diff --git "a/checkpoint-12610/trainer_state.json" "b/checkpoint-12610/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-12610/trainer_state.json"
@@ -0,0 +1,75769 @@
+{
+  "best_metric": 1.7273346185684204,
+  "best_model_checkpoint": "friends-emotions-detection-distilbert/checkpoint-12610",
+  "epoch": 10.0,
+  "eval_steps": 1,
+  "global_step": 12610,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 1.9939,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 1.9856,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-08,
+      "loss": 1.9595,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 2.0589,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-07,
+      "loss": 1.9149,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-07,
+      "loss": 2.1951,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 1.9876,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 2.0001,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 1.9878,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-07,
+      "loss": 1.8335,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1999999999999998e-07,
+      "loss": 1.8948,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-07,
+      "loss": 1.9302,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-07,
+      "loss": 1.9672,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8e-07,
+      "loss": 1.8529,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-07,
+      "loss": 1.9343,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 1.9058,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 1.9241,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 1.7796,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7999999999999996e-07,
+      "loss": 1.8189,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-07,
+      "loss": 1.9471,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1999999999999995e-07,
+      "loss": 2.0079,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3999999999999997e-07,
+      "loss": 2.1171,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6e-07,
+      "loss": 1.8559,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-07,
+      "loss": 1.948,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-07,
+      "loss": 2.0374,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-07,
+      "loss": 1.9085,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-07,
+      "loss": 2.1379,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6e-07,
+      "loss": 2.0649,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-07,
+      "loss": 2.0955,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-07,
+      "loss": 1.9793,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2e-07,
+      "loss": 1.8227,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4e-07,
+      "loss": 1.9413,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6e-07,
+      "loss": 2.0417,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 2.0501,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-07,
+      "loss": 1.9378,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2e-07,
+      "loss": 2.0919,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-07,
+      "loss": 2.0209,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 1.8542,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 1.9246,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8e-07,
+      "loss": 1.9533,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.199999999999999e-07,
+      "loss": 1.8934,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 1.9767,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.599999999999999e-07,
+      "loss": 1.9004,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 1.9053,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-07,
+      "loss": 1.9159,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.2e-07,
+      "loss": 1.917,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.399999999999999e-07,
+      "loss": 2.0859,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.6e-07,
+      "loss": 2.1052,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8e-07,
+      "loss": 2.0266,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-06,
+      "loss": 1.9958,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999602701628923e-07,
+      "loss": 1.9924,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999205403257847e-07,
+      "loss": 1.9535,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998808104886768e-07,
+      "loss": 1.9451,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998410806515692e-07,
+      "loss": 1.9937,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998013508144615e-07,
+      "loss": 1.8513,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997616209773539e-07,
+      "loss": 1.9444,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997218911402462e-07,
+      "loss": 1.9108,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996821613031386e-07,
+      "loss": 1.9823,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99642431466031e-07,
+      "loss": 1.9248,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996027016289233e-07,
+      "loss": 2.0078,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995629717918157e-07,
+      "loss": 1.8844,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99523241954708e-07,
+      "loss": 2.0238,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994835121176004e-07,
+      "loss": 2.1078,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994437822804925e-07,
+      "loss": 1.9282,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99404052443385e-07,
+      "loss": 2.095,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993643226062773e-07,
+      "loss": 1.8654,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993245927691696e-07,
+      "loss": 2.0082,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99284862932062e-07,
+      "loss": 1.9403,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992451330949543e-07,
+      "loss": 1.956,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.992054032578467e-07,
+      "loss": 1.924,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.991656734207388e-07,
+      "loss": 2.0717,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.991259435836312e-07,
+      "loss": 1.9318,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.990862137465235e-07,
+      "loss": 1.9811,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99046483909416e-07,
+      "loss": 2.036,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.990067540723083e-07,
+      "loss": 1.9768,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.989670242352006e-07,
+      "loss": 1.9835,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.989272943980928e-07,
+      "loss": 1.8921,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988875645609851e-07,
+      "loss": 1.9406,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988478347238775e-07,
+      "loss": 1.947,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988081048867698e-07,
+      "loss": 1.8193,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987683750496622e-07,
+      "loss": 1.921,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987286452125546e-07,
+      "loss": 1.902,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98688915375447e-07,
+      "loss": 1.8517,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986491855383393e-07,
+      "loss": 1.8167,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986094557012316e-07,
+      "loss": 1.9099,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98569725864124e-07,
+      "loss": 2.07,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985299960270163e-07,
+      "loss": 1.9212,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984902661899087e-07,
+      "loss": 1.8857,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984505363528009e-07,
+      "loss": 2.0047,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984108065156932e-07,
+      "loss": 1.8163,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983710766785856e-07,
+      "loss": 1.8511,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98331346841478e-07,
+      "loss": 1.9148,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.982916170043703e-07,
+      "loss": 1.8387,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.982518871672626e-07,
+      "loss": 1.8561,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982121573301548e-07,
+      "loss": 2.1154,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981724274930471e-07,
+      "loss": 2.0489,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981326976559395e-07,
+      "loss": 2.0388,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980929678188319e-07,
+      "loss": 1.9023,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980532379817242e-07,
+      "loss": 1.7938,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980135081446166e-07,
+      "loss": 1.8856,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97973778307509e-07,
+      "loss": 2.0962,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979340484704013e-07,
+      "loss": 1.9795,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978943186332934e-07,
+      "loss": 1.9034,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978545887961858e-07,
+      "loss": 1.9769,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978148589590782e-07,
+      "loss": 1.836,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977751291219705e-07,
+      "loss": 1.9086,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977353992848629e-07,
+      "loss": 2.2109,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976956694477552e-07,
+      "loss": 1.7998,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976559396106476e-07,
+      "loss": 1.9399,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9761620977354e-07,
+      "loss": 1.9368,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975764799364323e-07,
+      "loss": 2.0369,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975367500993247e-07,
+      "loss": 1.9413,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974970202622168e-07,
+      "loss": 2.0008,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974572904251092e-07,
+      "loss": 1.8842,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974175605880015e-07,
+      "loss": 1.8761,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973778307508939e-07,
+      "loss": 2.0078,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973381009137862e-07,
+      "loss": 1.9454,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972983710766786e-07,
+      "loss": 1.9756,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972586412395707e-07,
+      "loss": 1.9928,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.97218911402463e-07,
+      "loss": 1.8563,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.971791815653555e-07,
+      "loss": 1.8781,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.971394517282478e-07,
+      "loss": 1.9188,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.970997218911402e-07,
+      "loss": 2.0287,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.970599920540325e-07,
+      "loss": 1.8604,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.970202622169249e-07,
+      "loss": 1.9672,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.969805323798172e-07,
+      "loss": 2.0831,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.969408025427096e-07,
+      "loss": 1.94,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96901072705602e-07,
+      "loss": 1.911,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.968613428684943e-07,
+      "loss": 2.1335,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.968216130313865e-07,
+      "loss": 2.0112,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.967818831942788e-07,
+      "loss": 1.9501,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.967421533571712e-07,
+      "loss": 1.9625,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.967024235200635e-07,
+      "loss": 1.8892,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.96662693682956e-07,
+      "loss": 1.9257,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.966229638458483e-07,
+      "loss": 1.9985,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.965832340087406e-07,
+      "loss": 1.9197,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.965435041716328e-07,
+      "loss": 2.0355,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.965037743345251e-07,
+      "loss": 1.9683,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.964640444974175e-07,
+      "loss": 1.9058,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.964243146603098e-07,
+      "loss": 2.0196,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.963845848232022e-07,
+      "loss": 2.0302,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.963448549860945e-07,
+      "loss": 1.9425,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.963051251489867e-07,
+      "loss": 1.9413,
+      "step": 143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.96265395311879e-07,
+      "loss": 1.9455,
+      "step": 144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.962256654747714e-07,
+      "loss": 1.8606,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.961859356376638e-07,
+      "loss": 2.0125,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.961462058005561e-07,
+      "loss": 1.8996,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.961064759634485e-07,
+      "loss": 2.1354,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.960667461263408e-07,
+      "loss": 2.0735,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.960270162892332e-07,
+      "loss": 2.0049,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.959872864521256e-07,
+      "loss": 1.9037,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.95947556615018e-07,
+      "loss": 2.0549,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.959078267779103e-07,
+      "loss": 1.9128,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.958680969408026e-07,
+      "loss": 1.9445,
+      "step": 154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.958283671036948e-07,
+      "loss": 1.9004,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.957886372665871e-07,
+      "loss": 2.0512,
+      "step": 156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.957489074294795e-07,
+      "loss": 1.9541,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.957091775923718e-07,
+      "loss": 1.9748,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.956694477552642e-07,
+      "loss": 1.9086,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.956297179181566e-07,
+      "loss": 2.0272,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.955899880810487e-07,
+      "loss": 1.9396,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.95550258243941e-07,
+      "loss": 2.0224,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.955105284068334e-07,
+      "loss": 1.8664,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.954707985697258e-07,
+      "loss": 1.9885,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.954310687326181e-07,
+      "loss": 2.086,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.953913388955105e-07,
+      "loss": 1.8709,
+      "step": 166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.953516090584029e-07,
