      "loss": 3.3347,
+      "step": 2782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006784829536360347,
+      "loss": 3.1824,
+      "step": 2783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006782638019506917,
+      "loss": 3.252,
+      "step": 2784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006780446110250765,
+      "loss": 3.2716,
+      "step": 2785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006778253809074384,
+      "loss": 3.3218,
+      "step": 2786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006776061116460353,
+      "loss": 3.2824,
+      "step": 2787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006773868032891335,
+      "loss": 3.2143,
+      "step": 2788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006771674558850088,
+      "loss": 3.2272,
+      "step": 2789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006769480694819447,
+      "loss": 3.345,
+      "step": 2790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006767286441282336,
+      "loss": 3.3204,
+      "step": 2791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006765091798721763,
+      "loss": 3.2337,
+      "step": 2792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006762896767620826,
+      "loss": 3.2564,
+      "step": 2793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006760701348462703,
+      "loss": 3.2738,
+      "step": 2794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006758505541730661,
+      "loss": 3.4115,
+      "step": 2795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006756309347908052,
+      "loss": 3.1912,
+      "step": 2796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006754112767478313,
+      "loss": 3.2432,
+      "step": 2797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006751915800924963,
+      "loss": 3.315,
+      "step": 2798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006749718448731611,
+      "loss": 3.2842,
+      "step": 2799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006747520711381948,
+      "loss": 3.1246,
+      "step": 2800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000674532258935975,
+      "loss": 3.1839,
+      "step": 2801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006743124083148878,
+      "loss": 3.2736,
+      "step": 2802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006740925193233281,
+      "loss": 3.2739,
+      "step": 2803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006738725920096983,
+      "loss": 3.3954,
+      "step": 2804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006736526264224101,
+      "loss": 3.3439,
+      "step": 2805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006734326226098835,
+      "loss": 3.2505,
+      "step": 2806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006732125806205468,
+      "loss": 3.356,
+      "step": 2807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006729925005028365,
+      "loss": 3.2247,
+      "step": 2808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006727723823051979,
+      "loss": 3.4164,
+      "step": 2809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000672552226076084,
+      "loss": 3.1969,
+      "step": 2810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006723320318639572,
+      "loss": 3.2439,
+      "step": 2811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006721117997172874,
+      "loss": 3.2822,
+      "step": 2812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006718915296845531,
+      "loss": 3.2659,
+      "step": 2813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006716712218142413,
+      "loss": 3.1938,
+      "step": 2814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006714508761548472,
+      "loss": 3.3545,
+      "step": 2815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006712304927548742,
+      "loss": 3.2272,
+      "step": 2816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006710100716628344,
+      "loss": 3.2744,
+      "step": 2817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006707896129272477,
+      "loss": 3.2233,
+      "step": 2818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006705691165966426,
+      "loss": 3.4276,
+      "step": 2819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006703485827195557,
+      "loss": 3.241,
+      "step": 2820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006701280113445324,
+      "loss": 3.2323,
+      "step": 2821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006699074025201253,
+      "loss": 3.2669,
+      "step": 2822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006696867562948961,
+      "loss": 3.2367,
+      "step": 2823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006694660727174146,
+      "loss": 3.328,
+      "step": 2824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006692453518362586,
+      "loss": 3.2613,
+      "step": 2825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006690245937000144,
+      "loss": 3.1883,
+      "step": 2826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006688037983572762,
+      "loss": 3.2014,
+      "step": 2827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006685829658566462,
+      "loss": 3.3016,
+      "step": 2828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006683620962467358,
+      "loss": 3.2019,
+      "step": 2829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006681411895761633,
+      "loss": 3.3221,
+      "step": 2830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006679202458935559,
+      "loss": 3.3052,
+      "step": 2831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006676992652475486,
+      "loss": 3.328,
+      "step": 2832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000667478247686785,
+      "loss": 3.2886,
+      "step": 2833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006672571932599164,
+      "loss": 3.1662,
+      "step": 2834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006670361020156024,
+      "loss": 3.1904,
+      "step": 2835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006668149740025104,
+      "loss": 3.3312,
+      "step": 2836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006665938092693165,
+      "loss": 3.2363,
+      "step": 2837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006663726078647045,
+      "loss": 3.2995,
+      "step": 2838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000666151369837366,
+      "loss": 3.3215,
+      "step": 2839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006659300952360011,
+      "loss": 3.2241,
+      "step": 2840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006657087841093179,
+      "loss": 3.1086,
+      "step": 2841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006654874365060325,
+      "loss": 3.2566,
+      "step": 2842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006652660524748688,
+      "loss": 3.2437,
+      "step": 2843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000665044632064559,
+      "loss": 3.2176,
+      "step": 2844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000664823175323843,
+      "loss": 3.1715,
+      "step": 2845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006646016823014694,
+      "loss": 3.3201,
+      "step": 2846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006643801530461938,
+      "loss": 3.0813,
+      "step": 2847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006641585876067806,
+      "loss": 3.2653,
+      "step": 2848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006639369860320016,
+      "loss": 3.3378,
+      "step": 2849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006637153483706369,
+      "loss": 3.2339,
+      "step": 2850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006634936746714745,
+      "loss": 3.2191,
+      "step": 2851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006632719649833101,
+      "loss": 3.2473,
+      "step": 2852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006630502193549475,
+      "loss": 3.3005,
+      "step": 2853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006628284378351985,
+      "loss": 3.3315,
+      "step": 2854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006626066204728826,
+      "loss": 3.3383,
+      "step": 2855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006623847673168276,
+      "loss": 3.1044,
+      "step": 2856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006621628784158685,
+      "loss": 3.256,
+      "step": 2857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006619409538188486,
+      "loss": 3.1375,
+      "step": 2858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006617189935746191,
+      "loss": 3.2039,
+      "step": 2859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006614969977320388,
+      "loss": 3.2952,
+      "step": 2860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006612749663399747,
+      "loss": 3.2014,
+      "step": 2861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006610528994473011,
+      "loss": 3.2924,
+      "step": 2862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006608307971029008,
+      "loss": 3.2879,
+      "step": 2863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006606086593556636,
+      "loss": 3.3133,
+      "step": 2864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006603864862544878,
+      "loss": 3.2257,
+      "step": 2865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006601642778482792,
+      "loss": 3.1595,
+      "step": 2866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006599420341859511,
+      "loss": 3.2108,
+      "step": 2867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006597197553164251,
+      "loss": 3.2792,
+      "step": 2868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006594974412886302,
+      "loss": 3.2907,
+      "step": 2869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006592750921515031,
+      "loss": 3.3022,
+      "step": 2870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006590527079539884,
+      "loss": 3.3651,
+      "step": 2871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006588302887450386,
+      "loss": 3.2856,
+      "step": 2872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006586078345736134,
+      "loss": 3.3225,
+      "step": 2873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006583853454886806,
+      "loss": 3.3379,
+      "step": 2874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006581628215392152,
+      "loss": 3.1857,
+      "step": 2875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006579402627742008,
+      "loss": 3.2187,
+      "step": 2876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006577176692426278,
+      "loss": 3.1648,
+      "step": 2877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006574950409934945,
+      "loss": 3.1616,
+      "step": 2878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006572723780758069,
+      "loss": 3.3054,
+      "step": 2879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006570496805385787,
+      "loss": 3.1509,
+      "step": 2880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000656826948430831,
+      "loss": 3.2289,
+      "step": 2881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006566041818015927,
+      "loss": 3.0798,
+      "step": 2882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006563813806999005,
+      "loss": 3.2209,
+      "step": 2883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000656158545174798,
+      "loss": 3.2085,
+      "step": 2884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006559356752753371,
+      "loss": 3.1369,
+      "step": 2885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000655712771050577,
+      "loss": 3.3144,
+      "step": 2886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006554898325495842,
+      "loss": 3.0628,
+      "step": 2887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006552668598214331,
+      "loss": 3.4049,
+      "step": 2888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006550438529152055,
+      "loss": 3.3048,
+      "step": 2889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006548208118799908,
+      "loss": 3.4122,
+      "step": 2890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006545977367648859,
+      "loss": 3.2144,
+      "step": 2891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006543746276189951,
+      "loss": 3.4288,
+      "step": 2892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006541514844914299,
+      "loss": 3.2563,
+      "step": 2893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006539283074313101,
+      "loss": 3.2746,
+      "step": 2894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006537050964877625,
+      "loss": 3.282,
+      "step": 2895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006534818517099209,
+      "loss": 3.1768,
+      "step": 2896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006532585731469274,
+      "loss": 3.2541,
+      "step": 2897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000653035260847931,
+      "loss": 3.3542,
+      "step": 2898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006528119148620882,
+      "loss": 3.3448,
+      "step": 2899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006525885352385631,
+      "loss": 3.3586,
+      "step": 2900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006523651220265269,
+      "loss": 3.2899,
+      "step": 2901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006521416752751586,
+      "loss": 3.2887,
+      "step": 2902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006519181950336441,
+      "loss": 3.2911,
+      "step": 2903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006516946813511774,
+      "loss": 3.3432,
+      "step": 2904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006514711342769588,
+      "loss": 3.2196,
+      "step": 2905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006512475538601968,
+      "loss": 3.2621,
+      "step": 2906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000651023940150107,
+      "loss": 3.2837,
+      "step": 2907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006508002931959125,
+      "loss": 3.2013,
+      "step": 2908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006505766130468431,
+      "loss": 3.2471,
+      "step": 2909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006503528997521365,
+      "loss": 3.2407,
+      "step": 2910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006501291533610377,
+      "loss": 3.1515,
+      "step": 2911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006499053739227989,
+      "loss": 3.2824,
+      "step": 2912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006496815614866791,
+      "loss": 3.3025,
+      "step": 2913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006494577161019451,
+      "loss": 3.13,
+      "step": 2914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006492338378178708,
+      "loss": 3.2099,
+      "step": 2915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006490099266837376,
+      "loss": 3.3331,
+      "step": 2916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006487859827488336,
+      "loss": 3.1048,
+      "step": 2917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006485620060624545,
+      "loss": 3.2354,
+      "step": 2918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006483379966739028,
+      "loss": 3.4434,
+      "step": 2919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006481139546324891,
+      "loss": 3.2768,
+      "step": 2920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006478898799875302,
+      "loss": 3.2981,
+      "step": 2921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006476657727883506,
+      "loss": 3.2779,
+      "step": 2922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006474416330842818,
+      "loss": 3.3172,
+      "step": 2923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006472174609246625,
+      "loss": 3.3083,
+      "step": 2924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006469932563588386,
+      "loss": 3.3294,
+      "step": 2925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006467690194361628,
+      "loss": 3.3147,
+      "step": 2926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006465447502059954,
+      "loss": 3.2609,
+      "step": 2927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006463204487177037,
+      "loss": 3.1807,
+      "step": 2928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006460961150206619,
+      "loss": 3.1629,
+      "step": 2929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006458717491642514,
+      "loss": 3.298,
+      "step": 2930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006456473511978607,
+      "loss": 3.3014,
+      "step": 2931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006454229211708853,
+      "loss": 3.2526,
+      "step": 2932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006451984591327278,
+      "loss": 3.4198,
+      "step": 2933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006449739651327979,
+      "loss": 3.1244,
+      "step": 2934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006447494392205122,
+      "loss": 3.3049,
+      "step": 2935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006445248814452944,
+      "loss": 3.2626,
+      "step": 2936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006443002918565754,
+      "loss": 3.06,
+      "step": 2937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006440756705037927,
+      "loss": 3.3338,
+      "step": 2938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006438510174363912,
+      "loss": 3.2738,
+      "step": 2939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006436263327038224,
+      "loss": 3.2552,
+      "step": 2940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006434016163555452,
+      "loss": 3.1693,
+      "step": 2941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006431768684410251,
+      "loss": 3.2715,
+      "step": 2942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006429520890097346,
+      "loss": 3.1939,
+      "step": 2943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006427272781111533,
+      "loss": 3.2692,
+      "step": 2944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006425024357947677,
+      "loss": 3.1504,
+      "step": 2945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000642277562110071,
+      "loss": 3.3287,
+      "step": 2946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006420526571065636,
+      "loss": 3.1353,
+      "step": 2947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006418277208337526,
+      "loss": 3.1197,
+      "step": 2948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000641602753341152,
+      "loss": 3.2429,
+      "step": 2949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006413777546782828,
+      "loss": 3.2656,
+      "step": 2950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006411527248946728,
+      "loss": 3.2815,
+      "step": 2951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006409276640398564,
+      "loss": 3.3028,
+      "step": 2952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006407025721633752,
+      "loss": 3.381,
+      "step": 2953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006404774493147777,
+      "loss": 3.3677,
+      "step": 2954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006402522955436187,
+      "loss": 3.2937,
+      "step": 2955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006400271108994602,
+      "loss": 3.1929,
+      "step": 2956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000639801895431871,
+      "loss": 3.1225,
+      "step": 2957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006395766491904262,
+      "loss": 3.3399,
+      "step": 2958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006393513722247089,
+      "loss": 3.2407,
+      "step": 2959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006391260645843074,
+      "loss": 3.2491,
+      "step": 2960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006389007263188176,
+      "loss": 3.2679,
+      "step": 2961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006386753574778421,
+      "loss": 3.2853,
+      "step": 2962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006384499581109903,
+      "loss": 3.2665,
+      "step": 2963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006382245282678781,
+      "loss": 3.2671,
+      "step": 2964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006379990679981281,
+      "loss": 3.2733,
+      "step": 2965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006377735773513697,
+      "loss": 3.1639,
+      "step": 2966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000637548056377239,
+      "loss": 3.1967,
+      "step": 2967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000637322505125379,
+      "loss": 3.1617,
+      "step": 2968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006370969236454385,
+      "loss": 3.2546,
+      "step": 2969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006368713119870742,
+      "loss": 3.253,
+      "step": 2970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006366456701999483,
+      "loss": 3.2741,
+      "step": 2971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006364199983337305,
+      "loss": 3.3579,
+      "step": 2972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006361942964380967,
+      "loss": 3.2905,
+      "step": 2973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006359685645627294,
+      "loss": 3.2896,
+      "step": 2974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006357428027573174,
+      "loss": 3.2191,
+      "step": 2975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006355170110715571,
+      "loss": 3.258,
+      "step": 2976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006352911895551507,
+      "loss": 3.1467,
+      "step": 2977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006350653382578068,
+      "loss": 3.2894,
+      "step": 2978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006348394572292411,
+      "loss": 3.249,
+      "step": 2979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006346135465191754,
+      "loss": 3.2811,
+      "step": 2980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006343876061773385,
+      "loss": 3.2736,
+      "step": 2981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006341616362534652,
+      "loss": 3.3147,
+      "step": 2982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006339356367972973,
+      "loss": 3.1584,
+      "step": 2983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006337096078585825,
+      "loss": 3.2012,
+      "step": 2984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006334835494870759,
+      "loss": 3.259,
+      "step": 2985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006332574617325381,
+      "loss": 3.1947,
+      "step": 2986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006330313446447368,
+      "loss": 3.2572,
+      "step": 2987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000632805198273446,
+      "loss": 3.2683,
+      "step": 2988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006325790226684459,
+      "loss": 3.2501,
+      "step": 2989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006323528178795236,
+      "loss": 3.2299,
+      "step": 2990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006321265839564722,
+      "loss": 3.2231,
+      "step": 2991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006319003209490914,
+      "loss": 3.3582,
+      "step": 2992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006316740289071872,
+      "loss": 3.2329,
+      "step": 2993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006314477078805723,
+      "loss": 3.3179,
+      "step": 2994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006312213579190655,
+      "loss": 3.2708,
+      "step": 2995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006309949790724922,
+      "loss": 3.2385,
+      "step": 2996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006307685713906834,
+      "loss": 3.2586,
+      "step": 2997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006305421349234774,
+      "loss": 3.0786,
+      "step": 2998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006303156697207187,
+      "loss": 3.266,
+      "step": 2999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006300891758322575,
+      "loss": 3.2404,
+      "step": 3000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006298626533079508,
+      "loss": 3.1514,
+      "step": 3001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006296361021976619,
+      "loss": 3.2389,
+      "step": 3002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006294095225512603,
+      "loss": 3.2675,
+      "step": 3003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006291829144186218,
+      "loss": 3.211,
+      "step": 3004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006289562778496285,
+      "loss": 3.3841,
+      "step": 3005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006287296128941684,
+      "loss": 3.3493,
+      "step": 3006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006285029196021365,
+      "loss": 3.2789,
+      "step": 3007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006282761980234332,
+      "loss": 3.2232,
+      "step": 3008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000628049448207966,
+      "loss": 3.2507,
+      "step": 3009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006278226702056476,
+      "loss": 3.3205,
+      "step": 3010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006275958640663976,
+      "loss": 3.3396,
+      "step": 3011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000627369029840142,
+      "loss": 3.2751,
+      "step": 3012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006271421675768123,
+      "loss": 3.3025,
+      "step": 3013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006269152773263467,
+      "loss": 3.3544,
+      "step": 3014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006266883591386891,
+      "loss": 3.2833,
+      "step": 3015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006264614130637899,
+      "loss": 3.2594,
+      "step": 3016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006262344391516056,
+      "loss": 3.234,
+      "step": 3017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006260074374520988,
+      "loss": 3.189,
+      "step": 3018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000625780408015238,
+      "loss": 3.3295,
+      "step": 3019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006255533508909982,
+      "loss": 3.2597,
+      "step": 3020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006253262661293602,
+      "loss": 3.3978,
+      "step": 3021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006250991537803111,
+      "loss": 3.2721,
+      "step": 3022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006248720138938438,
+      "loss": 3.2833,
+      "step": 3023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006246448465199576,
+      "loss": 3.2559,
+      "step": 3024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006244176517086572,
+      "loss": 3.2385,
+      "step": 3025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006241904295099545,
+      "loss": 3.3482,
+      "step": 3026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006239631799738658,
+      "loss": 3.1604,
+      "step": 3027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006237359031504153,
+      "loss": 3.1872,
+      "step": 3028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006235085990896317,
+      "loss": 3.22,
+      "step": 3029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006232812678415504,
+      "loss": 3.2486,
+      "step": 3030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006230539094562125,
+      "loss": 3.2382,
+      "step": 3031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006228265239836653,
+      "loss": 3.3039,
+      "step": 3032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006225991114739622,
+      "loss": 3.2392,
+      "step": 3033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006223716719771619,
+      "loss": 3.3457,
+      "step": 3034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006221442055433297,
+      "loss": 3.3264,
+      "step": 3035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006219167122225365,
+      "loss": 3.283,
+      "step": 3036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006216891920648593,
+      "loss": 3.2296,
+      "step": 3037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000621461645120381,
+      "loss": 3.3089,
+      "step": 3038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00062123407143919,
+      "loss": 3.1844,
+      "step": 3039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006210064710713811,
+      "loss": 3.174,
+      "step": 3040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006207788440670546,
+      "loss": 3.2376,
+      "step": 3041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006205511904763172,
+      "loss": 3.1327,
+      "step": 3042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000620323510349281,
+      "loss": 3.2842,
+      "step": 3043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006200958037360636,
+      "loss": 3.3093,
+      "step": 3044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006198680706867892,
+      "loss": 3.2023,
+      "step": 3045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006196403112515874,
+      "loss": 3.351,
+      "step": 3046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006194125254805938,
+      "loss": 3.2677,
+      "step": 3047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006191847134239495,
+      "loss": 3.2655,
+      "step": 3048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006189568751318018,
+      "loss": 3.2504,
+      "step": 3049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006187290106543032,
+      "loss": 3.2788,
+      "step": 3050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006185011200416128,
+      "loss": 3.1929,
+      "step": 3051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006182732033438945,
+      "loss": 3.3505,
+      "step": 3052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006180452606113185,
+      "loss": 3.3154,
+      "step": 3053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006178172918940608,
+      "loss": 3.2751,
+      "step": 3054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006175892972423029,
+      "loss": 3.1969,
+      "step": 3055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000617361276706232,
+      "loss": 3.4048,
+      "step": 3056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006171332303360411,
+      "loss": 3.2562,
+      "step": 3057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006169051581819288,
+      "loss": 3.3747,
+      "step": 3058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006166770602940994,
+      "loss": 3.1618,
+      "step": 3059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006164489367227633,
+      "loss": 3.2322,
+      "step": 3060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006162207875181353,
+      "loss": 3.3375,
+      "step": 3061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006159926127304373,
+      "loss": 3.2261,
+      "step": 3062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006157644124098963,
+      "loss": 3.2276,
+      "step": 3063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006155361866067445,
+      "loss": 3.1939,
+      "step": 3064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006153079353712201,
+      "loss": 3.2969,
+      "step": 3065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006150796587535669,
+      "loss": 3.2332,
+      "step": 3066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006148513568040344,
+      "loss": 3.3344,
+      "step": 3067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006146230295728771,
+      "loss": 3.2634,
+      "step": 3068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000614394677110356,
+      "loss": 3.2296,
+      "step": 3069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006141662994667368,
+      "loss": 3.1233,
+      "step": 3070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006139378966922908,
+      "loss": 3.2563,
+      "step": 3071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006137094688372956,
+      "loss": 3.3209,
+      "step": 3072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006134810159520337,
+      "loss": 3.1583,
+      "step": 3073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000613252538086793,
+      "loss": 3.2603,
+      "step": 3074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006130240352918674,
+      "loss": 3.2493,
+      "step": 3075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006127955076175559,
+      "loss": 3.1237,
+      "step": 3076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000612566955114163,
+      "loss": 3.3243,
+      "step": 3077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006123383778319991,
+      "loss": 3.2638,
+      "step": 3078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006121097758213793,
+      "loss": 3.2308,
+      "step": 3079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006118811491326249,
+      "loss": 3.2996,
+      "step": 3080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006116524978160619,
+      "loss": 3.3955,
+      "step": 3081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006114238219220223,
+      "loss": 3.2582,
+      "step": 3082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006111951215008437,
+      "loss": 3.263,
+      "step": 3083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006109663966028681,
+      "loss": 3.1795,
+      "step": 3084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006107376472784437,
+      "loss": 3.2425,
+      "step": 3085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000610508873577924,
+      "loss": 3.1689,
+      "step": 3086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006102800755516679,
+      "loss": 3.3359,
+      "step": 3087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006100512532500393,
+      "loss": 3.4719,
+      "step": 3088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006098224067234077,
+      "loss": 3.198,
+      "step": 3089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006095935360221476,
+      "loss": 3.2933,
+      "step": 3090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006093646411966396,
+      "loss": 3.2318,
+      "step": 3091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006091357222972687,
+      "loss": 3.217,
+      "step": 3092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006089067793744258,
+      "loss": 3.2589,
+      "step": 3093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006086778124785068,
+      "loss": 3.3251,
+      "step": 3094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006084488216599133,
+      "loss": 3.1981,
+      "step": 3095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006082198069690515,
+      "loss": 3.2685,
+      "step": 3096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006079907684563333,
+      "loss": 3.393,
+      "step": 3097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006077617061721756,
+      "loss": 3.2113,
+      "step": 3098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000607532620167001,
+      "loss": 3.1016,
+      "step": 3099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006073035104912367,
+      "loss": 3.2169,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006070743771953157,
+      "loss": 3.2454,
+      "step": 3101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006068452203296754,
+      "loss": 3.2152,
+      "step": 3102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006066160399447594,
+      "loss": 3.2339,
+      "step": 3103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006063868360910159,
+      "loss": 3.227,
+      "step": 3104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000606157608818898,
+      "loss": 3.1516,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006059283581788648,
+      "loss": 3.119,
+      "step": 3106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006056990842213796,
+      "loss": 3.2774,
+      "step": 3107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006054697869969114,
+      "loss": 3.3809,
+      "step": 3108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006052404665559341,
+      "loss": 3.297,
+      "step": 3109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000605011122948927,
+      "loss": 3.2112,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006047817562263744,
+      "loss": 3.2805,
+      "step": 3111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006045523664387652,
+      "loss": 3.3048,
+      "step": 3112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006043229536365941,
+      "loss": 3.3136,
+      "step": 3113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006040935178703604,
+      "loss": 3.1619,
+      "step": 3114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006038640591905685,
+      "loss": 3.2138,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006036345776477281,
+      "loss": 3.0883,
+      "step": 3116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006034050732923537,
+      "loss": 3.3112,
+      "step": 3117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006031755461749648,
+      "loss": 3.2973,
+      "step": 3118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006029459963460861,
+      "loss": 3.2762,
+      "step": 3119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006027164238562472,
+      "loss": 3.2457,
+      "step": 3120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006024868287559826,
+      "loss": 3.36,
+      "step": 3121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006022572110958319,
+      "loss": 3.2776,
+      "step": 3122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006020275709263397,
+      "loss": 3.1317,
+      "step": 3123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006017979082980553,
+      "loss": 3.3039,
+      "step": 3124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0006015682232615336,
+      "loss": 3.3179,
+      "step": 3125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0006013385158673333,
+      "loss": 3.2328,
+      "step": 3126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000601108786166019,
+      "loss": 3.2392,
+      "step": 3127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00060087903420816,
+      "loss": 3.2182,
+      "step": 3128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0006006492600443301,
+      "loss": 3.2096,
+      "step": 3129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0006004194637251085,
+      "loss": 3.2318,
+      "step": 3130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000600189645301079,
+      "loss": 3.3465,
+      "step": 3131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005999598048228302,
+      "loss": 3.3014,
+      "step": 3132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005997299423409559,
+      "loss": 3.0876,
+      "step": 3133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005995000579060545,
+      "loss": 3.2957,
+      "step": 3134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005992701515687291,
+      "loss": 3.2702,
+      "step": 3135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005990402233795877,
+      "loss": 3.3656,
+      "step": 3136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005988102733892434,
+      "loss": 3.1968,
+      "step": 3137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005985803016483138,
+      "loss": 3.2172,
+      "step": 3138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005983503082074213,
+      "loss": 3.3685,
+      "step": 3139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005981202931171933,
+      "loss": 3.2492,
+      "step": 3140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005978902564282616,
+      "loss": 3.3206,
+      "step": 3141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005976601981912632,
+      "loss": 3.2116,
+      "step": 3142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005974301184568395,
+      "loss": 3.2235,
+      "step": 3143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005972000172756366,
+      "loss": 3.2545,
+      "step": 3144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005969698946983055,
+      "loss": 3.2059,
+      "step": 3145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005967397507755022,
+      "loss": 3.3206,
+      "step": 3146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005965095855578869,
+      "loss": 3.1173,
+      "step": 3147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005962793990961243,
+      "loss": 3.2464,
+      "step": 3148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005960491914408845,
+      "loss": 3.1963,
+      "step": 3149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005958189626428419,
+      "loss": 3.2827,
+      "step": 3150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005955887127526754,
+      "loss": 3.2277,
+      "step": 3151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005953584418210689,
+      "loss": 3.2113,
+      "step": 3152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005951281498987105,
+      "loss": 3.2404,
+      "step": 3153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005948978370362933,
+      "loss": 3.1788,
+      "step": 3154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005946675032845148,
+      "loss": 3.2937,
+      "step": 3155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005944371486940772,
+      "loss": 3.2668,
+      "step": 3156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005942067733156871,
+      "loss": 3.2224,
+      "step": 3157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005939763772000559,
+      "loss": 3.2991,
+      "step": 3158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005937459603978997,
+      "loss": 3.3603,
+      "step": 3159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005935155229599386,
+      "loss": 3.2502,
+      "step": 3160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005932850649368978,
+      "loss": 3.2464,
+      "step": 3161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005930545863795067,
+      "loss": 3.188,
+      "step": 3162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005928240873384993,
+      "loss": 3.1663,
+      "step": 3163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005925935678646144,
+      "loss": 3.2398,
+      "step": 3164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005923630280085947,
+      "loss": 3.2748,
+      "step": 3165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005921324678211879,
+      "loss": 3.1523,
+      "step": 3166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000591901887353146,
+      "loss": 3.2155,
+      "step": 3167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005916712866552254,
+      "loss": 3.3291,
+      "step": 3168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005914406657781871,
+      "loss": 3.2218,
+      "step": 3169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005912100247727964,
+      "loss": 3.2184,
+      "step": 3170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005909793636898231,
+      "loss": 3.3097,
+      "step": 3171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005907486825800414,
+      "loss": 3.267,
+      "step": 3172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005905179814942301,
+      "loss": 3.2012,
+      "step": 3173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005902872604831719,
+      "loss": 3.315,
+      "step": 3174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005900565195976543,
+      "loss": 3.2357,
+      "step": 3175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005898257588884691,
+      "loss": 3.3202,
+      "step": 3176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005895949784064127,
+      "loss": 3.3149,
+      "step": 3177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005893641782022852,
+      "loss": 3.2979,
+      "step": 3178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005891333583268916,
+      "loss": 3.3004,
+      "step": 3179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000588902518831041,
+      "loss": 3.1125,
+      "step": 3180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005886716597655471,
+      "loss": 3.3378,
+      "step": 3181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005884407811812275,
+      "loss": 3.3147,
+      "step": 3182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005882098831289043,
+      "loss": 3.2988,
+      "step": 3183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005879789656594037,
+      "loss": 3.2934,
+      "step": 3184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005877480288235568,
+      "loss": 3.4488,
+      "step": 3185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005875170726721982,
+      "loss": 3.1786,
+      "step": 3186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005872860972561671,
+      "loss": 3.2047,
+      "step": 3187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005870551026263069,
+      "loss": 3.1891,
+      "step": 3188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 3.218,
+      "step": 3189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000586593055928494,
+      "loss": 3.2997,
+      "step": 3190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005863620039622491,
+      "loss": 3.0456,
+      "step": 3191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000586130932985591,
+      "loss": 3.2112,
+      "step": 3192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005858998430493841,
+      "loss": 3.242,
+      "step": 3193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005856687342044968,
+      "loss": 3.2181,
+      "step": 3194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005854376065018021,
+      "loss": 3.2327,
+      "step": 3195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000585206459992177,
+      "loss": 3.2053,
+      "step": 3196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005849752947265023,
+      "loss": 3.3143,
+      "step": 3197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005847441107556632,
+      "loss": 3.2194,
+      "step": 3198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005845129081305492,
+      "loss": 3.1319,
+      "step": 3199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005842816869020538,
+      "loss": 3.2735,
+      "step": 3200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005840504471210741,
+      "loss": 3.2891,
+      "step": 3201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005838191888385121,
+      "loss": 3.2134,
+      "step": 3202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005835879121052731,
+      "loss": 3.2739,
+      "step": 3203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005833566169722674,
+      "loss": 3.1864,
+      "step": 3204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005831253034904082,
+      "loss": 3.303,
+      "step": 3205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005828939717106137,
+      "loss": 3.2396,
+      "step": 3206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005826626216838053,
+      "loss": 3.1644,
+      "step": 3207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005824312534609094,
+      "loss": 3.4487,
+      "step": 3208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005821998670928556,
+      "loss": 3.2508,
+      "step": 3209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005819684626305776,
+      "loss": 3.2202,
+      "step": 3210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005817370401250136,
+      "loss": 3.2755,
+      "step": 3211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005815055996271052,
+      "loss": 3.1585,
+      "step": 3212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005812741411877982,
+      "loss": 3.3328,
+      "step": 3213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005810426648580425,
+      "loss": 3.1315,
+      "step": 3214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005808111706887913,
+      "loss": 3.316,
+      "step": 3215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005805796587310028,
+      "loss": 3.144,
+      "step": 3216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005803481290356382,
+      "loss": 3.2485,
+      "step": 3217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005801165816536628,
+      "loss": 3.3051,
+      "step": 3218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005798850166360461,
+      "loss": 3.2119,
+      "step": 3219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005796534340337614,
+      "loss": 3.1534,
+      "step": 3220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005794218338977853,
+      "loss": 3.3262,
+      "step": 3221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005791902162790992,
+      "loss": 3.2959,
+      "step": 3222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005789585812286877,
+      "loss": 3.2719,
+      "step": 3223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005787269287975392,
+      "loss": 3.2357,
+      "step": 3224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005784952590366464,
+      "loss": 3.235,
+      "step": 3225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005782635719970055,
+      "loss": 3.1888,
+      "step": 3226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005780318677296166,
+      "loss": 3.2985,
+      "step": 3227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005778001462854831,
+      "loss": 3.3284,
+      "step": 3228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005775684077156132,
+      "loss": 3.1872,
+      "step": 3229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005773366520710181,
+      "loss": 3.2582,
+      "step": 3230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005771048794027128,
+      "loss": 3.2592,
+      "step": 3231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005768730897617164,
+      "loss": 3.2764,
+      "step": 3232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005766412831990513,
+      "loss": 3.218,
+      "step": 3233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000576409459765744,
+      "loss": 3.3444,
+      "step": 3234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005761776195128248,
+      "loss": 3.2446,
+      "step": 3235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005759457624913269,
+      "loss": 3.1785,
+      "step": 3236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005757138887522884,
+      "loss": 3.2727,
+      "step": 3237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00057548199834675,
+      "loss": 3.301,
+      "step": 3238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005752500913257568,
+      "loss": 3.2736,
+      "step": 3239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005750181677403569,
+      "loss": 3.2852,
+      "step": 3240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005747862276416028,
+      "loss": 3.1935,
+      "step": 3241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005745542710805502,
+      "loss": 3.2986,
+      "step": 3242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005743222981082582,
+      "loss": 3.228,
+      "step": 3243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005740903087757902,
+      "loss": 3.2497,
+      "step": 3244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005738583031342123,
+      "loss": 3.3076,
+      "step": 3245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005736262812345951,
+      "loss": 3.3299,
+      "step": 3246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005733942431280123,
+      "loss": 3.2272,
+      "step": 3247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005731621888655409,
+      "loss": 3.2522,
+      "step": 3248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005729301184982621,
+      "loss": 3.2882,
+      "step": 3249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005726980320772602,
+      "loss": 3.1968,
+      "step": 3250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005724659296536234,
+      "loss": 3.2209,
+      "step": 3251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005722338112784429,
+      "loss": 3.1716,
+      "step": 3252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005720016770028138,
+      "loss": 3.2655,
+      "step": 3253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005717695268778344,
+      "loss": 3.1404,
+      "step": 3254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000571537360954607,
+      "loss": 3.1403,
+      "step": 3255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005713051792842371,
+      "loss": 3.2154,
+      "step": 3256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005710729819178333,
+      "loss": 3.318,
+      "step": 3257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005710729819178333,
+      "loss": 3.2924,
+      "step": 3258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005708407689065082,
+      "loss": 3.2685,
+      "step": 3259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005706085403013773,
+      "loss": 3.2382,
+      "step": 3260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005703762961535604,
+      "loss": 3.3208,
+      "step": 3261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005701440365141799,
+      "loss": 3.0944,
+      "step": 3262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005699117614343618,
+      "loss": 3.175,
+      "step": 3263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005696794709652358,
+      "loss": 3.1948,
+      "step": 3264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005694471651579346,
+      "loss": 3.1416,
+      "step": 3265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005692148440635946,
+      "loss": 3.1449,
+      "step": 3266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005689825077333552,
+      "loss": 3.3023,
+      "step": 3267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005687501562183596,
+      "loss": 3.358,
+      "step": 3268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005685177895697541,
+      "loss": 3.2512,
+      "step": 3269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005682854078386882,
+      "loss": 3.2648,
+      "step": 3270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000568053011076315,
+      "loss": 3.0966,
+      "step": 3271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005678205993337907,
+      "loss": 3.2642,
+      "step": 3272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005675881726622749,
+      "loss": 3.2958,
+      "step": 3273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005673557311129306,
+      "loss": 3.2105,
+      "step": 3274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005671232747369236,
+      "loss": 3.299,
+      "step": 3275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005668908035854236,
+      "loss": 3.2727,
+      "step": 3276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005666583177096029,
+      "loss": 3.1718,
+      "step": 3277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005664258171606379,
+      "loss": 3.1595,
+      "step": 3278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005661933019897074,
+      "loss": 3.1777,
+      "step": 3279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005659607722479938,
+      "loss": 3.2947,
+      "step": 3280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005657282279866826,
+      "loss": 3.187,
+      "step": 3281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005654956692569627,
+      "loss": 3.1988,
+      "step": 3282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000565263096110026,
+      "loss": 3.2905,
+      "step": 3283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005650305085970672,
+      "loss": 3.283,
+      "step": 3284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005647979067692851,
+      "loss": 3.2282,
+      "step": 3285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005645652906778808,
+      "loss": 3.232,
+      "step": 3286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000564332660374059,
+      "loss": 3.3272,
+      "step": 3287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005641000159090272,
+      "loss": 3.3957,
+      "step": 3288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005638673573339966,
+      "loss": 3.2655,
+      "step": 3289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005636346847001806,
+      "loss": 3.2619,
+      "step": 3290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005634019980587968,
+      "loss": 3.249,
+      "step": 3291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005631692974610648,
+      "loss": 3.2037,
+      "step": 3292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005629365829582077,
+      "loss": 3.2209,
+      "step": 3293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005627038546014519,
+      "loss": 3.1988,
+      "step": 3294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005624711124420269,
+      "loss": 3.33,
+      "step": 3295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005622383565311646,
+      "loss": 3.2762,
+      "step": 3296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005620055869201007,
+      "loss": 3.1339,
+      "step": 3297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005617728036600734,
+      "loss": 3.1705,
+      "step": 3298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005615400068023238,
+      "loss": 3.3136,
+      "step": 3299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005613071963980966,
+      "loss": 3.1917,
+      "step": 3300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000561074372498639,
+      "loss": 3.2731,
+      "step": 3301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005608415351552013,
+      "loss": 3.2817,
+      "step": 3302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005606086844190367,
+      "loss": 3.3092,
+      "step": 3303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005603758203414015,
+      "loss": 3.242,
+      "step": 3304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005601429429735547,
+      "loss": 3.2815,
+      "step": 3305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005599100523667587,
+      "loss": 3.2599,
+      "step": 3306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005596771485722781,
+      "loss": 3.1389,
+      "step": 3307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000559444231641381,
+      "loss": 3.2742,
+      "step": 3308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000559211301625338,
+      "loss": 3.2601,
+      "step": 3309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005589783585754232,
+      "loss": 3.3026,
+      "step": 3310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005587454025429126,
+      "loss": 3.2131,
+      "step": 3311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005585124335790857,
+      "loss": 3.1579,
+      "step": 3312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005582794517352249,
+      "loss": 3.2619,
+      "step": 3313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005580464570626152,
+      "loss": 3.3872,
+      "step": 3314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005578134496125445,
+      "loss": 3.2571,
+      "step": 3315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005575804294363034,
+      "loss": 3.344,
+      "step": 3316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005573473965851855,
+      "loss": 3.3545,
+      "step": 3317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005571143511104873,
+      "loss": 3.1947,
+      "step": 3318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005568812930635076,
+      "loss": 3.2134,
+      "step": 3319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000556648222495548,
+      "loss": 3.2624,
+      "step": 3320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005564151394579136,
+      "loss": 3.2663,
+      "step": 3321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005561820440019116,
+      "loss": 3.147,
+      "step": 3322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000555948936178852,
+      "loss": 3.2098,
+      "step": 3323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005557158160400476,
+      "loss": 3.2802,
+      "step": 3324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005554826836368139,
+      "loss": 3.2689,
+      "step": 3325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005552495390204691,
+      "loss": 3.0766,
+      "step": 3326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005550163822423341,
+      "loss": 3.1545,
+      "step": 3327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005547832133537327,
+      "loss": 3.347,
+      "step": 3328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005545500324059907,
+      "loss": 3.2382,
+      "step": 3329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005543168394504374,
+      "loss": 3.2939,
+      "step": 3330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005540836345384042,
+      "loss": 3.2832,
+      "step": 3331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005538504177212251,
+      "loss": 3.1392,
+      "step": 3332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000553617189050237,
+      "loss": 3.1596,
+      "step": 3333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005533839485767795,
+      "loss": 3.2845,
+      "step": 3334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005531506963521943,
+      "loss": 3.3786,
+      "step": 3335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005529174324278261,
+      "loss": 3.2478,
+      "step": 3336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005526841568550223,
+      "loss": 3.0776,
+      "step": 3337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005524508696851322,
+      "loss": 3.2214,
+      "step": 3338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005522175709695084,
+      "loss": 3.2402,
+      "step": 3339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005519842607595055,
+      "loss": 3.2881,
+      "step": 3340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005517509391064809,
+      "loss": 3.2784,
+      "step": 3341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005515176060617945,
+      "loss": 3.2176,
+      "step": 3342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005512842616768087,
+      "loss": 3.1566,
+      "step": 3343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005510509060028885,
+      "loss": 3.2161,
+      "step": 3344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000550817539091401,
+      "loss": 3.1768,
+      "step": 3345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005505841609937161,
+      "loss": 3.2538,
+      "step": 3346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005503507717612062,
+      "loss": 3.3809,
+      "step": 3347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000550117371445246,
+      "loss": 3.3108,
+      "step": 3348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005498839600972126,
+      "loss": 3.1988,
+      "step": 3349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005496505377684857,
+      "loss": 3.2589,
+      "step": 3350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005494171045104472,
+      "loss": 3.1137,
+      "step": 3351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005491836603744817,
+      "loss": 3.3186,
+      "step": 3352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000548950205411976,
+      "loss": 3.2261,
+      "step": 3353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005487167396743194,
+      "loss": 3.3226,
+      "step": 3354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005484832632129031,
+      "loss": 3.2397,
+      "step": 3355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005482497760791214,
+      "loss": 3.0879,
+      "step": 3356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005480162783243706,
+      "loss": 3.2543,
+      "step": 3357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005477827700000492,
+      "loss": 3.3057,
+      "step": 3358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000547549251157558,
+      "loss": 3.2111,
+      "step": 3359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005473157218483006,
+      "loss": 3.3418,
+      "step": 3360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005470821821236823,
+      "loss": 3.205,
+      "step": 3361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005468486320351113,
+      "loss": 3.363,
+      "step": 3362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005466150716339975,
+      "loss": 3.2649,
+      "step": 3363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005463815009717533,
+      "loss": 3.1846,
+      "step": 3364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005461479200997935,
+      "loss": 3.2673,
+      "step": 3365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005459143290695351,
+      "loss": 3.2658,
+      "step": 3366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005456807279323971,
+      "loss": 3.2788,
+      "step": 3367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005454471167398008,
+      "loss": 3.2777,
+      "step": 3368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00054521349554317,
+      "loss": 3.2331,
+      "step": 3369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005449798643939305,
+      "loss": 3.315,
+      "step": 3370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005447462233435102,
+      "loss": 3.2811,
+      "step": 3371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005445125724433393,
+      "loss": 3.2094,
+      "step": 3372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005442789117448501,
+      "loss": 3.236,
+      "step": 3373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005440452412994774,
+      "loss": 3.2111,
+      "step": 3374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005438115611586575,
+      "loss": 3.1314,
+      "step": 3375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005435778713738292,
+      "loss": 3.3152,
+      "step": 3376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005433441719964333,
+      "loss": 3.183,
+      "step": 3377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005431104630779133,
+      "loss": 3.2483,
+      "step": 3378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005428767446697139,
+      "loss": 3.2409,
+      "step": 3379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005426430168232826,
+      "loss": 3.1943,
+      "step": 3380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005424092795900682,
+      "loss": 3.1581,
+      "step": 3381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005421755330215223,
+      "loss": 3.1605,
+      "step": 3382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005419417771690987,
+      "loss": 3.3837,
+      "step": 3383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005417080120842523,
+      "loss": 3.1582,
+      "step": 3384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005414742378184408,
+      "loss": 3.2823,
+      "step": 3385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005412404544231235,
+      "loss": 3.2681,
+      "step": 3386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005410066619497623,
+      "loss": 3.1736,
+      "step": 3387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005407728604498204,
+      "loss": 3.238,
+      "step": 3388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005405390499747633,
+      "loss": 3.2183,
+      "step": 3389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005403052305760586,
+      "loss": 3.1151,
+      "step": 3390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005400714023051754,
+      "loss": 3.3011,
+      "step": 3391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005398375652135858,
+      "loss": 3.2663,
+      "step": 3392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005396037193527626,
+      "loss": 3.1158,
+      "step": 3393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005393698647741809,
+      "loss": 3.0783,
+      "step": 3394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005391360015293183,
+      "loss": 3.0968,
+      "step": 3395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005389021296696537,
+      "loss": 3.1548,
+      "step": 3396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005386682492466681,
+      "loss": 3.2438,
+      "step": 3397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005384343603118443,
+      "loss": 3.1949,
+      "step": 3398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005382004629166671,
+      "loss": 3.2351,
+      "step": 3399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005379665571126231,
+      "loss": 3.2981,
+      "step": 3400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005377326429512008,
+      "loss": 3.1211,
+      "step": 3401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005374987204838906,
+      "loss": 3.2632,
+      "step": 3402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005372647897621842,
+      "loss": 3.2347,
+      "step": 3403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005370308508375758,
+      "loss": 3.1656,
+      "step": 3404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005367969037615613,
+      "loss": 3.1788,
+      "step": 3405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005365629485856381,
+      "loss": 3.2396,
+      "step": 3406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005363289853613053,
+      "loss": 3.2726,
+      "step": 3407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005360950141400643,
+      "loss": 3.2178,
+      "step": 3408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005358610349734178,
+      "loss": 3.3275,
+      "step": 3409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005356270479128706,
+      "loss": 3.1435,
+      "step": 3410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000535393053009929,
+      "loss": 3.2059,
+      "step": 3411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005351590503161007,
+      "loss": 3.2206,
+      "step": 3412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005349250398828958,
+      "loss": 3.2107,
+      "step": 3413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005346910217618259,
+      "loss": 3.2472,
+      "step": 3414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000534456996004404,
+      "loss": 3.1822,
+      "step": 3415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005342229626621447,
+      "loss": 3.3078,
+      "step": 3416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005339889217865649,
+      "loss": 3.1884,
+      "step": 3417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005337548734291826,
+      "loss": 3.3543,
+      "step": 3418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005335208176415179,
+      "loss": 3.1568,
+      "step": 3419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000533286754475092,
+      "loss": 3.2582,
+      "step": 3420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000533052683981428,
+      "loss": 3.3259,
+      "step": 3421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005328186062120508,
+      "loss": 3.1561,
+      "step": 3422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005325845212184866,
+      "loss": 3.2334,
+      "step": 3423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005323504290522631,
+      "loss": 3.2394,
+      "step": 3424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005321163297649099,
+      "loss": 3.3106,
+      "step": 3425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005318822234079584,
+      "loss": 3.2746,
+      "step": 3426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005316481100329408,
+      "loss": 3.2677,
+      "step": 3427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005314139896913912,
+      "loss": 3.3079,
+      "step": 3428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005311798624348457,
+      "loss": 3.2361,
+      "step": 3429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000530945728314841,
+      "loss": 3.3234,
+      "step": 3430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005307115873829161,
+      "loss": 3.1643,
+      "step": 3431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005304774396906111,
+      "loss": 3.2006,
+      "step": 3432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005302432852894677,
+      "loss": 3.3604,
+      "step": 3433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005300091242310292,
+      "loss": 3.1586,
+      "step": 3434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005297749565668402,
+      "loss": 3.1498,
+      "step": 3435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005295407823484467,
+      "loss": 3.2096,
+      "step": 3436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005293066016273963,
+      "loss": 3.2743,
+      "step": 3437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005290724144552379,
+      "loss": 3.1144,
+      "step": 3438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005288382208835221,
+      "loss": 3.2211,
+      "step": 3439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005286040209638005,
+      "loss": 3.2673,
+      "step": 3440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005283698147476263,
+      "loss": 3.2618,
+      "step": 3441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005281356022865542,
+      "loss": 3.2454,
+      "step": 3442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005279013836321399,
+      "loss": 3.2549,
+      "step": 3443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000527667158835941,
+      "loss": 3.3002,
+      "step": 3444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005274329279495162,
+      "loss": 3.2493,
+      "step": 3445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005271986910244253,
+      "loss": 3.1588,
+      "step": 3446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005269644481122295,
+      "loss": 3.1964,
+      "step": 3447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005267301992644922,
+      "loss": 3.1834,
+      "step": 3448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005264959445327765,
+      "loss": 3.2016,
+      "step": 3449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005262616839686481,
+      "loss": 3.1399,
+      "step": 3450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005260274176236734,
+      "loss": 3.2123,
+      "step": 3451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005257931455494203,
+      "loss": 3.1565,
+      "step": 3452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005255588677974579,
+      "loss": 3.1769,
+      "step": 3453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005253245844193564,
+      "loss": 3.1369,
+      "step": 3454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005250902954666876,
+      "loss": 3.1813,
+      "step": 3455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005248560009910239,
+      "loss": 3.203,
+      "step": 3456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005246217010439398,
+      "loss": 3.1443,
+      "step": 3457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005243873956770101,
+      "loss": 3.1865,
+      "step": 3458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005241530849418115,
+      "loss": 3.0818,
+      "step": 3459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005239187688899213,
+      "loss": 3.1461,
+      "step": 3460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005236844475729185,
+      "loss": 3.2421,
+      "step": 3461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000523450121042383,
+      "loss": 3.1737,
+      "step": 3462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005232157893498957,
+      "loss": 3.0103,
+      "step": 3463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000522981452547039,
+      "loss": 3.0988,
+      "step": 3464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000522747110685396,
+      "loss": 3.2677,
+      "step": 3465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005225127638165514,
+      "loss": 3.1246,
+      "step": 3466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005222784119920907,
+      "loss": 3.2292,
+      "step": 3467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005220440552636003,
+      "loss": 3.3352,
+      "step": 3468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005218096936826681,
+      "loss": 3.1104,
+      "step": 3469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005215753273008827,
+      "loss": 3.2108,
+      "step": 3470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005213409561698343,
+      "loss": 3.1207,
+      "step": 3471
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005211065803411135,
+      "loss": 3.3316,
+      "step": 3472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005208721998663122,
+      "loss": 3.1799,
+      "step": 3473
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005206378147970235,
+      "loss": 3.1977,
+      "step": 3474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005204034251848413,
+      "loss": 3.1366,
+      "step": 3475
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005201690310813605,
+      "loss": 3.2945,
+      "step": 3476
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000519934632538177,
+      "loss": 3.1545,
+      "step": 3477
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005197002296068878,
+      "loss": 3.3732,
+      "step": 3478
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005194658223390909,
+      "loss": 3.1622,
+      "step": 3479
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005192314107863848,
+      "loss": 3.1235,
+      "step": 3480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005189969950003697,
+      "loss": 3.2994,
+      "step": 3481
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005187625750326458,
+      "loss": 3.2308,
+      "step": 3482
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005185281509348153,
+      "loss": 3.2322,
+      "step": 3483
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005182937227584805,
+      "loss": 3.113,
+      "step": 3484
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005180592905552447,
+      "loss": 3.1934,
+      "step": 3485
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005178248543767122,
+      "loss": 3.1936,
+      "step": 3486
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005175904142744885,
+      "loss": 3.1288,
+      "step": 3487
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005173559703001796,
+      "loss": 3.1299,
+      "step": 3488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005171215225053921,
+      "loss": 3.1182,
+      "step": 3489
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005168870709417342,
+      "loss": 3.1636,
+      "step": 3490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000516652615660814,
+      "loss": 3.2376,
+      "step": 3491
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005164181567142415,
+      "loss": 3.1495,
+      "step": 3492
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005161836941536263,
+      "loss": 3.232,
+      "step": 3493
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00051594922803058,
+      "loss": 3.1684,
+      "step": 3494
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005157147583967139,
+      "loss": 3.2577,
+      "step": 3495
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005154802853036408,
+      "loss": 3.1508,
+      "step": 3496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000515245808802974,
+      "loss": 3.2368,
+      "step": 3497
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005150113289463275,
+      "loss": 3.2307,
+      "step": 3498
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005147768457853162,
+      "loss": 3.1892,
+      "step": 3499
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005145423593715557,
+      "loss": 3.2918,
+      "step": 3500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005143078697566622,
+      "loss": 3.2671,
+      "step": 3501
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005140733769922525,
+      "loss": 3.3621,
+      "step": 3502
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005138388811299446,
+      "loss": 3.2079,
+      "step": 3503
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005136043822213566,
+      "loss": 3.1294,
+      "step": 3504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005133698803181075,
+      "loss": 3.272,
+      "step": 3505
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005131353754718171,
+      "loss": 3.0887,
+      "step": 3506
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005129008677341057,
+      "loss": 3.1982,
+      "step": 3507
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000512666357156594,
+      "loss": 3.2666,
+      "step": 3508
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005124318437909038,
+      "loss": 3.2653,
+      "step": 3509
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005121973276886575,
+      "loss": 3.3024,
+      "step": 3510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005119628089014774,
+      "loss": 3.1919,
+      "step": 3511
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005117282874809873,
+      "loss": 3.1976,
+      "step": 3512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000511493763478811,
+      "loss": 3.2923,
+      "step": 3513
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005112592369465731,
+      "loss": 3.1581,
+      "step": 3514
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005110247079358986,
+      "loss": 3.2226,
+      "step": 3515
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005107901764984132,
+      "loss": 3.2401,
+      "step": 3516
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005105556426857429,
+      "loss": 3.1795,
+      "step": 3517
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005103211065495146,
+      "loss": 3.1341,
+      "step": 3518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005100865681413556,
+      "loss": 3.1452,
+      "step": 3519
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005098520275128934,
+      "loss": 3.2214,
+      "step": 3520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005096174847157562,
+      "loss": 3.2096,
+      "step": 3521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005093829398015728,
+      "loss": 3.1336,
+      "step": 3522
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005091483928219723,
+      "loss": 3.1596,
+      "step": 3523
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005089138438285841,
+      "loss": 3.1607,
+      "step": 3524
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005086792928730383,
+      "loss": 3.1573,
+      "step": 3525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005084447400069655,
+      "loss": 3.2674,
+      "step": 3526
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005082101852819967,
+      "loss": 3.1944,
+      "step": 3527
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005079756287497629,
+      "loss": 3.073,
+      "step": 3528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000507741070461896,
+      "loss": 3.2572,
+      "step": 3529
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000507506510470028,
+      "loss": 3.1718,
+      "step": 3530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005072719488257914,
+      "loss": 3.406,
+      "step": 3531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005070373855808189,
+      "loss": 3.2454,
+      "step": 3532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005068028207867439,
+      "loss": 3.1961,
+      "step": 3533
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005065682544951995,
+      "loss": 3.0997,
+      "step": 3534
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005063336867578201,
+      "loss": 3.1942,
+      "step": 3535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005060991176262393,
+      "loss": 3.1659,
+      "step": 3536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005058645471520922,
+      "loss": 3.2644,
+      "step": 3537
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000505629975387013,
+      "loss": 3.1847,
+      "step": 3538
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005053954023826368,
+      "loss": 3.1335,
+      "step": 3539
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005051608281905994,
+      "loss": 3.2163,
+      "step": 3540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005049262528625359,
+      "loss": 3.1759,
+      "step": 3541
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005046916764500824,
+      "loss": 3.1575,
+      "step": 3542
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005044570990048747,
+      "loss": 3.1792,
+      "step": 3543
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005042225205785493,
+      "loss": 3.2795,
+      "step": 3544
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005039879412227428,
+      "loss": 3.1508,
+      "step": 3545
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005037533609890916,
+      "loss": 3.2232,
+      "step": 3546
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005035187799292329,
+      "loss": 3.1986,
+      "step": 3547
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005032841980948036,
+      "loss": 3.2635,
+      "step": 3548
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005030496155374412,
+      "loss": 3.1836,
+      "step": 3549
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000502815032308783,
+      "loss": 3.1746,
+      "step": 3550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005025804484604664,
+      "loss": 3.2842,
+      "step": 3551
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005023458640441294,
+      "loss": 3.29,
+      "step": 3552
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005021112791114096,
+      "loss": 3.2743,
+      "step": 3553
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005018766937139452,
+      "loss": 3.2754,
+      "step": 3554
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005016421079033739,
+      "loss": 3.3299,
+      "step": 3555
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005014075217313341,
+      "loss": 3.1686,
+      "step": 3556
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005011729352494641,
+      "loss": 3.1705,
+      "step": 3557
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000500938348509402,
+      "loss": 3.1121,
+      "step": 3558
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000500703761562786,
+      "loss": 3.1733,
+      "step": 3559
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005004691744612547,
+      "loss": 3.2115,
+      "step": 3560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005002345872564467,
+      "loss": 3.1998,
+      "step": 3561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 3.0932,
+      "step": 3562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004997654127435536,
+      "loss": 3.2241,
+      "step": 3563
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004995308255387453,
+      "loss": 3.2415,
+      "step": 3564
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004992962384372141,
+      "loss": 3.2524,
+      "step": 3565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004990616514905981,
+      "loss": 3.2116,
+      "step": 3566
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004988270647505361,
+      "loss": 3.096,
+      "step": 3567
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004985924782686658,
+      "loss": 3.2048,
+      "step": 3568
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004983578920966261,
+      "loss": 3.1418,
+      "step": 3569
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004981233062860548,
+      "loss": 3.1074,
+      "step": 3570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004978887208885904,
+      "loss": 3.2035,
+      "step": 3571
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004976541359558707,
+      "loss": 3.3873,
+      "step": 3572
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004974195515395337,
+      "loss": 3.3064,
+      "step": 3573
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004971849676912172,
+      "loss": 3.1625,
+      "step": 3574
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004969503844625588,
+      "loss": 3.2012,
+      "step": 3575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004967158019051965,
+      "loss": 3.1308,
+      "step": 3576
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004964812200707672,
+      "loss": 3.2408,
+      "step": 3577
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004962466390109085,
+      "loss": 3.258,
+      "step": 3578
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004960120587772573,
+      "loss": 3.2491,
+      "step": 3579
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004957774794214508,
+      "loss": 3.137,
+      "step": 3580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004955429009951255,
+      "loss": 3.2677,
+      "step": 3581
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004953083235499177,
+      "loss": 3.2497,
+      "step": 3582
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004950737471374643,
+      "loss": 3.1953,
+      "step": 3583
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004948391718094006,
+      "loss": 3.2716,
+      "step": 3584
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004946045976173632,
+      "loss": 3.1777,
+      "step": 3585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000494370024612987,
+      "loss": 3.144,
+      "step": 3586
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000494135452847908,
+      "loss": 3.2341,
+      "step": 3587
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004939008823737606,
+      "loss": 3.2268,
+      "step": 3588
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00049366631324218,
+      "loss": 3.3093,
+      "step": 3589
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004934317455048005,
+      "loss": 3.191,
+      "step": 3590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004931971792132562,
+      "loss": 3.2268,
+      "step": 3591
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004929626144191812,
+      "loss": 3.2901,
+      "step": 3592
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004927280511742086,
+      "loss": 3.3023,
+      "step": 3593
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004924934895299721,
+      "loss": 3.3352,
+      "step": 3594
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000492258929538104,
+      "loss": 3.2554,
+      "step": 3595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004920243712502371,
+      "loss": 3.1866,
+      "step": 3596
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004917898147180033,
+      "loss": 3.2505,
+      "step": 3597
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004915552599930346,
+      "loss": 3.1956,
+      "step": 3598
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004913207071269618,
+      "loss": 3.1773,
+      "step": 3599
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000491086156171416,
+      "loss": 3.1524,
+      "step": 3600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004908516071780281,
+      "loss": 3.1307,
+      "step": 3601
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004906170601984272,
+      "loss": 3.1272,
+      "step": 3602
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004903825152842438,
+      "loss": 3.2893,
+      "step": 3603
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004901479724871066,
+      "loss": 3.2318,
+      "step": 3604
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004899134318586445,
+      "loss": 3.2093,
+      "step": 3605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004896788934504853,
+      "loss": 3.0065,
+      "step": 3606
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004894443573142572,
+      "loss": 3.2493,
+      "step": 3607
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000489209823501587,
+      "loss": 3.2217,
+      "step": 3608
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004889752920641014,
+      "loss": 3.2456,
+      "step": 3609
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004887407630534271,
+      "loss": 3.1262,
+      "step": 3610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000488506236521189,
+      "loss": 3.2429,
+      "step": 3611
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048827171251901273,
+      "loss": 3.1723,
+      "step": 3612
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004880371910985225,
+      "loss": 3.2008,
+      "step": 3613
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004878026723113427,
+      "loss": 3.1396,
+      "step": 3614
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048756815620909613,
+      "loss": 3.2738,
+      "step": 3615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048733364284340617,
+      "loss": 3.116,
+      "step": 3616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004870991322658944,
+      "loss": 3.1744,
+      "step": 3617
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048686462452818296,
+      "loss": 3.2628,
+      "step": 3618
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048663011968189265,
+      "loss": 3.2704,
+      "step": 3619
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004863956177786435,
+      "loss": 3.1901,
+      "step": 3620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004861611188700555,
+      "loss": 3.3276,
+      "step": 3621
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048592662300774736,
+      "loss": 3.2386,
+      "step": 3622
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004856921302433379,
+      "loss": 3.1664,
+      "step": 3623
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004854576406284443,
+      "loss": 3.305,
+      "step": 3624
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004852231542146839,
+      "loss": 3.2886,
+      "step": 3625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004849886710536725,
+      "loss": 3.0892,
+      "step": 3626
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004847541911970261,
+      "loss": 3.2835,
+      "step": 3627
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004845197146963594,
+      "loss": 3.1733,
+      "step": 3628
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048428524160328625,
+      "loss": 3.3403,
+      "step": 3629
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048405077196942017,
+      "loss": 3.2257,
+      "step": 3630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004838163058463736,
+      "loss": 3.1959,
+      "step": 3631
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048358184328575873,
+      "loss": 3.1408,
+      "step": 3632
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000483347384339186,
+      "loss": 3.1444,
+      "step": 3633
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004831129290582659,
+      "loss": 3.2625,
+      "step": 3634
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004828784774946078,
+      "loss": 3.2941,
+      "step": 3635
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004826440296998206,
+      "loss": 3.2095,
+      "step": 3636
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004824095857255117,
+      "loss": 3.1675,
+      "step": 3637
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048217514562328784,
+      "loss": 3.2552,
+      "step": 3638
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004819407094447555,
+      "loss": 3.1501,
+      "step": 3639
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004817062772415196,
+      "loss": 3.1004,
+      "step": 3640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048147184906518483,
+      "loss": 3.1336,
+      "step": 3641
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048123742496735414,
+      "loss": 3.1509,
+      "step": 3642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004810030049996304,
+      "loss": 3.1597,
+      "step": 3643
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004807685892136151,
+      "loss": 3.2088,
+      "step": 3644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004805341776609092,
+      "loss": 3.1144,
+      "step": 3645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00048029977039311237,
+      "loss": 3.1333,
+      "step": 3646
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00048006536746182313,
+      "loss": 3.2032,
+      "step": 3647
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047983096891863965,
+      "loss": 3.2688,
+      "step": 3648
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047959657481515874,
+      "loss": 3.2184,
+      "step": 3649
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047936218520297665,
+      "loss": 3.2813,
+      "step": 3650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004791278001336878,
+      "loss": 3.2475,
+      "step": 3651
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004788934196588866,
+      "loss": 3.1451,
+      "step": 3652
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047865904383016567,
+      "loss": 3.1412,
+      "step": 3653
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004784246726991173,
+      "loss": 3.155,
+      "step": 3654
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004781903063173321,
+      "loss": 3.257,
+      "step": 3655
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047795594473639975,
+      "loss": 3.2712,
+      "step": 3656
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004777215880079095,
+      "loss": 3.2999,
+      "step": 3657
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004774872361834486,
+      "loss": 3.2229,
+      "step": 3658
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004772528893146041,
+      "loss": 3.3174,
+      "step": 3659
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047701854745296106,
+      "loss": 3.2488,
+      "step": 3660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047678421065010436,
+      "loss": 3.179,
+      "step": 3661
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000476549878957617,
+      "loss": 3.1016,
+      "step": 3662
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047631555242708156,
+      "loss": 3.238,
+      "step": 3663
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004760812311100788,
+      "loss": 3.0342,
+      "step": 3664
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004758469150581886,
+      "loss": 3.0674,
+      "step": 3665
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004756126043229901,
+      "loss": 3.1496,
+      "step": 3666
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004753782989560603,
+      "loss": 3.1733,
+      "step": 3667
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047514399900897627,
+      "loss": 3.3232,
+      "step": 3668
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004749097045333126,
+      "loss": 3.3586,
+      "step": 3669
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004746754155806437,
+      "loss": 3.2001,
+      "step": 3670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047444113220254213,
+      "loss": 3.1849,
+      "step": 3671
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004742068544505798,
+      "loss": 3.2893,
+      "step": 3672
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004739725823763267,
+      "loss": 3.0684,
+      "step": 3673
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000473738316031352,
+      "loss": 3.11,
+      "step": 3674
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047350405546722367,
+      "loss": 3.2859,
+      "step": 3675
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047326980073550797,
+      "loss": 3.223,
+      "step": 3676
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047303555188777047,
+      "loss": 3.3546,
+      "step": 3677
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047280130897557485,
+      "loss": 3.1524,
+      "step": 3678
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004725670720504839,
+      "loss": 3.1957,
+      "step": 3679
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047233284116405893,
+      "loss": 3.1913,
+      "step": 3680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004720986163678602,
+      "loss": 3.0721,
+      "step": 3681
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000471864397713446,
+      "loss": 3.1537,
+      "step": 3682
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004716301852523738,
+      "loss": 3.2208,
+      "step": 3683
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047139597903619976,
+      "loss": 3.2736,
+      "step": 3684
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000471161779116478,
+      "loss": 3.2384,
+      "step": 3685
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004709275855447621,
+      "loss": 3.3967,
+      "step": 3686
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047069339837260373,
+      "loss": 3.2931,
+      "step": 3687
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004704592176515534,
+      "loss": 3.2945,
+      "step": 3688
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047022504343315985,
+      "loss": 3.1712,
+      "step": 3689
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046999087576897095,
+      "loss": 3.2604,
+      "step": 3690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046975671471053245,
+      "loss": 3.2514,
+      "step": 3691
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000469522560309389,
+      "loss": 3.1834,
+      "step": 3692
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046928841261708416,
+      "loss": 3.1217,
+      "step": 3693
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004690542716851591,
+      "loss": 3.1033,
+      "step": 3694
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004688201375651545,
+      "loss": 3.0763,
+      "step": 3695
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004685860103086087,
+      "loss": 3.0679,
+      "step": 3696
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046835188996705935,
+      "loss": 3.2249,
+      "step": 3697
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004681177765920416,
+      "loss": 3.0226,
+      "step": 3698
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004678836702350901,
+      "loss": 3.2229,
+      "step": 3699
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000467649570947737,
+      "loss": 3.2604,
+      "step": 3700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004674154787815135,
+      "loss": 3.2413,
+      "step": 3701
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004671813937879494,
+      "loss": 3.2905,
+      "step": 3702
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046694731601857204,
+      "loss": 3.1761,
+      "step": 3703
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046671324552490813,
+      "loss": 3.2753,
+      "step": 3704
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004664791823584821,
+      "loss": 3.2358,
+      "step": 3705
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004662451265708174,
+      "loss": 3.0456,
+      "step": 3706
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046601107821343514,
+      "loss": 3.3786,
+      "step": 3707
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046577703733785536,
+      "loss": 3.1959,
+      "step": 3708
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046554300399559613,
+      "loss": 3.192,
+      "step": 3709
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004653089782381742,
+      "loss": 3.324,
+      "step": 3710
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004650749601171043,
+      "loss": 3.1865,
+      "step": 3711
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046484094968389934,
+      "loss": 3.109,
+      "step": 3712
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046460694699007116,
+      "loss": 3.2007,
+      "step": 3713
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004643729520871293,
+      "loss": 3.2807,
+      "step": 3714
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046413896502658214,
+      "loss": 3.0879,
+      "step": 3715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004639049858599357,
+      "loss": 3.2074,
+      "step": 3716
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004636710146386947,
+      "loss": 3.278,
+      "step": 3717
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046343705141436195,
+      "loss": 3.2555,
+      "step": 3718
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004632030962384388,
+      "loss": 3.1672,
+      "step": 3719
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004629691491624243,
+      "loss": 3.2779,
+      "step": 3720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046273521023781596,
+      "loss": 3.172,
+      "step": 3721
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004625012795161097,
+      "loss": 3.1836,
+      "step": 3722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004622673570487992,
+      "loss": 3.3293,
+      "step": 3723
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000462033442887377,
+      "loss": 3.2018,
+      "step": 3724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004617995370833329,
+      "loss": 3.3641,
+      "step": 3725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046156563968815573,
+      "loss": 3.2679,
+      "step": 3726
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046133175075333183,
+      "loss": 3.191,
+      "step": 3727
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046109787033034636,
+      "loss": 3.2499,
+      "step": 3728
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046086399847068184,
+      "loss": 3.2342,
+      "step": 3729
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046063013522581916,
+      "loss": 3.1732,
+      "step": 3730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004603962806472376,
+      "loss": 3.2129,
+      "step": 3731
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046016243478641425,
+      "loss": 3.1685,
+      "step": 3732
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045992859769482453,
+      "loss": 3.0929,
+      "step": 3733
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004596947694239415,
+      "loss": 3.1812,
+      "step": 3734
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004594609500252368,
+      "loss": 3.1994,
+      "step": 3735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045922713955017967,
+      "loss": 3.087,
+      "step": 3736
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045899333805023786,
+      "loss": 3.1594,
+      "step": 3737
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004587595455768765,
+      "loss": 3.1735,
+      "step": 3738
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004585257621815593,
+      "loss": 3.1604,
+      "step": 3739
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004582919879157479,
+      "loss": 3.2697,
+      "step": 3740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045805822283090137,
+      "loss": 3.2114,
+      "step": 3741
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004578244669784777,
+      "loss": 3.0601,
+      "step": 3742
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045759072040993187,
+      "loss": 3.1277,
+      "step": 3743
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004573569831767176,
+      "loss": 3.2905,
+      "step": 3744
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045712325533028603,
+      "loss": 3.2718,
+      "step": 3745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004568895369220868,
+      "loss": 3.278,
+      "step": 3746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004566558280035667,
+      "loss": 3.1588,
+      "step": 3747
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045642212862617086,
+      "loss": 3.1912,
+      "step": 3748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045618843884134277,
+      "loss": 3.1865,
+      "step": 3749
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004559547587005227,
+      "loss": 3.1859,
+      "step": 3750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045572108825514996,
+      "loss": 3.1617,
+      "step": 3751
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045548742755666073,
+      "loss": 3.2566,
+      "step": 3752
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045525377665648985,
+      "loss": 3.2713,
+      "step": 3753
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004550201356060695,
+      "loss": 3.1732,
+      "step": 3754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045478650445683014,
+      "loss": 3.1581,
+      "step": 3755
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045455288326019935,
+      "loss": 3.1202,
+      "step": 3756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045431927206760305,
+      "loss": 3.1627,
+      "step": 3757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004540856709304651,
+      "loss": 3.1361,
+      "step": 3758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004538520799002065,
+      "loss": 3.0894,
+      "step": 3759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045361849902824676,
+      "loss": 3.2309,
+      "step": 3760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004533849283660025,
+      "loss": 3.0817,
+      "step": 3761
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045315136796488876,
+      "loss": 3.0932,
+      "step": 3762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045291781787631756,
+      "loss": 3.2852,
+      "step": 3763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045268427815169955,
+      "loss": 3.2032,
+      "step": 3764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004524507488424421,
+      "loss": 3.1599,
+      "step": 3765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000452217229999951,
+      "loss": 3.2816,
+      "step": 3766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004519837216756296,
+      "loss": 3.2345,
+      "step": 3767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004517502239208786,
+      "loss": 3.3493,
+      "step": 3768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045151673678709694,
+      "loss": 3.1288,
+      "step": 3769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004512832603256807,
+      "loss": 3.2525,
+      "step": 3770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045104979458802404,
+      "loss": 3.2619,
+      "step": 3771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045081633962551823,
+      "loss": 3.2968,
+      "step": 3772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004505828954895529,
+      "loss": 3.0708,
+      "step": 3773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045034946223151447,
+      "loss": 3.2054,
+      "step": 3774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004501160399027874,
+      "loss": 3.1687,
+      "step": 3775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004498826285547542,
+      "loss": 3.3402,
+      "step": 3776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00044964922823879385,
+      "loss": 3.2077,
+      "step": 3777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000449415839006284,
+      "loss": 3.2226,
+      "step": 3778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00044918246090859906,
+      "loss": 3.1346,
+      "step": 3779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00044894909399711166,
+      "loss": 3.1409,
+      "step": 3780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044871573832319124,
+      "loss": 3.2565,
+      "step": 3781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004484823939382056,
+      "loss": 3.1807,
+      "step": 3782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004482490608935193,
+      "loss": 3.209,
+      "step": 3783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004480157392404947,
+      "loss": 3.2073,
+      "step": 3784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004477824290304919,
+      "loss": 3.1481,
+      "step": 3785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044754913031486794,
+      "loss": 3.1632,
+      "step": 3786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004473158431449779,
+      "loss": 3.0496,
+      "step": 3787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004470825675721738,
+      "loss": 3.2307,
+      "step": 3788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044684930364780577,
+      "loss": 3.2482,
+      "step": 3789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004466160514232206,
+      "loss": 3.1386,
+      "step": 3790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000446382810949763,
+      "loss": 3.3337,
+      "step": 3791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004461495822787751,
+      "loss": 3.2076,
+      "step": 3792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044591636546159597,
+      "loss": 3.0852,
+      "step": 3793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044568316054956284,
+      "loss": 3.0665,
+      "step": 3794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044544996759400936,
+      "loss": 3.2644,
+      "step": 3795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044521678664626745,
+      "loss": 3.284,
+      "step": 3796
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004449836177576658,
+      "loss": 3.2861,
+      "step": 3797
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000444750460979531,
+      "loss": 3.2307,
+      "step": 3798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044451731636318615,
+      "loss": 3.2224,
+      "step": 3799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044428418395995245,
+      "loss": 3.1316,
+      "step": 3800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044405106382114796,
+      "loss": 3.2926,
+      "step": 3801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004438179559980884,
+      "loss": 3.2539,
+      "step": 3802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004435848605420865,
+      "loss": 3.1522,
+      "step": 3803
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004433517775044521,
+      "loss": 3.2026,
+      "step": 3804
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044311870693649263,
+      "loss": 3.2937,
+      "step": 3805
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044288564888951277,
+      "loss": 3.2174,
+      "step": 3806
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044265260341481456,
+      "loss": 3.2023,
+      "step": 3807
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044241957056369665,
+      "loss": 3.0363,
+      "step": 3808
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044218655038745557,
+      "loss": 3.2262,
+      "step": 3809
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004419535429373848,
+      "loss": 3.1189,
+      "step": 3810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044172054826477526,
+      "loss": 3.1864,
+      "step": 3811
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044148756642091447,
+      "loss": 3.304,
+      "step": 3812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004412545974570876,
+      "loss": 3.2396,
+      "step": 3813
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044102164142457707,
+      "loss": 3.3437,
+      "step": 3814
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000440788698374662,
+      "loss": 3.2134,
+      "step": 3815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004405557683586192,
+      "loss": 3.2442,
+      "step": 3816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000440322851427722,
+      "loss": 3.2183,
+      "step": 3817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00044008994763324147,
+      "loss": 3.405,
+      "step": 3818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043985705702644524,
+      "loss": 3.1572,
+      "step": 3819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043962417965859866,
+      "loss": 3.142,
+      "step": 3820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043939131558096346,
+      "loss": 3.1039,
+      "step": 3821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004391584648447988,
+      "loss": 3.2459,