+      "loss": 2.0106,
+      "step": 167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.953118792212952e-07,
+      "loss": 1.9621,
+      "step": 168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.952721493841874e-07,
+      "loss": 2.0775,
+      "step": 169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.952324195470797e-07,
+      "loss": 1.862,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95192689709972e-07,
+      "loss": 1.9014,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951529598728644e-07,
+      "loss": 1.9834,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951132300357568e-07,
+      "loss": 2.0536,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950735001986492e-07,
+      "loss": 1.8459,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950337703615415e-07,
+      "loss": 1.9606,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.949940405244339e-07,
+      "loss": 2.0349,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.949543106873262e-07,
+      "loss": 1.951,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.949145808502186e-07,
+      "loss": 2.085,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.948748510131107e-07,
+      "loss": 1.9694,
+      "step": 179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94835121176003e-07,
+      "loss": 2.0496,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.947953913388954e-07,
+      "loss": 1.9079,
+      "step": 181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.947556615017878e-07,
+      "loss": 1.9124,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.947159316646802e-07,
+      "loss": 1.8977,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946762018275725e-07,
+      "loss": 1.9823,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946364719904647e-07,
+      "loss": 1.8298,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94596742153357e-07,
+      "loss": 1.9642,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945570123162494e-07,
+      "loss": 1.8245,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945172824791417e-07,
+      "loss": 1.955,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94477552642034e-07,
+      "loss": 1.9409,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944378228049265e-07,
+      "loss": 1.9721,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943980929678188e-07,
+      "loss": 1.9786,
+      "step": 191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943583631307112e-07,
+      "loss": 2.0162,
+      "step": 192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943186332936035e-07,
+      "loss": 1.8919,
+      "step": 193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.942789034564959e-07,
+      "loss": 1.9921,
+      "step": 194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.942391736193882e-07,
+      "loss": 1.8131,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.941994437822804e-07,
+      "loss": 1.9646,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.941597139451727e-07,
+      "loss": 1.8317,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.94119984108065e-07,
+      "loss": 1.8939,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.940802542709575e-07,
+      "loss": 2.0341,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.940405244338498e-07,
+      "loss": 2.0678,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.940007945967422e-07,
+      "loss": 2.0643,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.939610647596345e-07,
+      "loss": 2.0298,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.939213349225267e-07,
+      "loss": 2.0577,
+      "step": 203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.93881605085419e-07,
+      "loss": 1.9393,
+      "step": 204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.938418752483114e-07,
+      "loss": 1.8592,
+      "step": 205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.938021454112038e-07,
+      "loss": 1.914,
+      "step": 206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.937624155740961e-07,
+      "loss": 2.0687,
+      "step": 207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.937226857369885e-07,
+      "loss": 1.9992,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.936829558998806e-07,
+      "loss": 1.9916,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.93643226062773e-07,
+      "loss": 1.8721,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.936034962256653e-07,
+      "loss": 1.8727,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.935637663885577e-07,
+      "loss": 1.8998,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.9352403655145e-07,
+      "loss": 1.9624,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.934843067143424e-07,
+      "loss": 1.8723,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.934445768772348e-07,
+      "loss": 2.0029,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.934048470401271e-07,
+      "loss": 2.0338,
+      "step": 216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.933651172030195e-07,
+      "loss": 1.8399,
+      "step": 217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.933253873659118e-07,
+      "loss": 2.0021,
+      "step": 218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.932856575288042e-07,
+      "loss": 2.0594,
+      "step": 219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.932459276916966e-07,
+      "loss": 1.8831,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.932061978545887e-07,
+      "loss": 1.9431,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.93166468017481e-07,
+      "loss": 2.0797,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.931267381803734e-07,
+      "loss": 1.9123,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.930870083432658e-07,
+      "loss": 1.9744,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.930472785061581e-07,
+      "loss": 2.016,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.930075486690505e-07,
+      "loss": 2.0081,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.929678188319426e-07,
+      "loss": 1.8908,
+      "step": 227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.92928088994835e-07,
+      "loss": 1.9197,
+      "step": 228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.928883591577274e-07,
+      "loss": 1.9589,
+      "step": 229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.928486293206197e-07,
+      "loss": 1.9588,
+      "step": 230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.92808899483512e-07,
+      "loss": 1.9038,
+      "step": 231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.927691696464044e-07,
+      "loss": 1.7868,
+      "step": 232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.927294398092968e-07,
+      "loss": 2.0093,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.926897099721891e-07,
+      "loss": 2.0058,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.926499801350813e-07,
+      "loss": 1.9166,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.926102502979736e-07,
+      "loss": 1.9315,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.92570520460866e-07,
+      "loss": 1.9653,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.925307906237584e-07,
+      "loss": 2.0577,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.924910607866507e-07,
+      "loss": 1.9122,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.92451330949543e-07,
+      "loss": 1.8963,
+      "step": 240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.924116011124354e-07,
+      "loss": 1.8411,
+      "step": 241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.923718712753278e-07,
+      "loss": 1.9513,
+      "step": 242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.923321414382201e-07,
+      "loss": 1.8834,
+      "step": 243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.922924116011125e-07,
+      "loss": 2.0184,
+      "step": 244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.922526817640047e-07,
+      "loss": 2.0616,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.92212951926897e-07,
+      "loss": 1.9919,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.921732220897894e-07,
+      "loss": 1.9199,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.921334922526817e-07,
+      "loss": 1.9977,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.92093762415574e-07,
+      "loss": 1.9237,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.920540325784664e-07,
+      "loss": 2.0264,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.920143027413586e-07,
+      "loss": 1.8865,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.91974572904251e-07,
+      "loss": 1.8102,
+      "step": 252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.919348430671433e-07,
+      "loss": 1.9444,
+      "step": 253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.918951132300357e-07,
+      "loss": 1.9662,
+      "step": 254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.91855383392928e-07,
+      "loss": 1.9391,
+      "step": 255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.918156535558204e-07,
+      "loss": 1.869,
+      "step": 256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.917759237187127e-07,
+      "loss": 1.9532,
+      "step": 257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.91736193881605e-07,
+      "loss": 1.8698,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.916964640444975e-07,
+      "loss": 1.9792,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.916567342073898e-07,
+      "loss": 1.9589,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.916170043702822e-07,
+      "loss": 1.8578,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.915772745331743e-07,
+      "loss": 1.9473,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.915375446960667e-07,
+      "loss": 1.8903,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.91497814858959e-07,
+      "loss": 2.0141,
+      "step": 264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.914580850218514e-07,
+      "loss": 1.992,
+      "step": 265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.914183551847437e-07,
+      "loss": 1.9439,
+      "step": 266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.91378625347636e-07,
+      "loss": 1.8903,
+      "step": 267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.913388955105285e-07,
+      "loss": 1.9223,
+      "step": 268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.912991656734206e-07,
+      "loss": 1.7784,
+      "step": 269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.91259435836313e-07,
+      "loss": 1.9361,
+      "step": 270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.912197059992053e-07,
+      "loss": 1.9896,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.911799761620977e-07,
+      "loss": 1.9881,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.9114024632499e-07,
+      "loss": 1.8724,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.911005164878824e-07,
+      "loss": 1.9586,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.910607866507745e-07,
+      "loss": 1.9036,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.91021056813667e-07,
+      "loss": 2.0436,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.909813269765593e-07,
+      "loss": 1.8587,
+      "step": 277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.909415971394516e-07,
+      "loss": 1.9937,
+      "step": 278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.90901867302344e-07,
+      "loss": 1.9465,
+      "step": 279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.908621374652363e-07,
+      "loss": 1.8958,
+      "step": 280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.908224076281287e-07,
+      "loss": 1.8753,
+      "step": 281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.90782677791021e-07,
+      "loss": 1.9564,
+      "step": 282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.907429479539134e-07,
+      "loss": 2.003,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.907032181168058e-07,
+      "loss": 1.9045,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.906634882796981e-07,
+      "loss": 1.8653,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.906237584425905e-07,
+      "loss": 1.9566,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.905840286054826e-07,
+      "loss": 1.9692,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.90544298768375e-07,
+      "loss": 2.0327,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.905045689312673e-07,
+      "loss": 1.877,
+      "step": 289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.904648390941597e-07,
+      "loss": 1.8807,
+      "step": 290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.90425109257052e-07,
+      "loss": 2.0274,
+      "step": 291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.903853794199444e-07,
+      "loss": 1.8738,
+      "step": 292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.903456495828366e-07,
+      "loss": 1.9325,
+      "step": 293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.90305919745729e-07,
+      "loss": 1.938,
+      "step": 294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.902661899086213e-07,
+      "loss": 1.8546,
+      "step": 295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.902264600715136e-07,
+      "loss": 1.8637,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.90186730234406e-07,
+      "loss": 1.8703,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.901470003972983e-07,
+      "loss": 1.8791,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.901072705601907e-07,
+      "loss": 1.8308,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.90067540723083e-07,
+      "loss": 1.8896,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.900278108859752e-07,
+      "loss": 1.8874,
+      "step": 301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.899880810488676e-07,
+      "loss": 1.9858,
+      "step": 302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.8994835121176e-07,
+      "loss": 1.9785,
+      "step": 303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.899086213746523e-07,
+      "loss": 1.9309,
+      "step": 304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.898688915375446e-07,
+      "loss": 1.8182,
+      "step": 305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.89829161700437e-07,
+      "loss": 2.0217,
+      "step": 306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.897894318633294e-07,
+      "loss": 1.8632,
+      "step": 307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.897497020262217e-07,
+      "loss": 1.9521,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.89709972189114e-07,
+      "loss": 1.9477,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.896702423520064e-07,
+      "loss": 1.839,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.896305125148986e-07,
+      "loss": 1.8047,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.89590782677791e-07,
+      "loss": 1.9742,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.895510528406833e-07,
+      "loss": 1.8992,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.895113230035756e-07,
+      "loss": 1.8465,
+      "step": 314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.89471593166468e-07,
+      "loss": 2.1728,
+      "step": 315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.894318633293604e-07,
+      "loss": 1.8934,
+      "step": 316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.893921334922525e-07,
+      "loss": 1.9571,
+      "step": 317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.893524036551449e-07,
+      "loss": 1.8808,
+      "step": 318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.893126738180372e-07,
+      "loss": 2.0435,
+      "step": 319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.892729439809296e-07,
+      "loss": 1.9911,
+      "step": 320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.89233214143822e-07,
+      "loss": 1.8452,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.891934843067143e-07,
+      "loss": 1.7983,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.891537544696067e-07,
+      "loss": 1.9785,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.89114024632499e-07,
+      "loss": 1.9635,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.890742947953914e-07,
+      "loss": 1.8127,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.890345649582837e-07,
+      "loss": 1.9964,
+      "step": 326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.88994835121176e-07,
+      "loss": 1.9381,
+      "step": 327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.889551052840682e-07,
+      "loss": 1.8482,
+      "step": 328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.889153754469606e-07,
+      "loss": 1.7938,
+      "step": 329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.88875645609853e-07,
+      "loss": 2.0034,
+      "step": 330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.888359157727453e-07,
+      "loss": 2.0412,
+      "step": 331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.887961859356377e-07,
+      "loss": 1.8853,
+      "step": 332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.8875645609853e-07,
+      "loss": 1.9114,
+      "step": 333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.887167262614224e-07,
+      "loss": 2.0332,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.886769964243145e-07,
+      "loss": 1.8421,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.886372665872069e-07,
+      "loss": 1.9433,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.885975367500992e-07,
+      "loss": 1.8543,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.885578069129916e-07,
+      "loss": 1.9701,
+      "step": 338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.88518077075884e-07,
+      "loss": 2.0351,
+      "step": 339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.884783472387763e-07,
+      "loss": 1.824,
+      "step": 340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.884386174016685e-07,
+      "loss": 1.9997,
+      "step": 341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.883988875645608e-07,
+      "loss": 1.9146,
+      "step": 342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.883591577274532e-07,
+      "loss": 1.9601,
+      "step": 343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.883194278903455e-07,
+      "loss": 1.9189,
+      "step": 344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.88279698053238e-07,
+      "loss": 2.0264,
+      "step": 345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.882399682161303e-07,
+      "loss": 2.0457,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.882002383790226e-07,
+      "loss": 2.0337,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.88160508541915e-07,
+      "loss": 1.869,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.881207787048073e-07,
+      "loss": 2.015,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.880810488676997e-07,
+      "loss": 1.9904,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.88041319030592e-07,
+      "loss": 1.9945,
+      "step": 351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.880015891934844e-07,
+      "loss": 1.902,
+      "step": 352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.879618593563765e-07,
+      "loss": 1.8769,
+      "step": 353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.87922129519269e-07,
+      "loss": 1.8459,
+      "step": 354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.878823996821613e-07,
+      "loss": 1.8281,
+      "step": 355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.878426698450536e-07,
+      "loss": 1.9995,
+      "step": 356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.87802940007946e-07,
+      "loss": 1.965,
+      "step": 357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.877632101708383e-07,
+      "loss": 1.9413,
+      "step": 358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.877234803337305e-07,
+      "loss": 2.0052,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.876837504966228e-07,
+      "loss": 1.9289,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.876440206595152e-07,
+      "loss": 1.8254,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.876042908224076e-07,
+      "loss": 1.9471,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.875645609853e-07,
+      "loss": 1.8788,
+      "step": 363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.875248311481923e-07,
+      "loss": 1.9451,
+      "step": 364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.874851013110846e-07,
+      "loss": 1.9495,
+      "step": 365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.87445371473977e-07,
+      "loss": 1.9503,
+      "step": 366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.874056416368691e-07,
+      "loss": 1.9974,
+      "step": 367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.873659117997615e-07,
+      "loss": 2.0199,
+      "step": 368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.873261819626538e-07,
+      "loss": 1.9658,
+      "step": 369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.872864521255462e-07,
+      "loss": 1.9672,
+      "step": 370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.872467222884386e-07,
+      "loss": 1.8505,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.87206992451331e-07,
+      "loss": 2.0523,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.871672626142233e-07,
+      "loss": 1.9627,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.871275327771156e-07,
+      "loss": 1.8625,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.87087802940008e-07,
+      "loss": 1.9445,
+      "step": 375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.870480731029004e-07,
+      "loss": 1.8922,
+      "step": 376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.870083432657925e-07,
+      "loss": 1.8766,
+      "step": 377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.869686134286849e-07,
+      "loss": 1.9108,
+      "step": 378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.869288835915772e-07,
+      "loss": 1.9021,
+      "step": 379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.868891537544696e-07,
+      "loss": 2.0169,
+      "step": 380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.86849423917362e-07,
+      "loss": 1.95,
+      "step": 381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.868096940802543e-07,
+      "loss": 1.96,
+      "step": 382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.867699642431464e-07,
+      "loss": 2.0145,
+      "step": 383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.867302344060388e-07,
+      "loss": 1.9322,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.866905045689312e-07,
+      "loss": 1.9244,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.866507747318235e-07,
+      "loss": 1.8778,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.866110448947159e-07,
+      "loss": 1.9744,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.865713150576082e-07,
+      "loss": 1.9286,
+      "step": 388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.865315852205006e-07,
+      "loss": 1.9479,
+      "step": 389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.86491855383393e-07,
+      "loss": 1.9292,
+      "step": 390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.864521255462853e-07,
+      "loss": 2.0158,
+      "step": 391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.864123957091777e-07,
+      "loss": 1.9738,
+      "step": 392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.8637266587207e-07,
+      "loss": 2.0293,
+      "step": 393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.863329360349622e-07,
+      "loss": 1.9132,
+      "step": 394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.862932061978545e-07,
+      "loss": 2.0238,
+      "step": 395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.862534763607469e-07,
+      "loss": 1.8801,
+      "step": 396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.862137465236392e-07,
+      "loss": 1.8887,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.861740166865316e-07,
+      "loss": 1.8154,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.86134286849424e-07,
+      "loss": 1.9637,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.860945570123163e-07,
+      "loss": 1.9022,
+      "step": 400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.860548271752085e-07,
+      "loss": 1.9167,
+      "step": 401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.860150973381008e-07,
+      "loss": 1.8802,
+      "step": 402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.859753675009932e-07,
+      "loss": 1.9748,
+      "step": 403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.859356376638855e-07,
+      "loss": 1.9969,
+      "step": 404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.858959078267779e-07,
+      "loss": 1.9072,
+      "step": 405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.858561779896702e-07,
+      "loss": 1.8214,
+      "step": 406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.858164481525624e-07,
+      "loss": 1.8049,
+      "step": 407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.857767183154547e-07,
+      "loss": 1.8715,
+      "step": 408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.85736988478347e-07,
+      "loss": 1.8524,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.856972586412395e-07,
+      "loss": 1.9336,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.856575288041318e-07,
+      "loss": 2.0815,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.856177989670242e-07,
+      "loss": 1.9225,
+      "step": 412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.855780691299165e-07,
+      "loss": 1.9211,
+      "step": 413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.855383392928089e-07,
+      "loss": 1.8992,
+      "step": 414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.854986094557013e-07,
+      "loss": 1.9675,
+      "step": 415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.854588796185936e-07,
+      "loss": 1.946,
+      "step": 416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.85419149781486e-07,
+      "loss": 2.1467,
+      "step": 417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.853794199443783e-07,
+      "loss": 1.9362,
+      "step": 418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.853396901072705e-07,
+      "loss": 1.9601,
+      "step": 419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.852999602701628e-07,
+      "loss": 1.8867,
+      "step": 420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.852602304330552e-07,
+      "loss": 1.9617,
+      "step": 421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.852205005959475e-07,
+      "loss": 1.9962,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.8518077075884e-07,
+      "loss": 1.9922,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.851410409217323e-07,
+      "loss": 1.9611,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.851013110846244e-07,
+      "loss": 1.9938,
+      "step": 425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.850615812475168e-07,
+      "loss": 1.844,
+      "step": 426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.850218514104091e-07,
+      "loss": 1.9348,
+      "step": 427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.849821215733015e-07,
+      "loss": 1.9566,
+      "step": 428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.849423917361938e-07,
+      "loss": 1.912,
+      "step": 429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.849026618990862e-07,
+      "loss": 2.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.848629320619786e-07,
+      "loss": 1.8951,
+      "step": 431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.84823202224871e-07,
+      "loss": 1.9256,
+      "step": 432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.84783472387763e-07,
+      "loss": 1.9727,
+      "step": 433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.847437425506554e-07,
+      "loss": 1.8219,
+      "step": 434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.847040127135478e-07,
+      "loss": 1.9137,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.846642828764401e-07,
+      "loss": 1.8745,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.846245530393325e-07,
+      "loss": 1.8399,
+      "step": 437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.845848232022248e-07,
+      "loss": 1.9562,
+      "step": 438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.845450933651172e-07,
+      "loss": 1.8382,
+      "step": 439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.845053635280096e-07,
+      "loss": 1.9937,
+      "step": 440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.84465633690902e-07,
+      "loss": 1.8476,
+      "step": 441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.844259038537943e-07,
+      "loss": 1.9028,
+      "step": 442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.843861740166864e-07,
+      "loss": 1.9377,
+      "step": 443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.843464441795788e-07,
+      "loss": 1.939,
+      "step": 444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.843067143424711e-07,
+      "loss": 1.9262,
+      "step": 445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.842669845053635e-07,
+      "loss": 1.9284,
+      "step": 446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.842272546682559e-07,
+      "loss": 1.8766,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.841875248311482e-07,
+      "loss": 1.9771,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.841477949940404e-07,
+      "loss": 1.9376,
+      "step": 449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.841080651569327e-07,
+      "loss": 1.9466,
+      "step": 450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.84068335319825e-07,
+      "loss": 1.9785,
+      "step": 451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.840286054827174e-07,
+      "loss": 1.8211,
+      "step": 452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.839888756456098e-07,
+      "loss": 1.9767,
+      "step": 453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.839491458085021e-07,
+      "loss": 2.0384,
+      "step": 454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.839094159713945e-07,
+      "loss": 1.9907,
+      "step": 455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.838696861342869e-07,
+      "loss": 1.9485,