+      "step": 3822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043892562750136115,
+      "loss": 3.1952,
+      "step": 3823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043869280360190347,
+      "loss": 3.0562,
+      "step": 3824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004384599931976764,
+      "loss": 3.1578,
+      "step": 3825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004382271963399268,
+      "loss": 3.323,
+      "step": 3826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004379944130798994,
+      "loss": 3.174,
+      "step": 3827
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004377616434688353,
+      "loss": 3.1091,
+      "step": 3828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004375288875579732,
+      "loss": 3.0464,
+      "step": 3829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004372961453985481,
+      "loss": 3.3275,
+      "step": 3830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004370634170417924,
+      "loss": 3.2356,
+      "step": 3831
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043683070253893553,
+      "loss": 3.1687,
+      "step": 3832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043659800194120334,
+      "loss": 3.1444,
+      "step": 3833
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043636531529981943,
+      "loss": 3.1932,
+      "step": 3834
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004361326426660035,
+      "loss": 3.4213,
+      "step": 3835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004358999840909728,
+      "loss": 3.0843,
+      "step": 3836
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000435667339625941,
+      "loss": 3.2171,
+      "step": 3837
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004354347093221194,
+      "loss": 3.2208,
+      "step": 3838
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043520209323071507,
+      "loss": 3.1032,
+      "step": 3839
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043496949140293283,
+      "loss": 3.249,
+      "step": 3840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043473690388997434,
+      "loss": 3.2737,
+      "step": 3841
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043450433074303734,
+      "loss": 3.2754,
+      "step": 3842
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004342717720133174,
+      "loss": 3.2744,
+      "step": 3843
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004340392277520061,
+      "loss": 3.3137,
+      "step": 3844
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004338066980102926,
+      "loss": 3.2754,
+      "step": 3845
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000433574182839362,
+      "loss": 3.2811,
+      "step": 3846
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043334168229039704,
+      "loss": 3.2642,
+      "step": 3847
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004331091964145766,
+      "loss": 3.0624,
+      "step": 3848
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043287672526307644,
+      "loss": 3.2067,
+      "step": 3849
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043264426888706966,
+      "loss": 3.3398,
+      "step": 3850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004324118273377251,
+      "loss": 3.236,
+      "step": 3851
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004321794006662093,
+      "loss": 3.1633,
+      "step": 3852
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043194698892368497,
+      "loss": 3.1391,
+      "step": 3853
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004317145921613119,
+      "loss": 3.0568,
+      "step": 3854
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004314822104302459,
+      "loss": 3.2879,
+      "step": 3855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043124984378164046,
+      "loss": 3.1896,
+      "step": 3856
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004310174922666449,
+      "loss": 3.2834,
+      "step": 3857
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004307851559364055,
+      "loss": 3.2652,
+      "step": 3858
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004305528348420656,
+      "loss": 3.1892,
+      "step": 3859
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043032052903476433,
+      "loss": 3.229,
+      "step": 3860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004300882385656383,
+      "loss": 3.2305,
+      "step": 3861
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042985596348582014,
+      "loss": 3.203,
+      "step": 3862
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004296237038464397,
+      "loss": 3.2652,
+      "step": 3863
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042939145969862264,
+      "loss": 3.1965,
+      "step": 3864
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000429159231093492,
+      "loss": 3.163,
+      "step": 3865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042892701808216696,
+      "loss": 3.1824,
+      "step": 3866
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042869482071576306,
+      "loss": 3.0142,
+      "step": 3867
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042846263904539305,
+      "loss": 3.2287,
+      "step": 3868
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004282304731221656,
+      "loss": 3.2401,
+      "step": 3869
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004279983229971863,
+      "loss": 3.1882,
+      "step": 3870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042776618872155715,
+      "loss": 3.3685,
+      "step": 3871
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004275340703463767,
+      "loss": 3.2122,
+      "step": 3872
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004273019679227397,
+      "loss": 3.232,
+      "step": 3873
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042706988150173784,
+      "loss": 3.1538,
+      "step": 3874
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042683781113445925,
+      "loss": 3.0751,
+      "step": 3875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042660575687198787,
+      "loss": 3.2469,
+      "step": 3876
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042637371876540507,
+      "loss": 3.2119,
+      "step": 3877
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042614169686578773,
+      "loss": 3.2187,
+      "step": 3878
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042590969122420995,
+      "loss": 3.1348,
+      "step": 3879
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004256777018917418,
+      "loss": 3.1702,
+      "step": 3880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042544572891945,
+      "loss": 3.0849,
+      "step": 3881
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004252137723583972,
+      "loss": 3.167,
+      "step": 3882
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004249818322596431,
+      "loss": 3.1909,
+      "step": 3883
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042474990867424347,
+      "loss": 3.3824,
+      "step": 3884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042451800165325005,
+      "loss": 3.1535,
+      "step": 3885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042428611124771184,
+      "loss": 3.2029,
+      "step": 3886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004240542375086731,
+      "loss": 3.207,
+      "step": 3887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004238223804871754,
+      "loss": 3.1492,
+      "step": 3888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042359054023425593,
+      "loss": 3.1949,
+      "step": 3889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004233587168009488,
+      "loss": 3.0165,
+      "step": 3890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004231269102382837,
+      "loss": 3.0848,
+      "step": 3891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042289512059728727,
+      "loss": 3.1501,
+      "step": 3892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004226633479289819,
+      "loss": 3.0703,
+      "step": 3893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042243159228438687,
+      "loss": 3.2244,
+      "step": 3894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004221998537145169,
+      "loss": 3.0597,
+      "step": 3895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042196813227038355,
+      "loss": 3.1418,
+      "step": 3896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004217364280029946,
+      "loss": 3.0592,
+      "step": 3897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042150474096335356,
+      "loss": 3.3565,
+      "step": 3898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004212730712024609,
+      "loss": 3.2946,
+      "step": 3899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004210414187713124,
+      "loss": 3.2278,
+      "step": 3900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042080978372090086,
+      "loss": 3.3059,
+      "step": 3901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004205781661022146,
+      "loss": 3.2639,
+      "step": 3902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042034656596623876,
+      "loss": 3.2645,
+      "step": 3903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042011498336395393,
+      "loss": 3.2686,
+      "step": 3904
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041988341834633714,
+      "loss": 3.2458,
+      "step": 3905
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000419651870964362,
+      "loss": 3.1302,
+      "step": 3906
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004194203412689972,
+      "loss": 3.1789,
+      "step": 3907
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004191888293112087,
+      "loss": 3.2982,
+      "step": 3908
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041895733514195766,
+      "loss": 3.268,
+      "step": 3909
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041872585881220186,
+      "loss": 3.1106,
+      "step": 3910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041849440037289477,
+      "loss": 3.1786,
+      "step": 3911
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041826295987498654,
+      "loss": 3.2415,
+      "step": 3912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041803153736942253,
+      "loss": 3.2482,
+      "step": 3913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004178001329071445,
+      "loss": 3.1382,
+      "step": 3914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004175687465390908,
+      "loss": 3.2586,
+      "step": 3915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041733737831619466,
+      "loss": 3.0783,
+      "step": 3916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004171060282893865,
+      "loss": 3.0757,
+      "step": 3917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004168746965095918,
+      "loss": 3.1406,
+      "step": 3918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041664338302773276,
+      "loss": 3.2691,
+      "step": 3919
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004164120878947267,
+      "loss": 3.1951,
+      "step": 3920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000416180811161488,
+      "loss": 3.2827,
+      "step": 3921
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000415949552878926,
+      "loss": 3.2308,
+      "step": 3922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004157183130979463,
+      "loss": 3.3029,
+      "step": 3923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004154870918694509,
+      "loss": 3.2606,
+      "step": 3924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004152558892443368,
+      "loss": 3.2146,
+      "step": 3925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041502470527349786,
+      "loss": 3.2293,
+      "step": 3926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041479354000782307,
+      "loss": 3.161,
+      "step": 3927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004145623934981979,
+      "loss": 3.2265,
+      "step": 3928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041433126579550316,
+      "loss": 3.2566,
+      "step": 3929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004141001569506161,
+      "loss": 3.2367,
+      "step": 3930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041386906701440914,
+      "loss": 3.1939,
+      "step": 3931
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041363799603775093,
+      "loss": 3.2132,
+      "step": 3932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041340694407150625,
+      "loss": 3.0105,
+      "step": 3933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 3.1917,
+      "step": 3934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041294489737369325,
+      "loss": 3.179,
+      "step": 3935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004127139027438329,
+      "loss": 3.1875,
+      "step": 3936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004124829273278019,
+      "loss": 3.2307,
+      "step": 3937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004122519711764431,
+      "loss": 3.1339,
+      "step": 3938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004120210343405963,
+      "loss": 3.131,
+      "step": 3939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004117901168710959,
+      "loss": 3.1833,
+      "step": 3940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041155921881877254,
+      "loss": 3.298,
+      "step": 3941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041132834023445305,
+      "loss": 3.2729,
+      "step": 3942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000411097481168959,
+      "loss": 3.1716,
+      "step": 3943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004108666416731085,
+      "loss": 3.1331,
+      "step": 3944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004106358217977148,
+      "loss": 3.2906,
+      "step": 3945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041040502159358746,
+      "loss": 3.1609,
+      "step": 3946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041017424111153077,
+      "loss": 3.2043,
+      "step": 3947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00040994348040234574,
+      "loss": 3.0853,
+      "step": 3948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004097127395168283,
+      "loss": 3.3268,
+      "step": 3949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00040948201850577007,
+      "loss": 3.097,
+      "step": 3950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004092513174199587,
+      "loss": 3.21,
+      "step": 3951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004090206363101769,
+      "loss": 3.1494,
+      "step": 3952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004087899752272037,
+      "loss": 3.1524,
+      "step": 3953
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004085593342218129,
+      "loss": 3.1724,
+      "step": 3954
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040832871334477465,
+      "loss": 3.0849,
+      "step": 3955
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040809811264685405,
+      "loss": 3.1859,
+      "step": 3956
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004078675321788121,
+      "loss": 3.3641,
+      "step": 3957
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004076369719914055,
+      "loss": 3.2084,
+      "step": 3958
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004074064321353857,
+      "loss": 3.2282,
+      "step": 3959
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040717591266150086,
+      "loss": 3.3031,
+      "step": 3960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004069454136204934,
+      "loss": 3.2404,
+      "step": 3961
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040671493506310233,
+      "loss": 3.1711,
+      "step": 3962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004064844770400614,
+      "loss": 3.2755,
+      "step": 3963
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004062540396021005,
+      "loss": 3.1002,
+      "step": 3964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004060236227999441,
+      "loss": 3.2025,
+      "step": 3965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040579322668431297,
+      "loss": 3.1782,
+      "step": 3966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040556285130592304,
+      "loss": 3.3421,
+      "step": 3967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040533249671548526,
+      "loss": 3.1477,
+      "step": 3968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004051021629637068,
+      "loss": 3.1892,
+      "step": 3969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040487185010128945,
+      "loss": 3.3576,
+      "step": 3970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004046415581789312,
+      "loss": 3.1145,
+      "step": 3971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004044112872473246,
+      "loss": 3.2326,
+      "step": 3972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004041810373571582,
+      "loss": 3.2882,
+      "step": 3973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040395080855911553,
+      "loss": 3.1865,
+      "step": 3974
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004037206009038758,
+      "loss": 3.2908,
+      "step": 3975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040349041444211344,
+      "loss": 3.1152,
+      "step": 3976
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004032602492244979,
+      "loss": 3.1498,
+      "step": 3977
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040303010530169447,
+      "loss": 3.4095,
+      "step": 3978
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004027999827243635,
+      "loss": 3.0859,
+      "step": 3979
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004025698815431607,
+      "loss": 3.0825,
+      "step": 3980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040233980180873685,
+      "loss": 3.2347,
+      "step": 3981
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004021097435717385,
+      "loss": 3.2511,
+      "step": 3982
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004018797068828068,
+      "loss": 3.2395,
+      "step": 3983
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004016496917925787,
+      "loss": 3.2503,
+      "step": 3984
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004014196983516862,
+      "loss": 3.2179,
+      "step": 3985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004011897266107567,
+      "loss": 3.1806,
+      "step": 3986
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004009597766204124,
+      "loss": 3.1287,
+      "step": 3987
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000400729848431271,
+      "loss": 3.284,
+      "step": 3988
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004004999420939456,
+      "loss": 3.312,
+      "step": 3989
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004002700576590441,
+      "loss": 3.3213,
+      "step": 3990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004000401951771699,
+      "loss": 3.3023,
+      "step": 3991
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039981035469892107,
+      "loss": 3.2983,
+      "step": 3992
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003995805362748916,
+      "loss": 3.236,
+      "step": 3993
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003993507399556699,
+      "loss": 3.3492,
+      "step": 3994
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003991209657918402,
+      "loss": 3.223,
+      "step": 3995
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039889121383398113,
+      "loss": 3.1532,
+      "step": 3996
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039866148413266676,
+      "loss": 3.1948,
+      "step": 3997
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039843177673846666,
+      "loss": 3.2222,
+      "step": 3998
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003982020917019446,
+      "loss": 3.2088,
+      "step": 3999
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039797242907366025,
+      "loss": 3.1355,
+      "step": 4000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000397742788904168,
+      "loss": 3.0896,
+      "step": 4001
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039751317124401745,
+      "loss": 3.2837,
+      "step": 4002
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003972835761437528,
+      "loss": 3.2005,
+      "step": 4003
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039705400365391396,
+      "loss": 3.1852,
+      "step": 4004
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003968244538250353,
+      "loss": 3.1617,
+      "step": 4005
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003965949267076464,
+      "loss": 3.2276,
+      "step": 4006
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039636542235227206,
+      "loss": 3.2348,
+      "step": 4007
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039613594080943154,
+      "loss": 3.2677,
+      "step": 4008
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003959064821296398,
+      "loss": 3.2916,
+      "step": 4009
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003956770463634059,
+      "loss": 3.1391,
+      "step": 4010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003954476335612349,
+      "loss": 3.1435,
+      "step": 4011
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039521824377362566,
+      "loss": 3.1791,
+      "step": 4012
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003949888770510731,
+      "loss": 3.2232,
+      "step": 4013
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000394759533444066,
+      "loss": 3.1358,
+      "step": 4014
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003945302130030888,
+      "loss": 3.1231,
+      "step": 4015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039430091577862073,
+      "loss": 3.2749,
+      "step": 4016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003940716418211354,
+      "loss": 3.2106,
+      "step": 4017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039384239118110204,
+      "loss": 3.2774,
+      "step": 4018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039361316390898416,
+      "loss": 3.2353,
+      "step": 4019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003933839600552407,
+      "loss": 3.2145,
+      "step": 4020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039315477967032456,
+      "loss": 3.2081,
+      "step": 4021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003929256228046845,
+      "loss": 3.2771,
+      "step": 4022
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039269648950876343,
+      "loss": 3.1912,
+      "step": 4023
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039246737983299906,
+      "loss": 3.2578,
+      "step": 4024
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003922382938278245,
+      "loss": 3.3033,
+      "step": 4025
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039200923154366686,
+      "loss": 3.1755,
+      "step": 4026
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003917801930309486,
+      "loss": 3.1325,
+      "step": 4027
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003915511783400867,
+      "loss": 3.1023,
+      "step": 4028
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039132218752149317,
+      "loss": 3.2588,
+      "step": 4029
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039109322062557425,
+      "loss": 3.1979,
+      "step": 4030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039086427770273135,
+      "loss": 3.1973,
+      "step": 4031
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039063535880336063,
+      "loss": 3.298,
+      "step": 4032
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039040646397785244,
+      "loss": 3.1923,
+      "step": 4033
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039017759327659264,
+      "loss": 3.178,
+      "step": 4034
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038994874674996085,
+      "loss": 3.3328,
+      "step": 4035
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038971992444833213,
+      "loss": 3.0656,
+      "step": 4036
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003894911264220758,
+      "loss": 3.1726,
+      "step": 4037
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038926235272155636,
+      "loss": 3.1708,
+      "step": 4038
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000389033603397132,
+      "loss": 3.17,
+      "step": 4039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003888048784991565,
+      "loss": 3.2044,
+      "step": 4040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003885761780779777,
+      "loss": 3.2153,
+      "step": 4041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038834750218393814,
+      "loss": 3.1924,
+      "step": 4042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038811885086737543,
+      "loss": 3.1763,
+      "step": 4043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038789022417862076,
+      "loss": 3.1785,
+      "step": 4044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000387661622168001,
+      "loss": 3.1306,
+      "step": 4045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038743304488583687,
+      "loss": 3.1417,
+      "step": 4046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038720449238244416,
+      "loss": 3.1771,
+      "step": 4047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038697596470813257,
+      "loss": 3.2698,
+      "step": 4048
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038674746191320696,
+      "loss": 3.1484,
+      "step": 4049
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038651898404796646,
+      "loss": 3.1724,
+      "step": 4050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003862905311627044,
+      "loss": 3.1,
+      "step": 4051
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038606210330770923,
+      "loss": 3.0924,
+      "step": 4052
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003858337005332634,
+      "loss": 3.2093,
+      "step": 4053
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003856053228896442,
+      "loss": 3.2267,
+      "step": 4054
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003853769704271228,
+      "loss": 3.2123,
+      "step": 4055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038514864319596576,
+      "loss": 3.2561,
+      "step": 4056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038492034124643304,
+      "loss": 3.1676,
+      "step": 4057
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038469206462877993,
+      "loss": 3.1073,
+      "step": 4058
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003844638133932557,
+      "loss": 3.2115,
+      "step": 4059
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003842355875901038,
+      "loss": 3.2581,
+      "step": 4060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003840073872695627,
+      "loss": 3.2943,
+      "step": 4061
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003837792124818646,
+      "loss": 3.2224,
+      "step": 4062
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000383551063277237,
+      "loss": 3.2126,
+      "step": 4063
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003833229397059005,
+      "loss": 3.1695,
+      "step": 4064
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003830948418180713,
+      "loss": 3.2221,
+      "step": 4065
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038286676966395897,
+      "loss": 3.3321,
+      "step": 4066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000382638723293768,
+      "loss": 3.3215,
+      "step": 4067
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038241070275769727,
+      "loss": 3.2907,
+      "step": 4068
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038218270810593924,
+      "loss": 3.1451,
+      "step": 4069
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038195473938868157,
+      "loss": 3.1538,
+      "step": 4070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038172679665610554,
+      "loss": 3.1464,
+      "step": 4071
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003814988799583874,
+      "loss": 3.1446,
+      "step": 4072
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038127098934569674,
+      "loss": 3.1271,
+      "step": 4073
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003810431248681983,
+      "loss": 3.1067,
+      "step": 4074
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038081528657605044,
+      "loss": 3.2711,
+      "step": 4075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003805874745194063,
+      "loss": 3.3308,
+      "step": 4076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038035968874841256,
+      "loss": 3.1709,
+      "step": 4077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038013192931321093,
+      "loss": 3.2316,
+      "step": 4078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037990419626393655,
+      "loss": 3.1083,
+      "step": 4079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003796764896507192,
+      "loss": 3.1901,
+      "step": 4080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003794488095236829,
+      "loss": 3.1927,
+      "step": 4081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037922115593294533,
+      "loss": 3.114,
+      "step": 4082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000378993528928619,
+      "loss": 3.2058,
+      "step": 4083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037876592856081,
+      "loss": 3.1745,
+      "step": 4084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003785383548796191,
+      "loss": 3.2863,
+      "step": 4085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003783108079351406,
+      "loss": 3.2427,
+      "step": 4086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003780832877774635,
+      "loss": 3.179,
+      "step": 4087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003778557944566704,
+      "loss": 3.1784,
+      "step": 4088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003776283280228381,
+      "loss": 3.1608,
+      "step": 4089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037740088852603797,
+      "loss": 3.1194,
+      "step": 4090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037717347601633466,
+      "loss": 3.2013,
+      "step": 4091
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037694609054378755,
+      "loss": 3.1917,
+      "step": 4092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037671873215844965,
+      "loss": 3.1769,
+      "step": 4093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037649140091036845,
+      "loss": 3.2586,
+      "step": 4094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037626409684958474,
+      "loss": 3.2098,
+      "step": 4095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003760368200261342,
+      "loss": 3.1913,
+      "step": 4096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037580957049004583,
+      "loss": 3.1873,
+      "step": 4097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003755823482913428,
+      "loss": 3.17,
+      "step": 4098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037535515348004265,
+      "loss": 3.2382,
+      "step": 4099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003751279861061562,
+      "loss": 3.2544,
+      "step": 4100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003749008462196889,
+      "loss": 3.2093,
+      "step": 4101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037467373387063967,
+      "loss": 3.1262,
+      "step": 4102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037444664910900186,
+      "loss": 3.2361,
+      "step": 4103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037421959198476195,
+      "loss": 3.0531,
+      "step": 4104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003739925625479013,
+      "loss": 3.1786,
+      "step": 4105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037376556084839466,
+      "loss": 3.1437,
+      "step": 4106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003735385869362102,
+      "loss": 3.345,
+      "step": 4107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003733116408613112,
+      "loss": 3.1622,
+      "step": 4108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037308472267365343,
+      "loss": 3.1459,
+      "step": 4109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003728578324231877,
+      "loss": 3.2328,
+      "step": 4110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000372630970159858,
+      "loss": 3.1024,
+      "step": 4111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037240413593360244,
+      "loss": 3.3138,
+      "step": 4112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003721773297943525,
+      "loss": 3.162,
+      "step": 4113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037195055179203417,
+      "loss": 3.289,
+      "step": 4114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037172380197656693,
+      "loss": 3.2632,
+      "step": 4115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003714970803978636,
+      "loss": 3.1751,
+      "step": 4116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003712703871058317,
+      "loss": 3.2248,
+      "step": 4117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003710437221503716,
+      "loss": 3.1523,
+      "step": 4118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003708170855813782,
+      "loss": 3.1638,
+      "step": 4119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003705904774487396,
+      "loss": 3.1285,
+      "step": 4120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037036389780233813,
+      "loss": 3.1076,
+      "step": 4121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003701373466920492,
+      "loss": 3.2363,
+      "step": 4122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00036991082416774256,
+      "loss": 3.0701,
+      "step": 4123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003696843302792815,
+      "loss": 3.0883,
+      "step": 4124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003694578650765226,
+      "loss": 3.2974,
+      "step": 4125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003692314286093167,
+      "loss": 3.1886,
+      "step": 4126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003690050209275079,
+      "loss": 3.106,
+      "step": 4127
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003687786420809345,
+      "loss": 3.2187,
+      "step": 4128
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036855229211942757,
+      "loss": 3.1785,
+      "step": 4129
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036832597109281285,
+      "loss": 3.0497,
+      "step": 4130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003680996790509087,
+      "loss": 3.0127,
+      "step": 4131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003678734160435279,
+      "loss": 3.0846,
+      "step": 4132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036764718212047666,
+      "loss": 3.1767,
+      "step": 4133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003674209773315542,
+      "loss": 3.102,
+      "step": 4134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036719480172655417,
+      "loss": 3.1547,
+      "step": 4135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003669686553552632,
+      "loss": 3.1923,
+      "step": 4136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000366742538267462,
+      "loss": 3.3023,
+      "step": 4137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036651645051292417,
+      "loss": 3.1525,
+      "step": 4138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036629039214141755,
+      "loss": 3.2204,
+      "step": 4139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036606436320270287,
+      "loss": 3.1223,
+      "step": 4140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003658383637465349,
+      "loss": 3.3409,
+      "step": 4141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036561239382266173,
+      "loss": 3.1399,
+      "step": 4142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036538645348082466,
+      "loss": 3.2694,
+      "step": 4143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036516054277075906,
+      "loss": 3.268,
+      "step": 4144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036493466174219313,
+      "loss": 3.1023,
+      "step": 4145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003647088104448494,
+      "loss": 3.2836,
+      "step": 4146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003644829889284428,
+      "loss": 3.3069,
+      "step": 4147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003642571972426826,
+      "loss": 3.2055,
+      "step": 4148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003640314354372708,
+      "loss": 3.231,
+      "step": 4149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036380570356190343,
+      "loss": 3.177,
+      "step": 4150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003635800016662696,
+      "loss": 3.145,
+      "step": 4151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003633543298000517,
+      "loss": 3.1336,
+      "step": 4152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000363128688012926,
+      "loss": 3.2226,
+      "step": 4153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036290307635456143,
+      "loss": 3.2366,
+      "step": 4154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036267749487462126,
+      "loss": 3.1169,
+      "step": 4155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036245194362276095,
+      "loss": 3.1966,
+      "step": 4156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003622264226486304,
+      "loss": 3.1723,
+      "step": 4157
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003620009320018719,
+      "loss": 3.0829,
+      "step": 4158
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000361775471732122,
+      "loss": 3.2862,
+      "step": 4159
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003615500418890099,
+      "loss": 3.1118,
+      "step": 4160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00036132464252215804,
+      "loss": 3.1997,
+      "step": 4161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003610992736811827,
+      "loss": 3.2473,
+      "step": 4162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003608739354156928,
+      "loss": 3.0605,
+      "step": 4163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003606486277752914,
+      "loss": 3.2096,
+      "step": 4164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00036042335080957373,
+      "loss": 3.1528,
+      "step": 4165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00036019810456812916,
+      "loss": 3.2216,
+      "step": 4166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035997288910053984,
+      "loss": 3.1796,
+      "step": 4167
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003597477044563814,
+      "loss": 3.2269,
+      "step": 4168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035952255068522233,
+      "loss": 3.3176,
+      "step": 4169
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035929742783662476,
+      "loss": 3.2202,
+      "step": 4170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035907233596014365,
+      "loss": 3.241,
+      "step": 4171
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003588472751053272,
+      "loss": 3.2567,
+      "step": 4172
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035862224532171727,
+      "loss": 3.1592,
+      "step": 4173
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035839724665884796,
+      "loss": 3.2479,
+      "step": 4174
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035817227916624744,
+      "loss": 3.1329,
+      "step": 4175
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003579473428934364,
+      "loss": 3.1178,
+      "step": 4176
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003577224378899291,
+      "loss": 3.1493,
+      "step": 4177
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035749756420523236,
+      "loss": 3.0414,
+      "step": 4178
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003572727218888467,
+      "loss": 3.1915,
+      "step": 4179
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003570479109902656,
+      "loss": 3.1355,
+      "step": 4180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035682313155897507,
+      "loss": 3.1475,
+      "step": 4181
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035659838364445503,
+      "loss": 3.3453,
+      "step": 4182
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003563736672961777,
+      "loss": 3.0462,
+      "step": 4183
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035614898256360896,
+      "loss": 3.2848,
+      "step": 4184
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003559243294962073,
+      "loss": 3.2593,
+      "step": 4185
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035569970814342477,
+      "loss": 3.2589,
+      "step": 4186
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003554751185547056,
+      "loss": 3.3488,
+      "step": 4187
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035525056077948794,
+      "loss": 3.1833,
+      "step": 4188
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003550260348672023,
+      "loss": 3.2401,
+      "step": 4189
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035480154086727227,
+      "loss": 3.1203,
+      "step": 4190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035457707882911497,
+      "loss": 3.1669,
+      "step": 4191
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003543526488021394,
+      "loss": 3.1721,
+      "step": 4192
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035412825083574865,
+      "loss": 3.0992,
+      "step": 4193
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003539038849793381,
+      "loss": 3.172,
+      "step": 4194
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003536795512822964,
+      "loss": 3.1138,
+      "step": 4195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035345524979400457,
+      "loss": 3.151,
+      "step": 4196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035323098056383727,
+      "loss": 3.1909,
+      "step": 4197
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003530067436411617,
+      "loss": 3.1957,
+      "step": 4198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035278253907533764,
+      "loss": 3.1767,
+      "step": 4199
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003525583669157184,
+      "loss": 3.298,
+      "step": 4200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003523342272116495,
+      "loss": 3.1666,
+      "step": 4201
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003521101200124699,
+      "loss": 3.0668,
+      "step": 4202
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003518860453675109,
+      "loss": 3.1048,
+      "step": 4203
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003516620033260972,
+      "loss": 3.0785,
+      "step": 4204
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003514379939375457,
+      "loss": 3.2453,
+      "step": 4205
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003512140172511665,
+      "loss": 3.1934,
+      "step": 4206
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035099007331626265,
+      "loss": 3.1732,
+      "step": 4207
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003507661621821292,
+      "loss": 3.1776,
+      "step": 4208
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035054228389805504,
+      "loss": 3.1875,
+      "step": 4209
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000350318438513321,
+      "loss": 3.0945,
+      "step": 4210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003500946260772013,
+      "loss": 3.1285,
+      "step": 4211
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003498708466389622,
+      "loss": 3.247,
+      "step": 4212
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003496471002478635,
+      "loss": 3.1806,
+      "step": 4213
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000349423386953157,
+      "loss": 3.1901,
+      "step": 4214
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034919970680408766,
+      "loss": 3.1186,
+      "step": 4215
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034897605984989314,
+      "loss": 3.1894,
+      "step": 4216
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003487524461398033,
+      "loss": 3.1216,
+      "step": 4217
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034852886572304134,
+      "loss": 3.1955,
+      "step": 4218
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003483053186488227,
+      "loss": 3.2126,
+      "step": 4219
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034808180496635586,
+      "loss": 3.1729,
+      "step": 4220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003478583247248414,
+      "loss": 3.2972,
+      "step": 4221
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003476348779734732,
+      "loss": 3.2511,
+      "step": 4222
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034741146476143696,
+      "loss": 3.248,
+      "step": 4223
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034718808513791187,
+      "loss": 3.1516,
+      "step": 4224
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034696473915206915,
+      "loss": 3.2058,
+      "step": 4225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034674142685307266,
+      "loss": 3.2724,
+      "step": 4226
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034651814829007914,
+      "loss": 3.1246,
+      "step": 4227
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003462949035122376,
+      "loss": 3.195,
+      "step": 4228
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034607169256868994,
+      "loss": 3.3511,
+      "step": 4229
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003458485155085701,
+      "loss": 3.1965,
+      "step": 4230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034562537238100506,
+      "loss": 3.2688,
+      "step": 4231
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000345402263235114,
+      "loss": 3.1221,
+      "step": 4232
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034517918812000924,
+      "loss": 3.2228,
+      "step": 4233
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034495614708479464,
+      "loss": 3.184,
+      "step": 4234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034473314017856706,
+      "loss": 3.1453,
+      "step": 4235
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034451016745041597,
+      "loss": 3.1863,
+      "step": 4236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003442872289494231,
+      "loss": 3.2089,
+      "step": 4237
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034406432472466306,
+      "loss": 3.1962,
+      "step": 4238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000343841454825202,
+      "loss": 3.1547,
+      "step": 4239
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003436186193000996,
+      "loss": 3.3114,
+      "step": 4240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034339581819840717,
+      "loss": 3.2018,
+      "step": 4241
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034317305156916913,
+      "loss": 3.0251,
+      "step": 4242
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003429503194614215,
+      "loss": 3.1709,
+      "step": 4243
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034272762192419327,
+      "loss": 3.1243,
+      "step": 4244
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003425049590065057,
+      "loss": 3.2077,
+      "step": 4245
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034228233075737224,
+      "loss": 3.1441,
+      "step": 4246
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003420597372257993,
+      "loss": 3.2513,
+      "step": 4247
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003418371784607848,
+      "loss": 3.0965,
+      "step": 4248
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034161465451131956,
+      "loss": 3.2459,
+      "step": 4249
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003413921654263866,
+      "loss": 3.1875,
+      "step": 4250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003411697112549615,
+      "loss": 3.1533,
+      "step": 4251
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003409472920460117,
+      "loss": 3.1681,
+      "step": 4252
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034072490784849705,
+      "loss": 3.2463,
+      "step": 4253
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034050255871137004,
+      "loss": 3.137,
+      "step": 4254
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000340280244683575,
+      "loss": 3.2839,
+      "step": 4255
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034005796581404907,
+      "loss": 3.1489,
+      "step": 4256
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000339835722151721,
+      "loss": 3.1893,
+      "step": 4257
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003396135137455123,
+      "loss": 3.2583,
+      "step": 4258
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033939134064433634,
+      "loss": 3.1624,
+      "step": 4259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033916920289709935,
+      "loss": 3.0282,
+      "step": 4260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033894710055269885,
+      "loss": 3.1511,
+      "step": 4261
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003387250336600254,
+      "loss": 3.2936,
+      "step": 4262
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033850300226796124,
+      "loss": 3.2264,
+      "step": 4263
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033828100642538096,
+      "loss": 3.0628,
+      "step": 4264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033805904618115155,
+      "loss": 3.2084,
+      "step": 4265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003378371215841316,
+      "loss": 3.2241,
+      "step": 4266
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003376152326831725,
+      "loss": 3.154,
+      "step": 4267
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003373933795271172,
+      "loss": 3.1831,
+      "step": 4268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033717156216480157,
+      "loss": 3.1617,
+      "step": 4269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003369497806450526,
+      "loss": 3.2192,
+      "step": 4270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033672803501669,
+      "loss": 3.2653,
+      "step": 4271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033650632532852576,
+      "loss": 3.0386,
+      "step": 4272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033628465162936316,
+      "loss": 3.2432,
+      "step": 4273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003360630139679986,
+      "loss": 3.1577,
+      "step": 4274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033584141239321953,
+      "loss": 3.1083,
+      "step": 4275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033561984695380624,
+      "loss": 3.1469,
+      "step": 4276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003353983176985306,
+      "loss": 3.1334,
+      "step": 4277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003351768246761571,
+      "loss": 3.2648,
+      "step": 4278
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033495536793544113,
+      "loss": 3.2822,
+      "step": 4279
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003347339475251313,
+      "loss": 3.234,
+      "step": 4280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033451256349396777,
+      "loss": 3.1696,
+      "step": 4281
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033429121589068214,
+      "loss": 3.1158,
+      "step": 4282
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033406990476399895,
+      "loss": 3.3167,
+      "step": 4283
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000333848630162634,
+      "loss": 3.1526,
+      "step": 4284
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033362739213529567,
+      "loss": 3.1677,
+      "step": 4285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033340619073068343,
+      "loss": 3.1768,
+      "step": 4286
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003331850259974897,
+      "loss": 3.2288,
+      "step": 4287
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003329638979843977,
+      "loss": 3.1919,
+      "step": 4288
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003327428067400836,
+      "loss": 3.2671,
+      "step": 4289
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033252175231321515,
+      "loss": 3.2077,
+      "step": 4290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003323007347524515,
+      "loss": 3.2547,
+      "step": 4291
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003320797541064443,
+      "loss": 3.2235,
+      "step": 4292
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003318588104238368,
+      "loss": 3.2003,
+      "step": 4293
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003316379037532644,
+      "loss": 3.1254,
+      "step": 4294
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033141703414335365,
+      "loss": 3.0023,
+      "step": 4295
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000331196201642724,
+      "loss": 3.0829,
+      "step": 4296
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033097540629998563,
+      "loss": 3.2302,
+      "step": 4297
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003307546481637413,
+      "loss": 3.1724,
+      "step": 4298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003305339272825855,
+      "loss": 3.1656,
+      "step": 4299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00033031324370510394,
+      "loss": 3.1136,
+      "step": 4300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00033009259747987486,
+      "loss": 3.1695,
+      "step": 4301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003298719886554677,
+      "loss": 3.2487,
+      "step": 4302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003296514172804442,
+      "loss": 3.1768,
+      "step": 4303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003294308834033574,
+      "loss": 3.2315,
+      "step": 4304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003292103870727524,
+      "loss": 3.2935,
+      "step": 4305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003289899283371657,
+      "loss": 3.16,
+      "step": 4306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032876950724512575,
+      "loss": 3.1158,
+      "step": 4307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000328549123845153,
+      "loss": 3.1635,
+      "step": 4308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003283287781857588,
+      "loss": 3.1948,
+      "step": 4309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000328108470315447,
+      "loss": 3.2165,
+      "step": 4310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003278882002827127,
+      "loss": 3.1096,
+      "step": 4311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003276679681360429,
+      "loss": 3.2466,
+      "step": 4312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000327447773923916,
+      "loss": 3.1804,
+      "step": 4313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032722761769480224,
+      "loss": 3.1581,
+      "step": 4314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032700749949716345,
+      "loss": 3.2719,
+      "step": 4315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032678741937945324,
+      "loss": 3.0779,
+      "step": 4316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032656737739011655,
+      "loss": 3.2199,
+      "step": 4317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003263473735775899,
+      "loss": 3.1606,
+      "step": 4318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003261274079903018,
+      "loss": 3.1951,
+      "step": 4319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032590748067667203,
+      "loss": 3.2043,
+      "step": 4320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003256875916851122,
+      "loss": 3.0775,
+      "step": 4321
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000325467741064025,
+      "loss": 3.1817,
+      "step": 4322
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032524792886180526,
+      "loss": 3.2576,
+      "step": 4323
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003250281551268389,
+      "loss": 3.1821,
+      "step": 4324
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003248084199075038,
+      "loss": 3.1587,
+      "step": 4325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032458872325216893,
+      "loss": 3.2003,
+      "step": 4326
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032436906520919484,
+      "loss": 3.2474,
+      "step": 4327
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032414944582693394,
+      "loss": 3.1876,
+      "step": 4328
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032392986515372967,
+      "loss": 3.1252,
+      "step": 4329
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032371032323791755,
+      "loss": 3.0773,
+      "step": 4330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032349082012782373,
+      "loss": 3.0792,
+      "step": 4331
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003232713558717665,
+      "loss": 3.2049,
+      "step": 4332
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003230519305180553,
+      "loss": 3.1554,
+      "step": 4333
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032283254411499124,
+      "loss": 3.3109,
+      "step": 4334
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032261319671086653,
+      "loss": 3.1282,
+      "step": 4335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003223938883539649,
+      "loss": 3.2183,
+      "step": 4336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003221746190925618,
+      "loss": 3.1729,
+      "step": 4337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003219553889749235,
+      "loss": 3.1048,
+      "step": 4338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032173619804930844,
+      "loss": 3.0901,
+      "step": 4339
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032151704636396535,
+      "loss": 3.0831,
+      "step": 4340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032129793396713536,