+      "step": 456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.838299562971792e-07,
+      "loss": 1.9684,
+      "step": 457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.837902264600716e-07,
+      "loss": 1.9107,
+      "step": 458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.83750496622964e-07,
+      "loss": 1.9717,
+      "step": 459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.83710766785856e-07,
+      "loss": 1.9336,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.836710369487484e-07,
+      "loss": 1.8043,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.836313071116408e-07,
+      "loss": 2.0931,
+      "step": 462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.835915772745332e-07,
+      "loss": 1.9039,
+      "step": 463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.835518474374255e-07,
+      "loss": 1.9784,
+      "step": 464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.835121176003179e-07,
+      "loss": 2.0358,
+      "step": 465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.834723877632102e-07,
+      "loss": 1.8741,
+      "step": 466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.834326579261024e-07,
+      "loss": 2.0103,
+      "step": 467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.833929280889947e-07,
+      "loss": 1.9008,
+      "step": 468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.83353198251887e-07,
+      "loss": 1.8339,
+      "step": 469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.833134684147795e-07,
+      "loss": 1.9521,
+      "step": 470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.832737385776718e-07,
+      "loss": 1.8511,
+      "step": 471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.832340087405642e-07,
+      "loss": 1.9972,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.831942789034563e-07,
+      "loss": 1.8603,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.831545490663487e-07,
+      "loss": 1.8678,
+      "step": 474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.83114819229241e-07,
+      "loss": 1.906,
+      "step": 475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.830750893921334e-07,
+      "loss": 1.8767,
+      "step": 476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.830353595550257e-07,
+      "loss": 1.9254,
+      "step": 477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.82995629717918e-07,
+      "loss": 2.0302,
+      "step": 478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.829558998808105e-07,
+      "loss": 1.9694,
+      "step": 479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.829161700437028e-07,
+      "loss": 1.8923,
+      "step": 480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.828764402065952e-07,
+      "loss": 1.798,
+      "step": 481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.828367103694875e-07,
+      "loss": 1.929,
+      "step": 482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.827969805323799e-07,
+      "loss": 2.0036,
+      "step": 483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.827572506952722e-07,
+      "loss": 1.9261,
+      "step": 484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.827175208581644e-07,
+      "loss": 2.0113,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.826777910210568e-07,
+      "loss": 1.9546,
+      "step": 486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.826380611839491e-07,
+      "loss": 1.8594,
+      "step": 487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.825983313468415e-07,
+      "loss": 1.9393,
+      "step": 488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.825586015097338e-07,
+      "loss": 1.9124,
+      "step": 489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.825188716726262e-07,
+      "loss": 1.9541,
+      "step": 490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.824791418355183e-07,
+      "loss": 1.8209,
+      "step": 491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.824394119984107e-07,
+      "loss": 1.9832,
+      "step": 492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.82399682161303e-07,
+      "loss": 1.9283,
+      "step": 493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.823599523241954e-07,
+      "loss": 1.8501,
+      "step": 494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.823202224870878e-07,
+      "loss": 2.0278,
+      "step": 495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.822804926499801e-07,
+      "loss": 1.964,
+      "step": 496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.822407628128725e-07,
+      "loss": 2.0869,
+      "step": 497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.822010329757648e-07,
+      "loss": 1.893,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.82161303138657e-07,
+      "loss": 2.0217,
+      "step": 499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.821215733015493e-07,
+      "loss": 1.8436,
+      "step": 500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.820818434644417e-07,
+      "loss": 1.9225,
+      "step": 501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.82042113627334e-07,
+      "loss": 2.0621,
+      "step": 502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.820023837902264e-07,
+      "loss": 2.0437,
+      "step": 503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.819626539531188e-07,
+      "loss": 1.8886,
+      "step": 504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.819229241160111e-07,
+      "loss": 1.9169,
+      "step": 505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.818831942789035e-07,
+      "loss": 1.9858,
+      "step": 506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.818434644417958e-07,
+      "loss": 1.7866,
+      "step": 507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.818037346046882e-07,
+      "loss": 2.0305,
+      "step": 508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.817640047675803e-07,
+      "loss": 1.9415,
+      "step": 509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.817242749304727e-07,
+      "loss": 1.9046,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.81684545093365e-07,
+      "loss": 1.973,
+      "step": 511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.816448152562574e-07,
+      "loss": 2.0504,
+      "step": 512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.816050854191498e-07,
+      "loss": 1.9361,
+      "step": 513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.815653555820421e-07,
+      "loss": 2.074,
+      "step": 514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.815256257449343e-07,
+      "loss": 1.8996,
+      "step": 515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.814858959078266e-07,
+      "loss": 1.8366,
+      "step": 516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.81446166070719e-07,
+      "loss": 1.8699,
+      "step": 517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.814064362336114e-07,
+      "loss": 2.0343,
+      "step": 518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.813667063965037e-07,
+      "loss": 2.0099,
+      "step": 519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.81326976559396e-07,
+      "loss": 2.0409,
+      "step": 520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.812872467222884e-07,
+      "loss": 1.9059,
+      "step": 521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.812475168851808e-07,
+      "loss": 1.9265,
+      "step": 522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.812077870480731e-07,
+      "loss": 1.9077,
+      "step": 523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.811680572109655e-07,
+      "loss": 1.9027,
+      "step": 524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.811283273738579e-07,
+      "loss": 2.0527,
+      "step": 525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.8108859753675e-07,
+      "loss": 1.8853,
+      "step": 526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.810488676996424e-07,
+      "loss": 1.9094,
+      "step": 527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.810091378625347e-07,
+      "loss": 1.9559,
+      "step": 528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.80969408025427e-07,
+      "loss": 2.0685,
+      "step": 529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.809296781883194e-07,
+      "loss": 1.8992,
+      "step": 530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.808899483512118e-07,
+      "loss": 1.8903,
+      "step": 531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.808502185141042e-07,
+      "loss": 1.9172,
+      "step": 532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.808104886769963e-07,
+      "loss": 1.8177,
+      "step": 533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.807707588398887e-07,
+      "loss": 1.9552,
+      "step": 534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.80731029002781e-07,
+      "loss": 1.8304,
+      "step": 535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.806912991656734e-07,
+      "loss": 1.9456,
+      "step": 536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.806515693285657e-07,
+      "loss": 1.8393,
+      "step": 537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.80611839491458e-07,
+      "loss": 1.9776,
+      "step": 538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.805721096543502e-07,
+      "loss": 1.9408,
+      "step": 539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.805323798172426e-07,
+      "loss": 1.9294,
+      "step": 540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.80492649980135e-07,
+      "loss": 1.9953,
+      "step": 541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.804529201430273e-07,
+      "loss": 1.976,
+      "step": 542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.804131903059197e-07,
+      "loss": 1.796,
+      "step": 543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.80373460468812e-07,
+      "loss": 1.9132,
+      "step": 544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.803337306317044e-07,
+      "loss": 1.9968,
+      "step": 545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.802940007945967e-07,
+      "loss": 1.9197,
+      "step": 546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.80254270957489e-07,
+      "loss": 1.9385,
+      "step": 547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.802145411203815e-07,
+      "loss": 1.9228,
+      "step": 548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.801748112832738e-07,
+      "loss": 2.0309,
+      "step": 549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.801350814461662e-07,
+      "loss": 1.824,
+      "step": 550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.800953516090583e-07,
+      "loss": 1.9874,
+      "step": 551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.800556217719507e-07,
+      "loss": 1.9885,
+      "step": 552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.80015891934843e-07,
+      "loss": 1.9486,
+      "step": 553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.799761620977354e-07,
+      "loss": 1.8808,
+      "step": 554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.799364322606277e-07,
+      "loss": 1.9367,
+      "step": 555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.798967024235201e-07,
+      "loss": 1.8033,
+      "step": 556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.798569725864123e-07,
+      "loss": 1.8431,
+      "step": 557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.798172427493046e-07,
+      "loss": 2.0143,
+      "step": 558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.79777512912197e-07,
+      "loss": 1.9333,
+      "step": 559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.797377830750893e-07,
+      "loss": 1.9387,
+      "step": 560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.796980532379817e-07,
+      "loss": 1.9048,
+      "step": 561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.79658323400874e-07,
+      "loss": 2.1124,
+      "step": 562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.796185935637664e-07,
+      "loss": 2.0698,
+      "step": 563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.795788637266588e-07,
+      "loss": 1.9818,
+      "step": 564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.79539133889551e-07,
+      "loss": 1.9534,
+      "step": 565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.794994040524433e-07,
+      "loss": 1.8784,
+      "step": 566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.794596742153356e-07,
+      "loss": 2.0794,
+      "step": 567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.79419944378228e-07,
+      "loss": 1.7938,
+      "step": 568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.793802145411203e-07,
+      "loss": 1.8349,
+      "step": 569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.793404847040127e-07,
+      "loss": 1.952,
+      "step": 570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.79300754866905e-07,
+      "loss": 1.8962,
+      "step": 571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.792610250297974e-07,
+      "loss": 1.9813,
+      "step": 572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.792212951926898e-07,
+      "loss": 1.9427,
+      "step": 573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.791815653555821e-07,
+      "loss": 1.8765,
+      "step": 574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.791418355184743e-07,
+      "loss": 1.8016,
+      "step": 575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.791021056813666e-07,
+      "loss": 1.8986,
+      "step": 576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.79062375844259e-07,
+      "loss": 1.9325,
+      "step": 577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.790226460071513e-07,
+      "loss": 1.9532,
+      "step": 578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.789829161700437e-07,
+      "loss": 1.8181,
+      "step": 579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.78943186332936e-07,
+      "loss": 1.8846,
+      "step": 580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.789034564958282e-07,
+      "loss": 1.9419,
+      "step": 581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.788637266587206e-07,
+      "loss": 1.871,
+      "step": 582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.78823996821613e-07,
+      "loss": 1.924,
+      "step": 583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.787842669845053e-07,
+      "loss": 1.8543,
+      "step": 584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.787445371473976e-07,
+      "loss": 2.0399,
+      "step": 585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.7870480731029e-07,
+      "loss": 1.9795,
+      "step": 586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.786650774731824e-07,
+      "loss": 1.8735,
+      "step": 587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.786253476360747e-07,
+      "loss": 1.8733,
+      "step": 588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.78585617798967e-07,
+      "loss": 2.0824,
+      "step": 589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.785458879618594e-07,
+      "loss": 2.0089,
+      "step": 590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.785061581247518e-07,
+      "loss": 1.9216,
+      "step": 591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.78466428287644e-07,
+      "loss": 1.8349,
+      "step": 592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.784266984505363e-07,
+      "loss": 1.9309,
+      "step": 593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.783869686134286e-07,
+      "loss": 1.9402,
+      "step": 594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.78347238776321e-07,
+      "loss": 1.8443,
+      "step": 595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.783075089392134e-07,
+      "loss": 1.8755,
+      "step": 596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.782677791021057e-07,
+      "loss": 1.8965,
+      "step": 597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.78228049264998e-07,
+      "loss": 1.8281,
+      "step": 598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.781883194278902e-07,
+      "loss": 1.9863,
+      "step": 599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.781485895907826e-07,
+      "loss": 2.0248,
+      "step": 600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.78108859753675e-07,
+      "loss": 1.8433,
+      "step": 601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.780691299165673e-07,
+      "loss": 1.8771,
+      "step": 602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.780294000794597e-07,
+      "loss": 1.9374,
+      "step": 603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.77989670242352e-07,
+      "loss": 1.7842,
+      "step": 604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.779499404052442e-07,
+      "loss": 1.9321,
+      "step": 605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.779102105681365e-07,
+      "loss": 1.8442,
+      "step": 606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.778704807310289e-07,
+      "loss": 1.9969,
+      "step": 607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.778307508939212e-07,
+      "loss": 1.8927,
+      "step": 608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.777910210568136e-07,
+      "loss": 2.0356,
+      "step": 609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.77751291219706e-07,
+      "loss": 1.942,
+      "step": 610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.777115613825983e-07,
+      "loss": 1.9844,
+      "step": 611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.776718315454907e-07,
+      "loss": 1.9403,
+      "step": 612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.77632101708383e-07,
+      "loss": 1.9433,
+      "step": 613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.775923718712754e-07,
+      "loss": 1.9457,
+      "step": 614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.775526420341677e-07,
+      "loss": 1.9826,
+      "step": 615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.7751291219706e-07,
+      "loss": 1.8582,
+      "step": 616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.774731823599522e-07,
+      "loss": 1.9316,
+      "step": 617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.774334525228446e-07,
+      "loss": 1.9005,
+      "step": 618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.77393722685737e-07,
+      "loss": 1.8678,
+      "step": 619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.773539928486293e-07,
+      "loss": 1.8607,
+      "step": 620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.773142630115217e-07,
+      "loss": 1.8345,
+      "step": 621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.77274533174414e-07,
+      "loss": 1.9142,
+      "step": 622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.772348033373062e-07,
+      "loss": 2.035,
+      "step": 623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.771950735001985e-07,
+      "loss": 1.8536,
+      "step": 624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.771553436630909e-07,
+      "loss": 2.1025,
+      "step": 625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.771156138259833e-07,
+      "loss": 1.8499,
+      "step": 626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.770758839888756e-07,
+      "loss": 1.9898,
+      "step": 627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.77036154151768e-07,
+      "loss": 1.8918,
+      "step": 628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.769964243146603e-07,
+      "loss": 1.8571,
+      "step": 629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.769566944775527e-07,
+      "loss": 1.9408,
+      "step": 630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.769169646404448e-07,
+      "loss": 1.9248,
+      "step": 631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.768772348033372e-07,
+      "loss": 1.9276,
+      "step": 632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.768375049662295e-07,
+      "loss": 1.9575,
+      "step": 633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.76797775129122e-07,
+      "loss": 1.9095,
+      "step": 634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.767580452920143e-07,
+      "loss": 2.0108,
+      "step": 635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.767183154549066e-07,
+      "loss": 2.0013,
+      "step": 636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76678585617799e-07,
+      "loss": 1.8717,
+      "step": 637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.766388557806913e-07,
+      "loss": 2.0514,
+      "step": 638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.765991259435837e-07,
+      "loss": 1.9473,
+      "step": 639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76559396106476e-07,
+      "loss": 1.8329,
+      "step": 640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.765196662693682e-07,
+      "loss": 1.9138,
+      "step": 641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.764799364322606e-07,
+      "loss": 2.0154,
+      "step": 642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76440206595153e-07,
+      "loss": 1.842,
+      "step": 643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.764004767580453e-07,
+      "loss": 1.7989,
+      "step": 644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.763607469209376e-07,
+      "loss": 1.9082,
+      "step": 645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7632101708383e-07,
+      "loss": 1.9519,
+      "step": 646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762812872467221e-07,
+      "loss": 2.0122,
+      "step": 647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762415574096145e-07,
+      "loss": 1.8993,
+      "step": 648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762018275725068e-07,
+      "loss": 1.9115,
+      "step": 649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761620977353992e-07,
+      "loss": 1.9672,
+      "step": 650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761223678982916e-07,
+      "loss": 1.9863,
+      "step": 651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76082638061184e-07,
+      "loss": 2.028,
+      "step": 652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760429082240763e-07,
+      "loss": 1.8824,
+      "step": 653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760031783869686e-07,
+      "loss": 2.0,
+      "step": 654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75963448549861e-07,
+      "loss": 1.9239,
+      "step": 655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759237187127534e-07,
+      "loss": 1.9488,
+      "step": 656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758839888756457e-07,
+      "loss": 1.911,
+      "step": 657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758442590385379e-07,
+      "loss": 2.0497,
+      "step": 658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758045292014302e-07,
+      "loss": 1.9792,
+      "step": 659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757647993643226e-07,
+      "loss": 2.0176,
+      "step": 660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75725069527215e-07,
+      "loss": 1.8196,
+      "step": 661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756853396901073e-07,
+      "loss": 1.8831,
+      "step": 662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.756456098529996e-07,
+      "loss": 1.9924,
+      "step": 663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.756058800158918e-07,
+      "loss": 2.0029,
+      "step": 664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.755661501787841e-07,
+      "loss": 1.8841,
+      "step": 665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.755264203416765e-07,
+      "loss": 2.0399,
+      "step": 666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.754866905045689e-07,
+      "loss": 1.9512,
+      "step": 667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.754469606674612e-07,
+      "loss": 1.9246,
+      "step": 668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.754072308303536e-07,
+      "loss": 1.9055,
+      "step": 669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.753675009932457e-07,
+      "loss": 1.8716,
+      "step": 670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.75327771156138e-07,
+      "loss": 1.9665,
+      "step": 671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.752880413190304e-07,
+      "loss": 1.907,
+      "step": 672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.752483114819228e-07,
+      "loss": 1.986,
+      "step": 673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.752085816448152e-07,
+      "loss": 1.8793,
+      "step": 674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.751688518077075e-07,
+      "loss": 1.7673,
+      "step": 675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.751291219705999e-07,
+      "loss": 1.9529,
+      "step": 676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.750893921334922e-07,
+      "loss": 1.8037,
+      "step": 677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.750496622963846e-07,
+      "loss": 1.8954,
+      "step": 678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.75009932459277e-07,
+      "loss": 1.9507,
+      "step": 679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.749702026221693e-07,
+      "loss": 1.8704,
+      "step": 680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.749304727850617e-07,
+      "loss": 1.9089,
+      "step": 681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.748907429479538e-07,
+      "loss": 2.0412,
+      "step": 682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.748510131108462e-07,
+      "loss": 1.9246,
+      "step": 683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.748112832737385e-07,
+      "loss": 1.8655,
+      "step": 684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.747715534366309e-07,
+      "loss": 1.9129,
+      "step": 685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.747318235995232e-07,
+      "loss": 1.8483,
+      "step": 686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.746920937624156e-07,
+      "loss": 1.964,
+      "step": 687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.746523639253077e-07,
+      "loss": 1.9019,
+      "step": 688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.746126340882e-07,
+      "loss": 1.8555,
+      "step": 689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.745729042510925e-07,
+      "loss": 1.9675,
+      "step": 690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.745331744139848e-07,
+      "loss": 1.9868,
+      "step": 691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.744934445768772e-07,
+      "loss": 1.9312,
+      "step": 692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.744537147397695e-07,
+      "loss": 1.9501,
+      "step": 693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.744139849026619e-07,
+      "loss": 2.0861,
+      "step": 694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.743742550655542e-07,
+      "loss": 1.8651,
+      "step": 695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.743345252284466e-07,
+      "loss": 1.9067,
+      "step": 696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.742947953913388e-07,
+      "loss": 1.9083,
+      "step": 697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.742550655542311e-07,
+      "loss": 1.8829,
+      "step": 698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.742153357171235e-07,
+      "loss": 2.0149,
+      "step": 699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.741756058800158e-07,
+      "loss": 1.9337,
+      "step": 700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.741358760429082e-07,
+      "loss": 1.8083,
+      "step": 701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.740961462058005e-07,
+      "loss": 1.8688,
+      "step": 702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.74056416368693e-07,
+      "loss": 2.0058,
+      "step": 703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.740166865315853e-07,
+      "loss": 1.8016,
+      "step": 704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.739769566944776e-07,
+      "loss": 1.8823,
+      "step": 705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.739372268573698e-07,
+      "loss": 1.9091,
+      "step": 706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.738974970202621e-07,
+      "loss": 1.8745,
+      "step": 707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.738577671831545e-07,
+      "loss": 1.9723,
+      "step": 708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.738180373460468e-07,
+      "loss": 1.8843,
+      "step": 709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.737783075089392e-07,
+      "loss": 2.0583,
+      "step": 710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.737385776718315e-07,
+      "loss": 1.8581,
+      "step": 711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.736988478347237e-07,
+      "loss": 1.9179,
+      "step": 712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.73659117997616e-07,
+      "loss": 2.1619,
+      "step": 713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.736193881605084e-07,
+      "loss": 2.0044,
+      "step": 714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.735796583234008e-07,
+      "loss": 2.0633,
+      "step": 715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.735399284862931e-07,
+      "loss": 1.824,
+      "step": 716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.735001986491855e-07,
+      "loss": 2.0407,
+      "step": 717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.734604688120778e-07,
+      "loss": 1.8118,
+      "step": 718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.734207389749702e-07,
+      "loss": 1.9075,
+      "step": 719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.733810091378626e-07,
+      "loss": 1.8082,
+      "step": 720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.73341279300755e-07,
+      "loss": 2.0126,
+      "step": 721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.733015494636473e-07,
+      "loss": 1.9122,
+      "step": 722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.732618196265396e-07,
+      "loss": 1.873,
+      "step": 723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.732220897894318e-07,
+      "loss": 1.9521,
+      "step": 724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.731823599523241e-07,
+      "loss": 1.8686,
+      "step": 725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.731426301152165e-07,
+      "loss": 1.9086,