+      "loss": 3.1518,
+      "step": 4341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003210788609070504,
+      "loss": 3.0302,
+      "step": 4342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032085982723193394,
+      "loss": 3.1618,
+      "step": 4343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032064083299000067,
+      "loss": 3.1068,
+      "step": 4344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00032042187822945643,
+      "loss": 3.3139,
+      "step": 4345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003202029629984991,
+      "loss": 3.1033,
+      "step": 4346
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031998408734531673,
+      "loss": 3.1995,
+      "step": 4347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003197652513180898,
+      "loss": 3.2368,
+      "step": 4348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000319546454964989,
+      "loss": 3.2304,
+      "step": 4349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003193276983341773,
+      "loss": 3.081,
+      "step": 4350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031910898147380806,
+      "loss": 3.1436,
+      "step": 4351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031889030443202674,
+      "loss": 3.1729,
+      "step": 4352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000318671667256969,
+      "loss": 3.2123,
+      "step": 4353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031845306999676273,
+      "loss": 3.2762,
+      "step": 4354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003182345126995265,
+      "loss": 3.1385,
+      "step": 4355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003180159954133699,
+      "loss": 3.1416,
+      "step": 4356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003177975181863945,
+      "loss": 3.2634,
+      "step": 4357
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031757908106669195,
+      "loss": 3.3148,
+      "step": 4358
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003173606841023461,
+      "loss": 3.2412,
+      "step": 4359
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003171423273414312,
+      "loss": 3.0294,
+      "step": 4360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003169240108320135,
+      "loss": 3.0657,
+      "step": 4361
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031670573462214927,
+      "loss": 3.2326,
+      "step": 4362
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003164874987598868,
+      "loss": 3.2638,
+      "step": 4363
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031626930329326533,
+      "loss": 3.1991,
+      "step": 4364
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003160511482703147,
+      "loss": 3.0151,
+      "step": 4365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003158330337390565,
+      "loss": 3.2689,
+      "step": 4366
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003156149597475029,
+      "loss": 3.2691,
+      "step": 4367
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031539692634365783,
+      "loss": 3.1791,
+      "step": 4368
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031517893357551534,
+      "loss": 3.1898,
+      "step": 4369
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031496098149106147,
+      "loss": 3.1655,
+      "step": 4370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003147430701382724,
+      "loss": 3.2061,
+      "step": 4371
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003145251995651162,
+      "loss": 3.1195,
+      "step": 4372
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031430736981955145,
+      "loss": 3.2595,
+      "step": 4373
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031408958094952756,
+      "loss": 3.0659,
+      "step": 4374
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003138718330029857,
+      "loss": 3.0981,
+      "step": 4375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003136541260278574,
+      "loss": 3.0637,
+      "step": 4376
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003134364600720654,
+      "loss": 3.1966,
+      "step": 4377
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003132188351835232,
+      "loss": 3.0912,
+      "step": 4378
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003130012514101357,
+      "loss": 3.2594,
+      "step": 4379
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003127837087997981,
+      "loss": 3.0932,
+      "step": 4380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003125662074003972,
+      "loss": 3.3003,
+      "step": 4381
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031234874725981045,
+      "loss": 3.1608,
+      "step": 4382
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000312131328425906,
+      "loss": 3.1476,
+      "step": 4383
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031191395094654317,
+      "loss": 3.1319,
+      "step": 4384
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003116966148695721,
+      "loss": 3.112,
+      "step": 4385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003114793202428342,
+      "loss": 3.2105,
+      "step": 4386
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031126206711416083,
+      "loss": 3.1714,
+      "step": 4387
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003110448555313752,
+      "loss": 3.1959,
+      "step": 4388
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031082768554229064,
+      "loss": 3.1005,
+      "step": 4389
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000310610557194712,
+      "loss": 3.1843,
+      "step": 4390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003103934705364344,
+      "loss": 3.056,
+      "step": 4391
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031017642561524385,
+      "loss": 3.1208,
+      "step": 4392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030995942247891747,
+      "loss": 3.0345,
+      "step": 4393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030974246117522285,
+      "loss": 3.1842,
+      "step": 4394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030952554175191894,
+      "loss": 3.1379,
+      "step": 4395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030930866425675453,
+      "loss": 3.2579,
+      "step": 4396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003090918287374701,
+      "loss": 3.0501,
+      "step": 4397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030887503524179637,
+      "loss": 3.2008,
+      "step": 4398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003086582838174551,
+      "loss": 3.1131,
+      "step": 4399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003084415745121586,
+      "loss": 3.2033,
+      "step": 4400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003082249073736097,
+      "loss": 3.0802,
+      "step": 4401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003080082824495024,
+      "loss": 3.2029,
+      "step": 4402
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003077916997875211,
+      "loss": 3.2674,
+      "step": 4403
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003075751594353413,
+      "loss": 3.1842,
+      "step": 4404
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030735866144062843,
+      "loss": 3.1468,
+      "step": 4405
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030714220585103933,
+      "loss": 3.118,
+      "step": 4406
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030692579271422124,
+      "loss": 3.1837,
+      "step": 4407
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030670942207781205,
+      "loss": 3.0979,
+      "step": 4408
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030649309398944033,
+      "loss": 3.1452,
+      "step": 4409
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003062768084967249,
+      "loss": 3.2495,
+      "step": 4410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003060605656472759,
+      "loss": 3.1266,
+      "step": 4411
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003058443654886935,
+      "loss": 3.1904,
+      "step": 4412
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030562820806856906,
+      "loss": 3.1553,
+      "step": 4413
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030541209343448375,
+      "loss": 3.0798,
+      "step": 4414
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030519602163400994,
+      "loss": 3.1815,
+      "step": 4415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030497999271471033,
+      "loss": 3.0954,
+      "step": 4416
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003047640067241385,
+      "loss": 3.101,
+      "step": 4417
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030454806370983794,
+      "loss": 3.1149,
+      "step": 4418
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030433216371934314,
+      "loss": 3.2756,
+      "step": 4419
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030411630680017923,
+      "loss": 3.2381,
+      "step": 4420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003039004929998613,
+      "loss": 3.1258,
+      "step": 4421
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003036847223658958,
+      "loss": 3.2076,
+      "step": 4422
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030346899494577864,
+      "loss": 3.2223,
+      "step": 4423
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030325331078699707,
+      "loss": 3.1679,
+      "step": 4424
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003030376699370284,
+      "loss": 3.1179,
+      "step": 4425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030282207244334083,
+      "loss": 3.1891,
+      "step": 4426
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030260651835339224,
+      "loss": 3.1348,
+      "step": 4427
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030239100771463156,
+      "loss": 3.1525,
+      "step": 4428
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030217554057449826,
+      "loss": 3.1918,
+      "step": 4429
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003019601169804216,
+      "loss": 3.0102,
+      "step": 4430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030174473697982206,
+      "loss": 3.1344,
+      "step": 4431
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003015294006201097,
+      "loss": 3.1769,
+      "step": 4432
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030131410794868564,
+      "loss": 3.1818,
+      "step": 4433
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003010988590129409,
+      "loss": 3.0946,
+      "step": 4434
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003008836538602575,
+      "loss": 3.1929,
+      "step": 4435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030066849253800705,
+      "loss": 3.2949,
+      "step": 4436
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030045337509355187,
+      "loss": 3.136,
+      "step": 4437
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00030023830157424505,
+      "loss": 3.1546,
+      "step": 4438
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00030002327202742897,
+      "loss": 3.1719,
+      "step": 4439
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002998082865004374,
+      "loss": 3.1127,
+      "step": 4440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029959334504059377,
+      "loss": 3.0611,
+      "step": 4441
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002993784476952123,
+      "loss": 3.1246,
+      "step": 4442
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029916359451159677,
+      "loss": 3.0966,
+      "step": 4443
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002989487855370421,
+      "loss": 3.1859,
+      "step": 4444
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029873402081883276,
+      "loss": 3.2014,
+      "step": 4445
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002985193004042438,
+      "loss": 3.1886,
+      "step": 4446
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029830462434054085,
+      "loss": 3.2714,
+      "step": 4447
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002980899926749788,
+      "loss": 3.1641,
+      "step": 4448
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029787540545480397,
+      "loss": 3.2276,
+      "step": 4449
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029766086272725194,
+      "loss": 3.1944,
+      "step": 4450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002974463645395492,
+      "loss": 3.1607,
+      "step": 4451
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002972319109389118,
+      "loss": 3.22,
+      "step": 4452
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002970175019725465,
+      "loss": 3.1153,
+      "step": 4453
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029680313768764985,
+      "loss": 3.079,
+      "step": 4454
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002965888181314088,
+      "loss": 3.2596,
+      "step": 4455
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002963745433510005,
+      "loss": 3.2678,
+      "step": 4456
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029616031339359185,
+      "loss": 3.1018,
+      "step": 4457
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002959461283063404,
+      "loss": 3.1338,
+      "step": 4458
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029573198813639343,
+      "loss": 3.1867,
+      "step": 4459
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002955178929308888,
+      "loss": 3.2485,
+      "step": 4460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002953038427369537,
+      "loss": 3.0609,
+      "step": 4461
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029508983760170637,
+      "loss": 3.1474,
+      "step": 4462
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029487587757225407,
+      "loss": 3.0304,
+      "step": 4463
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002946619626956952,
+      "loss": 3.0586,
+      "step": 4464
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002944480930191176,
+      "loss": 3.2352,
+      "step": 4465
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002942342685895989,
+      "loss": 3.1061,
+      "step": 4466
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029402048945420756,
+      "loss": 3.0077,
+      "step": 4467
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002938067556600015,
+      "loss": 3.0736,
+      "step": 4468
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000293593067254029,
+      "loss": 3.1296,
+      "step": 4469
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029337942428332784,
+      "loss": 3.1422,
+      "step": 4470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002931658267949264,
+      "loss": 3.3099,
+      "step": 4471
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029295227483584253,
+      "loss": 3.0691,
+      "step": 4472
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002927387684530846,
+      "loss": 3.2537,
+      "step": 4473
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002925253076936505,
+      "loss": 3.2675,
+      "step": 4474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029231189260452807,
+      "loss": 3.202,
+      "step": 4475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029209852323269536,
+      "loss": 3.1449,
+      "step": 4476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002918851996251203,
+      "loss": 3.1791,
+      "step": 4477
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029167192182876075,
+      "loss": 3.1496,
+      "step": 4478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002914586898905642,
+      "loss": 3.1588,
+      "step": 4479
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002912455038574686,
+      "loss": 3.1174,
+      "step": 4480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000291032363776401,
+      "loss": 3.2206,
+      "step": 4481
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029081926969427917,
+      "loss": 3.1654,
+      "step": 4482
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002906062216580105,
+      "loss": 3.1472,
+      "step": 4483
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029039321971449173,
+      "loss": 3.168,
+      "step": 4484
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029018026391061023,
+      "loss": 3.3089,
+      "step": 4485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028996735429324255,
+      "loss": 3.1458,
+      "step": 4486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028975449090925566,
+      "loss": 3.1569,
+      "step": 4487
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002895416738055057,
+      "loss": 3.1133,
+      "step": 4488
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002893289030288393,
+      "loss": 3.1959,
+      "step": 4489
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028911617862609223,
+      "loss": 3.2427,
+      "step": 4490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028890350064409066,
+      "loss": 3.1281,
+      "step": 4491
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002886908691296504,
+      "loss": 3.2941,
+      "step": 4492
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002884782841295764,
+      "loss": 3.1643,
+      "step": 4493
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028826574569066455,
+      "loss": 3.1858,
+      "step": 4494
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002880532538596992,
+      "loss": 3.2509,
+      "step": 4495
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028784080868345547,
+      "loss": 3.191,
+      "step": 4496
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028762841020869747,
+      "loss": 3.0864,
+      "step": 4497
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002874160584821798,
+      "loss": 3.1358,
+      "step": 4498
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028720375355064575,
+      "loss": 3.2339,
+      "step": 4499
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002869914954608293,
+      "loss": 3.144,
+      "step": 4500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002867792842594538,
+      "loss": 3.2085,
+      "step": 4501
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028656711999323176,
+      "loss": 3.2408,
+      "step": 4502
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002863550027088662,
+      "loss": 2.9579,
+      "step": 4503
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000286142932453049,
+      "loss": 3.2115,
+      "step": 4504
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028593090927246244,
+      "loss": 3.1262,
+      "step": 4505
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028571893321377776,
+      "loss": 3.2156,
+      "step": 4506
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028550700432365645,
+      "loss": 3.2541,
+      "step": 4507
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028529512264874893,
+      "loss": 3.2622,
+      "step": 4508
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028508328823569567,
+      "loss": 3.2184,
+      "step": 4509
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028487150113112706,
+      "loss": 3.0903,
+      "step": 4510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002846597613816622,
+      "loss": 3.2262,
+      "step": 4511
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002844480690339106,
+      "loss": 3.2081,
+      "step": 4512
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002842364241344707,
+      "loss": 3.1509,
+      "step": 4513
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000284024826729931,
+      "loss": 3.0985,
+      "step": 4514
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028381327686686917,
+      "loss": 3.2135,
+      "step": 4515
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028360177459185265,
+      "loss": 3.0244,
+      "step": 4516
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002833903199514383,
+      "loss": 3.0926,
+      "step": 4517
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028317891299217235,
+      "loss": 3.0954,
+      "step": 4518
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028296755376059106,
+      "loss": 3.2463,
+      "step": 4519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002827562423032195,
+      "loss": 3.1249,
+      "step": 4520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002825449786665729,
+      "loss": 3.2133,
+      "step": 4521
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028233376289715507,
+      "loss": 3.1463,
+      "step": 4522
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028212259504146043,
+      "loss": 3.1391,
+      "step": 4523
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002819114751459718,
+      "loss": 3.2367,
+      "step": 4524
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028170040325716216,
+      "loss": 3.0864,
+      "step": 4525
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028148937942149347,
+      "loss": 3.1327,
+      "step": 4526
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002812784036854172,
+      "loss": 3.2885,
+      "step": 4527
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028106747609537485,
+      "loss": 3.2478,
+      "step": 4528
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028085659669779616,
+      "loss": 3.0744,
+      "step": 4529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002806457655391015,
+      "loss": 3.163,
+      "step": 4530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028043498266569944,
+      "loss": 3.2394,
+      "step": 4531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000280224248123989,
+      "loss": 3.1724,
+      "step": 4532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028001356196035766,
+      "loss": 3.2768,
+      "step": 4533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002798029242211828,
+      "loss": 3.1896,
+      "step": 4534
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002795923349528311,
+      "loss": 3.079,
+      "step": 4535
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002793817942016585,
+      "loss": 3.1793,
+      "step": 4536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00027917130201401025,
+      "loss": 3.1879,
+      "step": 4537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002789608584362205,
+      "loss": 3.2226,
+      "step": 4538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00027875046351461365,
+      "loss": 3.0989,
+      "step": 4539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002785401172955022,
+      "loss": 3.2153,
+      "step": 4540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027832981982518925,
+      "loss": 3.2679,
+      "step": 4541
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027811957114996577,
+      "loss": 3.2148,
+      "step": 4542
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002779093713161131,
+      "loss": 3.3309,
+      "step": 4543
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002776992203699014,
+      "loss": 3.2397,
+      "step": 4544
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002774891183575902,
+      "loss": 3.2486,
+      "step": 4545
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002772790653254278,
+      "loss": 3.1203,
+      "step": 4546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002770690613196525,
+      "loss": 3.2269,
+      "step": 4547
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002768591063864911,
+      "loss": 3.1521,
+      "step": 4548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027664920057215975,
+      "loss": 3.2765,
+      "step": 4549
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027643934392286427,
+      "loss": 3.1786,
+      "step": 4550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000276229536484799,
+      "loss": 3.1451,
+      "step": 4551
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002760197783041477,
+      "loss": 3.2161,
+      "step": 4552
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027581006942708363,
+      "loss": 3.2235,
+      "step": 4553
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027560040989976894,
+      "loss": 3.1718,
+      "step": 4554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027539079976835444,
+      "loss": 3.1093,
+      "step": 4555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027518123907898105,
+      "loss": 3.0281,
+      "step": 4556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002749717278777779,
+      "loss": 3.0718,
+      "step": 4557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027476226621086353,
+      "loss": 3.2629,
+      "step": 4558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027455285412434597,
+      "loss": 3.1003,
+      "step": 4559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002743434916643216,
+      "loss": 3.1979,
+      "step": 4560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002741341788768764,
+      "loss": 3.0718,
+      "step": 4561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002739249158080855,
+      "loss": 3.2098,
+      "step": 4562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002737157025040129,
+      "loss": 3.1909,
+      "step": 4563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002735065390107112,
+      "loss": 3.2881,
+      "step": 4564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027329742537422297,
+      "loss": 3.258,
+      "step": 4565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002730883616405791,
+      "loss": 3.1274,
+      "step": 4566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027287934785579936,
+      "loss": 3.2338,
+      "step": 4567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002726703840658934,
+      "loss": 3.2258,
+      "step": 4568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027246147031685886,
+      "loss": 3.2384,
+      "step": 4569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000272252606654683,
+      "loss": 3.0686,
+      "step": 4570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002720437931253419,
+      "loss": 3.3101,
+      "step": 4571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002718350297748008,
+      "loss": 3.0962,
+      "step": 4572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027162631664901323,
+      "loss": 3.0311,
+      "step": 4573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002714176537939225,
+      "loss": 3.1796,
+      "step": 4574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002712090412554604,
+      "loss": 3.1838,
+      "step": 4575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00027100047907954733,
+      "loss": 3.3009,
+      "step": 4576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00027079196731209353,
+      "loss": 3.1507,
+      "step": 4577
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00027058350599899706,
+      "loss": 3.1066,
+      "step": 4578
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002703750951861458,
+      "loss": 3.1002,
+      "step": 4579
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002701667349194159,
+      "loss": 3.0537,
+      "step": 4580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000269958425244673,
+      "loss": 3.2584,
+      "step": 4581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002697501662077707,
+      "loss": 3.0661,
+      "step": 4582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026954195785455244,
+      "loss": 3.1509,
+      "step": 4583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026933380023084976,
+      "loss": 3.0836,
+      "step": 4584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026912569338248315,
+      "loss": 3.1636,
+      "step": 4585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002689176373552622,
+      "loss": 3.2025,
+      "step": 4586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002687096321949853,
+      "loss": 3.2062,
+      "step": 4587
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026850167794743965,
+      "loss": 3.0706,
+      "step": 4588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026829377465840075,
+      "loss": 3.14,
+      "step": 4589
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002680859223736336,
+      "loss": 3.1051,
+      "step": 4590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026787812113889144,
+      "loss": 3.2444,
+      "step": 4591
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002676703709999165,
+      "loss": 3.0053,
+      "step": 4592
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002674626720024399,
+      "loss": 3.1174,
+      "step": 4593
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026725502419218085,
+      "loss": 3.1158,
+      "step": 4594
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002670474276148481,
+      "loss": 3.1198,
+      "step": 4595
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002668398823161389,
+      "loss": 3.1602,
+      "step": 4596
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026663238834173897,
+      "loss": 3.0769,
+      "step": 4597
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002664249457373228,
+      "loss": 3.1638,
+      "step": 4598
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026621755454855385,
+      "loss": 3.1523,
+      "step": 4599
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026601021482108373,
+      "loss": 3.2395,
+      "step": 4600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002658029266005534,
+      "loss": 3.1136,
+      "step": 4601
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000265595689932592,
+      "loss": 3.0708,
+      "step": 4602
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026538850486281705,
+      "loss": 3.2708,
+      "step": 4603
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026518137143683563,
+      "loss": 3.1723,
+      "step": 4604
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002649742897002426,
+      "loss": 3.272,
+      "step": 4605
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026476725969862226,
+      "loss": 3.1536,
+      "step": 4606
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002645602814775464,
+      "loss": 3.1652,
+      "step": 4607
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002643533550825766,
+      "loss": 3.2234,
+      "step": 4608
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026414648055926197,
+      "loss": 3.0745,
+      "step": 4609
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002639396579531412,
+      "loss": 3.1498,
+      "step": 4610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026373288730974085,
+      "loss": 3.2241,
+      "step": 4611
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026352616867457597,
+      "loss": 3.1241,
+      "step": 4612
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002633195020931507,
+      "loss": 3.1765,
+      "step": 4613
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002631128876109575,
+      "loss": 3.1491,
+      "step": 4614
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026290632527347746,
+      "loss": 3.1721,
+      "step": 4615
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002626998151261798,
+      "loss": 3.3024,
+      "step": 4616
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002624933572145227,
+      "loss": 3.1746,
+      "step": 4617
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002622869515839524,
+      "loss": 3.0736,
+      "step": 4618
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002620805982799042,
+      "loss": 3.1496,
+      "step": 4619
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026187429734780143,
+      "loss": 3.1626,
+      "step": 4620
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002616680488330558,
+      "loss": 3.1653,
+      "step": 4621
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026146185278106803,
+      "loss": 3.1045,
+      "step": 4622
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026125570923722673,
+      "loss": 3.1786,
+      "step": 4623
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026104961824690963,
+      "loss": 3.3294,
+      "step": 4624
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002608435798554819,
+      "loss": 3.1639,
+      "step": 4625
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002606375941082981,
+      "loss": 3.2586,
+      "step": 4626
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002604316610507005,
+      "loss": 3.1593,
+      "step": 4627
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002602257807280203,
+      "loss": 3.1559,
+      "step": 4628
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002600199531855768,
+      "loss": 3.1758,
+      "step": 4629
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002598141784686775,
+      "loss": 3.3047,
+      "step": 4630
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025960845662261855,
+      "loss": 3.2599,
+      "step": 4631
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025940278769268467,
+      "loss": 3.1531,
+      "step": 4632
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002591971717241487,
+      "loss": 3.2921,
+      "step": 4633
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025899160876227147,
+      "loss": 3.0657,
+      "step": 4634
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025878609885230275,
+      "loss": 3.1355,
+      "step": 4635
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002585806420394802,
+      "loss": 3.149,
+      "step": 4636
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002583752383690301,
+      "loss": 3.1341,
+      "step": 4637
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025816988788616665,
+      "loss": 3.116,
+      "step": 4638
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025796459063609267,
+      "loss": 3.1596,
+      "step": 4639
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025775934666399936,
+      "loss": 3.2647,
+      "step": 4640
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025755415601506573,
+      "loss": 3.2792,
+      "step": 4641
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002573490187344596,
+      "loss": 3.3075,
+      "step": 4642
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025714393486733636,
+      "loss": 3.2197,
+      "step": 4643
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002569389044588404,
+      "loss": 3.1239,
+      "step": 4644
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002567339275541037,
+      "loss": 3.0941,
+      "step": 4645
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002565290041982471,
+      "loss": 3.1523,
+      "step": 4646
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025632413443637885,
+      "loss": 3.1606,
+      "step": 4647
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025611931831359615,
+      "loss": 3.1903,
+      "step": 4648
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002559145558749842,
+      "loss": 3.0856,
+      "step": 4649
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025570984716561594,
+      "loss": 3.0458,
+      "step": 4650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002555051922305532,
+      "loss": 3.185,
+      "step": 4651
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025530059111484514,
+      "loss": 3.1185,
+      "step": 4652
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025509604386353,
+      "loss": 3.1567,
+      "step": 4653
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002548915505216333,
+      "loss": 3.2298,
+      "step": 4654
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025468711113416944,
+      "loss": 3.0566,
+      "step": 4655
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025448272574614015,
+      "loss": 3.2068,
+      "step": 4656
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002542783944025361,
+      "loss": 3.1994,
+      "step": 4657
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002540741171483356,
+      "loss": 3.1709,
+      "step": 4658
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025386989402850486,
+      "loss": 3.1098,
+      "step": 4659
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025366572508799884,
+      "loss": 3.0496,
+      "step": 4660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025346161037175977,
+      "loss": 3.1481,
+      "step": 4661
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025325754992471883,
+      "loss": 3.2612,
+      "step": 4662
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002530535437917942,
+      "loss": 3.2384,
+      "step": 4663
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025284959201789313,
+      "loss": 3.1182,
+      "step": 4664
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025264569464791023,
+      "loss": 3.2053,
+      "step": 4665
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025244185172672827,
+      "loss": 3.1141,
+      "step": 4666
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002522380632992185,
+      "loss": 3.3139,
+      "step": 4667
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520343294102394,
+      "loss": 3.0521,
+      "step": 4668
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518306501046381,
+      "loss": 3.1451,
+      "step": 4669
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025162702542724926,
+      "loss": 3.1,
+      "step": 4670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002514234554228959,
+      "loss": 3.2367,
+      "step": 4671
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002512199401363886,
+      "loss": 3.0206,
+      "step": 4672
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025101647961252635,
+      "loss": 3.147,
+      "step": 4673
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002508130738960955,
+      "loss": 3.2224,
+      "step": 4674
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002506097230318709,
+      "loss": 3.2643,
+      "step": 4675
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025040642706461533,
+      "loss": 3.1765,
+      "step": 4676
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002502031860390788,
+      "loss": 3.1892,
+      "step": 4677
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 3.3372,
+      "step": 4678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024979686899210515,
+      "loss": 3.2664,
+      "step": 4679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002495937930601084,
+      "loss": 3.1426,
+      "step": 4680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024939077224871164,
+      "loss": 3.1751,
+      "step": 4681
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024918780660260507,
+      "loss": 3.1972,
+      "step": 4682
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024898489616646607,
+      "loss": 3.1065,
+      "step": 4683
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024878204098496045,
+      "loss": 3.1524,
+      "step": 4684
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024857924110274185,
+      "loss": 3.1291,
+      "step": 4685
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024837649656445115,
+      "loss": 3.3093,
+      "step": 4686
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002481738074147178,
+      "loss": 3.2211,
+      "step": 4687
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002479711736981584,
+      "loss": 3.03,
+      "step": 4688
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024776859545937787,
+      "loss": 3.2136,
+      "step": 4689
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024756607274296845,
+      "loss": 3.0667,
+      "step": 4690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002473636055935105,
+      "loss": 3.1628,
+      "step": 4691
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024716119405557215,
+      "loss": 3.2409,
+      "step": 4692
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002469588381737093,
+      "loss": 3.1614,
+      "step": 4693
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002467565379924653,
+      "loss": 3.1096,
+      "step": 4694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002465542935563712,
+      "loss": 3.1729,
+      "step": 4695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002463521049099465,
+      "loss": 3.2557,
+      "step": 4696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002461499720976974,
+      "loss": 3.1858,
+      "step": 4697
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002459478951641188,
+      "loss": 3.2453,
+      "step": 4698
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002457458741536924,
+      "loss": 3.1961,
+      "step": 4699
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002455439091108883,
+      "loss": 3.1637,
+      "step": 4700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024534200008016397,
+      "loss": 3.1196,
+      "step": 4701
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024514014710596464,
+      "loss": 3.2624,
+      "step": 4702
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024493835023272313,
+      "loss": 3.0744,
+      "step": 4703
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002447366095048597,
+      "loss": 3.2355,
+      "step": 4704
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002445349249667828,
+      "loss": 3.1455,
+      "step": 4705
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024433329666288773,
+      "loss": 3.2215,
+      "step": 4706
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002441317246375584,
+      "loss": 3.1803,
+      "step": 4707
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024393020893516537,
+      "loss": 3.2079,
+      "step": 4708
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024372874960006742,
+      "loss": 3.0395,
+      "step": 4709
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002435273466766107,
+      "loss": 3.2556,
+      "step": 4710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024332600020912916,
+      "loss": 3.0952,
+      "step": 4711
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000243124710241944,
+      "loss": 3.0976,
+      "step": 4712
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024292347681936395,
+      "loss": 3.1068,
+      "step": 4713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024272229998568574,
+      "loss": 3.0871,
+      "step": 4714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024252117978519304,
+      "loss": 3.1564,
+      "step": 4715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024232011626215783,
+      "loss": 3.0593,
+      "step": 4716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002421191094608387,
+      "loss": 3.1595,
+      "step": 4717
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002419181594254824,
+      "loss": 3.1244,
+      "step": 4718
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024171726620032309,
+      "loss": 3.2051,
+      "step": 4719
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024151642982958243,
+      "loss": 3.1475,
+      "step": 4720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024131565035746932,
+      "loss": 3.1166,
+      "step": 4721
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024111492782818013,
+      "loss": 3.2291,
+      "step": 4722
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002409142622858992,
+      "loss": 3.1165,
+      "step": 4723
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024071365377479766,
+      "loss": 3.1898,
+      "step": 4724
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024051310233903466,
+      "loss": 3.1328,
+      "step": 4725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024031260802275623,
+      "loss": 3.177,
+      "step": 4726
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024011217087009635,
+      "loss": 3.0635,
+      "step": 4727
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023991179092517612,
+      "loss": 3.2857,
+      "step": 4728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023971146823210438,
+      "loss": 3.2692,
+      "step": 4729
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023951120283497668,
+      "loss": 3.1513,
+      "step": 4730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002393109947778768,
+      "loss": 3.1462,
+      "step": 4731
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023911084410487533,
+      "loss": 3.0955,
+      "step": 4732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002389107508600301,
+      "loss": 3.2386,
+      "step": 4733
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000238710715087387,
+      "loss": 3.0783,
+      "step": 4734
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023851073683097852,
+      "loss": 3.2868,
+      "step": 4735
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023831081613482496,
+      "loss": 3.2019,
+      "step": 4736
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023811095304293383,
+      "loss": 3.0669,
+      "step": 4737
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002379111475993001,
+      "loss": 3.0716,
+      "step": 4738
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002377113998479056,
+      "loss": 3.2232,
+      "step": 4739
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023751170983271996,
+      "loss": 3.1723,
+      "step": 4740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023731207759769985,
+      "loss": 3.2145,
+      "step": 4741
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023711250318678907,
+      "loss": 3.1971,
+      "step": 4742
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023691298664391903,
+      "loss": 3.2075,
+      "step": 4743
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023671352801300823,
+      "loss": 3.0842,
+      "step": 4744
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002365141273379627,
+      "loss": 3.0985,
+      "step": 4745
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023631478466267498,
+      "loss": 3.1266,
+      "step": 4746
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023611550003102582,
+      "loss": 3.2164,
+      "step": 4747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023591627348688228,
+      "loss": 3.2108,
+      "step": 4748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023571710507409944,
+      "loss": 3.1917,
+      "step": 4749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002355179948365189,
+      "loss": 3.177,
+      "step": 4750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023531894281796973,
+      "loss": 3.2635,
+      "step": 4751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002351199490622683,
+      "loss": 3.066,
+      "step": 4752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023492101361321805,
+      "loss": 3.2129,
+      "step": 4753
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023472213651460976,
+      "loss": 3.209,
+      "step": 4754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002345233178102209,
+      "loss": 3.3059,
+      "step": 4755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023432455754381675,
+      "loss": 3.1397,
+      "step": 4756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002341258557591489,
+      "loss": 3.1549,
+      "step": 4757
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023392721249995696,
+      "loss": 3.1978,
+      "step": 4758
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023372862780996707,
+      "loss": 3.2501,
+      "step": 4759
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002335301017328924,
+      "loss": 3.1446,
+      "step": 4760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023333163431243364,
+      "loss": 3.137,
+      "step": 4761
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002331332255922784,
+      "loss": 3.1342,
+      "step": 4762
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002329348756161015,
+      "loss": 3.1787,
+      "step": 4763
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002327365844275643,
+      "loss": 3.2193,
+      "step": 4764
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023253835207031603,
+      "loss": 3.2207,
+      "step": 4765
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023234017858799212,
+      "loss": 3.1829,
+      "step": 4766
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023214206402421585,
+      "loss": 3.0381,
+      "step": 4767
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002319440084225969,
+      "loss": 3.0953,
+      "step": 4768
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023174601182673204,
+      "loss": 3.0973,
+      "step": 4769
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002315480742802054,
+      "loss": 3.2272,
+      "step": 4770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000231350195826588,
+      "loss": 3.1671,
+      "step": 4771
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023115237650943792,
+      "loss": 3.1089,
+      "step": 4772
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002309546163722997,
+      "loss": 3.2299,
+      "step": 4773
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002307569154587056,
+      "loss": 3.3094,
+      "step": 4774
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023055927381217413,
+      "loss": 3.1787,
+      "step": 4775
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002303616914762115,
+      "loss": 3.2224,
+      "step": 4776
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002301641684943102,
+      "loss": 3.2235,
+      "step": 4777
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022996670490994987,
+      "loss": 3.1893,
+      "step": 4778
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022976930076659718,
+      "loss": 3.1835,
+      "step": 4779
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022957195610770576,
+      "loss": 3.0309,
+      "step": 4780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002293746709767162,
+      "loss": 3.0467,
+      "step": 4781
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022917744541705543,
+      "loss": 3.2067,
+      "step": 4782
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022898027947213813,
+      "loss": 3.1588,
+      "step": 4783
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000228783173185365,
+      "loss": 3.1309,
+      "step": 4784
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022858612660012434,
+      "loss": 3.2307,
+      "step": 4785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002283891397597908,
+      "loss": 3.055,
+      "step": 4786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002281922127077259,
+      "loss": 3.2264,
+      "step": 4787
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022799534548727834,
+      "loss": 3.1983,
+      "step": 4788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022779853814178347,
+      "loss": 3.1529,
+      "step": 4789
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022760179071456356,
+      "loss": 3.0728,
+      "step": 4790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022740510324892733,
+      "loss": 3.2164,
+      "step": 4791
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002272084757881709,
+      "loss": 3.1414,
+      "step": 4792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002270119083755764,
+      "loss": 3.1329,
+      "step": 4793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022681540105441356,
+      "loss": 3.082,
+      "step": 4794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022661895386793824,
+      "loss": 3.0959,
+      "step": 4795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022642256685939338,
+      "loss": 3.199,
+      "step": 4796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022622624007200887,
+      "loss": 3.1239,
+      "step": 4797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022602997354900073,
+      "loss": 3.2793,
+      "step": 4798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002258337673335724,
+      "loss": 3.1101,
+      "step": 4799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022563762146891344,
+      "loss": 3.139,
+      "step": 4800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002254415359982006,
+      "loss": 3.1757,
+      "step": 4801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000225245510964597,
+      "loss": 3.1062,
+      "step": 4802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002250495464112528,
+      "loss": 3.2204,
+      "step": 4803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022485364238130436,
+      "loss": 3.2032,
+      "step": 4804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022465779891787512,
+      "loss": 3.2247,
+      "step": 4805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022446201606407534,
+      "loss": 3.2433,
+      "step": 4806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022426629386300124,
+      "loss": 3.0789,
+      "step": 4807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022407063235773645,
+      "loss": 3.2201,
+      "step": 4808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022387503159135064,
+      "loss": 3.087,
+      "step": 4809
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002236794916069007,
+      "loss": 3.1565,
+      "step": 4810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022348401244742943,
+      "loss": 3.1342,
+      "step": 4811
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022328859415596703,
+      "loss": 3.2827,
+      "step": 4812
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022309323677552944,
+      "loss": 3.1224,
+      "step": 4813
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022289794034911997,
+      "loss": 3.2083,
+      "step": 4814
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022270270491972833,
+      "loss": 3.0468,
+      "step": 4815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022250753053033024,
+      "loss": 3.082,
+      "step": 4816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022231241722388883,
+      "loss": 3.1906,
+      "step": 4817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022211736504335305,
+      "loss": 3.3461,
+      "step": 4818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000221922374031659,
+      "loss": 3.2171,
+      "step": 4819
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022172744423172865,
+      "loss": 3.2047,
+      "step": 4820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022153257568647124,
+      "loss": 3.15,
+      "step": 4821
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022133776843878184,
+      "loss": 3.1991,
+      "step": 4822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002211430225315425,
+      "loss": 3.1327,
+      "step": 4823
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002209483380076218,
+      "loss": 3.1133,
+      "step": 4824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022075371490987424,
+      "loss": 3.0977,
+      "step": 4825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022055915328114158,
+      "loss": 3.1069,
+      "step": 4826
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002203646531642512,
+      "loss": 3.1309,
+      "step": 4827
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022017021460201785,
+      "loss": 3.0878,
+      "step": 4828
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021997583763724187,
+      "loss": 3.1742,
+      "step": 4829
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021978152231271077,
+      "loss": 3.0155,
+      "step": 4830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021958726867119782,
+      "loss": 3.2554,
+      "step": 4831
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021939307675546322,
+      "loss": 3.2394,
+      "step": 4832
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021919894660825363,
+      "loss": 3.0463,
+      "step": 4833
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021900487827230148,
+      "loss": 3.134,
+      "step": 4834
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021881087179032643,
+      "loss": 3.1376,
+      "step": 4835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021861692720503368,
+      "loss": 3.1092,
+      "step": 4836
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021842304455911565,
+      "loss": 3.1688,
+      "step": 4837
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021822922389525036,
+      "loss": 3.1174,
+      "step": 4838
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002180354652561028,
+      "loss": 3.147,
+      "step": 4839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021784176868432377,
+      "loss": 3.1493,
+      "step": 4840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002176481342225508,
+      "loss": 3.0886,
+      "step": 4841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002174545619134079,
+      "loss": 3.1193,
+      "step": 4842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002172610517995046,
+      "loss": 3.0278,
+      "step": 4843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021706760392343778,
+      "loss": 3.2115,
+      "step": 4844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021687421832778964,
+      "loss": 3.1312,
+      "step": 4845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021668089505512956,
+      "loss": 3.0539,
+      "step": 4846
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021668089505512956,
+      "loss": 3.236,
+      "step": 4847
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002164876341480123,
+      "loss": 3.1148,
+      "step": 4848
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002162944356489797,
+      "loss": 3.144,
+      "step": 4849
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021610129960055946,
+      "loss": 2.9799,
+      "step": 4850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021590822604526576,
+      "loss": 3.0892,
+      "step": 4851
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021571521502559876,
+      "loss": 3.0612,
+      "step": 4852
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021552226658404466,
+      "loss": 3.1319,
+      "step": 4853
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002153293807630766,
+      "loss": 3.1463,
+      "step": 4854
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002151365576051532,
+      "loss": 3.0752,
+      "step": 4855
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021494379715272,
+      "loss": 3.2536,
+      "step": 4856
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002147510994482078,
+      "loss": 3.1966,
+      "step": 4857
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021455846453403456,
+      "loss": 3.1548,
+      "step": 4858
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021436589245260373,