+      "step": 726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.731029002781089e-07,
+      "loss": 1.9312,
+      "step": 727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.730631704410012e-07,
+      "loss": 1.908,
+      "step": 728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.730234406038936e-07,
+      "loss": 1.9421,
+      "step": 729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.729837107667857e-07,
+      "loss": 2.0309,
+      "step": 730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.72943980929678e-07,
+      "loss": 1.9476,
+      "step": 731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.729042510925704e-07,
+      "loss": 1.9484,
+      "step": 732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.728645212554628e-07,
+      "loss": 2.0161,
+      "step": 733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.728247914183551e-07,
+      "loss": 1.9348,
+      "step": 734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.727850615812475e-07,
+      "loss": 1.9309,
+      "step": 735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.727453317441396e-07,
+      "loss": 1.8878,
+      "step": 736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.72705601907032e-07,
+      "loss": 1.9483,
+      "step": 737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726658720699244e-07,
+      "loss": 1.8634,
+      "step": 738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726261422328167e-07,
+      "loss": 2.0258,
+      "step": 739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72586412395709e-07,
+      "loss": 1.9943,
+      "step": 740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725466825586014e-07,
+      "loss": 2.0185,
+      "step": 741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725069527214938e-07,
+      "loss": 1.8714,
+      "step": 742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724672228843862e-07,
+      "loss": 2.0003,
+      "step": 743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724274930472785e-07,
+      "loss": 1.9301,
+      "step": 744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723877632101709e-07,
+      "loss": 2.0216,
+      "step": 745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723480333730632e-07,
+      "loss": 1.9642,
+      "step": 746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723083035359556e-07,
+      "loss": 1.8887,
+      "step": 747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.722685736988477e-07,
+      "loss": 1.987,
+      "step": 748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.7222884386174e-07,
+      "loss": 1.9395,
+      "step": 749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.721891140246324e-07,
+      "loss": 1.8998,
+      "step": 750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.721493841875248e-07,
+      "loss": 1.9822,
+      "step": 751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.721096543504172e-07,
+      "loss": 1.914,
+      "step": 752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.720699245133095e-07,
+      "loss": 1.984,
+      "step": 753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.720301946762017e-07,
+      "loss": 1.9198,
+      "step": 754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.71990464839094e-07,
+      "loss": 1.9972,
+      "step": 755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.719507350019864e-07,
+      "loss": 1.9648,
+      "step": 756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.719110051648787e-07,
+      "loss": 1.9498,
+      "step": 757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.71871275327771e-07,
+      "loss": 1.9009,
+      "step": 758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.718315454906635e-07,
+      "loss": 1.8997,
+      "step": 759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.717918156535558e-07,
+      "loss": 1.9118,
+      "step": 760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.717520858164482e-07,
+      "loss": 2.1055,
+      "step": 761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.717123559793405e-07,
+      "loss": 2.0381,
+      "step": 762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.716726261422327e-07,
+      "loss": 1.8458,
+      "step": 763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.71632896305125e-07,
+      "loss": 1.9987,
+      "step": 764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.715931664680174e-07,
+      "loss": 1.9152,
+      "step": 765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.715534366309097e-07,
+      "loss": 2.045,
+      "step": 766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.715137067938021e-07,
+      "loss": 2.0366,
+      "step": 767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.714739769566945e-07,
+      "loss": 1.9909,
+      "step": 768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.714342471195868e-07,
+      "loss": 1.8972,
+      "step": 769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.713945172824792e-07,
+      "loss": 1.9568,
+      "step": 770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.713547874453715e-07,
+      "loss": 1.9665,
+      "step": 771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.713150576082637e-07,
+      "loss": 1.7975,
+      "step": 772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.71275327771156e-07,
+      "loss": 1.8454,
+      "step": 773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.712355979340484e-07,
+      "loss": 1.9061,
+      "step": 774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.711958680969408e-07,
+      "loss": 1.8312,
+      "step": 775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.711561382598331e-07,
+      "loss": 1.8745,
+      "step": 776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.711164084227255e-07,
+      "loss": 1.8234,
+      "step": 777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.710766785856176e-07,
+      "loss": 1.8999,
+      "step": 778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.7103694874851e-07,
+      "loss": 1.8954,
+      "step": 779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.709972189114023e-07,
+      "loss": 1.8238,
+      "step": 780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.709574890742947e-07,
+      "loss": 1.9458,
+      "step": 781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.70917759237187e-07,
+      "loss": 1.9284,
+      "step": 782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.708780294000794e-07,
+      "loss": 2.037,
+      "step": 783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.708382995629718e-07,
+      "loss": 1.9925,
+      "step": 784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.707985697258641e-07,
+      "loss": 2.0499,
+      "step": 785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.707588398887565e-07,
+      "loss": 2.09,
+      "step": 786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.707191100516488e-07,
+      "loss": 1.8208,
+      "step": 787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.706793802145412e-07,
+      "loss": 1.905,
+      "step": 788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.706396503774333e-07,
+      "loss": 1.9434,
+      "step": 789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.705999205403257e-07,
+      "loss": 1.837,
+      "step": 790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.70560190703218e-07,
+      "loss": 2.0141,
+      "step": 791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.705204608661104e-07,
+      "loss": 1.8757,
+      "step": 792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.704807310290028e-07,
+      "loss": 1.7809,
+      "step": 793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.704410011918951e-07,
+      "loss": 1.9743,
+      "step": 794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.704012713547875e-07,
+      "loss": 1.9374,
+      "step": 795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.703615415176796e-07,
+      "loss": 2.0142,
+      "step": 796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.70321811680572e-07,
+      "loss": 1.8585,
+      "step": 797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.702820818434644e-07,
+      "loss": 1.9988,
+      "step": 798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.702423520063567e-07,
+      "loss": 1.8637,
+      "step": 799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.70202622169249e-07,
+      "loss": 1.9345,
+      "step": 800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.701628923321414e-07,
+      "loss": 1.8577,
+      "step": 801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.701231624950336e-07,
+      "loss": 1.9521,
+      "step": 802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.70083432657926e-07,
+      "loss": 1.9863,
+      "step": 803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.700437028208183e-07,
+      "loss": 1.8516,
+      "step": 804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.700039729837106e-07,
+      "loss": 1.9691,
+      "step": 805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.69964243146603e-07,
+      "loss": 1.9986,
+      "step": 806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.699245133094954e-07,
+      "loss": 1.8986,
+      "step": 807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.698847834723877e-07,
+      "loss": 1.919,
+      "step": 808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.6984505363528e-07,
+      "loss": 1.7968,
+      "step": 809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.698053237981724e-07,
+      "loss": 1.853,
+      "step": 810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.697655939610648e-07,
+      "loss": 1.9149,
+      "step": 811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.697258641239572e-07,
+      "loss": 1.8302,
+      "step": 812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.696861342868495e-07,
+      "loss": 2.0495,
+      "step": 813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.696464044497417e-07,
+      "loss": 1.9163,
+      "step": 814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.69606674612634e-07,
+      "loss": 1.9702,
+      "step": 815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.695669447755264e-07,
+      "loss": 1.9897,
+      "step": 816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.695272149384187e-07,
+      "loss": 1.9333,
+      "step": 817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.69487485101311e-07,
+      "loss": 1.8077,
+      "step": 818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.694477552642034e-07,
+      "loss": 1.9988,
+      "step": 819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.694080254270956e-07,
+      "loss": 1.9571,
+      "step": 820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.69368295589988e-07,
+      "loss": 1.9773,
+      "step": 821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.693285657528803e-07,
+      "loss": 2.0275,
+      "step": 822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.692888359157727e-07,
+      "loss": 1.8462,
+      "step": 823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.69249106078665e-07,
+      "loss": 2.0327,
+      "step": 824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.692093762415574e-07,
+      "loss": 1.9071,
+      "step": 825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.691696464044497e-07,
+      "loss": 1.9605,
+      "step": 826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.69129916567342e-07,
+      "loss": 1.8469,
+      "step": 827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.690901867302345e-07,
+      "loss": 1.9554,
+      "step": 828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.690504568931266e-07,
+      "loss": 1.8828,
+      "step": 829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.69010727056019e-07,
+      "loss": 1.9683,
+      "step": 830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.689709972189113e-07,
+      "loss": 1.8495,
+      "step": 831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.689312673818037e-07,
+      "loss": 1.9399,
+      "step": 832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.68891537544696e-07,
+      "loss": 1.9475,
+      "step": 833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.688518077075884e-07,
+      "loss": 1.9245,
+      "step": 834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.688120778704807e-07,
+      "loss": 1.9398,
+      "step": 835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.68772348033373e-07,
+      "loss": 1.8936,
+      "step": 836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.687326181962655e-07,
+      "loss": 1.9261,
+      "step": 837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.686928883591576e-07,
+      "loss": 1.9126,
+      "step": 838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.6865315852205e-07,
+      "loss": 1.7621,
+      "step": 839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.686134286849423e-07,
+      "loss": 1.9036,
+      "step": 840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.685736988478347e-07,
+      "loss": 1.9081,
+      "step": 841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.68533969010727e-07,
+      "loss": 2.0486,
+      "step": 842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.684942391736194e-07,
+      "loss": 1.8845,
+      "step": 843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.684545093365115e-07,
+      "loss": 1.9066,
+      "step": 844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.68414779499404e-07,
+      "loss": 1.9185,
+      "step": 845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.683750496622963e-07,
+      "loss": 2.0187,
+      "step": 846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.683353198251886e-07,
+      "loss": 1.9627,
+      "step": 847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.68295589988081e-07,
+      "loss": 1.8788,
+      "step": 848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.682558601509733e-07,
+      "loss": 1.8734,
+      "step": 849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.682161303138657e-07,
+      "loss": 1.8744,
+      "step": 850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.68176400476758e-07,
+      "loss": 1.8962,
+      "step": 851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.681366706396504e-07,
+      "loss": 1.9549,
+      "step": 852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.680969408025428e-07,
+      "loss": 2.0312,
+      "step": 853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.680572109654351e-07,
+      "loss": 1.7357,
+      "step": 854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.680174811283273e-07,
+      "loss": 1.9939,
+      "step": 855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.679777512912196e-07,
+      "loss": 1.9126,
+      "step": 856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.67938021454112e-07,
+      "loss": 1.8827,
+      "step": 857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.678982916170043e-07,
+      "loss": 1.9234,
+      "step": 858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.678585617798967e-07,
+      "loss": 1.8146,
+      "step": 859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.67818831942789e-07,
+      "loss": 1.8852,
+      "step": 860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.677791021056814e-07,