+      "loss": 3.2177,
+      "step": 4859
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021417338324630558,
+      "loss": 3.1875,
+      "step": 4860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021398093695751587,
+      "loss": 3.1237,
+      "step": 4861
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021378855362859663,
+      "loss": 3.1576,
+      "step": 4862
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021359623330189653,
+      "loss": 3.186,
+      "step": 4863
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002134039760197497,
+      "loss": 3.1699,
+      "step": 4864
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021321178182447708,
+      "loss": 3.1358,
+      "step": 4865
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021301965075838498,
+      "loss": 3.1838,
+      "step": 4866
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002128275828637664,
+      "loss": 3.0968,
+      "step": 4867
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021263557818290024,
+      "loss": 3.1526,
+      "step": 4868
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021244363675805168,
+      "loss": 3.0926,
+      "step": 4869
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002122517586314715,
+      "loss": 3.1612,
+      "step": 4870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002120599438453968,
+      "loss": 3.0952,
+      "step": 4871
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002118681924420509,
+      "loss": 3.124,
+      "step": 4872
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002116765044636429,
+      "loss": 3.2084,
+      "step": 4873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002114848799523683,
+      "loss": 3.3064,
+      "step": 4874
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021129331895040803,
+      "loss": 3.1184,
+      "step": 4875
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021110182149992963,
+      "loss": 3.047,
+      "step": 4876
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002109103876430864,
+      "loss": 3.1922,
+      "step": 4877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021071901742201782,
+      "loss": 3.1065,
+      "step": 4878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002105277108788491,
+      "loss": 3.1321,
+      "step": 4879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021033646805569133,
+      "loss": 3.1027,
+      "step": 4880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021014528899464214,
+      "loss": 3.148,
+      "step": 4881
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020995417373778435,
+      "loss": 3.1258,
+      "step": 4882
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020976312232718765,
+      "loss": 3.115,
+      "step": 4883
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020957213480490673,
+      "loss": 3.2527,
+      "step": 4884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002093812112129828,
+      "loss": 3.0708,
+      "step": 4885
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020919035159344302,
+      "loss": 3.1619,
+      "step": 4886
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002089995559883004,
+      "loss": 3.0303,
+      "step": 4887
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020880882443955352,
+      "loss": 3.0762,
+      "step": 4888
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020861815698918707,
+      "loss": 3.2248,
+      "step": 4889
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020842755367917194,
+      "loss": 3.1526,
+      "step": 4890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020823701455146431,
+      "loss": 3.1506,
+      "step": 4891
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002080465396480069,
+      "loss": 3.2438,
+      "step": 4892
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020785612901072763,
+      "loss": 3.2399,
+      "step": 4893
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020766578268154072,
+      "loss": 3.1575,
+      "step": 4894
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002074755007023461,
+      "loss": 3.1926,
+      "step": 4895
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020728528311502974,
+      "loss": 3.0671,
+      "step": 4896
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020709512996146314,
+      "loss": 3.05,
+      "step": 4897
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020690504128350335,
+      "loss": 3.2301,
+      "step": 4898
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020671501712299413,
+      "loss": 3.1826,
+      "step": 4899
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020652505752176398,
+      "loss": 3.0602,
+      "step": 4900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020633516252162803,
+      "loss": 3.1879,
+      "step": 4901
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020614533216438673,
+      "loss": 3.1159,
+      "step": 4902
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020595556649182677,
+      "loss": 3.2745,
+      "step": 4903
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002057658655457198,
+      "loss": 3.1159,
+      "step": 4904
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020557622936782412,
+      "loss": 3.2713,
+      "step": 4905
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020538665799988304,
+      "loss": 3.1568,
+      "step": 4906
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020519715148362584,
+      "loss": 3.0868,
+      "step": 4907
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020500770986076796,
+      "loss": 3.1231,
+      "step": 4908
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020481833317300975,
+      "loss": 3.0659,
+      "step": 4909
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020462902146203794,
+      "loss": 3.2051,
+      "step": 4910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020443977476952469,
+      "loss": 3.2312,
+      "step": 4911
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020425059313712807,
+      "loss": 3.0831,
+      "step": 4912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020406147660649132,
+      "loss": 3.1583,
+      "step": 4913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020387242521924392,
+      "loss": 3.03,
+      "step": 4914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020368343901700054,
+      "loss": 3.1212,
+      "step": 4915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020349451804136192,
+      "loss": 3.0455,
+      "step": 4916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020330566233391417,
+      "loss": 3.1794,
+      "step": 4917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002031168719362289,
+      "loss": 2.9961,
+      "step": 4918
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020292814688986372,
+      "loss": 3.1809,
+      "step": 4919
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020273948723636165,
+      "loss": 3.2329,
+      "step": 4920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002025508930172516,
+      "loss": 3.2684,
+      "step": 4921
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020236236427404735,
+      "loss": 3.1181,
+      "step": 4922
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020217390104824924,
+      "loss": 3.2072,
+      "step": 4923
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020198550338134218,
+      "loss": 3.2338,
+      "step": 4924
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002017971713147977,
+      "loss": 3.1974,
+      "step": 4925
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020160890489007193,
+      "loss": 3.1437,
+      "step": 4926
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020142070414860702,
+      "loss": 3.1719,
+      "step": 4927
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020123256913183064,
+      "loss": 3.1549,
+      "step": 4928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020104449988115597,
+      "loss": 3.0832,
+      "step": 4929
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000200856496437982,
+      "loss": 3.2204,
+      "step": 4930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020066855884369244,
+      "loss": 3.2044,
+      "step": 4931
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020048068713965743,
+      "loss": 3.3526,
+      "step": 4932
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020029288136723177,
+      "loss": 3.0929,
+      "step": 4933
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002001051415677566,
+      "loss": 3.2433,
+      "step": 4934
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019991746778255786,
+      "loss": 3.1197,
+      "step": 4935
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019972986005294707,
+      "loss": 3.0751,
+      "step": 4936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001995423184202214,
+      "loss": 3.1883,
+      "step": 4937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019935484292566346,
+      "loss": 3.0802,
+      "step": 4938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019916743361054145,
+      "loss": 3.0523,
+      "step": 4939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019898009051610844,
+      "loss": 3.1174,
+      "step": 4940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019879281368360359,
+      "loss": 3.15,
+      "step": 4941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019860560315425086,
+      "loss": 3.2163,
+      "step": 4942
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001984184589692602,
+      "loss": 3.1737,
+      "step": 4943
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019823138116982652,
+      "loss": 3.1692,
+      "step": 4944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001980443697971302,
+      "loss": 3.146,
+      "step": 4945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001978574248923371,
+      "loss": 3.1343,
+      "step": 4946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019767054649659849,
+      "loss": 3.1502,
+      "step": 4947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019748373465105107,
+      "loss": 3.0044,
+      "step": 4948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019729698939681645,
+      "loss": 3.2191,
+      "step": 4949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019711031077500218,
+      "loss": 3.2671,
+      "step": 4950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019692369882670053,
+      "loss": 3.0781,
+      "step": 4951
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019673715359298972,
+      "loss": 3.1761,
+      "step": 4952
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019655067511493263,
+      "loss": 3.2169,
+      "step": 4953
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019636426343357794,
+      "loss": 3.1785,
+      "step": 4954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001961779185899597,
+      "loss": 3.2021,
+      "step": 4955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019599164062509666,
+      "loss": 3.2001,
+      "step": 4956
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019580542957999353,
+      "loss": 3.1405,
+      "step": 4957
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019561928549563967,
+      "loss": 3.2605,
+      "step": 4958
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019543320841301028,
+      "loss": 3.0594,
+      "step": 4959
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019524719837306514,
+      "loss": 3.1016,
+      "step": 4960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019506125541675018,
+      "loss": 3.0993,
+      "step": 4961
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019487537958499552,
+      "loss": 3.1634,
+      "step": 4962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001946895709187173,
+      "loss": 3.1763,
+      "step": 4963
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019450382945881674,
+      "loss": 3.2498,
+      "step": 4964
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001943181552461798,
+      "loss": 3.1754,
+      "step": 4965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001941325483216783,
+      "loss": 3.099,
+      "step": 4966
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019394700872616856,
+      "loss": 3.1702,
+      "step": 4967
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019376153650049283,
+      "loss": 3.1546,
+      "step": 4968
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019357613168547773,
+      "loss": 2.973,
+      "step": 4969
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001933907943219358,
+      "loss": 3.1945,
+      "step": 4970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019320552445066408,
+      "loss": 3.1176,
+      "step": 4971
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001930203221124452,
+      "loss": 3.2118,
+      "step": 4972
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019283518734804694,
+      "loss": 3.2346,
+      "step": 4973
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019265012019822165,
+      "loss": 3.1857,
+      "step": 4974
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019246512070370764,
+      "loss": 3.3577,
+      "step": 4975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019228018890522742,
+      "loss": 3.162,
+      "step": 4976
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019209532484348947,
+      "loss": 3.1579,
+      "step": 4977
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019191052855918657,
+      "loss": 3.116,
+      "step": 4978
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019172580009299735,
+      "loss": 3.0441,
+      "step": 4979
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019154113948558466,
+      "loss": 3.2342,
+      "step": 4980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001913565467775971,
+      "loss": 3.1608,
+      "step": 4981
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019117202200966832,
+      "loss": 3.2481,
+      "step": 4982
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019098756522241634,
+      "loss": 3.0637,
+      "step": 4983
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019080317645644507,
+      "loss": 3.1949,
+      "step": 4984
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019061885575234272,
+      "loss": 3.0564,
+      "step": 4985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019043460315068306,
+      "loss": 3.1937,
+      "step": 4986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019025041869202448,
+      "loss": 3.218,
+      "step": 4987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001900663024169107,
+      "loss": 2.9805,
+      "step": 4988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018988225436587002,
+      "loss": 3.0926,
+      "step": 4989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018969827457941614,
+      "loss": 3.0241,
+      "step": 4990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018951436309804764,
+      "loss": 3.1611,
+      "step": 4991
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018933051996224782,
+      "loss": 3.2516,
+      "step": 4992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018914674521248532,
+      "loss": 3.1694,
+      "step": 4993
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001889630388892131,
+      "loss": 3.0599,
+      "step": 4994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018877940103287,
+      "loss": 3.0429,
+      "step": 4995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018859583168387878,
+      "loss": 3.1682,
+      "step": 4996
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018841233088264797,
+      "loss": 2.9952,
+      "step": 4997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018822889866957027,
+      "loss": 3.1523,
+      "step": 4998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018804553508502386,
+      "loss": 3.1523,
+      "step": 4999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001878622401693717,
+      "loss": 3.0525,
+      "step": 5000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018767901396296127,
+      "loss": 3.3394,
+      "step": 5001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018749585650612555,
+      "loss": 3.1009,
+      "step": 5002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001873127678391816,
+      "loss": 3.0961,
+      "step": 5003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018712974800243215,
+      "loss": 3.0347,
+      "step": 5004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018694679703616402,
+      "loss": 3.1738,
+      "step": 5005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001867639149806495,
+      "loss": 3.2247,
+      "step": 5006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018658110187614536,
+      "loss": 3.1574,
+      "step": 5007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001863983577628935,
+      "loss": 3.1298,
+      "step": 5008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001862156826811202,
+      "loss": 3.1571,
+      "step": 5009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001860330766710367,
+      "loss": 3.2516,
+      "step": 5010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018585053977283934,
+      "loss": 3.0246,
+      "step": 5011
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018566807202670865,
+      "loss": 3.2718,
+      "step": 5012
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018548567347281077,
+      "loss": 3.3197,
+      "step": 5013
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018530334415129568,
+      "loss": 3.2324,
+      "step": 5014
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018512108410229878,
+      "loss": 3.178,
+      "step": 5015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018493889336593995,
+      "loss": 3.091,
+      "step": 5016
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018475677198232415,
+      "loss": 3.0006,
+      "step": 5017
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018457471999154063,
+      "loss": 3.2251,
+      "step": 5018
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018439273743366325,
+      "loss": 3.1379,
+      "step": 5019
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018421082434875132,
+      "loss": 3.1103,
+      "step": 5020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018402898077684804,
+      "loss": 3.1625,
+      "step": 5021
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000183847206757982,
+      "loss": 3.0827,
+      "step": 5022
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018366550233216584,
+      "loss": 3.1919,
+      "step": 5023
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018348386753939733,
+      "loss": 3.2462,
+      "step": 5024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018330230241965883,
+      "loss": 3.1634,
+      "step": 5025
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001831208070129175,
+      "loss": 3.1601,
+      "step": 5026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018293938135912474,
+      "loss": 3.1203,
+      "step": 5027
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001827580254982167,
+      "loss": 3.1295,
+      "step": 5028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018257673947011456,
+      "loss": 3.1419,
+      "step": 5029
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018239552331472358,
+      "loss": 3.1273,
+      "step": 5030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001822143770719342,
+      "loss": 3.2123,
+      "step": 5031
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000182033300781621,
+      "loss": 3.2782,
+      "step": 5032
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018185229448364337,
+      "loss": 3.0794,
+      "step": 5033
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001816713582178453,
+      "loss": 3.2066,
+      "step": 5034
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018149049202405553,
+      "loss": 3.2021,
+      "step": 5035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018130969594208702,
+      "loss": 3.1035,
+      "step": 5036
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001811289700117373,
+      "loss": 3.2057,
+      "step": 5037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001809483142727889,
+      "loss": 3.0486,
+      "step": 5038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001807677287650083,
+      "loss": 3.0078,
+      "step": 5039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018058721352814721,
+      "loss": 3.0885,
+      "step": 5040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018040676860194116,
+      "loss": 2.9607,
+      "step": 5041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001802263940261107,
+      "loss": 3.0293,
+      "step": 5042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018004608984036069,
+      "loss": 3.1549,
+      "step": 5043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001798658560843807,
+      "loss": 3.1456,
+      "step": 5044
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017968569279784463,
+      "loss": 3.0543,
+      "step": 5045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001795056000204105,
+      "loss": 3.1443,
+      "step": 5046
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001793255777917217,
+      "loss": 3.0805,
+      "step": 5047
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017914562615140505,
+      "loss": 3.1715,
+      "step": 5048
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017896574513907287,
+      "loss": 3.1215,
+      "step": 5049
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017878593479432104,
+      "loss": 3.2094,
+      "step": 5050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001786061951567303,
+      "loss": 3.1217,
+      "step": 5051
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017842652626586598,
+      "loss": 3.2132,
+      "step": 5052
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017824692816127775,
+      "loss": 3.1405,
+      "step": 5053
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017806740088249933,
+      "loss": 3.1858,
+      "step": 5054
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017788794446904906,
+      "loss": 3.1055,
+      "step": 5055
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017770855896043002,
+      "loss": 3.0878,
+      "step": 5056
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017752924439612912,
+      "loss": 3.104,
+      "step": 5057
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017735000081561797,
+      "loss": 3.2313,
+      "step": 5058
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017717082825835268,
+      "loss": 3.1515,
+      "step": 5059
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017699172676377363,
+      "loss": 2.9477,
+      "step": 5060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001768126963713052,
+      "loss": 3.1655,
+      "step": 5061
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017663373712035668,
+      "loss": 3.1422,
+      "step": 5062
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017645484905032128,
+      "loss": 3.131,
+      "step": 5063
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017627603220057657,
+      "loss": 3.0871,
+      "step": 5064
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017609728661048492,
+      "loss": 3.0861,
+      "step": 5065
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017591861231939216,
+      "loss": 3.2191,
+      "step": 5066
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017574000936662927,
+      "loss": 3.0206,
+      "step": 5067
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001755614777915111,
+      "loss": 3.1692,
+      "step": 5068
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017538301763333702,
+      "loss": 3.0007,
+      "step": 5069
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017520462893139022,
+      "loss": 3.0352,
+      "step": 5070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017502631172493878,
+      "loss": 3.1687,
+      "step": 5071
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017484806605323457,
+      "loss": 3.1945,
+      "step": 5072
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001746698919555137,
+      "loss": 3.1269,
+      "step": 5073
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017449178947099702,
+      "loss": 3.2012,
+      "step": 5074
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017431375863888898,
+      "loss": 3.0409,
+      "step": 5075
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001741357994983787,
+      "loss": 3.1089,
+      "step": 5076
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017395791208863948,
+      "loss": 3.2255,
+      "step": 5077
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001737800964488288,
+      "loss": 3.0954,
+      "step": 5078
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000173602352618088,
+      "loss": 3.0814,
+      "step": 5079
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017342468063554322,
+      "loss": 3.1429,
+      "step": 5080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017324708054030431,
+      "loss": 3.0447,
+      "step": 5081
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017306955237146522,
+      "loss": 3.0767,
+      "step": 5082
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017289209616810469,
+      "loss": 3.1377,
+      "step": 5083
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017271471196928485,
+      "loss": 3.1865,
+      "step": 5084
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017253739981405253,
+      "loss": 3.1529,
+      "step": 5085
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017236015974143848,
+      "loss": 3.1991,
+      "step": 5086
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017218299179045787,
+      "loss": 3.1605,
+      "step": 5087
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017200589600010934,
+      "loss": 3.0345,
+      "step": 5088
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017182887240937644,
+      "loss": 3.0794,
+      "step": 5089
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017165192105722622,
+      "loss": 3.2268,
+      "step": 5090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017147504198260994,
+      "loss": 3.1163,
+      "step": 5091
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017129823522446337,
+      "loss": 3.0638,
+      "step": 5092
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017112150082170568,
+      "loss": 3.1321,
+      "step": 5093
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017094483881324073,
+      "loss": 3.2566,
+      "step": 5094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017076824923795614,
+      "loss": 3.2599,
+      "step": 5095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017059173213472385,
+      "loss": 3.1829,
+      "step": 5096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017041528754239927,
+      "loss": 3.0713,
+      "step": 5097
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017023891549982256,
+      "loss": 3.1495,
+      "step": 5098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017006261604581724,
+      "loss": 3.2171,
+      "step": 5099
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016988638921919147,
+      "loss": 3.2043,
+      "step": 5100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016971023505873707,
+      "loss": 3.1246,
+      "step": 5101
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001695341536032297,
+      "loss": 3.2132,
+      "step": 5102
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016935814489142935,
+      "loss": 3.0695,
+      "step": 5103
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016918220896208004,
+      "loss": 3.1494,
+      "step": 5104
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016900634585390967,
+      "loss": 3.2162,
+      "step": 5105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016883055560562978,
+      "loss": 3.0525,
+      "step": 5106
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001686548382559364,
+      "loss": 3.0624,
+      "step": 5107
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000168479193843509,
+      "loss": 3.1044,
+      "step": 5108
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016830362240701164,
+      "loss": 3.066,
+      "step": 5109
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016812812398509154,
+      "loss": 3.0577,
+      "step": 5110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001679526986163804,
+      "loss": 3.0099,
+      "step": 5111
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016777734633949393,
+      "loss": 3.1244,
+      "step": 5112
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016760206719303107,
+      "loss": 3.3101,
+      "step": 5113
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016742686121557542,
+      "loss": 3.0983,
+      "step": 5114
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016725172844569392,
+      "loss": 3.1898,
+      "step": 5115
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001670766689219379,
+      "loss": 3.192,
+      "step": 5116
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016690168268284196,
+      "loss": 3.2193,
+      "step": 5117
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016672676976692524,
+      "loss": 3.1499,
+      "step": 5118
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016655193021269005,
+      "loss": 3.0394,
+      "step": 5119
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016637716405862307,
+      "loss": 3.2218,
+      "step": 5120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001662024713431949,
+      "loss": 3.0087,
+      "step": 5121
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016602785210485926,
+      "loss": 3.0303,
+      "step": 5122
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016585330638205453,
+      "loss": 3.2245,
+      "step": 5123
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016567883421320228,
+      "loss": 3.2196,
+      "step": 5124
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016550443563670848,
+      "loss": 3.2211,
+      "step": 5125
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016533011069096217,
+      "loss": 3.1984,
+      "step": 5126
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016515585941433692,
+      "loss": 3.2623,
+      "step": 5127
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001649816818451894,
+      "loss": 3.1305,
+      "step": 5128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001648075780218607,
+      "loss": 3.1474,
+      "step": 5129
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016463354798267537,
+      "loss": 3.1244,
+      "step": 5130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016445959176594155,
+      "loss": 3.1782,
+      "step": 5131
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016428570940995154,
+      "loss": 3.2158,
+      "step": 5132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016411190095298077,
+      "loss": 3.0847,
+      "step": 5133
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016393816643328924,
+      "loss": 3.1619,
+      "step": 5134
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016376450588911985,
+      "loss": 3.1613,
+      "step": 5135
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016359091935869984,
+      "loss": 3.2884,
+      "step": 5136
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016341740688023966,
+      "loss": 3.1123,
+      "step": 5137
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016324396849193374,
+      "loss": 3.1471,
+      "step": 5138
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016307060423196047,
+      "loss": 3.1733,
+      "step": 5139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001628973141384812,
+      "loss": 3.1481,
+      "step": 5140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016272409824964168,
+      "loss": 3.1135,
+      "step": 5141
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016255095660357076,
+      "loss": 3.1053,
+      "step": 5142
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016237788923838148,
+      "loss": 3.3294,
+      "step": 5143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001622048961921699,
+      "loss": 3.1823,
+      "step": 5144
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001620319775030164,
+      "loss": 3.2064,
+      "step": 5145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016185913320898438,
+      "loss": 3.1228,
+      "step": 5146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016168636334812125,
+      "loss": 3.2693,
+      "step": 5147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016151366795845817,
+      "loss": 3.1474,
+      "step": 5148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016134104707800928,
+      "loss": 3.2661,
+      "step": 5149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000161168500744773,
+      "loss": 3.0609,
+      "step": 5150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016099602899673083,
+      "loss": 3.0265,
+      "step": 5151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001608236318718483,
+      "loss": 3.1289,
+      "step": 5152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000160651309408074,
+      "loss": 3.0658,
+      "step": 5153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016047906164334065,
+      "loss": 3.2073,
+      "step": 5154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016030688861556392,
+      "loss": 3.0244,
+      "step": 5155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016013479036264356,
+      "loss": 3.2324,
+      "step": 5156
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001599627669224628,
+      "loss": 3.1937,
+      "step": 5157
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015979081833288796,
+      "loss": 3.1276,
+      "step": 5158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015961894463176941,
+      "loss": 3.1273,
+      "step": 5159
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001594471458569406,
+      "loss": 3.108,
+      "step": 5160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015927542204621886,
+      "loss": 3.1275,
+      "step": 5161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015910377323740473,
+      "loss": 3.1144,
+      "step": 5162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001589321994682823,
+      "loss": 3.2121,
+      "step": 5163
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015876070077661952,
+      "loss": 3.17,
+      "step": 5164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015858927720016709,
+      "loss": 3.1923,
+      "step": 5165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015841792877665994,
+      "loss": 3.182,
+      "step": 5166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015824665554381578,
+      "loss": 2.9975,
+      "step": 5167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001580754575393364,
+      "loss": 3.1815,
+      "step": 5168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015790433480090632,
+      "loss": 3.1293,
+      "step": 5169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001577332873661943,
+      "loss": 3.1856,
+      "step": 5170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001575623152728518,
+      "loss": 3.1244,
+      "step": 5171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015739141855851413,
+      "loss": 3.1271,
+      "step": 5172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015722059726080002,
+      "loss": 3.2803,
+      "step": 5173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015704985141731115,
+      "loss": 3.0374,
+      "step": 5174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015687918106563326,
+      "loss": 3.2125,
+      "step": 5175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015670858624333478,
+      "loss": 3.234,
+      "step": 5176
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001565380669879682,
+      "loss": 3.1614,
+      "step": 5177
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015636762333706868,
+      "loss": 3.1564,
+      "step": 5178
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015619725532815533,
+      "loss": 3.1641,
+      "step": 5179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015602696299873016,
+      "loss": 3.0884,
+      "step": 5180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015585674638627883,
+      "loss": 3.1131,
+      "step": 5181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015568660552827036,
+      "loss": 3.1779,
+      "step": 5182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015551654046215668,
+      "loss": 3.0806,
+      "step": 5183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015534655122537367,
+      "loss": 3.0105,
+      "step": 5184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015517663785533982,
+      "loss": 3.0174,
+      "step": 5185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015500680038945752,
+      "loss": 3.0993,
+      "step": 5186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001548370388651119,
+      "loss": 3.3187,
+      "step": 5187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015466735331967206,
+      "loss": 3.094,
+      "step": 5188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001544977437904897,
+      "loss": 3.1944,
+      "step": 5189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001543282103149001,
+      "loss": 3.2544,
+      "step": 5190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001541587529302218,
+      "loss": 3.1072,
+      "step": 5191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015398937167375676,
+      "loss": 3.1428,
+      "step": 5192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015382006658278986,
+      "loss": 3.0977,
+      "step": 5193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001536508376945891,
+      "loss": 3.0319,
+      "step": 5194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001534816850464063,
+      "loss": 3.1062,
+      "step": 5195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001533126086754758,
+      "loss": 3.1383,
+      "step": 5196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001531436086190159,
+      "loss": 3.2505,
+      "step": 5197
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015297468491422722,
+      "loss": 3.1718,
+      "step": 5198
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015280583759829435,
+      "loss": 3.265,
+      "step": 5199
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015263706670838467,
+      "loss": 3.1504,
+      "step": 5200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015246837228164906,
+      "loss": 3.1196,
+      "step": 5201
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015229975435522108,
+      "loss": 3.0123,
+      "step": 5202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015213121296621767,
+      "loss": 3.0216,
+      "step": 5203
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015196274815173916,
+      "loss": 3.1087,
+      "step": 5204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015179435994886858,
+      "loss": 3.2677,
+      "step": 5205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015162604839467266,
+      "loss": 3.2803,
+      "step": 5206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015145781352620053,
+      "loss": 3.0644,
+      "step": 5207
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015128965538048505,
+      "loss": 3.3286,
+      "step": 5208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015112157399454207,
+      "loss": 3.2309,
+      "step": 5209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015095356940537054,
+      "loss": 3.03,
+      "step": 5210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015078564164995223,
+      "loss": 3.0838,
+      "step": 5211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001506177907652521,
+      "loss": 3.1295,
+      "step": 5212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015045001678821863,
+      "loss": 3.182,
+      "step": 5213
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001502823197557826,
+      "loss": 3.2146,
+      "step": 5214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015011469970485853,
+      "loss": 3.2906,
+      "step": 5215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014994715667234366,
+      "loss": 3.1762,
+      "step": 5216
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001497796906951186,
+      "loss": 3.1908,
+      "step": 5217
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001496123018100463,
+      "loss": 3.0245,
+      "step": 5218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014944499005397371,
+      "loss": 3.162,
+      "step": 5219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014927775546372994,
+      "loss": 3.0931,
+      "step": 5220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014911059807612743,
+      "loss": 3.0925,
+      "step": 5221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001489435179279619,
+      "loss": 3.0957,
+      "step": 5222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014877651505601158,
+      "loss": 3.0923,
+      "step": 5223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000148609589497038,
+      "loss": 3.1732,
+      "step": 5224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014844274128778562,
+      "loss": 3.0558,
+      "step": 5225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001482759704649821,
+      "loss": 3.292,
+      "step": 5226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001481092770653374,
+      "loss": 3.1974,
+      "step": 5227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001479426611255453,
+      "loss": 3.253,
+      "step": 5228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014777612268228175,
+      "loss": 3.1121,
+      "step": 5229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014760966177220598,
+      "loss": 3.0561,
+      "step": 5230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014744327843196043,
+      "loss": 3.1365,
+      "step": 5231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014727697269816987,
+      "loss": 3.2947,
+      "step": 5232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014711074460744246,
+      "loss": 3.1352,
+      "step": 5233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014694459419636906,
+      "loss": 3.3585,
+      "step": 5234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014677852150152376,
+      "loss": 3.2597,
+      "step": 5235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014661252655946285,
+      "loss": 3.0481,
+      "step": 5236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014644660940672628,
+      "loss": 3.0324,
+      "step": 5237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014628077007983637,
+      "loss": 3.287,
+      "step": 5238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014611500861529825,
+      "loss": 3.145,
+      "step": 5239
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014594932504960053,
+      "loss": 3.1928,
+      "step": 5240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014578371941921381,
+      "loss": 3.1074,
+      "step": 5241
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014561819176059227,
+      "loss": 3.1368,
+      "step": 5242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014545274211017262,
+      "loss": 3.1952,
+      "step": 5243
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014528737050437452,
+      "loss": 2.9415,
+      "step": 5244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014512207697960005,
+      "loss": 3.1715,
+      "step": 5245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014495686157223477,
+      "loss": 3.1022,
+      "step": 5246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014479172431864645,
+      "loss": 3.1012,
+      "step": 5247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014462666525518576,
+      "loss": 3.1904,
+      "step": 5248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014446168441818665,
+      "loss": 3.1091,
+      "step": 5249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014429678184396506,
+      "loss": 3.257,
+      "step": 5250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014413195756882036,
+      "loss": 3.1502,
+      "step": 5251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014396721162903442,
+      "loss": 3.169,
+      "step": 5252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014380254406087207,
+      "loss": 3.1429,
+      "step": 5253
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001436379549005804,
+      "loss": 3.1733,
+      "step": 5254
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001434734441843899,
+      "loss": 3.2021,
+      "step": 5255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001433090119485132,
+      "loss": 3.1254,
+      "step": 5256
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014314465822914585,
+      "loss": 3.1646,
+      "step": 5257
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014298038306246642,
+      "loss": 3.0537,
+      "step": 5258
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014281618648463563,
+      "loss": 3.2739,
+      "step": 5259
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014265206853179735,
+      "loss": 3.1048,
+      "step": 5260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014248802924007802,
+      "loss": 3.1322,
+      "step": 5261
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014232406864558695,
+      "loss": 3.2404,
+      "step": 5262
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014216018678441555,
+      "loss": 3.1883,
+      "step": 5263
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014199638369263856,
+      "loss": 3.0812,
+      "step": 5264
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014183265940631302,
+      "loss": 3.0961,
+      "step": 5265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014166901396147851,
+      "loss": 3.1166,
+      "step": 5266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014150544739415755,
+      "loss": 3.1536,
+      "step": 5267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014134195974035524,
+      "loss": 3.0609,
+      "step": 5268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014117855103605936,
+      "loss": 3.0596,
+      "step": 5269
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014101522131723993,
+      "loss": 3.2519,
+      "step": 5270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014085197061985022,
+      "loss": 3.3013,
+      "step": 5271
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014068879897982534,
+      "loss": 3.1412,
+      "step": 5272
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014052570643308376,
+      "loss": 3.1466,
+      "step": 5273
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014036269301552606,
+      "loss": 3.2268,
+      "step": 5274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014019975876303532,
+      "loss": 3.2264,
+      "step": 5275
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014003690371147754,
+      "loss": 3.1332,
+      "step": 5276
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013987412789670124,
+      "loss": 3.1265,
+      "step": 5277
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013971143135453745,
+      "loss": 3.1487,
+      "step": 5278
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013954881412079945,
+      "loss": 3.0509,
+      "step": 5279
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013938627623128364,
+      "loss": 3.1947,
+      "step": 5280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001392238177217683,
+      "loss": 3.2064,
+      "step": 5281
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013906143862801485,
+      "loss": 3.1461,
+      "step": 5282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013889913898576677,
+      "loss": 3.1508,
+      "step": 5283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013873691883075018,
+      "loss": 2.9904,
+      "step": 5284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001385747781986738,
+      "loss": 3.0992,
+      "step": 5285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013841271712522886,
+      "loss": 3.2263,
+      "step": 5286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013825073564608907,
+      "loss": 3.1719,
+      "step": 5287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001380888337969104,
+      "loss": 3.0805,
+      "step": 5288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013792701161333164,
+      "loss": 3.124,
+      "step": 5289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001377652691309736,
+      "loss": 3.1584,
+      "step": 5290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013760360638544013,
+      "loss": 3.0419,
+      "step": 5291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013744202341231675,
+      "loss": 3.0817,
+      "step": 5292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013728052024717237,
+      "loss": 3.176,
+      "step": 5293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001371190969255574,
+      "loss": 3.2068,
+      "step": 5294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013695775348300532,
+      "loss": 3.1377,
+      "step": 5295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001367964899550319,
+      "loss": 3.1348,
+      "step": 5296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000136635306377135,
+      "loss": 3.2,
+      "step": 5297
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001364742027847954,
+      "loss": 3.1097,
+      "step": 5298
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013631317921347562,
+      "loss": 3.1914,
+      "step": 5299
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013615223569862135,
+      "loss": 3.2135,
+      "step": 5300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013599137227565984,
+      "loss": 3.1536,
+      "step": 5301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013583058898000139,
+      "loss": 3.0547,
+      "step": 5302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013566988584703816,
+      "loss": 3.1536,
+      "step": 5303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013550926291214498,
+      "loss": 3.1743,
+      "step": 5304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013534872021067913,
+      "loss": 3.1049,
+      "step": 5305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013518825777797971,
+      "loss": 3.0741,
+      "step": 5306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013502787564936874,
+      "loss": 3.2195,
+      "step": 5307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013486757386015004,
+      "loss": 3.0371,
+      "step": 5308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013470735244561026,
+      "loss": 3.129,
+      "step": 5309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013454721144101782,
+      "loss": 3.0907,
+      "step": 5310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000134387150881624,
+      "loss": 3.2764,
+      "step": 5311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013422717080266178,
+      "loss": 3.1031,
+      "step": 5312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013406727123934693,
+      "loss": 3.288,
+      "step": 5313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001339074522268774,
+      "loss": 3.2172,
+      "step": 5314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013374771380043305,
+      "loss": 3.1777,
+      "step": 5315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013358805599517654,
+      "loss": 3.1742,
+      "step": 5316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013342847884625214,
+      "loss": 2.9946,
+      "step": 5317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013326898238878715,
+      "loss": 3.1068,
+      "step": 5318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013310956665789025,
+      "loss": 3.1322,
+      "step": 5319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013295023168865305,
+      "loss": 3.0843,
+      "step": 5320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013279097751614928,
+      "loss": 3.0899,
+      "step": 5321
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001326318041754343,
+      "loss": 3.1408,
+      "step": 5322
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013247271170154645,
+      "loss": 3.166,
+      "step": 5323
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013231370012950567,
+      "loss": 3.1185,
+      "step": 5324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001321547694943146,
+      "loss": 3.1159,
+      "step": 5325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013199591983095748,
+      "loss": 3.2484,
+      "step": 5326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013183715117440144,
+      "loss": 3.2348,
+      "step": 5327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013167846355959502,
+      "loss": 3.1438,
+      "step": 5328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013151985702146942,
+      "loss": 3.1333,
+      "step": 5329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000131361331594938,
+      "loss": 3.0621,
+      "step": 5330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013120288731489592,
+      "loss": 3.2499,
+      "step": 5331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013104452421622092,
+      "loss": 3.0724,
+      "step": 5332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013088624233377239,
+      "loss": 3.0089,
+      "step": 5333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013072804170239228,
+      "loss": 3.1741,
+      "step": 5334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001305699223569043,
+      "loss": 3.0532,
+      "step": 5335
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001304118843321146,
+      "loss": 3.2063,
+      "step": 5336
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013025392766281103,
+      "loss": 3.1054,
+      "step": 5337
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013009605238376387,
+      "loss": 3.1746,
+      "step": 5338
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012993825852972558,
+      "loss": 3.2553,
+      "step": 5339
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012978054613543017,
+      "loss": 3.2672,
+      "step": 5340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012962291523559426,
+      "loss": 3.104,
+      "step": 5341
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012946536586491618,
+      "loss": 3.1966,
+      "step": 5342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012930789805807658,
+      "loss": 3.1915,
+      "step": 5343
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001291505118497378,
+      "loss": 3.1312,
+      "step": 5344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001289932072745447,
+      "loss": 3.1569,
+      "step": 5345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012883598436712367,
+      "loss": 3.0799,
+      "step": 5346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012867884316208344,
+      "loss": 3.0748,
+      "step": 5347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012852178369401486,
+      "loss": 3.1858,
+      "step": 5348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012836480599749027,
+      "loss": 3.1225,
+      "step": 5349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012820791010706472,
+      "loss": 3.0755,
+      "step": 5350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001280510960572745,
+      "loss": 3.0321,
+      "step": 5351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001278943638826386,
+      "loss": 3.1258,
+      "step": 5352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012773771361765734,
+      "loss": 3.0121,
+      "step": 5353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012758114529681357,
+      "loss": 3.2094,
+      "step": 5354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012742465895457166,
+      "loss": 3.1055,
+      "step": 5355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012726825462537822,
+      "loss": 3.2514,
+      "step": 5356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001271119323436618,
+      "loss": 3.0504,
+      "step": 5357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012695569214383256,
+      "loss": 3.0434,
+      "step": 5358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012679953406028317,
+      "loss": 3.1434,
+      "step": 5359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012664345812738752,
+      "loss": 3.1578,
+      "step": 5360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012648746437950208,
+      "loss": 3.0658,
+      "step": 5361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001263315528509647,
+      "loss": 3.0547,
+      "step": 5362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012617572357609562,
+      "loss": 3.1036,
+      "step": 5363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012601997658919645,
+      "loss": 3.1853,
+      "step": 5364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000125864311924551,
+      "loss": 3.1244,
+      "step": 5365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001257087296164252,
+      "loss": 3.211,
+      "step": 5366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012555322969906617,
+      "loss": 2.9349,
+      "step": 5367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012539781220670366,
+      "loss": 3.3625,
+      "step": 5368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012524247717354852,
+      "loss": 3.1634,
+      "step": 5369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012508722463379425,
+      "loss": 3.0458,
+      "step": 5370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012493205462161533,
+      "loss": 3.073,
+      "step": 5371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012477696717116875,
+      "loss": 3.1837,
+      "step": 5372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001246219623165931,
+      "loss": 3.2856,
+      "step": 5373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012446704009200894,
+      "loss": 3.2806,
+      "step": 5374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012431220053151832,
+      "loss": 3.0416,
+      "step": 5375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012415744366920506,
+      "loss": 3.0895,
+      "step": 5376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001240027695391353,
+      "loss": 3.1812,
+      "step": 5377
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012384817817535638,
+      "loss": 3.1306,
+      "step": 5378
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000123693669611898,
+      "loss": 2.9844,
+      "step": 5379
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012353924388277092,
+      "loss": 3.1007,
+      "step": 5380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012338490102196825,
+      "loss": 3.0952,
+      "step": 5381
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012323064106346476,
+      "loss": 3.2096,
+      "step": 5382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012307646404121692,
+      "loss": 3.05,
+      "step": 5383
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012292236998916263,
+      "loss": 3.1101,
+      "step": 5384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012276835894122214,
+      "loss": 3.0353,
+      "step": 5385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012261443093129691,
+      "loss": 3.0664,
+      "step": 5386
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001224605859932702,
+      "loss": 3.0404,
+      "step": 5387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001224605859932702,
+      "loss": 3.0942,
+      "step": 5388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001223068241610073,
+      "loss": 3.2222,
+      "step": 5389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012215314546835482,
+      "loss": 3.1991,
+      "step": 5390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012199954994914125,
+      "loss": 3.2577,
+      "step": 5391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012184603763717683,
+      "loss": 3.1187,
+      "step": 5392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012169260856625358,
+      "loss": 3.0335,
+      "step": 5393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001215392627701447,
+      "loss": 3.0217,
+      "step": 5394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012138600028260577,
+      "loss": 3.2296,
+      "step": 5395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012123282113737338,
+      "loss": 3.0723,
+      "step": 5396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012107972536816597,
+      "loss": 3.1711,
+      "step": 5397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012092671300868402,
+      "loss": 3.0709,
+      "step": 5398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012077378409260903,
+      "loss": 3.1338,
+      "step": 5399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012062093865360457,
+      "loss": 3.1171,
+      "step": 5400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012046817672531568,
+      "loss": 3.1967,
+      "step": 5401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012031549834136923,
+      "loss": 3.0787,
+      "step": 5402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012016290353537318,
+      "loss": 2.9692,
+      "step": 5403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001200103923409177,
+      "loss": 3.1746,
+      "step": 5404
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011985796479157424,
+      "loss": 3.1496,
+      "step": 5405
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001197056209208956,
+      "loss": 3.1575,
+      "step": 5406
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011955336076241679,
+      "loss": 3.1932,
+      "step": 5407
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001194011843496537,
+      "loss": 3.0708,
+      "step": 5408
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011924909171610433,
+      "loss": 3.0858,
+      "step": 5409
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011909708289524801,
+      "loss": 3.1443,
+      "step": 5410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011894515792054572,
+      "loss": 3.0889,
+      "step": 5411
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001187933168254397,
+      "loss": 3.1304,
+      "step": 5412
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011864155964335422,
+      "loss": 3.2459,
+      "step": 5413
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001184898864076946,
+      "loss": 3.1676,
+      "step": 5414
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011833829715184774,
+      "loss": 3.1648,
+      "step": 5415
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011818679190918247,
+      "loss": 3.0778,
+      "step": 5416
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011803537071304855,
+      "loss": 3.0615,
+      "step": 5417
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011788403359677769,
+      "loss": 3.3043,
+      "step": 5418
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011773278059368292,
+      "loss": 3.3182,
+      "step": 5419
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011758161173705894,
+      "loss": 3.0587,
+      "step": 5420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011743052706018142,
+      "loss": 3.1818,
+      "step": 5421
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001172795265963082,
+      "loss": 3.0737,
+      "step": 5422
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011712861037867795,
+      "loss": 3.0625,
+      "step": 5423
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 3.1656,
+      "step": 5424
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001168270308150094,
+      "loss": 3.0614,
+      "step": 5425
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011667636753535632,
+      "loss": 3.2706,
+      "step": 5426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011652578863471664,
+      "loss": 3.0948,
+      "step": 5427
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011637529414623621,
+      "loss": 3.1239,
+      "step": 5428
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011622488410304288,
+      "loss": 3.0686,
+      "step": 5429
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011607455853824539,
+      "loss": 3.053,
+      "step": 5430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011592431748493432,
+      "loss": 3.1994,
+      "step": 5431
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011577416097618138,
+      "loss": 3.2508,
+      "step": 5432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011562408904503952,
+      "loss": 3.2567,
+      "step": 5433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011547410172454342,
+      "loss": 3.2176,
+      "step": 5434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011532419904770908,
+      "loss": 3.2951,
+      "step": 5435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011517438104753386,
+      "loss": 3.0762,
+      "step": 5436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011502464775699617,
+      "loss": 3.2401,
+      "step": 5437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011487499920905625,
+      "loss": 3.0554,
+      "step": 5438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001147254354366552,
+      "loss": 3.1294,
+      "step": 5439
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011457595647271602,
+      "loss": 3.0859,
+      "step": 5440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011442656235014248,
+      "loss": 3.1748,
+      "step": 5441
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011427725310181981,
+      "loss": 3.0671,
+      "step": 5442
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001141280287606149,
+      "loss": 3.0693,
+      "step": 5443
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011397888935937561,
+      "loss": 3.2975,
+      "step": 5444
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011382983493093146,
+      "loss": 3.0962,
+      "step": 5445
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011368086550809264,
+      "loss": 3.0998,
+      "step": 5446
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011353198112365132,
+      "loss": 3.2048,
+      "step": 5447
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011338318181038037,
+      "loss": 3.1093,
+      "step": 5448
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011323446760103445,
+      "loss": 2.993,
+      "step": 5449
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011308583852834914,
+      "loss": 3.2219,
+      "step": 5450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011293729462504115,
+      "loss": 3.0682,
+      "step": 5451
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001127888359238089,
+      "loss": 3.1374,
+      "step": 5452
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011264046245733178,
+      "loss": 3.1466,
+      "step": 5453
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011249217425827063,
+      "loss": 3.083,
+      "step": 5454
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011234397135926705,
+      "loss": 3.1164,
+      "step": 5455
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011219585379294444,
+      "loss": 3.1787,
+      "step": 5456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011204782159190685,
+      "loss": 3.2134,
+      "step": 5457
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011189987478874014,
+      "loss": 3.0724,
+      "step": 5458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011175201341601088,
+      "loss": 3.1622,
+      "step": 5459
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011160423750626692,
+      "loss": 3.2536,
+      "step": 5460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011145654709203746,
+      "loss": 3.1974,
+      "step": 5461
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011130894220583293,
+      "loss": 3.0959,
+      "step": 5462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011116142288014486,
+      "loss": 3.1825,
+      "step": 5463
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011101398914744565,
+      "loss": 3.0801,
+      "step": 5464
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001108666410401894,
+      "loss": 3.0428,
+      "step": 5465
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011071937859081077,
+      "loss": 3.0085,
+      "step": 5466
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011057220183172623,
+      "loss": 2.9627,
+      "step": 5467
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011042511079533273,
+      "loss": 3.2802,
+      "step": 5468
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011027810551400868,
+      "loss": 3.1886,
+      "step": 5469
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011013118602011357,
+      "loss": 3.0375,
+      "step": 5470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010998435234598814,
+      "loss": 3.2455,
+      "step": 5471
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010983760452395414,
+      "loss": 3.0744,
+      "step": 5472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010969094258631418,
+      "loss": 3.283,
+      "step": 5473
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010954436656535249,
+      "loss": 3.1899,
+      "step": 5474
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010939787649333372,
+      "loss": 3.2315,
+      "step": 5475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010925147240250427,
+      "loss": 3.1046,
+      "step": 5476
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010910515432509104,
+      "loss": 3.1273,
+      "step": 5477
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001089589222933024,
+      "loss": 3.1227,
+      "step": 5478
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010881277633932779,
+      "loss": 3.1894,
+      "step": 5479
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010866671649533722,
+      "loss": 3.1886,
+      "step": 5480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010852074279348235,
+      "loss": 3.1275,
+      "step": 5481
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010837485526589535,
+      "loss": 3.0769,
+      "step": 5482
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010822905394468995,
+      "loss": 3.0361,
+      "step": 5483
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010808333886196037,
+      "loss": 3.137,
+      "step": 5484
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010793771004978236,
+      "loss": 3.1749,
+      "step": 5485
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010779216754021215,
+      "loss": 3.1162,
+      "step": 5486
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010764671136528741,
+      "loss": 3.1732,
+      "step": 5487
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010750134155702673,
+      "loss": 3.1083,
+      "step": 5488
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010735605814742933,
+      "loss": 3.1178,
+      "step": 5489
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000107210861168476,
+      "loss": 3.1654,
+      "step": 5490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010706575065212792,
+      "loss": 3.0944,
+      "step": 5491
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001069207266303277,
+      "loss": 3.1974,
+      "step": 5492
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010677578913499852,
+      "loss": 3.0679,
+      "step": 5493
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010663093819804493,
+      "loss": 3.1568,
+      "step": 5494
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010648617385135196,
+      "loss": 3.1797,
+      "step": 5495
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001063414961267859,
+      "loss": 3.0465,
+      "step": 5496
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010619690505619417,
+      "loss": 3.1455,
+      "step": 5497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010605240067140443,
+      "loss": 3.1708,
+      "step": 5498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010590798300422599,
+      "loss": 3.1423,
+      "step": 5499
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001057636520864485,
+      "loss": 3.0129,
+      "step": 5500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010561940794984298,
+      "loss": 3.0769,
+      "step": 5501
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010547525062616093,
+      "loss": 3.2102,
+      "step": 5502
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010533118014713522,
+      "loss": 3.1989,
+      "step": 5503
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010518719654447894,
+      "loss": 3.2395,
+      "step": 5504
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010504329984988676,
+      "loss": 3.0653,
+      "step": 5505
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010489949009503385,
+      "loss": 3.1303,
+      "step": 5506
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001047557673115761,
+      "loss": 3.1334,
+      "step": 5507
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010461213153115079,
+      "loss": 3.02,
+      "step": 5508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010446858278537535,
+      "loss": 3.1165,
+      "step": 5509
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010432512110584869,
+      "loss": 2.9439,
+      "step": 5510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010418174652415002,
+      "loss": 3.1927,
+      "step": 5511
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010403845907183989,
+      "loss": 3.1363,
+      "step": 5512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010389525878045919,
+      "loss": 3.0676,
+      "step": 5513
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010375214568152997,
+      "loss": 3.1188,
+      "step": 5514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010360911980655502,
+      "loss": 3.055,
+      "step": 5515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010346618118701768,
+      "loss": 3.0288,
+      "step": 5516
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010332332985438247,
+      "loss": 3.2292,
+      "step": 5517
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010318056584009433,
+      "loss": 3.0528,
+      "step": 5518
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010303788917557938,
+      "loss": 3.0817,
+      "step": 5519
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000102895299892244,
+      "loss": 3.1629,
+      "step": 5520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010275279802147591,
+      "loss": 3.1764,
+      "step": 5521
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010261038359464304,
+      "loss": 3.17,
+      "step": 5522
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010246805664309439,
+      "loss": 3.0638,
+      "step": 5523
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010232581719815982,
+      "loss": 3.0679,
+      "step": 5524
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010218366529114948,
+      "loss": 3.1037,
+      "step": 5525
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001020416009533548,
+      "loss": 3.0619,
+      "step": 5526
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010189962421604738,
+      "loss": 3.1645,
+      "step": 5527
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010175773511048009,
+      "loss": 3.0997,
+      "step": 5528
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010161593366788596,
+      "loss": 3.2458,
+      "step": 5529
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010147421991947914,
+      "loss": 3.2387,
+      "step": 5530
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010133259389645427,
+      "loss": 3.2158,
+      "step": 5531
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010119105562998699,
+      "loss": 3.0364,
+      "step": 5532
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010104960515123307,
+      "loss": 3.1223,
+      "step": 5533
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010090824249132929,
+      "loss": 3.3009,
+      "step": 5534
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010076696768139326,
+      "loss": 3.0936,
+      "step": 5535
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010062578075252282,
+      "loss": 3.1831,
+      "step": 5536
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010048468173579695,
+      "loss": 3.1967,
+      "step": 5537
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010034367066227474,
+      "loss": 3.0485,
+      "step": 5538
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010020274756299641,
+      "loss": 3.2266,
+      "step": 5539
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010006191246898255,
+      "loss": 3.1436,
+      "step": 5540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.992116541123464e-05,
+      "loss": 3.2538,
+      "step": 5541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.97805064207345e-05,
+      "loss": 3.0057,
+      "step": 5542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.963993552844436e-05,
+      "loss": 3.0449,
+      "step": 5543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.949945276530781e-05,
+      "loss": 3.1463,
+      "step": 5544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.935905816224817e-05,
+      "loss": 3.1752,
+      "step": 5545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.921875175017003e-05,
+      "loss": 3.0902,
+      "step": 5546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.907853355995817e-05,
+      "loss": 3.1715,
+      "step": 5547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.893840362247809e-05,
+      "loss": 3.0875,
+      "step": 5548
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.879836196857595e-05,
+      "loss": 3.0317,
+      "step": 5549
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.865840862907837e-05,
+      "loss": 3.1437,
+      "step": 5550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.851854363479257e-05,
+      "loss": 3.1242,
+      "step": 5551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.837876701650605e-05,
+      "loss": 3.0736,
+      "step": 5552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.823907880498744e-05,
+      "loss": 3.2624,
+      "step": 5553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.809947903098526e-05,
+      "loss": 3.1708,
+      "step": 5554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.795996772522919e-05,
+      "loss": 3.0965,
+      "step": 5555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.782054491842879e-05,
+      "loss": 2.9626,
+      "step": 5556
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.768121064127461e-05,
+      "loss": 3.0937,
+      "step": 5557
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.754196492443763e-05,
+      "loss": 3.1123,
+      "step": 5558
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.740280779856936e-05,
+      "loss": 3.1006,
+      "step": 5559
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.726373929430155e-05,
+      "loss": 3.1795,
+      "step": 5560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.712475944224658e-05,
+      "loss": 3.1755,
+      "step": 5561
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.69858682729976e-05,
+      "loss": 3.047,
+      "step": 5562
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.684706581712766e-05,
+      "loss": 3.2352,
+      "step": 5563
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.670835210519091e-05,
+      "loss": 3.1047,
+      "step": 5564
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.656972716772144e-05,
+      "loss": 3.0115,
+      "step": 5565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.643119103523413e-05,
+      "loss": 3.0072,
+      "step": 5566
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.629274373822422e-05,
+      "loss": 3.131,
+      "step": 5567
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.615438530716752e-05,
+      "loss": 3.1,
+      "step": 5568
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.601611577251984e-05,
+      "loss": 3.0952,
+      "step": 5569
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.587793516471805e-05,
+      "loss": 3.197,
+      "step": 5570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.573984351417897e-05,
+      "loss": 3.2412,
+      "step": 5571
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.56018408512998e-05,
+      "loss": 3.1781,
+      "step": 5572
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.546392720645868e-05,
+      "loss": 3.1386,
+      "step": 5573
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.53261026100135e-05,
+      "loss": 3.1958,
+      "step": 5574
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.518836709230299e-05,
+      "loss": 3.049,
+      "step": 5575
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.505072068364623e-05,
+      "loss": 3.1813,
+      "step": 5576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.491316341434264e-05,
+      "loss": 3.0863,
+      "step": 5577
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.477569531467172e-05,
+      "loss": 3.0816,
+      "step": 5578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.46383164148939e-05,
+      "loss": 3.2418,
+      "step": 5579
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.450102674524952e-05,
+      "loss": 3.2126,
+      "step": 5580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.43638263359593e-05,
+      "loss": 2.9755,
+      "step": 5581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.422671521722459e-05,
+      "loss": 3.0607,
+      "step": 5582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.408969341922696e-05,
+      "loss": 3.0805,
+      "step": 5583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.395276097212841e-05,
+      "loss": 3.1012,
+      "step": 5584
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.381591790607081e-05,
+      "loss": 3.2185,
+      "step": 5585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.367916425117712e-05,
+      "loss": 3.2632,
+      "step": 5586
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.354250003754983e-05,
+      "loss": 3.08,
+      "step": 5587
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.340592529527236e-05,
+      "loss": 3.011,
+      "step": 5588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.326944005440819e-05,
+      "loss": 3.0774,
+      "step": 5589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.31330443450008e-05,
+      "loss": 3.0464,
+      "step": 5590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.299673819707454e-05,
+      "loss": 3.1466,
+      "step": 5591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.286052164063369e-05,
+      "loss": 3.0501,
+      "step": 5592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.272439470566301e-05,
+      "loss": 3.0369,
+      "step": 5593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.258835742212718e-05,
+      "loss": 3.0716,
+      "step": 5594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.245240981997155e-05,
+      "loss": 3.1176,
+      "step": 5595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231655192912136e-05,
+      "loss": 3.1047,
+      "step": 5596
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.218078377948257e-05,
+      "loss": 3.1662,
+      "step": 5597
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204510540094096e-05,
+      "loss": 3.0296,
+      "step": 5598
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190951682336251e-05,
+      "loss": 3.1674,
+      "step": 5599
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177401807659386e-05,
+      "loss": 3.1993,
+      "step": 5600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163860919046152e-05,
+      "loss": 3.0883,
+      "step": 5601
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150329019477254e-05,
+      "loss": 3.1375,
+      "step": 5602
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.136806111931367e-05,
+      "loss": 3.0646,
+      "step": 5603
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.123292199385246e-05,
+      "loss": 3.1663,
+      "step": 5604
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.109787284813615e-05,
+      "loss": 3.2081,
+      "step": 5605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.096291371189258e-05,
+      "loss": 3.3023,
+      "step": 5606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.082804461482952e-05,
+      "loss": 3.2478,
+      "step": 5607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.069326558663487e-05,
+      "loss": 3.1789,
+      "step": 5608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.055857665697693e-05,
+      "loss": 3.1448,
+      "step": 5609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.042397785550405e-05,
+      "loss": 3.1976,
+      "step": 5610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.028946921184494e-05,
+      "loss": 3.0925,
+      "step": 5611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.015505075560798e-05,
+      "loss": 3.1302,
+      "step": 5612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.002072251638221e-05,
+      "loss": 3.2994,
+      "step": 5613
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.98864845237365e-05,
+      "loss": 3.2657,
+      "step": 5614
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.975233680722e-05,
+      "loss": 3.0787,
+      "step": 5615
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.961827939636197e-05,
+      "loss": 3.2262,
+      "step": 5616
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.948431232067156e-05,
+      "loss": 3.1494,
+      "step": 5617
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.93504356096384e-05,
+      "loss": 3.2401,
+      "step": 5618
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.921664929273204e-05,
+      "loss": 3.0623,
+      "step": 5619
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.908295339940231e-05,
+      "loss": 3.1681,
+      "step": 5620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.894934795907872e-05,
+      "loss": 3.1344,
+      "step": 5621
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.881583300117136e-05,
+      "loss": 3.1869,
+      "step": 5622
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.868240855506993e-05,
+      "loss": 3.0666,
+      "step": 5623
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.854907465014478e-05,
+      "loss": 3.1513,
+      "step": 5624
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.841583131574576e-05,
+      "loss": 3.2698,
+      "step": 5625
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.828267858120293e-05,
+      "loss": 3.1768,
+      "step": 5626
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.81496164758267e-05,
+      "loss": 2.9771,
+      "step": 5627
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.801664502890722e-05,
+      "loss": 3.2075,
+      "step": 5628
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.7883764269715e-05,
+      "loss": 3.0786,
+      "step": 5629
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.775097422750011e-05,
+      "loss": 3.1317,
+      "step": 5630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.761827493149321e-05,
+      "loss": 3.1709,
+      "step": 5631
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.748566641090433e-05,
+      "loss": 3.0478,
+      "step": 5632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.735314869492428e-05,
+      "loss": 3.2399,
+      "step": 5633
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.722072181272311e-05,
+      "loss": 3.0607,
+      "step": 5634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.708838579345147e-05,
+      "loss": 3.0165,
+      "step": 5635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.695614066623991e-05,
+      "loss": 3.1985,
+      "step": 5636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.682398646019857e-05,
+      "loss": 3.1545,
+      "step": 5637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.66919232044181e-05,
+      "loss": 3.0805,
+      "step": 5638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.655995092796865e-05,
+      "loss": 3.1113,
+      "step": 5639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.642806965990079e-05,
+      "loss": 3.0384,
+      "step": 5640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.629627942924473e-05,
+      "loss": 3.0908,
+      "step": 5641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.61645802650109e-05,
+      "loss": 3.1961,
+      "step": 5642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.603297219618934e-05,
+      "loss": 3.1144,
+      "step": 5643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.590145525175031e-05,
+      "loss": 3.1511,
+      "step": 5644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.577002946064416e-05,
+      "loss": 3.1327,
+      "step": 5645
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.563869485180059e-05,
+      "loss": 3.0826,
+      "step": 5646
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.550745145412997e-05,
+      "loss": 3.2003,
+      "step": 5647
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.537629929652186e-05,
+      "loss": 3.1246,
+      "step": 5648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.524523840784643e-05,
+      "loss": 3.1919,
+      "step": 5649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.511426881695311e-05,
+      "loss": 3.2014,
+      "step": 5650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.49833905526718e-05,
+      "loss": 3.2336,
+      "step": 5651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.485260364381186e-05,
+      "loss": 3.1385,
+      "step": 5652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.472190811916274e-05,
+      "loss": 3.1219,
+      "step": 5653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.459130400749404e-05,
+      "loss": 3.0361,
+      "step": 5654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.446079133755458e-05,
+      "loss": 3.1297,
+      "step": 5655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.433037013807365e-05,
+      "loss": 3.0088,
+      "step": 5656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.420004043776003e-05,
+      "loss": 3.069,
+      "step": 5657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.406980226530276e-05,
+      "loss": 3.0421,
+      "step": 5658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.393965564937017e-05,
+      "loss": 3.129,
+      "step": 5659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.380960061861098e-05,
+      "loss": 3.079,
+      "step": 5660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.367963720165334e-05,
+      "loss": 3.3014,
+      "step": 5661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.354976542710558e-05,
+      "loss": 3.1775,
+      "step": 5662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.341998532355565e-05,
+      "loss": 3.191,
+      "step": 5663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.329029691957124e-05,
+      "loss": 3.0846,
+      "step": 5664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.31607002437002e-05,
+      "loss": 3.0621,
+      "step": 5665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.303119532446973e-05,
+      "loss": 3.1856,
+      "step": 5666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.290178219038725e-05,
+      "loss": 3.0607,
+      "step": 5667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.277246086993961e-05,
+      "loss": 3.2026,
+      "step": 5668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.264323139159385e-05,
+      "loss": 3.2192,
+      "step": 5669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.251409378379638e-05,
+      "loss": 3.1192,
+      "step": 5670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.23850480749736e-05,
+      "loss": 3.0606,
+      "step": 5671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.225609429353187e-05,
+      "loss": 3.207,
+      "step": 5672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.212723246785681e-05,
+      "loss": 3.1878,
+      "step": 5673
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.199846262631438e-05,
+      "loss": 3.1628,
+      "step": 5674
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.186978479724971e-05,
+      "loss": 3.042,
+      "step": 5675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.17411990089883e-05,
+      "loss": 3.0535,
+      "step": 5676
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.161270528983477e-05,
+      "loss": 3.1311,
+      "step": 5677
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.148430366807403e-05,
+      "loss": 3.1216,
+      "step": 5678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.135599417197021e-05,
+      "loss": 3.1101,
+      "step": 5679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.122777682976745e-05,
+      "loss": 3.1245,
+      "step": 5680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.109965166968975e-05,
+      "loss": 3.1276,
+      "step": 5681
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.097161871994042e-05,
+      "loss": 3.1533,
+      "step": 5682
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.084367800870285e-05,
+      "loss": 3.1673,
+      "step": 5683
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.07158295641397e-05,
+      "loss": 3.1783,
+      "step": 5684
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.058807341439389e-05,
+      "loss": 3.1357,
+      "step": 5685
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.046040958758744e-05,
+      "loss": 3.2349,
+      "step": 5686
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.033283811182246e-05,
+      "loss": 3.1471,
+      "step": 5687
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.02053590151805e-05,
+      "loss": 3.1883,
+      "step": 5688
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.007797232572306e-05,
+      "loss": 3.0984,
+      "step": 5689
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.995067807149093e-05,
+      "loss": 3.0762,
+      "step": 5690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.982347628050468e-05,
+      "loss": 3.0534,
+      "step": 5691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.969636698076476e-05,
+      "loss": 3.1992,
+      "step": 5692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.956935020025081e-05,
+      "loss": 3.1692,
+      "step": 5693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.944242596692264e-05,
+      "loss": 3.2134,
+      "step": 5694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.93155943087192e-05,
+      "loss": 3.2619,
+      "step": 5695
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.918885525355934e-05,
+      "loss": 3.098,
+      "step": 5696
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.906220882934146e-05,
+      "loss": 3.0488,
+      "step": 5697
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.893565506394374e-05,
+      "loss": 3.2197,
+      "step": 5698
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.880919398522362e-05,
+      "loss": 3.0536,
+      "step": 5699
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.86828256210182e-05,
+      "loss": 3.1299,
+      "step": 5700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.855654999914457e-05,
+      "loss": 2.9909,
+      "step": 5701
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.843036714739887e-05,
+      "loss": 3.1417,
+      "step": 5702
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.830427709355725e-05,
+      "loss": 3.1359,
+      "step": 5703
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.817827986537507e-05,
+      "loss": 3.1367,
+      "step": 5704
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.805237549058752e-05,
+      "loss": 3.122,
+      "step": 5705
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.792656399690923e-05,
+      "loss": 3.246,
+      "step": 5706
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.780084541203464e-05,
+      "loss": 3.2403,
+      "step": 5707
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.767521976363735e-05,
+      "loss": 3.1602,
+      "step": 5708
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.754968707937054e-05,
+      "loss": 3.2944,
+      "step": 5709
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.742424738686732e-05,
+      "loss": 3.204,
+      "step": 5710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.729890071373984e-05,
+      "loss": 3.1059,
+      "step": 5711
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.717364708758023e-05,
+      "loss": 3.0879,
+      "step": 5712
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.704848653595964e-05,
+      "loss": 3.0976,
+      "step": 5713
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.692341908642914e-05,
+      "loss": 3.0776,
+      "step": 5714
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.67984447665192e-05,
+      "loss": 3.0764,
+      "step": 5715
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.667356360373984e-05,
+      "loss": 3.1327,
+      "step": 5716
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.654877562558038e-05,
+      "loss": 3.1778,
+      "step": 5717
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.642408085950964e-05,
+      "loss": 3.342,
+      "step": 5718
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.629947933297626e-05,
+      "loss": 3.0878,
+      "step": 5719
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.617497107340788e-05,
+      "loss": 3.0365,
+      "step": 5720
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.60505561082121e-05,
+      "loss": 3.1462,
+      "step": 5721
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.592623446477543e-05,
+      "loss": 3.2017,
+      "step": 5722
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.580200617046434e-05,
+      "loss": 3.0974,
+      "step": 5723
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.567787125262449e-05,
+      "loss": 3.0495,
+      "step": 5724
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.555382973858116e-05,
+      "loss": 3.0326,
+      "step": 5725
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.542988165563892e-05,
+      "loss": 3.174,
+      "step": 5726
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.530602703108158e-05,
+      "loss": 3.2053,
+      "step": 5727
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.518226589217286e-05,
+      "loss": 3.0889,
+      "step": 5728
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.505859826615551e-05,
+      "loss": 3.1619,
+      "step": 5729
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.4935024180252e-05,
+      "loss": 3.1477,
+      "step": 5730
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.481154366166382e-05,
+      "loss": 2.9796,
+      "step": 5731
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.468815673757218e-05,
+      "loss": 3.2387,
+      "step": 5732
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.456486343513764e-05,
+      "loss": 3.0035,
+      "step": 5733
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.444166378150013e-05,
+      "loss": 3.064,
+      "step": 5734
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.431855780377894e-05,
+      "loss": 3.0552,
+      "step": 5735
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.419554552907259e-05,
+      "loss": 3.1756,
+      "step": 5736
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.407262698445932e-05,
+      "loss": 3.1005,
+      "step": 5737
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.394980219699632e-05,
+      "loss": 3.1331,
+      "step": 5738
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.382707119372051e-05,
+      "loss": 3.0275,
+      "step": 5739
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.370443400164794e-05,
+      "loss": 3.1569,
+      "step": 5740
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.358189064777432e-05,
+      "loss": 3.1197,
+      "step": 5741
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.345944115907421e-05,
+      "loss": 3.0455,
+      "step": 5742
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.333708556250195e-05,
+      "loss": 3.1007,
+      "step": 5743
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.321482388499096e-05,
+      "loss": 3.1533,
+      "step": 5744
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.30926561534539e-05,
+      "loss": 3.1551,
+      "step": 5745
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.297058239478316e-05,
+      "loss": 3.1592,
+      "step": 5746
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.284860263585003e-05,
+      "loss": 3.2132,
+      "step": 5747
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.272671690350529e-05,
+      "loss": 3.2199,
+      "step": 5748
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.260492522457906e-05,
+      "loss": 3.2039,
+      "step": 5749
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.248322762588084e-05,
+      "loss": 3.1203,
+      "step": 5750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.236162413419895e-05,
+      "loss": 3.0179,
+      "step": 5751
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.224011477630166e-05,
+      "loss": 3.0848,
+      "step": 5752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.211869957893591e-05,
+      "loss": 3.0705,
+      "step": 5753
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.199737856882843e-05,
+      "loss": 3.0928,
+      "step": 5754
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.187615177268486e-05,
+      "loss": 3.1805,
+      "step": 5755
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.175501921719007e-05,
+      "loss": 3.1842,
+      "step": 5756
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.163398092900852e-05,
+      "loss": 3.2316,
+      "step": 5757
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.151303693478362e-05,
+      "loss": 3.0629,
+      "step": 5758
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.139218726113838e-05,
+      "loss": 3.2306,
+      "step": 5759
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.127143193467444e-05,
+      "loss": 3.1869,
+      "step": 5760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.115077098197337e-05,
+      "loss": 3.2043,
+      "step": 5761
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.10302044295954e-05,
+      "loss": 3.1551,
+      "step": 5762
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.090973230408033e-05,
+      "loss": 3.1294,
+      "step": 5763
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.078935463194707e-05,
+      "loss": 3.1149,
+      "step": 5764
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.066907143969353e-05,
+      "loss": 3.2406,
+      "step": 5765
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.054888275379712e-05,
+      "loss": 3.0397,
+      "step": 5766
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.042878860071439e-05,
+      "loss": 3.2545,
+      "step": 5767
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.030878900688114e-05,
+      "loss": 3.0968,
+      "step": 5768
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.018888399871204e-05,
+      "loss": 3.0862,
+      "step": 5769
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.006907360260129e-05,
+      "loss": 3.0908,
+      "step": 5770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.9949357844922e-05,
+      "loss": 3.0943,
+      "step": 5771
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.982973675202675e-05,
+      "loss": 3.109,
+      "step": 5772
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.971021035024694e-05,
+      "loss": 3.1406,
+      "step": 5773
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.959077866589326e-05,
+      "loss": 3.1201,
+      "step": 5774
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.947144172525566e-05,
+      "loss": 3.1079,
+      "step": 5775
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.935219955460309e-05,
+      "loss": 3.0473,
+      "step": 5776
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.923305218018394e-05,
+      "loss": 3.0841,
+      "step": 5777
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.911399962822518e-05,
+      "loss": 3.1898,
+      "step": 5778
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.899504192493344e-05,
+      "loss": 3.1737,
+      "step": 5779
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.887617909649407e-05,
+      "loss": 3.0332,
+      "step": 5780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.875741116907191e-05,
+      "loss": 3.0799,
+      "step": 5781
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.863873816881061e-05,
+      "loss": 3.2181,
+      "step": 5782
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.852016012183299e-05,
+      "loss": 3.1488,
+      "step": 5783
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.840167705424105e-05,
+      "loss": 3.129,
+      "step": 5784
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.828328899211584e-05,
+      "loss": 3.2644,
+      "step": 5785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.81649959615176e-05,
+      "loss": 3.1889,
+      "step": 5786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.80467979884854e-05,
+      "loss": 3.1024,
+      "step": 5787
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.792869509903777e-05,
+      "loss": 3.2234,
+      "step": 5788
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.781068731917173e-05,
+      "loss": 3.0658,
+      "step": 5789
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.769277467486407e-05,
+      "loss": 3.2215,
+      "step": 5790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.757495719206996e-05,
+      "loss": 3.1552,
+      "step": 5791
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.745723489672412e-05,
+      "loss": 3.0855,
+      "step": 5792
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.733960781474019e-05,
+      "loss": 3.1169,
+      "step": 5793
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.722207597201064e-05,
+      "loss": 3.0435,
+      "step": 5794
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.71046393944073e-05,
+      "loss": 3.286,
+      "step": 5795
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.698729810778065e-05,
+      "loss": 3.081,
+      "step": 5796
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.687005213796071e-05,
+      "loss": 3.1921,
+      "step": 5797
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.675290151075591e-05,
+      "loss": 3.1434,
+      "step": 5798
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.663584625195424e-05,
+      "loss": 3.104,
+      "step": 5799
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.651888638732228e-05,
+      "loss": 3.2131,
+      "step": 5800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.640202194260586e-05,
+      "loss": 3.0813,
+      "step": 5801
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.628525294352988e-05,
+      "loss": 3.1544,
+      "step": 5802
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.616857941579779e-05,
+      "loss": 3.1577,
+      "step": 5803
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.60520013850926e-05,
+      "loss": 3.0804,
+      "step": 5804
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.593551887707577e-05,
+      "loss": 3.123,
+      "step": 5805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.581913191738826e-05,
+      "loss": 3.1361,
+      "step": 5806
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.570284053164944e-05,
+      "loss": 3.2301,
+      "step": 5807
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.558664474545817e-05,
+      "loss": 3.0878,
+      "step": 5808
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.547054458439178e-05,
+      "loss": 3.0973,
+      "step": 5809
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.535454007400688e-05,
+      "loss": 3.0193,
+      "step": 5810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.523863123983909e-05,
+      "loss": 3.0989,
+      "step": 5811
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.51228181074025e-05,
+      "loss": 3.1453,
+      "step": 5812
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.500710070219079e-05,
+      "loss": 3.1774,
+      "step": 5813
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.48914790496759e-05,
+      "loss": 3.0116,
+      "step": 5814
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.477595317530933e-05,
+      "loss": 3.2688,
+      "step": 5815
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.466052310452092e-05,
+      "loss": 3.0009,
+      "step": 5816
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.454518886271981e-05,
+      "loss": 3.2751,
+      "step": 5817
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.442995047529388e-05,
+      "loss": 3.1206,
+      "step": 5818
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.431480796760991e-05,
+      "loss": 3.1342,
+      "step": 5819
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.419976136501376e-05,
+      "loss": 3.0335,
+      "step": 5820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.40848106928299e-05,
+      "loss": 3.0244,
+      "step": 5821
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.39699559763619e-05,
+      "loss": 2.9725,
+      "step": 5822
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.3855197240892e-05,
+      "loss": 3.1518,
+      "step": 5823
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.374053451168166e-05,
+      "loss": 3.0842,
+      "step": 5824
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.362596781397068e-05,
+      "loss": 3.0639,
+      "step": 5825
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.351149717297833e-05,
+      "loss": 3.0486,
+      "step": 5826
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.339712261390212e-05,
+      "loss": 3.0668,
+      "step": 5827
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.328284416191892e-05,
+      "loss": 3.1715,
+      "step": 5828
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.316866184218434e-05,
+      "loss": 3.2332,
+      "step": 5829
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.305457567983247e-05,
+      "loss": 3.2108,
+      "step": 5830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.294058569997674e-05,
+      "loss": 3.1008,
+      "step": 5831
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.282669192770896e-05,
+      "loss": 3.0863,
+      "step": 5832
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.271289438810013e-05,
+      "loss": 3.0901,
+      "step": 5833
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.259919310619977e-05,
+      "loss": 3.1385,
+      "step": 5834
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.248558810703647e-05,
+      "loss": 3.1787,
+      "step": 5835
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.237207941561734e-05,
+      "loss": 3.0745,
+      "step": 5836
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.225866705692856e-05,
+      "loss": 3.3117,
+      "step": 5837
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.214535105593505e-05,
+      "loss": 3.0915,
+      "step": 5838
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.203213143758035e-05,
+      "loss": 3.0393,
+      "step": 5839
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.191900822678698e-05,
+      "loss": 3.1505,
+      "step": 5840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.180598144845606e-05,
+      "loss": 3.1008,
+      "step": 5841
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.169305112746776e-05,
+      "loss": 3.1517,
+      "step": 5842
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.158021728868062e-05,
+      "loss": 3.0368,
+      "step": 5843
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.146747995693224e-05,
+      "loss": 3.1443,
+      "step": 5844
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.135483915703888e-05,
+      "loss": 3.1922,
+      "step": 5845
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.124229491379574e-05,
+      "loss": 3.1315,
+      "step": 5846
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.112984725197645e-05,
+      "loss": 3.0991,
+      "step": 5847
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.1017496196333454e-05,
+      "loss": 3.1582,
+      "step": 5848
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.090524177159812e-05,
+      "loss": 3.0246,
+      "step": 5849
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.079308400248029e-05,
+      "loss": 3.1043,
+      "step": 5850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.068102291366884e-05,
+      "loss": 3.1463,
+      "step": 5851
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.056905852983102e-05,
+      "loss": 3.1716,
+      "step": 5852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.0457190875612964e-05,
+      "loss": 3.0289,
+      "step": 5853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.034541997563964e-05,
+      "loss": 3.1747,
+      "step": 5854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.0233745854514636e-05,
+      "loss": 3.1044,
+      "step": 5855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.012216853682001e-05,
+      "loss": 3.0298,
+      "step": 5856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.0010688047116735e-05,
+      "loss": 3.2529,
+      "step": 5857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.989930440994451e-05,
+      "loss": 3.1578,
+      "step": 5858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.978801764982145e-05,
+      "loss": 3.126,
+      "step": 5859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.967682779124478e-05,
+      "loss": 3.2123,
+      "step": 5860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.956573485868988e-05,
+      "loss": 3.1169,
+      "step": 5861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.945473887661118e-05,
+      "loss": 3.1828,
+      "step": 5862
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.934383986944158e-05,
+      "loss": 3.1016,
+      "step": 5863
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.9233037861592896e-05,
+      "loss": 3.104,
+      "step": 5864
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.9122332877455276e-05,
+      "loss": 3.1052,
+      "step": 5865
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.901172494139739e-05,
+      "loss": 3.179,
+      "step": 5866
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.890121407776716e-05,
+      "loss": 3.2006,
+      "step": 5867
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.879080031089046e-05,
+      "loss": 3.1153,
+      "step": 5868
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.8680483665072335e-05,
+      "loss": 3.2712,
+      "step": 5869
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.857026416459593e-05,
+      "loss": 3.1874,
+      "step": 5870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.84601418337235e-05,
+      "loss": 3.1764,
+      "step": 5871
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.8350116696695596e-05,
+      "loss": 3.0982,
+      "step": 5872
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.824018877773163e-05,
+      "loss": 3.0777,
+      "step": 5873
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.813035810102935e-05,
+      "loss": 3.2982,
+      "step": 5874
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.8020624690765076e-05,
+      "loss": 3.0828,
+      "step": 5875
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7910988571094e-05,
+      "loss": 3.1596,
+      "step": 5876
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.78014497661497e-05,
+      "loss": 3.2105,
+      "step": 5877
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.76920083000444e-05,
+      "loss": 3.2571,
+      "step": 5878
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.758266419686881e-05,
+      "loss": 3.2297,
+      "step": 5879
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.747341748069229e-05,
+      "loss": 3.1762,
+      "step": 5880
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7364268175562786e-05,
+      "loss": 2.9494,
+      "step": 5881
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.725521630550684e-05,
+      "loss": 3.1783,
+      "step": 5882
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.714626189452937e-05,
+      "loss": 3.1142,
+      "step": 5883
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.70374049666138e-05,
+      "loss": 3.1604,
+      "step": 5884
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6928645545722524e-05,
+      "loss": 3.3185,
+      "step": 5885
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6819983655795935e-05,
+      "loss": 3.0474,
+      "step": 5886
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.671141932075335e-05,
+      "loss": 3.0614,
+      "step": 5887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.660295256449233e-05,
+      "loss": 3.2309,
+      "step": 5888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.649458341088914e-05,
+      "loss": 3.1124,