+      "loss": 1.74,
+      "step": 861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.677393722685736e-07,
+      "loss": 1.9189,
+      "step": 862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.67699642431466e-07,
+      "loss": 1.8943,
+      "step": 863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.676599125943583e-07,
+      "loss": 1.9915,
+      "step": 864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.676201827572506e-07,
+      "loss": 1.9096,
+      "step": 865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.67580452920143e-07,
+      "loss": 1.8632,
+      "step": 866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.675407230830354e-07,
+      "loss": 1.8664,
+      "step": 867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.675009932459275e-07,
+      "loss": 1.961,
+      "step": 868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.674612634088199e-07,
+      "loss": 1.9052,
+      "step": 869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.674215335717122e-07,
+      "loss": 2.0796,
+      "step": 870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.673818037346046e-07,
+      "loss": 1.9263,
+      "step": 871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.67342073897497e-07,
+      "loss": 1.9247,
+      "step": 872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.673023440603893e-07,
+      "loss": 1.9834,
+      "step": 873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.672626142232816e-07,
+      "loss": 2.0157,
+      "step": 874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.67222884386174e-07,
+      "loss": 1.9688,
+      "step": 875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.671831545490664e-07,
+      "loss": 1.9986,
+      "step": 876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.671434247119587e-07,
+      "loss": 2.0432,
+      "step": 877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.67103694874851e-07,
+      "loss": 1.9467,
+      "step": 878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.670639650377434e-07,
+      "loss": 1.924,
+      "step": 879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.670242352006356e-07,
+      "loss": 1.8559,
+      "step": 880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.66984505363528e-07,
+      "loss": 1.9146,
+      "step": 881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.669447755264203e-07,
+      "loss": 2.1362,
+      "step": 882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.669050456893127e-07,
+      "loss": 1.8737,
+      "step": 883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.66865315852205e-07,
+      "loss": 1.9769,
+      "step": 884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.668255860150974e-07,
+      "loss": 1.8719,
+      "step": 885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.667858561779895e-07,
+      "loss": 1.9004,
+      "step": 886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.667461263408819e-07,
+      "loss": 1.9134,
+      "step": 887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.667063965037742e-07,
+      "loss": 2.005,
+      "step": 888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.666666666666666e-07,
+      "loss": 1.8722,
+      "step": 889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.66626936829559e-07,
+      "loss": 2.0813,
+      "step": 890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.665872069924513e-07,
+      "loss": 1.919,
+      "step": 891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.665474771553437e-07,
+      "loss": 2.0255,
+      "step": 892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.66507747318236e-07,
+      "loss": 1.8931,
+      "step": 893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.664680174811284e-07,
+      "loss": 1.8894,
+      "step": 894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.664282876440205e-07,
+      "loss": 1.9931,
+      "step": 895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.663885578069129e-07,
+      "loss": 1.8026,
+      "step": 896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.663488279698052e-07,
+      "loss": 1.9559,
+      "step": 897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.663090981326976e-07,
+      "loss": 1.8643,
+      "step": 898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.6626936829559e-07,
+      "loss": 1.8916,
+      "step": 899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.662296384584823e-07,
+      "loss": 1.916,
+      "step": 900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.661899086213747e-07,
+      "loss": 1.9349,
+      "step": 901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.66150178784267e-07,
+      "loss": 1.9235,
+      "step": 902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.661104489471594e-07,
+      "loss": 1.9373,
+      "step": 903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.660707191100515e-07,
+      "loss": 2.0437,
+      "step": 904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.660309892729439e-07,
+      "loss": 1.8671,
+      "step": 905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.659912594358362e-07,
+      "loss": 1.9461,
+      "step": 906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.659515295987286e-07,
+      "loss": 1.9854,
+      "step": 907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.65911799761621e-07,
+      "loss": 1.9143,
+      "step": 908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.658720699245133e-07,
+      "loss": 1.9802,
+      "step": 909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.658323400874055e-07,
+      "loss": 2.0035,
+      "step": 910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.657926102502978e-07,
+      "loss": 1.9404,
+      "step": 911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.657528804131902e-07,
+      "loss": 1.9221,
+      "step": 912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.657131505760825e-07,
+      "loss": 1.8617,
+      "step": 913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.65673420738975e-07,
+      "loss": 2.0481,
+      "step": 914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.656336909018673e-07,
+      "loss": 1.8811,
+      "step": 915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.655939610647596e-07,
+      "loss": 1.9316,
+      "step": 916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65554231227652e-07,
+      "loss": 2.0133,
+      "step": 917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.655145013905443e-07,
+      "loss": 2.0991,
+      "step": 918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.654747715534367e-07,
+      "loss": 1.8345,
+      "step": 919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65435041716329e-07,
+      "loss": 1.8902,
+      "step": 920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.653953118792212e-07,
+      "loss": 2.0724,
+      "step": 921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.653555820421135e-07,
+      "loss": 1.8676,
+      "step": 922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65315852205006e-07,
+      "loss": 1.8803,
+      "step": 923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.652761223678983e-07,
+      "loss": 1.9201,
+      "step": 924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.652363925307906e-07,
+      "loss": 1.929,
+      "step": 925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65196662693683e-07,
+      "loss": 1.9533,
+      "step": 926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.651569328565753e-07,
+      "loss": 1.8435,
+      "step": 927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.651172030194675e-07,
+      "loss": 1.9674,
+      "step": 928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.650774731823598e-07,
+      "loss": 1.927,
+      "step": 929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.650377433452522e-07,
+      "loss": 1.8825,
+      "step": 930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.649980135081446e-07,
+      "loss": 1.8594,
+      "step": 931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.64958283671037e-07,
+      "loss": 1.8346,
+      "step": 932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.649185538339293e-07,
+      "loss": 1.9859,
+      "step": 933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.648788239968214e-07,
+      "loss": 1.8723,
+      "step": 934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.648390941597138e-07,
+      "loss": 1.9988,
+      "step": 935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.647993643226061e-07,
+      "loss": 1.8978,
+      "step": 936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.647596344854985e-07,
+      "loss": 2.0967,
+      "step": 937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.647199046483909e-07,
+      "loss": 1.9328,
+      "step": 938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.646801748112832e-07,
+      "loss": 1.9099,
+      "step": 939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.646404449741756e-07,
+      "loss": 1.9026,
+      "step": 940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.64600715137068e-07,
+      "loss": 1.9945,
+      "step": 941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.645609852999603e-07,
+      "loss": 1.8932,
+      "step": 942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.645212554628526e-07,
+      "loss": 1.7823,
+      "step": 943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.64481525625745e-07,
+      "loss": 2.0072,
+      "step": 944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.644417957886374e-07,
+      "loss": 1.9903,
+      "step": 945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.644020659515295e-07,
+      "loss": 1.8555,
+      "step": 946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.643623361144219e-07,
+      "loss": 1.8699,
+      "step": 947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.643226062773142e-07,
+      "loss": 1.8776,
+      "step": 948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.642828764402066e-07,
+      "loss": 2.0889,
+      "step": 949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.64243146603099e-07,
+      "loss": 1.985,
+      "step": 950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.642034167659913e-07,
+      "loss": 1.9335,
+      "step": 951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.641636869288834e-07,
+      "loss": 1.9911,
+      "step": 952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.641239570917758e-07,
+      "loss": 2.0708,
+      "step": 953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.640842272546682e-07,
+      "loss": 1.8778,
+      "step": 954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.640444974175605e-07,
+      "loss": 1.8861,
+      "step": 955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.640047675804529e-07,
+      "loss": 1.9959,
+      "step": 956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.639650377433452e-07,
+      "loss": 2.0132,
+      "step": 957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.639253079062376e-07,
+      "loss": 1.9908,
+      "step": 958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.6388557806913e-07,
+      "loss": 1.8323,
+      "step": 959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.638458482320223e-07,
+      "loss": 1.9573,
+      "step": 960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.638061183949144e-07,
+      "loss": 1.9854,
+      "step": 961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.637663885578068e-07,
+      "loss": 1.8609,
+      "step": 962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.637266587206992e-07,
+      "loss": 1.8023,
+      "step": 963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.636869288835915e-07,
+      "loss": 1.8517,
+      "step": 964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.636471990464839e-07,
+      "loss": 1.9025,
+      "step": 965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.636074692093762e-07,
+      "loss": 2.0559,
+      "step": 966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.635677393722686e-07,
+      "loss": 1.8116,
+      "step": 967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.63528009535161e-07,
+      "loss": 1.8679,
+      "step": 968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.634882796980533e-07,
+      "loss": 1.8721,
+      "step": 969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.634485498609455e-07,
+      "loss": 1.9772,
+      "step": 970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.634088200238378e-07,
+      "loss": 1.8603,
+      "step": 971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.633690901867302e-07,
+      "loss": 1.8571,
+      "step": 972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.633293603496225e-07,
+      "loss": 1.8475,
+      "step": 973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.632896305125149e-07,
+      "loss": 1.7665,
+      "step": 974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.632499006754072e-07,
+      "loss": 1.9683,
+      "step": 975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.632101708382994e-07,
+      "loss": 1.957,
+      "step": 976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.631704410011917e-07,
+      "loss": 1.8609,
+      "step": 977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.631307111640841e-07,
+      "loss": 1.9617,
+      "step": 978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.630909813269765e-07,
+      "loss": 1.9077,
+      "step": 979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.630512514898688e-07,
+      "loss": 1.8809,
+      "step": 980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.630115216527612e-07,
+      "loss": 2.0476,
+      "step": 981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.629717918156535e-07,
+      "loss": 1.9656,
+      "step": 982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.62932061978546e-07,
+      "loss": 1.9136,
+      "step": 983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.628923321414383e-07,
+      "loss": 1.9652,
+      "step": 984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.628526023043306e-07,
+      "loss": 1.8859,
+      "step": 985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.62812872467223e-07,
+      "loss": 1.9796,
+      "step": 986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.627731426301151e-07,
+      "loss": 2.1159,
+      "step": 987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.627334127930075e-07,
+      "loss": 1.934,
+      "step": 988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.626936829558998e-07,
+      "loss": 1.9161,
+      "step": 989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.626539531187922e-07,
+      "loss": 1.924,
+      "step": 990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.626142232816845e-07,
+      "loss": 1.8966,
+      "step": 991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.62574493444577e-07,
+      "loss": 1.8656,
+      "step": 992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.625347636074693e-07,
+      "loss": 1.9463,
+      "step": 993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.624950337703614e-07,
+      "loss": 2.0539,
+      "step": 994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.624553039332538e-07,