+      "step": 5889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.638631188379856e-05,
+      "loss": 3.12,
+      "step": 5890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6278138007053944e-05,
+      "loss": 3.0623,
+      "step": 5891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.617006180446688e-05,
+      "loss": 3.1212,
+      "step": 5892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.606208329982754e-05,
+      "loss": 3.1292,
+      "step": 5893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.5954202516904864e-05,
+      "loss": 3.026,
+      "step": 5894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.584641947944591e-05,
+      "loss": 3.1871,
+      "step": 5895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.573873421117642e-05,
+      "loss": 3.0561,
+      "step": 5896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.563114673580061e-05,
+      "loss": 3.0537,
+      "step": 5897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.55236570770013e-05,
+      "loss": 3.1289,
+      "step": 5898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.54162652584394e-05,
+      "loss": 3.1259,
+      "step": 5899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.530897130375467e-05,
+      "loss": 3.1487,
+      "step": 5900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.5201775236565056e-05,
+      "loss": 3.0727,
+      "step": 5901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.509467708046706e-05,
+      "loss": 2.9689,
+      "step": 5902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.498767685903572e-05,
+      "loss": 3.2156,
+      "step": 5903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.4880774595824246e-05,
+      "loss": 3.1497,
+      "step": 5904
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.477397031436465e-05,
+      "loss": 3.1182,
+      "step": 5905
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.466726403816713e-05,
+      "loss": 3.2049,
+      "step": 5906
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.45606557907205e-05,
+      "loss": 3.1578,
+      "step": 5907
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.445414559549167e-05,
+      "loss": 3.0453,
+      "step": 5908
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.4347733475926365e-05,
+      "loss": 3.1014,
+      "step": 5909
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.4241419455448414e-05,
+      "loss": 3.1318,
+      "step": 5910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.41352035574601e-05,
+      "loss": 3.1573,
+      "step": 5911
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.4029085805342324e-05,
+      "loss": 3.1595,
+      "step": 5912
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.392306622245407e-05,
+      "loss": 3.114,
+      "step": 5913
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.38171448321329e-05,
+      "loss": 3.0953,
+      "step": 5914
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.3711321657694825e-05,
+      "loss": 3.1468,
+      "step": 5915
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.36055967224342e-05,
+      "loss": 3.0251,
+      "step": 5916
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.3499970049623483e-05,
+      "loss": 3.0255,
+      "step": 5917
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.3394441662513945e-05,
+      "loss": 3.0289,
+      "step": 5918
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.3289011584334855e-05,
+      "loss": 3.1095,
+      "step": 5919
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.318367983829392e-05,
+      "loss": 3.0044,
+      "step": 5920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.307844644757742e-05,
+      "loss": 3.2112,
+      "step": 5921
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.297331143534973e-05,
+      "loss": 3.0923,
+      "step": 5922
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.286827482475365e-05,
+      "loss": 3.1513,
+      "step": 5923
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.276333663891036e-05,
+      "loss": 3.1145,
+      "step": 5924
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.26584969009195e-05,
+      "loss": 3.043,
+      "step": 5925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.255375563385867e-05,
+      "loss": 3.1938,
+      "step": 5926
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.244911286078419e-05,
+      "loss": 3.0888,
+      "step": 5927
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.234456860473041e-05,
+      "loss": 3.1117,
+      "step": 5928
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.224012288871011e-05,
+      "loss": 3.0747,
+      "step": 5929
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.213577573571443e-05,
+      "loss": 3.1275,
+      "step": 5930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.2031527168712624e-05,
+      "loss": 3.2608,
+      "step": 5931
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.192737721065255e-05,
+      "loss": 3.0666,
+      "step": 5932
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.182332588446009e-05,
+      "loss": 3.1642,
+      "step": 5933
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.17193732130396e-05,
+      "loss": 3.0845,
+      "step": 5934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.1615519219273546e-05,
+      "loss": 3.1004,
+      "step": 5935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.151176392602291e-05,
+      "loss": 3.1509,
+      "step": 5936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.140810735612655e-05,
+      "loss": 3.0493,
+      "step": 5937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.13045495324021e-05,
+      "loss": 3.0786,
+      "step": 5938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.120109047764504e-05,
+      "loss": 3.1068,
+      "step": 5939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.109773021462921e-05,
+      "loss": 3.093,
+      "step": 5940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.099446876610692e-05,
+      "loss": 3.164,
+      "step": 5941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.089130615480841e-05,
+      "loss": 2.9833,
+      "step": 5942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.07882424034426e-05,
+      "loss": 3.155,
+      "step": 5943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.068527753469604e-05,
+      "loss": 3.0512,
+      "step": 5944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.058241157123411e-05,
+      "loss": 3.146,
+      "step": 5945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.047964453569992e-05,
+      "loss": 3.2512,
+      "step": 5946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.037697645071526e-05,
+      "loss": 3.1434,
+      "step": 5947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.027440733887972e-05,
+      "loss": 3.1983,
+      "step": 5948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.017193722277136e-05,
+      "loss": 3.0799,
+      "step": 5949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.006956612494651e-05,
+      "loss": 3.0886,
+      "step": 5950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.996729406793943e-05,
+      "loss": 3.1598,
+      "step": 5951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.986512107426283e-05,
+      "loss": 3.2882,
+      "step": 5952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9763047166407395e-05,
+      "loss": 3.1816,
+      "step": 5953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.966107236684225e-05,
+      "loss": 3.137,
+      "step": 5954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9559196698014455e-05,
+      "loss": 3.0786,
+      "step": 5955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.94574201823495e-05,
+      "loss": 3.2056,
+      "step": 5956
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9355742842250695e-05,
+      "loss": 3.1278,
+      "step": 5957
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.925416470009991e-05,
+      "loss": 3.0525,
+      "step": 5958
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.91526857782571e-05,
+      "loss": 3.178,
+      "step": 5959
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9051306099060055e-05,
+      "loss": 3.1103,
+      "step": 5960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.895002568482509e-05,
+      "loss": 3.1031,
+      "step": 5961
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.884884455784644e-05,
+      "loss": 3.085,
+      "step": 5962
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.8747762740396714e-05,
+      "loss": 3.1366,
+      "step": 5963
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.864678025472635e-05,
+      "loss": 3.0422,
+      "step": 5964
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.854589712306423e-05,
+      "loss": 3.0585,
+      "step": 5965
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.844511336761709e-05,
+      "loss": 3.0954,
+      "step": 5966
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.834442901057001e-05,
+      "loss": 3.1752,
+      "step": 5967
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.824384407408622e-05,
+      "loss": 3.1914,
+      "step": 5968
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.814335858030672e-05,
+      "loss": 3.0803,
+      "step": 5969
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.80429725513511e-05,
+      "loss": 3.0941,
+      "step": 5970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.794268600931651e-05,
+      "loss": 3.1277,
+      "step": 5971
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7842498976278794e-05,
+      "loss": 3.1871,
+      "step": 5972
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7742411474291425e-05,
+      "loss": 3.1268,
+      "step": 5973
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7642423525386225e-05,
+      "loss": 3.1671,
+      "step": 5974
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.754253515157286e-05,
+      "loss": 3.0443,
+      "step": 5975
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7442746374839365e-05,
+      "loss": 3.0991,
+      "step": 5976
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.734305721715182e-05,
+      "loss": 3.1885,
+      "step": 5977
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.724346770045401e-05,
+      "loss": 3.1831,
+      "step": 5978
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7143977846668294e-05,
+      "loss": 3.0665,
+      "step": 5979
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.704458767769471e-05,
+      "loss": 3.2454,
+      "step": 5980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6945297215411576e-05,
+      "loss": 3.177,
+      "step": 5981
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.684610648167503e-05,
+      "loss": 3.0679,
+      "step": 5982
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.674701549831961e-05,
+      "loss": 3.0118,
+      "step": 5983
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.664802428715753e-05,
+      "loss": 3.0512,
+      "step": 5984
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.654913286997925e-05,
+      "loss": 3.2122,
+      "step": 5985
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.64503412685533e-05,
+      "loss": 3.0368,
+      "step": 5986
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6351649504626e-05,
+      "loss": 3.086,
+      "step": 5987
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6253057599922044e-05,
+      "loss": 3.1825,
+      "step": 5988
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6154565576143724e-05,
+      "loss": 3.136,
+      "step": 5989
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6056173454971815e-05,
+      "loss": 3.2387,
+      "step": 5990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.595788125806466e-05,
+      "loss": 3.0936,
+      "step": 5991
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.585968900705889e-05,
+      "loss": 3.1251,
+      "step": 5992
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.5761596723569e-05,
+      "loss": 3.0895,
+      "step": 5993
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.566360442918754e-05,
+      "loss": 2.9366,
+      "step": 5994
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.556571214548516e-05,
+      "loss": 3.1494,
+      "step": 5995
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.546791989401011e-05,
+      "loss": 3.1151,
+      "step": 5996
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.537022769628923e-05,
+      "loss": 3.2048,
+      "step": 5997
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.527263557382666e-05,
+      "loss": 3.1502,
+      "step": 5998
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.517514354810509e-05,
+      "loss": 3.0745,
+      "step": 5999
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.5077751640584696e-05,
+      "loss": 3.1254,
+      "step": 6000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4980459872704e-05,
+      "loss": 3.1738,
+      "step": 6001
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4883268265879404e-05,
+      "loss": 3.0157,
+      "step": 6002
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4786176841504935e-05,
+      "loss": 3.2618,
+      "step": 6003
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.468918562095309e-05,
+      "loss": 3.0559,
+      "step": 6004
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.459229462557379e-05,
+      "loss": 3.0632,
+      "step": 6005
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.449550387669537e-05,
+      "loss": 3.1193,
+      "step": 6006
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4398813395623626e-05,
+      "loss": 3.1638,
+      "step": 6007
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.43022232036428e-05,
+      "loss": 3.1171,
+      "step": 6008
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4205733322014496e-05,
+      "loss": 3.0651,
+      "step": 6009
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.410934377197867e-05,
+      "loss": 3.1376,
+      "step": 6010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.401305457475313e-05,
+      "loss": 3.1768,
+      "step": 6011
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.391686575153331e-05,
+      "loss": 3.0957,
+      "step": 6012
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.382077732349299e-05,
+      "loss": 3.0247,
+      "step": 6013
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.372478931178336e-05,
+      "loss": 3.0169,
+      "step": 6014
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.362890173753392e-05,
+      "loss": 3.1424,
+      "step": 6015
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.353311462185183e-05,
+      "loss": 3.1256,
+      "step": 6016
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.343742798582229e-05,
+      "loss": 3.0539,
+      "step": 6017
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3341841850508104e-05,
+      "loss": 3.0544,
+      "step": 6018
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.324635623695028e-05,
+      "loss": 3.1493,
+      "step": 6019
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.315097116616767e-05,
+      "loss": 3.2077,
+      "step": 6020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.305568665915666e-05,
+      "loss": 3.0578,
+      "step": 6021
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.296050273689195e-05,
+      "loss": 3.156,
+      "step": 6022
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.286541942032562e-05,
+      "loss": 3.1038,
+      "step": 6023
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.277043673038816e-05,
+      "loss": 3.0752,
+      "step": 6024
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.267555468798734e-05,
+      "loss": 3.1069,
+      "step": 6025
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.2580773314009225e-05,
+      "loss": 3.2482,
+      "step": 6026
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.248609262931735e-05,
+      "loss": 3.0604,
+      "step": 6027
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.239151265475344e-05,
+      "loss": 3.0564,
+      "step": 6028
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.2297033411136886e-05,
+      "loss": 2.9924,
+      "step": 6029
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.220265491926489e-05,
+      "loss": 3.1359,
+      "step": 6030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.21083771999125e-05,
+      "loss": 2.9635,
+      "step": 6031
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.2014200273832404e-05,
+      "loss": 3.1017,
+      "step": 6032
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1920124161755555e-05,
+      "loss": 3.0713,
+      "step": 6033
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.182614888439018e-05,
+      "loss": 3.1888,
+      "step": 6034
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1732274462422725e-05,
+      "loss": 3.2342,
+      "step": 6035
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1638500916517165e-05,
+      "loss": 3.1207,
+      "step": 6036
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1544828267315425e-05,
+      "loss": 3.1116,
+      "step": 6037
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.145125653543719e-05,
+      "loss": 3.1723,
+      "step": 6038
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135778574147997e-05,
+      "loss": 3.1291,
+      "step": 6039
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1264415906018924e-05,
+      "loss": 3.0732,
+      "step": 6040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.117114704960695e-05,
+      "loss": 3.204,
+      "step": 6041
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.107797919277506e-05,
+      "loss": 3.134,
+      "step": 6042
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.098491235603163e-05,
+      "loss": 3.1801,
+      "step": 6043
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0891946559863055e-05,
+      "loss": 3.055,
+      "step": 6044
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.079908182473335e-05,
+      "loss": 3.1006,
+      "step": 6045
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0706318171084334e-05,
+      "loss": 3.1204,
+      "step": 6046
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.061365561933567e-05,
+      "loss": 3.1425,
+      "step": 6047
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.05210941898847e-05,
+      "loss": 3.1938,
+      "step": 6048
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.042863390310642e-05,
+      "loss": 3.1896,
+      "step": 6049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0336274779353535e-05,
+      "loss": 3.0589,
+      "step": 6050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.024401683895668e-05,
+      "loss": 3.1176,
+      "step": 6051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0151860102224044e-05,
+      "loss": 3.1769,
+      "step": 6052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.005980458944169e-05,
+      "loss": 3.0197,
+      "step": 6053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.996785032087324e-05,
+      "loss": 3.0603,
+      "step": 6054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.987599731676022e-05,
+      "loss": 3.2522,
+      "step": 6055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.978424559732158e-05,
+      "loss": 3.0098,
+      "step": 6056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.969259518275431e-05,
+      "loss": 3.1194,
+      "step": 6057
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.960104609323284e-05,
+      "loss": 3.0131,
+      "step": 6058
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.950959834890927e-05,
+      "loss": 3.0291,
+      "step": 6059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.941825196991378e-05,
+      "loss": 3.1139,
+      "step": 6060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.932700697635372e-05,
+      "loss": 3.0906,
+      "step": 6061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.923586338831453e-05,
+      "loss": 3.1001,
+      "step": 6062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.914482122585905e-05,
+      "loss": 3.2197,
+      "step": 6063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.905388050902808e-05,
+      "loss": 3.0445,
+      "step": 6064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.896304125783973e-05,
+      "loss": 3.2518,
+      "step": 6065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8872303492290154e-05,
+      "loss": 3.1282,
+      "step": 6066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.878166723235288e-05,
+      "loss": 3.0362,
+      "step": 6067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8691132497979066e-05,
+      "loss": 3.1454,
+      "step": 6068
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8600699309097876e-05,
+      "loss": 3.0578,
+      "step": 6069
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.851036768561572e-05,
+      "loss": 3.095,
+      "step": 6070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.842013764741686e-05,
+      "loss": 3.0728,
+      "step": 6071
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8330009214363194e-05,
+      "loss": 3.1818,
+      "step": 6072
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.823998240629434e-05,
+      "loss": 3.026,
+      "step": 6073
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.815005724302717e-05,
+      "loss": 3.0081,
+      "step": 6074
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.806023374435663e-05,
+      "loss": 3.0505,
+      "step": 6075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.797051193005507e-05,
+      "loss": 3.0276,
+      "step": 6076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.788089181987236e-05,
+      "loss": 3.1396,
+      "step": 6077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779137343353623e-05,
+      "loss": 3.2437,
+      "step": 6078
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.770195679075178e-05,
+      "loss": 3.1643,
+      "step": 6079
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.761264191120189e-05,
+      "loss": 3.1272,
+      "step": 6080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7523428814546954e-05,
+      "loss": 3.0861,
+      "step": 6081
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.743431752042509e-05,
+      "loss": 3.0726,
+      "step": 6082
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.734530804845177e-05,
+      "loss": 3.1536,
+      "step": 6083
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.725640041822026e-05,
+      "loss": 3.1487,
+      "step": 6084
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.716759464930125e-05,
+      "loss": 3.0443,
+      "step": 6085
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.70788907612431e-05,
+      "loss": 3.0403,
+      "step": 6086
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.699028877357186e-05,
+      "loss": 3.1099,
+      "step": 6087
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.690178870579075e-05,
+      "loss": 3.1775,
+      "step": 6088
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.681339057738103e-05,
+      "loss": 3.1721,
+      "step": 6089
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.672509440780125e-05,
+      "loss": 2.9877,
+      "step": 6090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6636900216487745e-05,
+      "loss": 3.2046,
+      "step": 6091
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.654880802285393e-05,
+      "loss": 3.1411,
+      "step": 6092
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.646081784629141e-05,
+      "loss": 2.9192,
+      "step": 6093
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.637292970616879e-05,
+      "loss": 3.0551,
+      "step": 6094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.6285143621832386e-05,
+      "loss": 3.1556,
+      "step": 6095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.619745961260623e-05,
+      "loss": 3.145,
+      "step": 6096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.610987769779156e-05,
+      "loss": 3.1005,
+      "step": 6097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.60223978966675e-05,
+      "loss": 2.9397,
+      "step": 6098
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.593502022849043e-05,
+      "loss": 2.9234,
+      "step": 6099
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.584774471249447e-05,
+      "loss": 3.0883,
+      "step": 6100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5760571367890904e-05,
+      "loss": 3.0439,
+      "step": 6101
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.567350021386895e-05,
+      "loss": 3.0982,
+      "step": 6102
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.558653126959499e-05,
+      "loss": 3.194,
+      "step": 6103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.549966455421305e-05,
+      "loss": 3.1463,
+      "step": 6104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5412900086844615e-05,
+      "loss": 3.0536,
+      "step": 6105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.532623788658873e-05,
+      "loss": 3.0922,
+      "step": 6106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5239677972522035e-05,
+      "loss": 3.0262,
+      "step": 6107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.515322036369822e-05,
+      "loss": 3.1892,
+      "step": 6108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.506686507914902e-05,
+      "loss": 3.082,
+      "step": 6109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4980612137883175e-05,
+      "loss": 3.1276,
+      "step": 6110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4894461558887256e-05,
+      "loss": 2.9493,
+      "step": 6111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4808413361125e-05,
+      "loss": 3.0368,
+      "step": 6112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4722467563537696e-05,
+      "loss": 3.164,
+      "step": 6113
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4636624185044266e-05,
+      "loss": 3.0932,
+      "step": 6114
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.455088324454092e-05,
+      "loss": 3.1401,
+      "step": 6115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.44652447609014e-05,
+      "loss": 2.9825,
+      "step": 6116
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4379708752976804e-05,
+      "loss": 3.115,
+      "step": 6117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.429427523959577e-05,
+      "loss": 3.1157,
+      "step": 6118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.420894423956422e-05,
+      "loss": 3.1178,
+      "step": 6119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4123715771665786e-05,
+      "loss": 3.1586,
+      "step": 6120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.403858985466129e-05,
+      "loss": 3.0353,
+      "step": 6121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3953566507288927e-05,
+      "loss": 3.0048,
+      "step": 6122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3868645748264534e-05,
+      "loss": 3.2191,
+      "step": 6123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.378382759628129e-05,
+      "loss": 3.108,
+      "step": 6124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3699112070009865e-05,
+      "loss": 3.2803,
+      "step": 6125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3614499188098e-05,
+      "loss": 3.0845,
+      "step": 6126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.352998896917131e-05,
+      "loss": 3.0518,
+      "step": 6127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.344558143183246e-05,
+      "loss": 3.0412,
+      "step": 6128
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3361276594661706e-05,
+      "loss": 3.0978,
+      "step": 6129
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.327707447621653e-05,
+      "loss": 3.2128,
+      "step": 6130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.319297509503205e-05,
+      "loss": 3.0835,
+      "step": 6131
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3108978469620407e-05,
+      "loss": 3.087,
+      "step": 6132
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3025084618471535e-05,
+      "loss": 3.0215,
+      "step": 6133
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.294129356005249e-05,
+      "loss": 3.1951,
+      "step": 6134
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2857605312807684e-05,
+      "loss": 3.1377,
+      "step": 6135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.27740198951591e-05,
+      "loss": 3.2303,
+      "step": 6136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2690537325505806e-05,
+      "loss": 3.073,
+      "step": 6137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.260715762222449e-05,
+      "loss": 3.1163,
+      "step": 6138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.252388080366903e-05,
+      "loss": 3.0205,
+      "step": 6139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.24407068881708e-05,
+      "loss": 3.0895,
+      "step": 6140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.235763589403829e-05,
+      "loss": 3.0907,
+      "step": 6141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227466783955757e-05,
+      "loss": 3.1009,
+      "step": 6142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.219180274299205e-05,
+      "loss": 3.1725,
+      "step": 6143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.210904062258219e-05,
+      "loss": 3.0727,
+      "step": 6144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.202638149654613e-05,
+      "loss": 3.1022,
+      "step": 6145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1943825383079106e-05,
+      "loss": 3.0993,
+      "step": 6146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1861372300353854e-05,
+      "loss": 3.2258,
+      "step": 6147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.177902226652024e-05,
+      "loss": 3.043,
+      "step": 6148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.16967752997056e-05,
+      "loss": 3.1962,
+      "step": 6149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.161463141801446e-05,
+      "loss": 3.167,
+      "step": 6150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.153259063952879e-05,
+      "loss": 3.1604,
+      "step": 6151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.145065298230782e-05,
+      "loss": 3.2486,
+      "step": 6152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.136881846438794e-05,
+      "loss": 3.0451,
+      "step": 6153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.128708710378308e-05,
+      "loss": 3.0778,
+      "step": 6154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.12054589184842e-05,
+      "loss": 3.2212,
+      "step": 6155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.112393392645985e-05,
+      "loss": 3.1389,
+      "step": 6156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.104251214565545e-05,
+      "loss": 3.0437,
+      "step": 6157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0961193593994186e-05,
+      "loss": 3.1085,
+      "step": 6158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.087997828937627e-05,
+      "loss": 3.2129,
+      "step": 6159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0798866249679036e-05,
+      "loss": 3.0632,
+      "step": 6160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.071785749275741e-05,
+      "loss": 3.1146,
+      "step": 6161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.063695203644329e-05,
+      "loss": 3.1002,
+      "step": 6162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0556149898546046e-05,
+      "loss": 3.1694,
+      "step": 6163
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0475451096852237e-05,
+      "loss": 3.1584,
+      "step": 6164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.039485564912564e-05,
+      "loss": 3.1646,
+      "step": 6165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0314363573107294e-05,
+      "loss": 3.0997,
+      "step": 6166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0233974886515468e-05,
+      "loss": 3.0548,
+      "step": 6167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 3.0232,
+      "step": 6168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0073507752371e-05,
+      "loss": 3.1437,
+      "step": 6169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9993429340141153e-05,
+      "loss": 3.0972,
+      "step": 6170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9913454387983318e-05,
+      "loss": 3.0373,
+      "step": 6171
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9833582913502132e-05,
+      "loss": 3.2651,
+      "step": 6172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.97538149342792e-05,
+      "loss": 3.1755,
+      "step": 6173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.967415046787353e-05,
+      "loss": 3.1516,
+      "step": 6174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9594589531821036e-05,
+      "loss": 3.145,
+      "step": 6175
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9515132143635215e-05,
+      "loss": 3.1525,
+      "step": 6176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9435778320806627e-05,
+      "loss": 3.1297,
+      "step": 6177
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9356528080802914e-05,
+      "loss": 3.1787,
+      "step": 6178
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9277381441069063e-05,
+      "loss": 3.1076,
+      "step": 6179
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9198338419027137e-05,
+      "loss": 3.047,
+      "step": 6180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.911939903207661e-05,
+      "loss": 3.1579,
+      "step": 6181
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9040563297593804e-05,
+      "loss": 3.1297,
+      "step": 6182
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8961831232932555e-05,
+      "loss": 3.2644,
+      "step": 6183
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8883202855423673e-05,
+      "loss": 3.1341,
+      "step": 6184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8804678182375198e-05,
+      "loss": 3.1868,
+      "step": 6185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8726257231072474e-05,
+      "loss": 3.2603,
+      "step": 6186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.864794001877774e-05,
+      "loss": 3.1676,
+      "step": 6187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8569726562730658e-05,
+      "loss": 3.1233,
+      "step": 6188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8491616880147843e-05,
+      "loss": 3.069,
+      "step": 6189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.841361098822326e-05,
+      "loss": 3.0773,
+      "step": 6190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.83357089041279e-05,
+      "loss": 3.0975,
+      "step": 6191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.825791064500993e-05,
+      "loss": 3.0335,
+      "step": 6192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.818021622799466e-05,
+      "loss": 3.0387,
+      "step": 6193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8102625670184566e-05,
+      "loss": 3.0713,
+      "step": 6194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8025138988659327e-05,
+      "loss": 3.1511,
+      "step": 6195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7947756200475517e-05,
+      "loss": 3.2123,
+      "step": 6196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7870477322667176e-05,
+      "loss": 3.1706,
+      "step": 6197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7793302372245145e-05,
+      "loss": 3.1443,
+      "step": 6198
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7716231366197665e-05,
+      "loss": 3.1652,
+      "step": 6199
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7639264321489888e-05,
+      "loss": 3.1763,
+      "step": 6200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7562401255064207e-05,
+      "loss": 3.1487,
+      "step": 6201
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7485642183840032e-05,
+      "loss": 3.0922,
+      "step": 6202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.740898712471396e-05,
+      "loss": 3.0595,
+      "step": 6203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.733243609455971e-05,
+      "loss": 3.0,
+      "step": 6204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7255989110227973e-05,
+      "loss": 3.0418,
+      "step": 6205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7179646188546738e-05,
+      "loss": 3.1038,
+      "step": 6206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.710340734632083e-05,
+      "loss": 3.206,
+      "step": 6207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.702727260033244e-05,
+      "loss": 3.0555,
+      "step": 6208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.695124196734061e-05,
+      "loss": 3.1806,
+      "step": 6209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6875315464081564e-05,
+      "loss": 3.0993,
+      "step": 6210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6799493107268646e-05,
+      "loss": 3.152,
+      "step": 6211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.672377491359229e-05,
+      "loss": 3.157,
+      "step": 6212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6648160899719888e-05,
+      "loss": 3.1223,
+      "step": 6213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.657265108229584e-05,
+      "loss": 3.2137,
+      "step": 6214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6497245477941967e-05,
+      "loss": 3.0486,
+      "step": 6215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6421944103256656e-05,
+      "loss": 3.1507,
+      "step": 6216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.634674697481576e-05,
+      "loss": 3.2027,
+      "step": 6217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6271654109171927e-05,
+      "loss": 3.0697,
+      "step": 6218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6196665522855045e-05,
+      "loss": 3.1612,
+      "step": 6219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6121781232371853e-05,
+      "loss": 3.2194,
+      "step": 6220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.604700125420645e-05,
+      "loss": 3.096,
+      "step": 6221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5972325604819492e-05,
+      "loss": 3.122,
+      "step": 6222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5897754300649112e-05,
+      "loss": 3.2408,
+      "step": 6223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5823287358110292e-05,
+      "loss": 2.9659,
+      "step": 6224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.574892479359492e-05,
+      "loss": 3.0851,
+      "step": 6225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5674666623472177e-05,
+      "loss": 3.1031,
+      "step": 6226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.560051286408793e-05,
+      "loss": 3.1885,
+      "step": 6227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5526463531765464e-05,
+      "loss": 3.2135,
+      "step": 6228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.545251864280479e-05,
+      "loss": 2.9322,
+      "step": 6229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.537867821348305e-05,
+      "loss": 3.1367,
+      "step": 6230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5304942260054253e-05,
+      "loss": 3.1428,
+      "step": 6231
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.523131079874963e-05,
+      "loss": 3.0451,
+      "step": 6232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.5157783845777216e-05,
+      "loss": 3.1494,
+      "step": 6233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.5084361417322067e-05,
+      "loss": 3.1537,
+      "step": 6234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.5011043529546363e-05,
+      "loss": 3.1055,
+      "step": 6235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4937830198589086e-05,
+      "loss": 3.0113,
+      "step": 6236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.486472144056634e-05,
+      "loss": 3.1561,
+      "step": 6237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4791717271571202e-05,
+      "loss": 3.2308,
+      "step": 6238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.471881770767376e-05,
+      "loss": 3.1286,
+      "step": 6239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.464602276492084e-05,
+      "loss": 3.0813,
+      "step": 6240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4573332459336628e-05,
+      "loss": 3.101,
+      "step": 6241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.450074680692188e-05,
+      "loss": 3.1187,
+      "step": 6242
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.442826582365454e-05,
+      "loss": 3.1196,
+      "step": 6243
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4355889525489505e-05,
+      "loss": 3.0235,
+      "step": 6244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.428361792835848e-05,
+      "loss": 3.2031,
+      "step": 6245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4211451048170296e-05,
+      "loss": 3.1032,
+      "step": 6246
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4139388900810688e-05,
+      "loss": 3.1277,
+      "step": 6247
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4067431502142413e-05,
+      "loss": 3.0769,
+      "step": 6248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3995578868004908e-05,
+      "loss": 3.0699,
+      "step": 6249
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3923831014214803e-05,
+      "loss": 3.084,
+      "step": 6250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3852187956565576e-05,
+      "loss": 3.1903,
+      "step": 6251
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.378064971082755e-05,
+      "loss": 3.1221,
+      "step": 6252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3709216292748194e-05,
+      "loss": 3.0171,
+      "step": 6253
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.363788771805159e-05,
+      "loss": 3.0426,
+      "step": 6254
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3566664002439066e-05,
+      "loss": 3.1029,
+      "step": 6255
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3495545161588694e-05,
+      "loss": 3.1745,
+      "step": 6256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3424531211155563e-05,
+      "loss": 3.1584,
+      "step": 6257
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.335362216677139e-05,
+      "loss": 3.1612,
+      "step": 6258
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3282818044045306e-05,
+      "loss": 3.223,
+      "step": 6259
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.321211885856278e-05,
+      "loss": 3.0827,
+      "step": 6260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.314152462588659e-05,
+      "loss": 3.1879,
+      "step": 6261
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3071035361556192e-05,
+      "loss": 3.0752,
+      "step": 6262
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3000651081088064e-05,
+      "loss": 3.1168,
+      "step": 6263
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2930371799975592e-05,
+      "loss": 3.2242,
+      "step": 6264
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2860197533688843e-05,
+      "loss": 3.1362,
+      "step": 6265
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2790128297675073e-05,
+      "loss": 3.2278,
+      "step": 6266
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.272016410735811e-05,
+      "loss": 3.1278,
+      "step": 6267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2650304978138912e-05,
+      "loss": 3.1338,
+      "step": 6268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2580550925395128e-05,
+      "loss": 3.1196,
+      "step": 6269
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2510901964481358e-05,
+      "loss": 3.14,
+      "step": 6270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2441358110729006e-05,
+      "loss": 3.1376,
+      "step": 6271
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.237191937944649e-05,
+      "loss": 3.0487,
+      "step": 6272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2302585785919094e-05,
+      "loss": 3.1273,
+      "step": 6273
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.22333573454086e-05,
+      "loss": 3.2614,
+      "step": 6274
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.21642340731541e-05,
+      "loss": 3.1334,
+      "step": 6275
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2095215984371197e-05,
+      "loss": 3.1012,
+      "step": 6276
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.202630309425263e-05,
+      "loss": 3.1209,
+      "step": 6277
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.195749541796771e-05,
+      "loss": 3.0694,
+      "step": 6278
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1888792970662706e-05,
+      "loss": 3.1122,
+      "step": 6279
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1820195767460803e-05,
+      "loss": 3.0099,
+      "step": 6280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1751703823461866e-05,
+      "loss": 3.0901,
+      "step": 6281
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1683317153742778e-05,
+      "loss": 3.1909,
+      "step": 6282
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1615035773356996e-05,
+      "loss": 3.1488,
+      "step": 6283
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1546859697335052e-05,
+      "loss": 3.1188,
+      "step": 6284
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1478788940684114e-05,
+      "loss": 3.1619,
+      "step": 6285
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1410823518388302e-05,
+      "loss": 3.1038,
+      "step": 6286
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.134296344540848e-05,
+      "loss": 3.1216,
+      "step": 6287
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1275208736682262e-05,
+      "loss": 3.0857,
+      "step": 6288
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1207559407124155e-05,
+      "loss": 3.1864,
+      "step": 6289
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.114001547162542e-05,
+      "loss": 3.0873,
+      "step": 6290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1072576945054324e-05,
+      "loss": 3.1745,
+      "step": 6291
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.100524384225555e-05,
+      "loss": 3.0182,
+      "step": 6292
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0938016178050912e-05,
+      "loss": 3.0848,
+      "step": 6293
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0870893967238792e-05,
+      "loss": 3.1946,
+      "step": 6294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0803877224594536e-05,
+      "loss": 3.2287,
+      "step": 6295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0736965964870124e-05,
+      "loss": 3.1475,
+      "step": 6296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.067016020279433e-05,
+      "loss": 3.1075,
+      "step": 6297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0603459953072834e-05,
+      "loss": 3.1745,
+      "step": 6298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0536865230387948e-05,
+      "loss": 3.1149,
+      "step": 6299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0470376049398942e-05,
+      "loss": 3.3075,
+      "step": 6300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0403992424741613e-05,
+      "loss": 3.0925,
+      "step": 6301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.033771437102877e-05,
+      "loss": 3.0459,
+      "step": 6302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0271541902849633e-05,
+      "loss": 3.1562,
+      "step": 6303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0205475034770604e-05,
+      "loss": 3.1506,
+      "step": 6304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.013951378133455e-05,
+      "loss": 3.2096,
+      "step": 6305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0073658157061137e-05,
+      "loss": 3.1631,
+      "step": 6306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0007908176446877e-05,
+      "loss": 3.1004,
+      "step": 6307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9942263853964915e-05,
+      "loss": 3.1105,
+      "step": 6308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.987672520406525e-05,
+      "loss": 3.1035,
+      "step": 6309
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9811292241174562e-05,
+      "loss": 3.1308,
+      "step": 6310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.974596497969622e-05,
+      "loss": 3.1432,
+      "step": 6311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9680743434010386e-05,
+      "loss": 3.1862,
+      "step": 6312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9615627618473964e-05,
+      "loss": 3.1164,
+      "step": 6313
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955061754742049e-05,
+      "loss": 3.0971,
+      "step": 6314
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9485713235160295e-05,
+      "loss": 3.1791,
+      "step": 6315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9420914695980563e-05,
+      "loss": 3.0816,
+      "step": 6316
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9356221944144824e-05,
+      "loss": 3.0991,
+      "step": 6317
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9291634993893803e-05,
+      "loss": 3.1261,
+      "step": 6318
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9227153859444456e-05,
+      "loss": 3.0616,
+      "step": 6319
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9162778554990822e-05,
+      "loss": 3.1082,
+      "step": 6320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9098509094703452e-05,
+      "loss": 3.1024,
+      "step": 6321
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.90343454927297e-05,
+      "loss": 3.0202,
+      "step": 6322
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8970287763193428e-05,
+      "loss": 3.0735,
+      "step": 6323
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8906335920195416e-05,
+      "loss": 3.1481,
+      "step": 6324
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.884248997781307e-05,
+      "loss": 3.1614,
+      "step": 6325
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.877874995010037e-05,
+      "loss": 2.9909,
+      "step": 6326
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.871511585108826e-05,
+      "loss": 3.1483,
+      "step": 6327
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8651587694783922e-05,
+      "loss": 3.1272,
+      "step": 6328
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8588165495171673e-05,
+      "loss": 3.0809,
+      "step": 6329
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.852484926621223e-05,
+      "loss": 3.0576,
+      "step": 6330
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8461639021843058e-05,
+      "loss": 3.0726,
+      "step": 6331
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8398534775978304e-05,
+      "loss": 3.1004,
+      "step": 6332
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8335536542508746e-05,
+      "loss": 3.1923,
+      "step": 6333
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.827264433530196e-05,
+      "loss": 3.1047,
+      "step": 6334
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.820985816820192e-05,
+      "loss": 3.2106,
+      "step": 6335
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8147178055029577e-05,
+      "loss": 3.0289,
+      "step": 6336
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8084604009582227e-05,
+      "loss": 3.0494,
+      "step": 6337
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.802213604563413e-05,
+      "loss": 3.1464,
+      "step": 6338
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7959774176935837e-05,
+      "loss": 3.1448,
+      "step": 6339
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7897518417214876e-05,
+      "loss": 3.0576,
+      "step": 6340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7835368780175164e-05,
+      "loss": 3.146,
+      "step": 6341
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7773325279497487e-05,
+      "loss": 3.1263,
+      "step": 6342
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.771138792883914e-05,
+      "loss": 3.1386,
+      "step": 6343
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7649556741833994e-05,
+      "loss": 3.1723,
+      "step": 6344
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7587831732092717e-05,
+      "loss": 3.3039,
+      "step": 6345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7526212913202444e-05,
+      "loss": 3.1389,
+      "step": 6346
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7464700298727042e-05,
+      "loss": 3.1866,
+      "step": 6347
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.740329390220685e-05,
+      "loss": 3.173,
+      "step": 6348
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.734199373715911e-05,
+      "loss": 3.0617,
+      "step": 6349
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7280799817077362e-05,
+      "loss": 3.067,
+      "step": 6350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7219712155431945e-05,
+      "loss": 2.9978,
+      "step": 6351
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7158730765669816e-05,
+      "loss": 3.0746,
+      "step": 6352
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7097855661214357e-05,
+      "loss": 3.005,
+      "step": 6353
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.70370868554659e-05,
+      "loss": 3.0737,
+      "step": 6354
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6976424361800912e-05,
+      "loss": 3.0536,
+      "step": 6355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6915868193572935e-05,
+      "loss": 3.0977,
+      "step": 6356
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6855418364111695e-05,
+      "loss": 3.182,
+      "step": 6357
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6795074886723827e-05,
+      "loss": 3.1073,
+      "step": 6358
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.673483777469237e-05,
+      "loss": 3.1384,
+      "step": 6359
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6674707041276936e-05,
+      "loss": 3.1393,
+      "step": 6360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6614682699713944e-05,
+      "loss": 3.2779,
+      "step": 6361
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6554764763216046e-05,
+      "loss": 3.1476,
+      "step": 6362
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6494953244972854e-05,
+      "loss": 3.0892,
+      "step": 6363
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.643524815815023e-05,
+      "loss": 3.1203,
+      "step": 6364
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.637564951589082e-05,
+      "loss": 3.0482,
+      "step": 6365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.631615733131364e-05,
+      "loss": 3.1882,
+      "step": 6366
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6256771617514487e-05,
+      "loss": 3.0241,
+      "step": 6367
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6197492387565627e-05,
+      "loss": 3.0305,
+      "step": 6368
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6138319654515897e-05,
+      "loss": 2.9972,
+      "step": 6369
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.607925343139066e-05,
+      "loss": 3.161,
+      "step": 6370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6020293731191794e-05,
+      "loss": 3.1261,
+      "step": 6371
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.596144056689791e-05,
+      "loss": 3.0534,
+      "step": 6372
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5902693951463877e-05,
+      "loss": 3.0869,
+      "step": 6373
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5844053897821454e-05,
+      "loss": 2.9584,
+      "step": 6374
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5785520418878597e-05,
+      "loss": 3.1653,
+      "step": 6375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.572709352752011e-05,
+      "loss": 3.0914,
+      "step": 6376
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.566877323660709e-05,
+      "loss": 3.119,
+      "step": 6377
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5610559558977446e-05,
+      "loss": 3.213,
+      "step": 6378
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.55524525074453e-05,
+      "loss": 3.121,
+      "step": 6379
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5494452094801494e-05,
+      "loss": 3.0992,
+      "step": 6380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5436558333813356e-05,
+      "loss": 3.1636,
+      "step": 6381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.537877123722464e-05,
+      "loss": 3.1216,
+      "step": 6382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.53210908177559e-05,
+      "loss": 3.1121,
+      "step": 6383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.526351708810386e-05,
+      "loss": 3.0118,
+      "step": 6384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5206050060942001e-05,
+      "loss": 3.245,
+      "step": 6385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5148689748920197e-05,
+      "loss": 3.0742,
+      "step": 6386
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5091436164664963e-05,
+      "loss": 3.1396,
+      "step": 6387
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.503428932077916e-05,
+      "loss": 3.1392,
+      "step": 6388
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4977249229842172e-05,
+      "loss": 3.0874,
+      "step": 6389
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4920315904410064e-05,
+      "loss": 3.083,
+      "step": 6390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4863489357015093e-05,
+      "loss": 3.099,
+      "step": 6391
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4806769600166359e-05,
+      "loss": 3.119,
+      "step": 6392
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4750156646349156e-05,
+      "loss": 3.1154,
+      "step": 6393
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4693650508025513e-05,
+      "loss": 3.1144,
+      "step": 6394
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4637251197633705e-05,
+      "loss": 3.168,
+      "step": 6395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4580958727588745e-05,
+      "loss": 3.0573,
+      "step": 6396
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524773110281885e-05,
+      "loss": 3.1149,
+      "step": 6397
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4468694358081013e-05,
+      "loss": 3.0852,
+      "step": 6398
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4412722483330532e-05,
+      "loss": 3.2055,
+      "step": 6399
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4356857498351029e-05,
+      "loss": 3.0752,
+      "step": 6400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.430109941544e-05,
+      "loss": 3.1552,
+      "step": 6401
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4245448246871017e-05,
+      "loss": 3.242,
+      "step": 6402
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4189904004894338e-05,
+      "loss": 3.1432,
+      "step": 6403
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.413446670173657e-05,
+      "loss": 3.1599,
+      "step": 6404
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4079136349601008e-05,
+      "loss": 3.2014,
+      "step": 6405
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4023912960667017e-05,
+      "loss": 3.1798,
+      "step": 6406
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3968796547090767e-05,
+      "loss": 3.1114,
+      "step": 6407
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3913787121004718e-05,
+      "loss": 3.1716,
+      "step": 6408
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3858884694517736e-05,
+      "loss": 3.2266,
+      "step": 6409
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3804089279715326e-05,
+      "loss": 3.1591,
+      "step": 6410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.374940088865917e-05,
+      "loss": 3.1218,
+      "step": 6411
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3694819533387636e-05,
+      "loss": 3.1426,
+      "step": 6412
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.364034522591534e-05,
+      "loss": 3.0111,
+      "step": 6413
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3585977978233577e-05,
+      "loss": 3.0267,
+      "step": 6414
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3531717802309829e-05,
+      "loss": 3.1814,
+      "step": 6415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3477564710088097e-05,
+      "loss": 3.2315,
+      "step": 6416
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3423518713488847e-05,
+      "loss": 3.1598,
+      "step": 6417
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.336957982440884e-05,
+      "loss": 3.0794,
+      "step": 6418
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3315748054721466e-05,
+      "loss": 3.2218,
+      "step": 6419
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3262023416276414e-05,
+      "loss": 3.0482,
+      "step": 6420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3208405920899835e-05,
+      "loss": 3.0691,
+      "step": 6421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3154895580394177e-05,
+      "loss": 3.224,
+      "step": 6422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3101492406538518e-05,
+      "loss": 3.112,
+      "step": 6423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3048196411088009e-05,
+      "loss": 3.0941,
+      "step": 6424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2995007605774656e-05,
+      "loss": 3.1237,
+      "step": 6425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2941926002306537e-05,
+      "loss": 3.0528,
+      "step": 6426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2888951612368082e-05,
+      "loss": 3.0388,
+      "step": 6427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2836084447620467e-05,
+      "loss": 3.0796,
+      "step": 6428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2783324519700935e-05,
+      "loss": 3.0084,
+      "step": 6429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2730671840223362e-05,
+      "loss": 3.0655,
+      "step": 6430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2678126420777814e-05,
+      "loss": 3.044,
+      "step": 6431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2625688272930925e-05,
+      "loss": 3.0145,
+      "step": 6432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2573357408225516e-05,
+      "loss": 3.1551,
+      "step": 6433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2521133838181043e-05,
+      "loss": 3.2641,
+      "step": 6434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2469017574293085e-05,
+      "loss": 3.1316,
+      "step": 6435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2417008628033688e-05,
+      "loss": 3.2041,
+      "step": 6436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2365107010851417e-05,
+      "loss": 3.115,
+      "step": 6437
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2313312734171078e-05,
+      "loss": 2.9428,
+      "step": 6438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2261625809393884e-05,
+      "loss": 3.0693,
+      "step": 6439
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2210046247897344e-05,
+      "loss": 3.0712,
+      "step": 6440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2158574061035432e-05,
+      "loss": 3.0785,
+      "step": 6441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2107209260138475e-05,
+      "loss": 3.0498,
+      "step": 6442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2055951856513147e-05,
+      "loss": 3.1652,
+      "step": 6443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2004801861442372e-05,
+      "loss": 3.0856,
+      "step": 6444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1953759286185584e-05,
+      "loss": 3.1266,
+      "step": 6445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1902824141978575e-05,
+      "loss": 3.0183,
+      "step": 6446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.185199644003332e-05,
+      "loss": 3.1794,
+      "step": 6447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1801276191538369e-05,
+      "loss": 3.1448,
+      "step": 6448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1750663407658402e-05,
+      "loss": 3.0938,
+      "step": 6449
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1700158099534618e-05,
+      "loss": 3.172,
+      "step": 6450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.164976027828446e-05,
+      "loss": 3.1913,
+      "step": 6451
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1599469955001718e-05,
+      "loss": 3.0513,
+      "step": 6452
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1549287140756592e-05,
+      "loss": 3.1635,
+      "step": 6453
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.149921184659547e-05,
+      "loss": 3.0089,
+      "step": 6454
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1449244083541199e-05,
+      "loss": 3.1717,
+      "step": 6455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1399383862592927e-05,
+      "loss": 3.2397,
+      "step": 6456
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1349631194726151e-05,
+      "loss": 3.073,
+      "step": 6457
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.129998609089261e-05,
+      "loss": 3.0997,
+      "step": 6458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.125044856202051e-05,
+      "loss": 3.1539,
+      "step": 6459
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1201018619014181e-05,
+      "loss": 3.1912,
+      "step": 6460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1151696272754475e-05,
+      "loss": 3.1343,
+      "step": 6461
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1102481534098374e-05,
+      "loss": 3.1613,
+      "step": 6462
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1053374413879269e-05,
+      "loss": 3.1971,
+      "step": 6463
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1004374922906847e-05,
+      "loss": 3.2249,
+      "step": 6464
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0955483071967144e-05,
+      "loss": 3.2099,
+      "step": 6465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0906698871822552e-05,
+      "loss": 3.1964,
+      "step": 6466
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0858022333211481e-05,
+      "loss": 3.04,
+      "step": 6467
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0809453466849028e-05,
+      "loss": 3.1474,
+      "step": 6468
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0760992283426252e-05,
+      "loss": 3.1436,
+      "step": 6469
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0712638793610785e-05,
+      "loss": 3.1062,
+      "step": 6470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0664393008046281e-05,
+      "loss": 3.1872,
+      "step": 6471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0616254937352964e-05,
+      "loss": 2.9585,
+      "step": 6472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0568224592127197e-05,
+      "loss": 3.2496,
+      "step": 6473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0520301982941572e-05,
+      "loss": 3.164,
+      "step": 6474
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0472487120345097e-05,
+      "loss": 3.2019,
+      "step": 6475
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0424780014863022e-05,
+      "loss": 3.1041,
+      "step": 6476
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0377180676996834e-05,
+      "loss": 3.1425,
+      "step": 6477
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0329689117224261e-05,
+      "loss": 3.0253,
+      "step": 6478
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0282305345999498e-05,
+      "loss": 3.1091,
+      "step": 6479
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0235029373752758e-05,
+      "loss": 3.1686,
+      "step": 6480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0187861210890714e-05,
+      "loss": 3.1576,
+      "step": 6481
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.014080086779634e-05,
+      "loss": 3.1219,
+      "step": 6482
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0093848354828572e-05,
+      "loss": 3.0322,
+      "step": 6483
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.004700368232303e-05,
+      "loss": 2.9939,
+      "step": 6484
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0000266860591245e-05,
+      "loss": 3.0838,
+      "step": 6485
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.953637899921209e-06,
+      "loss": 3.058,
+      "step": 6486
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.907116810577043e-06,
+      "loss": 3.1716,
+      "step": 6487
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.860703602799281e-06,
+      "loss": 3.1987,
+      "step": 6488
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.814398286804582e-06,
+      "loss": 3.1385,
+      "step": 6489
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.768200872785848e-06,
+      "loss": 3.2481,
+      "step": 6490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.722111370912334e-06,
+      "loss": 3.0922,
+      "step": 6491
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.67612979132948e-06,
+      "loss": 3.0689,
+      "step": 6492
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.63025614415891e-06,
+      "loss": 3.1911,
+      "step": 6493
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.584490439498605e-06,
+      "loss": 3.1391,
+      "step": 6494
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.538832687422728e-06,
+      "loss": 3.165,
+      "step": 6495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.49328289798157e-06,
+      "loss": 3.1902,
+      "step": 6496
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.447841081201947e-06,
+      "loss": 3.105,
+      "step": 6497
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.402507247086578e-06,
+      "loss": 3.1774,
+      "step": 6498
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.357281405614648e-06,
+      "loss": 3.2452,
+      "step": 6499
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.312163566741416e-06,
+      "loss": 3.134,
+      "step": 6500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.267153740398494e-06,
+      "loss": 3.061,
+      "step": 6501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.222251936493674e-06,
+      "loss": 3.1878,
+      "step": 6502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.17745816491089e-06,
+      "loss": 3.1199,
+      "step": 6503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.132772435510362e-06,
+      "loss": 3.0505,
+      "step": 6504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.088194758128555e-06,
+      "loss": 3.136,
+      "step": 6505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.043725142578118e-06,
+      "loss": 3.0947,
+      "step": 6506
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.999363598647892e-06,
+      "loss": 3.1023,
+      "step": 6507
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.95511013610295e-06,
+      "loss": 3.1275,
+      "step": 6508
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.91096476468467e-06,
+      "loss": 3.2256,
+      "step": 6509
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.866927494110388e-06,
+      "loss": 3.1785,
+      "step": 6510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.822998334073905e-06,
+      "loss": 3.0855,
+      "step": 6511
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.779177294245044e-06,
+      "loss": 3.127,
+      "step": 6512
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.73546438426992e-06,
+      "loss": 3.1186,
+      "step": 6513
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.691859613770836e-06,
+      "loss": 3.0505,
+      "step": 6514
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.648362992346336e-06,
+      "loss": 3.1687,
+      "step": 6515
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.604974529571042e-06,
+      "loss": 3.1518,
+      "step": 6516
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.561694234995754e-06,
+      "loss": 3.1206,
+      "step": 6517
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.518522118147742e-06,
+      "loss": 3.227,
+      "step": 6518
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.475458188530016e-06,
+      "loss": 3.1349,
+      "step": 6519
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.432502455622215e-06,
+      "loss": 3.1655,
+      "step": 6520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.389654928879831e-06,
+      "loss": 3.1374,
+      "step": 6521
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.346915617734707e-06,
+      "loss": 3.1228,
+      "step": 6522
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.304284531594819e-06,
+      "loss": 3.1044,
+      "step": 6523
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.261761679844327e-06,
+      "loss": 3.1756,
+      "step": 6524
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.21934707184352e-06,
+      "loss": 3.2662,
+      "step": 6525
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.177040716928986e-06,
+      "loss": 3.1271,
+      "step": 6526
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.134842624413385e-06,
+      "loss": 3.1252,
+      "step": 6527
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.092752803585512e-06,
+      "loss": 3.1733,
+      "step": 6528
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.0507712637104e-06,
+      "loss": 3.1694,
+      "step": 6529
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.008898014029209e-06,
+      "loss": 3.1329,
+      "step": 6530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.967133063759291e-06,
+      "loss": 3.2065,
+      "step": 6531
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.925476422094124e-06,
+      "loss": 3.2147,
+      "step": 6532
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.883928098203374e-06,
+      "loss": 3.2289,
+      "step": 6533
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.842488101232892e-06,
+      "loss": 3.1734,
+      "step": 6534
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.801156440304657e-06,
+      "loss": 3.1445,
+      "step": 6535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.759933124516727e-06,
+      "loss": 3.0373,
+      "step": 6536
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.718818162943397e-06,
+      "loss": 3.1655,
+      "step": 6537
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.67781156463515e-06,
+      "loss": 3.1298,
+      "step": 6538
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.636913338618379e-06,
+      "loss": 3.151,
+      "step": 6539
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.59612349389599e-06,
+      "loss": 3.1662,
+      "step": 6540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.5554420394467475e-06,
+      "loss": 3.1336,
+      "step": 6541
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.514868984225598e-06,
+      "loss": 3.1456,
+      "step": 6542
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.474404337163731e-06,
+      "loss": 3.1809,
+      "step": 6543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.434048107168523e-06,
+      "loss": 3.07,
+      "step": 6544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3938003031231994e-06,
+      "loss": 3.0673,
+      "step": 6545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.353660933887396e-06,
+      "loss": 3.3123,
+      "step": 6546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3136300082967124e-06,
+      "loss": 3.0501,
+      "step": 6547
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.273707535162988e-06,
+      "loss": 3.1938,
+      "step": 6548
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.233893523274193e-06,
+      "loss": 3.1424,
+      "step": 6549
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.1941879813943176e-06,
+      "loss": 3.1305,
+      "step": 6550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.154590918263482e-06,
+      "loss": 3.2114,
+      "step": 6551
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.115102342598101e-06,
+      "loss": 2.985,
+      "step": 6552
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.075722263090556e-06,
+      "loss": 2.9967,
+      "step": 6553
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.036450688409302e-06,
+      "loss": 3.2987,
+      "step": 6554
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.997287627199034e-06,
+      "loss": 3.1176,
+      "step": 6555
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.9582330880805235e-06,
+      "loss": 2.9769,
+      "step": 6556
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.9192870796506155e-06,
+      "loss": 3.0675,
+      "step": 6557
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.8804496104823425e-06,
+      "loss": 3.1121,
+      "step": 6558
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.841720689124698e-06,
+      "loss": 2.953,
+      "step": 6559
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.803100324102918e-06,
+      "loss": 3.047,
+      "step": 6560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.764588523918314e-06,
+      "loss": 3.0858,
+      "step": 6561
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.726185297048326e-06,
+      "loss": 3.0979,
+      "step": 6562
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.68789065194636e-06,
+      "loss": 3.1084,
+      "step": 6563
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.649704597042061e-06,
+      "loss": 3.1673,
+      "step": 6564
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.611627140741206e-06,
+      "loss": 3.0525,
+      "step": 6565
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.573658291425421e-06,
+      "loss": 3.1141,
+      "step": 6566
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.535798057452691e-06,
+      "loss": 3.0109,
+      "step": 6567
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.498046447156958e-06,
+      "loss": 3.0086,
+      "step": 6568
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.460403468848353e-06,
+      "loss": 3.152,
+      "step": 6569
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.422869130812914e-06,
+      "loss": 3.0007,
+      "step": 6570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.385443441312977e-06,
+      "loss": 3.0666,
+      "step": 6571
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.348126408586841e-06,
+      "loss": 3.1098,
+      "step": 6572
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.310918040848823e-06,
+      "loss": 3.2256,
+      "step": 6573
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.273818346289539e-06,
+      "loss": 3.0961,
+      "step": 6574
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236827333075401e-06,
+      "loss": 3.0497,
+      "step": 6575
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.199945009349173e-06,
+      "loss": 3.0716,
+      "step": 6576
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.163171383229527e-06,
+      "loss": 3.1237,
+      "step": 6577
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12650646281121e-06,
+      "loss": 3.0781,
+      "step": 6578
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.0899502561651554e-06,
+      "loss": 3.1141,
+      "step": 6579
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.053502771338204e-06,
+      "loss": 3.161,
+      "step": 6580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.017164016353438e-06,
+      "loss": 3.1361,
+      "step": 6581
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.980933999209792e-06,
+      "loss": 3.1931,
+      "step": 6582
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.9448127278824986e-06,
+      "loss": 3.1015,
+      "step": 6583
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.908800210322696e-06,
+      "loss": 3.1451,
+      "step": 6584
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.872896454457655e-06,
+      "loss": 3.1517,
+      "step": 6585
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.837101468190609e-06,
+      "loss": 3.0217,
+      "step": 6586
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.8014152594010324e-06,
+      "loss": 3.0352,
+      "step": 6587
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.76583783594431e-06,
+      "loss": 3.0468,
+      "step": 6588
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.730369205651842e-06,
+      "loss": 3.2401,
+      "step": 6589
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.695009376331217e-06,
+      "loss": 3.0762,
+      "step": 6590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.659758355765987e-06,
+      "loss": 3.1973,
+      "step": 6591
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.624616151715834e-06,
+      "loss": 3.0739,
+      "step": 6592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.589582771916291e-06,
+      "loss": 3.1586,
+      "step": 6593
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.5546582240791345e-06,
+      "loss": 3.0276,
+      "step": 6594
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.51984251589216e-06,
+      "loss": 3.2684,
+      "step": 6595
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.48513565501918e-06,
+      "loss": 3.1695,
+      "step": 6596
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.450537649099918e-06,
+      "loss": 3.2277,
+      "step": 6597
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.416048505750393e-06,
+      "loss": 3.1205,
+      "step": 6598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.38166823256242e-06,
+      "loss": 3.0206,
+      "step": 6599
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.347396837104057e-06,
+      "loss": 3.1656,
+      "step": 6600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.313234326919158e-06,
+      "loss": 3.0998,
+      "step": 6601
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.279180709527765e-06,
+      "loss": 3.1064,
+      "step": 6602
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.245235992425934e-06,
+      "loss": 3.1081,
+      "step": 6603
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.211400183085746e-06,
+      "loss": 3.1318,
+      "step": 6604
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.177673288955353e-06,
+      "loss": 3.0762,
+      "step": 6605
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.1440553174588174e-06,
+      "loss": 3.1076,
+      "step": 6606
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.110546275996275e-06,
+      "loss": 3.1163,
+      "step": 6607
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.077146171943936e-06,
+      "loss": 3.1445,
+      "step": 6608
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.0438550126539755e-06,
+      "loss": 3.1288,
+      "step": 6609
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.010672805454586e-06,
+      "loss": 3.1066,
+      "step": 6610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.977599557649981e-06,
+      "loss": 3.004,
+      "step": 6611
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.944635276520393e-06,
+      "loss": 3.1546,
+      "step": 6612
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.911779969322127e-06,
+      "loss": 3.2847,
+      "step": 6613
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.879033643287456e-06,
+      "loss": 3.2426,
+      "step": 6614
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.846396305624612e-06,
+      "loss": 3.0965,
+      "step": 6615
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.813867963517904e-06,
+      "loss": 3.1232,
+      "step": 6616
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.7814486241276045e-06,
+      "loss": 3.0834,
+      "step": 6617
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.749138294590005e-06,
+      "loss": 3.1261,
+      "step": 6618
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.716936982017472e-06,
+      "loss": 3.0876,
+      "step": 6619
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.684844693498225e-06,
+      "loss": 3.309,
+      "step": 6620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.652861436096556e-06,
+      "loss": 3.1672,
+      "step": 6621
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.62098721685289e-06,
+      "loss": 3.1924,
+      "step": 6622
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.589222042783447e-06,
+      "loss": 3.1482,
+      "step": 6623
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.55756592088058e-06,
+      "loss": 3.2945,
+      "step": 6624
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.526018858112546e-06,
+      "loss": 3.1823,
+      "step": 6625
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.49458086142357e-06,
+      "loss": 3.1074,
+      "step": 6626
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.463251937734059e-06,
+      "loss": 3.2325,
+      "step": 6627
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.43203209394022e-06,
+      "loss": 3.0032,
+      "step": 6628
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.400921336914276e-06,
+      "loss": 3.1205,
+      "step": 6629
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.369919673504585e-06,
+      "loss": 3.0627,
+      "step": 6630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.339027110535298e-06,
+      "loss": 3.1729,
+      "step": 6631
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.308243654806643e-06,
+      "loss": 3.0825,
+      "step": 6632
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.277569313094809e-06,
+      "loss": 2.9559,
+      "step": 6633
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.247004092152007e-06,
+      "loss": 3.1059,
+      "step": 6634
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.216547998706355e-06,
+      "loss": 3.069,
+      "step": 6635
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.186201039462045e-06,
+      "loss": 3.1254,
+      "step": 6636
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.155963221099124e-06,
+      "loss": 3.0965,
+      "step": 6637
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.125834550273766e-06,
+      "loss": 3.1988,
+      "step": 6638
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.095815033618e-06,
+      "loss": 3.0437,
+      "step": 6639
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.065904677739873e-06,
+      "loss": 3.048,
+      "step": 6640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.036103489223397e-06,
+      "loss": 3.1192,
+      "step": 6641
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.006411474628491e-06,
+      "loss": 3.0545,
+      "step": 6642
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.976828640491203e-06,
+      "loss": 3.0196,
+      "step": 6643
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.947354993323326e-06,
+      "loss": 3.1585,
+      "step": 6644
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.917990539612892e-06,
+      "loss": 3.1762,
+      "step": 6645
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.888735285823564e-06,
+      "loss": 3.1073,
+      "step": 6646
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.8595892383953005e-06,
+      "loss": 3.1317,
+      "step": 6647
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.830552403743803e-06,
+      "loss": 3.059,
+      "step": 6648
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.8016247882607937e-06,
+      "loss": 3.126,
+      "step": 6649
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7728063983139547e-06,
+      "loss": 3.0564,
+      "step": 6650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7440972402469355e-06,
+      "loss": 3.046,
+      "step": 6651
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.715497320379346e-06,
+      "loss": 3.2579,
+      "step": 6652
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.6870066450067075e-06,
+      "loss": 3.0391,
+      "step": 6653
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.6586252204005577e-06,
+      "loss": 3.1155,
+      "step": 6654
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.6303530528082883e-06,
+      "loss": 3.1711,
+      "step": 6655
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.6021901484533658e-06,
+      "loss": 3.0661,
+      "step": 6656
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5741365135351643e-06,
+      "loss": 3.1122,
+      "step": 6657
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5461921542288558e-06,
+      "loss": 3.1051,
+      "step": 6658
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518357076685852e-06,
+      "loss": 2.994,
+      "step": 6659
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4906312870331967e-06,
+      "loss": 3.1956,
+      "step": 6660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4630147913741193e-06,
+      "loss": 3.0257,
+      "step": 6661
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.435507595787646e-06,
+      "loss": 3.2278,
+      "step": 6662
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.408109706328766e-06,
+      "loss": 3.0691,
+      "step": 6663
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3808211290284885e-06,
+      "loss": 3.0541,
+      "step": 6664
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.353641869893731e-06,
+      "loss": 3.1371,
+      "step": 6665
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.326571934907263e-06,
+      "loss": 3.0868,
+      "step": 6666
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2996113300278186e-06,
+      "loss": 3.2443,
+      "step": 6667
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2727600611901497e-06,
+      "loss": 3.0647,
+      "step": 6668
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2460181343048623e-06,
+      "loss": 3.2229,
+      "step": 6669
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.219385555258525e-06,
+      "loss": 3.2333,
+      "step": 6670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1928623299136152e-06,
+      "loss": 3.1131,
+      "step": 6671
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.166448464108629e-06,
+      "loss": 3.1104,
+      "step": 6672
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1401439636577488e-06,
+      "loss": 3.1835,
+      "step": 6673
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1139488343513987e-06,
+      "loss": 3.1722,
+      "step": 6674
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0878630819556886e-06,
+      "loss": 3.1088,
+      "step": 6675
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.061886712212747e-06,
+      "loss": 3.1898,
+      "step": 6676
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0360197308406667e-06,
+      "loss": 3.2285,
+      "step": 6677
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0102621435333934e-06,
+      "loss": 3.1814,
+      "step": 6678
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.98461395596078e-06,
+      "loss": 3.0481,
+      "step": 6679
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.959075173768588e-06,
+      "loss": 2.9798,
+      "step": 6680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.933645802578655e-06,
+      "loss": 3.1877,
+      "step": 6681
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.9083258479885023e-06,
+      "loss": 3.1271,
+      "step": 6682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.883115315571727e-06,
+      "loss": 3.1703,
+      "step": 6683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.858014210877835e-06,
+      "loss": 3.0959,
+      "step": 6684
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.833022539432073e-06,
+      "loss": 3.1577,
+      "step": 6685
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8081403067358736e-06,
+      "loss": 3.0853,
+      "step": 6686
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7833675182663e-06,
+      "loss": 3.0689,
+      "step": 6687
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.758704179476601e-06,
+      "loss": 3.2256,
+      "step": 6688
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7341502957956564e-06,
+      "loss": 3.1787,
+      "step": 6689
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7097058726284207e-06,
+      "loss": 3.1872,
+      "step": 6690
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6853709153557003e-06,
+      "loss": 3.1374,
+      "step": 6691
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.661145429334322e-06,
+      "loss": 3.1215,
+      "step": 6692
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.637029419896797e-06,
+      "loss": 3.0955,
+      "step": 6693
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.613022892351713e-06,
+      "loss": 3.107,
+      "step": 6694
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.589125851983509e-06,
+      "loss": 3.2329,
+      "step": 6695
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5653383040524227e-06,
+      "loss": 3.164,
+      "step": 6696
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5416602537948196e-06,
+      "loss": 3.1893,
+      "step": 6697
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.518091706422698e-06,
+      "loss": 3.0646,
+      "step": 6698
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.494632667124186e-06,
+      "loss": 3.1814,
+      "step": 6699
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4712831410630964e-06,
+      "loss": 3.0469,
+      "step": 6700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.448043133379374e-06,
+      "loss": 2.9907,
+      "step": 6701
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.424912649188593e-06,
+      "loss": 3.115,
+      "step": 6702
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4018916935823475e-06,
+      "loss": 3.1553,
+      "step": 6703
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.378980271628195e-06,
+      "loss": 3.1778,
+      "step": 6704
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3561783883694897e-06,
+      "loss": 3.1571,
+      "step": 6705
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3334860488254395e-06,
+      "loss": 3.1781,
+      "step": 6706
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.310903257991215e-06,
+      "loss": 3.1866,
+      "step": 6707
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.2884300208378393e-06,
+      "loss": 3.1239,
+      "step": 6708
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.2660663423123005e-06,
+      "loss": 3.0597,
+      "step": 6709
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.243812227337272e-06,
+      "loss": 3.1686,
+      "step": 6710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.2216676808115566e-06,
+      "loss": 3.0893,
+      "step": 6711
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.1996327076096446e-06,
+      "loss": 3.0646,
+      "step": 6712
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.177707312581989e-06,
+      "loss": 3.0691,
+      "step": 6713
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.1558915005548964e-06,
+      "loss": 3.0052,
+      "step": 6714
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.1341852763306357e-06,
+      "loss": 3.2306,
+      "step": 6715
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.112588644687219e-06,
+      "loss": 3.0507,
+      "step": 6716
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0911016103786207e-06,
+      "loss": 3.0642,
+      "step": 6717
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.069724178134613e-06,
+      "loss": 3.0167,
+      "step": 6718
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0484563526609877e-06,
+      "loss": 3.0788,
+      "step": 6719
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.027298138639333e-06,
+      "loss": 3.1587,
+      "step": 6720
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.006249540726979e-06,
+      "loss": 3.1517,
+      "step": 6721
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.98531056355733e-06,
+      "loss": 3.1374,
+      "step": 6722
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9644812117395327e-06,
+      "loss": 3.1044,
+      "step": 6723
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.943761489858642e-06,
+      "loss": 3.0862,
+      "step": 6724
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.92315140247562e-06,
+      "loss": 3.0958,
+      "step": 6725
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9026509541272275e-06,
+      "loss": 3.0533,
+      "step": 6726
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8822601493261315e-06,
+      "loss": 3.114,
+      "step": 6727
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8619789925608533e-06,
+      "loss": 3.0877,
+      "step": 6728
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8418074882958213e-06,
+      "loss": 3.1191,
+      "step": 6729
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8217456409711508e-06,
+      "loss": 3.1173,
+      "step": 6730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8017934550030867e-06,
+      "loss": 3.0344,
+      "step": 6731
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.781950934783505e-06,
+      "loss": 3.1849,
+      "step": 6732
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7622180846803004e-06,
+      "loss": 3.0601,
+      "step": 6733
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7425949090371097e-06,
+      "loss": 3.0951,
+      "step": 6734
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7230814121735327e-06,
+      "loss": 3.0339,
+      "step": 6735
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7036775983849673e-06,
+      "loss": 3.1062,
+      "step": 6736
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6843834719426077e-06,
+      "loss": 3.1881,
+      "step": 6737
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6651990370936676e-06,
+      "loss": 3.0898,
+      "step": 6738
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.646124298061047e-06,
+      "loss": 3.1736,
+      "step": 6739
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6271592590435536e-06,
+      "loss": 3.0992,
+      "step": 6740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6083039242159036e-06,
+      "loss": 3.1851,
+      "step": 6741
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.589558297728555e-06,
+      "loss": 3.0764,
+      "step": 6742
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5709223837079845e-06,
+      "loss": 3.1067,
+      "step": 6743
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5523961862564107e-06,
+      "loss": 3.1616,
+      "step": 6744
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5339797094517939e-06,
+      "loss": 3.0898,
+      "step": 6745
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5156729573481687e-06,
+      "loss": 3.2208,
+      "step": 6746
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.497475933975312e-06,
+      "loss": 3.0507,
+      "step": 6747
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4793886433387416e-06,
+      "loss": 3.0989,
+      "step": 6748
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4614110894199394e-06,
+      "loss": 3.1607,
+      "step": 6749
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4435432761762956e-06,
+      "loss": 3.2414,
+      "step": 6750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.4257852075408307e-06,
+      "loss": 3.0513,
+      "step": 6751
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.4081368874226396e-06,
+      "loss": 3.1658,
+      "step": 6752
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905983197065597e-06,
+      "loss": 2.9693,
+      "step": 6753
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3731695082531692e-06,
+      "loss": 3.1665,
+      "step": 6754
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.355850456899066e-06,
+      "loss": 3.0752,
+      "step": 6755
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3386411694565892e-06,
+      "loss": 3.0448,
+      "step": 6756
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3215416497138755e-06,
+      "loss": 2.961,
+      "step": 6757
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.304551901435025e-06,
+      "loss": 3.0816,
+      "step": 6758
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2876719283598236e-06,
+      "loss": 3.2536,
+      "step": 6759
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2709017342039663e-06,
+      "loss": 3.1568,
+      "step": 6760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.254241322659111e-06,
+      "loss": 3.1017,
+      "step": 6761
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2376906973924906e-06,
+      "loss": 3.2607,
+      "step": 6762
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2212498620474132e-06,
+      "loss": 3.1339,
+      "step": 6763
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.204918820242873e-06,
+      "loss": 3.1992,
+      "step": 6764
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1886975755736606e-06,
+      "loss": 3.1328,
+      "step": 6765
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1725861316105869e-06,
+      "loss": 3.1167,
+      "step": 6766
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1565844919001479e-06,
+      "loss": 3.113,
+      "step": 6767
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1406926599646372e-06,
+      "loss": 3.0854,
+      "step": 6768
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1249106393023124e-06,
+      "loss": 3.0557,
+      "step": 6769
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1092384333871719e-06,
+      "loss": 3.1149,
+      "step": 6770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.093676045669012e-06,
+      "loss": 3.1887,
+      "step": 6771
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0782234795735369e-06,
+      "loss": 3.1836,
+      "step": 6772
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.062880738502303e-06,
+      "loss": 3.0745,
+      "step": 6773
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0476478258324984e-06,
+      "loss": 3.0747,
+      "step": 6774
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.032524744917329e-06,
+      "loss": 3.0798,
+      "step": 6775
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0175114990857438e-06,
+      "loss": 3.1467,
+      "step": 6776
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0026080916425428e-06,
+      "loss": 3.1378,
+      "step": 6777
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.878145258683802e-07,
+      "loss": 3.0692,
+      "step": 6778
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.731308050195954e-07,
+      "loss": 3.2619,
+      "step": 6779
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.585569323284915e-07,
+      "loss": 3.1351,
+      "step": 6780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.440929110031138e-07,
+      "loss": 3.068,
+      "step": 6781
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.297387442273597e-07,
+      "loss": 3.0807,
+      "step": 6782
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.15494435160924e-07,
+      "loss": 3.0431,
+      "step": 6783
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.013599869394096e-07,
+      "loss": 3.0765,
+      "step": 6784
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.873354026740499e-07,
+      "loss": 3.1644,
+      "step": 6785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.734206854520422e-07,
+      "loss": 3.1755,
+      "step": 6786
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.59615838336436e-07,
+      "loss": 3.0669,
+      "step": 6787
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.459208643659122e-07,
+      "loss": 2.9631,
+      "step": 6788
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.323357665551146e-07,
+      "loss": 3.0475,
+      "step": 6789
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.188605478944289e-07,
+      "loss": 3.0898,
+      "step": 6790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.054952113501491e-07,
+      "loss": 3.1258,
+      "step": 6791
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.92239759864255e-07,
+      "loss": 3.1756,
+      "step": 6792
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.79094196354635e-07,
+      "loss": 3.2198,
+      "step": 6793
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.660585237149742e-07,
+      "loss": 3.2482,
+      "step": 6794
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.531327448146441e-07,
+      "loss": 3.1899,
+      "step": 6795
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.403168624990353e-07,
+      "loss": 2.9881,
+      "step": 6796
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.276108795892245e-07,
+      "loss": 3.1387,
+      "step": 6797
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.150147988820854e-07,
+      "loss": 3.1051,
+      "step": 6798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.025286231502892e-07,
+      "loss": 3.3018,
+      "step": 6799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.901523551424705e-07,
+      "loss": 3.2559,
+      "step": 6800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.778859975828388e-07,
+      "loss": 3.2025,
+      "step": 6801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.657295531715679e-07,
+      "loss": 3.1552,
+      "step": 6802
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.536830245845726e-07,
+      "loss": 3.0959,
+      "step": 6803
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.417464144736207e-07,
+      "loss": 3.1548,
+      "step": 6804
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.299197254662214e-07,
+      "loss": 3.0946,
+      "step": 6805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.182029601657368e-07,
+      "loss": 3.0277,
+      "step": 6806
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.06596121151326e-07,
+      "loss": 3.0718,
+      "step": 6807
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.950992109779452e-07,
+      "loss": 3.1978,
+      "step": 6808
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.837122321763477e-07,
+      "loss": 3.0582,
+      "step": 6809
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.724351872530842e-07,
+      "loss": 3.1567,
+      "step": 6810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612680786905023e-07,
+      "loss": 3.3195,
+      "step": 6811
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.502109089467466e-07,
+      "loss": 3.101,
+      "step": 6812
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.392636804557593e-07,
+      "loss": 3.1405,
+      "step": 6813
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.284263956273904e-07,
+      "loss": 3.2588,
+      "step": 6814
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.176990568471207e-07,
+      "loss": 3.0991,
+      "step": 6815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.070816664762834e-07,
+      "loss": 3.1159,
+      "step": 6816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9657422685212e-07,
+      "loss": 3.0886,
+      "step": 6817
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.861767402874473e-07,
+      "loss": 3.0876,
+      "step": 6818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.758892090711009e-07,
+      "loss": 3.0841,
+      "step": 6819
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.657116354676583e-07,
+      "loss": 2.9826,
+      "step": 6820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.556440217173274e-07,
+      "loss": 3.1118,
+      "step": 6821
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.456863700363356e-07,
+      "loss": 3.1408,
+      "step": 6822
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3583868261654057e-07,
+      "loss": 3.197,
+      "step": 6823
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2610096162576383e-07,
+      "loss": 3.1698,
+      "step": 6824
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.1647320920740196e-07,
+      "loss": 2.9909,
+      "step": 6825
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0695542748081515e-07,
+      "loss": 3.0803,
+      "step": 6826
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.975476185411608e-07,
+      "loss": 3.0492,
+      "step": 6827
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.8824978445922697e-07,
+      "loss": 2.9804,
+      "step": 6828
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7906192728176525e-07,
+      "loss": 3.1853,
+      "step": 6829
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6998404903121343e-07,
+      "loss": 3.1189,
+      "step": 6830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.610161517058619e-07,
+      "loss": 3.1471,
+      "step": 6831
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.521582372797427e-07,
+      "loss": 3.1151,
+      "step": 6832
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434103077027406e-07,
+      "loss": 3.1341,
+      "step": 6833
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.347723649004264e-07,
+      "loss": 3.1743,
+      "step": 6834
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2624441077433454e-07,
+      "loss": 3.1439,
+      "step": 6835
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.178264472015191e-07,
+      "loss": 3.1221,
+      "step": 6836
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0951847603516437e-07,
+      "loss": 3.0627,
+      "step": 6837
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.013204991038632e-07,
+      "loss": 3.0999,
+      "step": 6838
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9323251821239406e-07,
+      "loss": 3.1567,
+      "step": 6839
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.852545351409996e-07,
+      "loss": 3.0508,
+      "step": 6840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.773865516458307e-07,
+      "loss": 3.2437,
+      "step": 6841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6962856945883516e-07,
+      "loss": 3.1639,
+      "step": 6842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6198059028781363e-07,
+      "loss": 3.188,
+      "step": 6843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.544426158161972e-07,
+      "loss": 3.2085,
+      "step": 6844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4701464770326976e-07,
+      "loss": 3.0681,
+      "step": 6845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.396966875841677e-07,
+      "loss": 2.9869,
+      "step": 6846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3248873706971373e-07,
+      "loss": 3.0449,
+      "step": 6847
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.25390797746583e-07,
+      "loss": 3.1544,
+      "step": 6848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1840287117713686e-07,
+      "loss": 3.0971,
+      "step": 6849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1152495889970036e-07,
+      "loss": 3.0133,
+      "step": 6850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0475706242822913e-07,
+      "loss": 3.162,
+      "step": 6851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9809918325247589e-07,
+      "loss": 3.0428,
+      "step": 6852
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.915513228380461e-07,
+      "loss": 3.2029,
+      "step": 6853
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8511348262623128e-07,
+      "loss": 3.1264,
+      "step": 6854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7878566403417563e-07,
+      "loss": 3.0131,
+      "step": 6855
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7256786845482041e-07,
+      "loss": 3.1825,
+      "step": 6856
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6646009725684864e-07,
+      "loss": 3.12,
+      "step": 6857
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6046235178474034e-07,
+      "loss": 3.0983,
+      "step": 6858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5457463335871725e-07,
+      "loss": 3.0628,
+      "step": 6859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.487969432747982e-07,
+      "loss": 3.1142,
+      "step": 6860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.431292828048547e-07,
+      "loss": 3.0811,
+      "step": 6861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3757165319644438e-07,
+      "loss": 3.0236,
+      "step": 6862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3212405567292195e-07,
+      "loss": 3.1595,
+      "step": 6863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2678649143349485e-07,
+      "loss": 3.1628,
+      "step": 6864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2155896165300107e-07,
+      "loss": 3.0549,
+      "step": 6865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1644146748224226e-07,
+      "loss": 3.0374,
+      "step": 6866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1143401004765075e-07,
+      "loss": 2.98,
+      "step": 6867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0653659045156694e-07,
+      "loss": 3.0992,
+      "step": 6868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0174920977190638e-07,
+      "loss": 3.0933,
+      "step": 6869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.707186906254827e-08,
+      "loss": 3.2008,
+      "step": 6870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.250456935316898e-08,
+      "loss": 3.1931,
+      "step": 6871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.804731164901991e-08,
+      "loss": 3.0202,
+      "step": 6872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.37000969313162e-08,
+      "loss": 3.1126,
+      "step": 6873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.946292615701456e-08,
+      "loss": 3.1199,
+      "step": 6874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.533580025875785e-08,
+      "loss": 3.0864,
+      "step": 6875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.131872014509711e-08,
+      "loss": 3.2039,
+      "step": 6876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.741168670021391e-08,
+      "loss": 3.1242,
+      "step": 6877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.361470078419806e-08,
+      "loss": 3.1363,
+      "step": 6878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.992776323282545e-08,
+      "loss": 3.0905,
+      "step": 6879
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.635087485772461e-08,
+      "loss": 3.2045,
+      "step": 6880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.288403644626572e-08,
+      "loss": 3.2134,
+      "step": 6881
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.952724876150505e-08,
+      "loss": 3.1003,
+      "step": 6882
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.628051254240706e-08,
+      "loss": 3.0644,
+      "step": 6883
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.314382850362231e-08,
+      "loss": 3.2093,
+      "step": 6884
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.011719733570951e-08,
+      "loss": 3.1352,
+      "step": 6885
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.720061970480249e-08,
+      "loss": 3.107,
+      "step": 6886
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.439409625294321e-08,
+      "loss": 3.2134,
+      "step": 6887
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.169762759797079e-08,
+      "loss": 3.1457,
+      "step": 6888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9111214333354952e-08,
+      "loss": 3.094,
+      "step": 6889
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.663485702847357e-08,
+      "loss": 3.2528,
+      "step": 6890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4268556228446147e-08,
+      "loss": 3.1685,
+      "step": 6891
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2012312454133822e-08,
+      "loss": 3.0823,
+      "step": 6892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9866126202250366e-08,
+      "loss": 3.0776,
+      "step": 6893
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7829997945084665e-08,
+      "loss": 3.1803,
+      "step": 6894
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5903928131000278e-08,
+      "loss": 3.1332,
+      "step": 6895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.408791718382485e-08,
+      "loss": 3.1009,
+      "step": 6896
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2381965503460712e-08,
+      "loss": 3.1051,
+      "step": 6897
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0786073465274271e-08,
+      "loss": 3.139,
+      "step": 6898
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.300241420706623e-09,
+      "loss": 3.0286,
+      "step": 6899
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.92446969671845e-09,
+      "loss": 3.1334,
+      "step": 6900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.658758596178594e-09,
+      "loss": 3.2288,
+      "step": 6901
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.503108397753031e-09,
+      "loss": 3.1797,
+      "step": 6902
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.457519355738349e-09,
+      "loss": 3.0901,
+      "step": 6903
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.5219917003948e-09,
+      "loss": 3.0597,
+      "step": 6904
+    },
+    {
+      "epoch": 2.0,
+      "step": 6904,
+      "total_flos": 8.419903045579571e+16,
+      "train_loss": 3.300872454810391,
+      "train_runtime": 32006.6725,
+      "train_samples_per_second": 13.809,
+      "train_steps_per_second": 0.216
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6904,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 2000,
+  "total_flos": 8.419903045579571e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}