+      "loss": 2.0054,
+      "step": 995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.624155740961461e-07,
+      "loss": 1.9091,
+      "step": 996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.623758442590385e-07,
+      "loss": 1.9473,
+      "step": 997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.623361144219308e-07,
+      "loss": 1.9933,
+      "step": 998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.622963845848232e-07,
+      "loss": 2.0606,
+      "step": 999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.622566547477153e-07,
+      "loss": 1.928,
+      "step": 1000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.622169249106077e-07,
+      "loss": 1.8474,
+      "step": 1001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.621771950735e-07,
+      "loss": 1.7884,
+      "step": 1002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.621374652363924e-07,
+      "loss": 1.7964,
+      "step": 1003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.620977353992848e-07,
+      "loss": 1.8332,
+      "step": 1004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.620580055621771e-07,
+      "loss": 1.8453,
+      "step": 1005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.620182757250695e-07,
+      "loss": 1.9258,
+      "step": 1006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.619785458879618e-07,
+      "loss": 1.9215,
+      "step": 1007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.619388160508542e-07,
+      "loss": 2.0291,
+      "step": 1008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.618990862137466e-07,
+      "loss": 1.9419,
+      "step": 1009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.61859356376639e-07,
+      "loss": 1.8739,
+      "step": 1010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.618196265395313e-07,
+      "loss": 1.9183,
+      "step": 1011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.617798967024234e-07,
+      "loss": 1.8646,
+      "step": 1012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.617401668653158e-07,
+      "loss": 2.0694,
+      "step": 1013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.617004370282081e-07,
+      "loss": 1.9739,
+      "step": 1014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.616607071911005e-07,
+      "loss": 1.9211,
+      "step": 1015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.616209773539929e-07,
+      "loss": 1.8552,
+      "step": 1016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.615812475168852e-07,
+      "loss": 1.9057,
+      "step": 1017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.615415176797774e-07,
+      "loss": 1.8726,
+      "step": 1018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.615017878426697e-07,
+      "loss": 1.9674,
+      "step": 1019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61462058005562e-07,
+      "loss": 2.0071,
+      "step": 1020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.614223281684544e-07,
+      "loss": 1.9842,
+      "step": 1021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.613825983313468e-07,
+      "loss": 1.8178,
+      "step": 1022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.613428684942392e-07,
+      "loss": 2.1006,
+      "step": 1023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.613031386571315e-07,
+      "loss": 1.961,
+      "step": 1024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.612634088200239e-07,
+      "loss": 1.8504,
+      "step": 1025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.612236789829162e-07,
+      "loss": 2.0366,
+      "step": 1026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.611839491458084e-07,
+      "loss": 1.8194,
+      "step": 1027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.611442193087007e-07,
+      "loss": 1.8645,
+      "step": 1028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.61104489471593e-07,
+      "loss": 1.8765,
+      "step": 1029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.610647596344854e-07,
+      "loss": 1.9724,
+      "step": 1030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.610250297973778e-07,
+      "loss": 1.7323,
+      "step": 1031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.609852999602702e-07,
+      "loss": 1.8878,
+      "step": 1032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.609455701231625e-07,
+      "loss": 1.9287,
+      "step": 1033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.609058402860549e-07,
+      "loss": 1.8021,
+      "step": 1034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.608661104489472e-07,
+      "loss": 1.8629,
+      "step": 1035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.608263806118394e-07,
+      "loss": 1.9082,
+      "step": 1036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.607866507747317e-07,
+      "loss": 1.9193,
+      "step": 1037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.60746920937624e-07,
+      "loss": 1.932,
+      "step": 1038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.607071911005165e-07,
+      "loss": 1.7738,
+      "step": 1039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.606674612634088e-07,
+      "loss": 1.9791,
+      "step": 1040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.606277314263012e-07,
+      "loss": 1.8201,
+      "step": 1041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.605880015891933e-07,
+      "loss": 2.0008,
+      "step": 1042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.605482717520857e-07,
+      "loss": 2.0208,
+      "step": 1043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.60508541914978e-07,
+      "loss": 1.7981,
+      "step": 1044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.604688120778704e-07,
+      "loss": 1.8262,
+      "step": 1045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.604290822407627e-07,
+      "loss": 2.045,
+      "step": 1046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.60389352403655e-07,
+      "loss": 2.0268,
+      "step": 1047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.603496225665475e-07,
+      "loss": 1.9463,
+      "step": 1048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.603098927294398e-07,
+      "loss": 1.941,
+      "step": 1049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.602701628923322e-07,
+      "loss": 1.9387,
+      "step": 1050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.602304330552245e-07,
+      "loss": 1.8949,
+      "step": 1051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.601907032181169e-07,
+      "loss": 1.891,
+      "step": 1052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.60150973381009e-07,
+      "loss": 1.7938,
+      "step": 1053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.601112435439014e-07,
+      "loss": 1.8944,
+      "step": 1054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.600715137067938e-07,
+      "loss": 1.8832,
+      "step": 1055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.600317838696861e-07,
+      "loss": 2.0639,
+      "step": 1056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.599920540325785e-07,
+      "loss": 1.8719,
+      "step": 1057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.599523241954708e-07,
+      "loss": 2.023,
+      "step": 1058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.599125943583632e-07,
+      "loss": 1.9251,
+      "step": 1059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.598728645212553e-07,
+      "loss": 1.9572,
+      "step": 1060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.598331346841477e-07,
+      "loss": 1.7227,
+      "step": 1061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.5979340484704e-07,
+      "loss": 1.8123,
+      "step": 1062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.597536750099324e-07,
+      "loss": 1.9403,
+      "step": 1063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.597139451728248e-07,
+      "loss": 2.0284,
+      "step": 1064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.596742153357171e-07,
+      "loss": 1.9277,
+      "step": 1065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.596344854986093e-07,
+      "loss": 1.7565,
+      "step": 1066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.595947556615016e-07,
+      "loss": 1.8782,
+      "step": 1067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.59555025824394e-07,
+      "loss": 1.8709,
+      "step": 1068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.595152959872863e-07,
+      "loss": 2.0184,
+      "step": 1069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.594755661501787e-07,
+      "loss": 1.9588,
+      "step": 1070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.59435836313071e-07,
+      "loss": 1.8704,
+      "step": 1071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.593961064759634e-07,
+      "loss": 1.8885,
+      "step": 1072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.593563766388558e-07,
+      "loss": 1.9107,
+      "step": 1073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.593166468017481e-07,
+      "loss": 1.943,
+      "step": 1074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.592769169646405e-07,
+      "loss": 1.8585,
+      "step": 1075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.592371871275328e-07,
+      "loss": 1.7368,
+      "step": 1076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.591974572904252e-07,
+      "loss": 2.0407,
+      "step": 1077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.591577274533173e-07,
+      "loss": 1.9958,
+      "step": 1078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.591179976162097e-07,
+      "loss": 1.9569,
+      "step": 1079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59078267779102e-07,
+      "loss": 1.9987,
+      "step": 1080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590385379419944e-07,
+      "loss": 1.8652,
+      "step": 1081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.589988081048868e-07,
+      "loss": 2.0248,
+      "step": 1082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.589590782677791e-07,
+      "loss": 2.0016,
+      "step": 1083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.589193484306713e-07,
+      "loss": 1.8683,
+      "step": 1084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588796185935636e-07,
+      "loss": 2.0304,
+      "step": 1085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58839888756456e-07,
+      "loss": 1.8374,
+      "step": 1086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588001589193484e-07,
+      "loss": 1.9001,
+      "step": 1087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587604290822407e-07,
+      "loss": 2.1462,
+      "step": 1088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58720699245133e-07,
+      "loss": 1.828,
+      "step": 1089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586809694080254e-07,
+      "loss": 1.9095,
+      "step": 1090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.586412395709178e-07,
+      "loss": 1.9063,
+      "step": 1091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.586015097338101e-07,
+      "loss": 1.9378,
+      "step": 1092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.585617798967023e-07,
+      "loss": 2.04,
+      "step": 1093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.585220500595947e-07,
+      "loss": 1.7869,
+      "step": 1094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.58482320222487e-07,
+      "loss": 1.88,
+      "step": 1095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.584425903853794e-07,
+      "loss": 1.8746,
+      "step": 1096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.584028605482717e-07,
+      "loss": 2.1145,
+      "step": 1097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.58363130711164e-07,
+      "loss": 1.7913,
+      "step": 1098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.583234008740564e-07,
+      "loss": 1.957,
+      "step": 1099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.582836710369488e-07,
+      "loss": 1.8606,
+      "step": 1100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.582439411998412e-07,
+      "loss": 1.9453,
+      "step": 1101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.582042113627333e-07,
+      "loss": 2.0659,
+      "step": 1102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.581644815256257e-07,
+      "loss": 1.8478,
+      "step": 1103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.58124751688518e-07,
+      "loss": 1.8192,
+      "step": 1104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.580850218514104e-07,
+      "loss": 1.965,
+      "step": 1105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.580452920143027e-07,
+      "loss": 1.778,
+      "step": 1106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.58005562177195e-07,
+      "loss": 1.8499,
+      "step": 1107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.579658323400872e-07,
+      "loss": 1.9071,
+      "step": 1108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.579261025029796e-07,
+      "loss": 1.9625,
+      "step": 1109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.57886372665872e-07,
+      "loss": 1.8123,
+      "step": 1110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.578466428287643e-07,
+      "loss": 1.9158,
+      "step": 1111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.578069129916567e-07,
+      "loss": 1.9987,
+      "step": 1112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.57767183154549e-07,
+      "loss": 2.0125,
+      "step": 1113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.577274533174414e-07,
+      "loss": 1.8929,
+      "step": 1114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.576877234803337e-07,
+      "loss": 1.9729,
+      "step": 1115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.57647993643226e-07,
+      "loss": 1.9232,
+      "step": 1116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.576082638061185e-07,
+      "loss": 1.9294,
+      "step": 1117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.575685339690108e-07,
+      "loss": 1.7804,
+      "step": 1118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.57528804131903e-07,
+      "loss": 1.7644,
+      "step": 1119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.574890742947953e-07,
+      "loss": 2.0387,
+      "step": 1120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.574493444576877e-07,
+      "loss": 1.9896,
+      "step": 1121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.5740961462058e-07,
+      "loss": 1.7879,
+      "step": 1122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.573698847834724e-07,
+      "loss": 1.8474,
+      "step": 1123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.573301549463648e-07,
+      "loss": 1.8956,
+      "step": 1124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.572904251092571e-07,
+      "loss": 1.8699,
+      "step": 1125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.572506952721493e-07,
+      "loss": 1.9919,
+      "step": 1126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.572109654350416e-07,
+      "loss": 2.0116,
+      "step": 1127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.57171235597934e-07,
+      "loss": 1.9244,
+      "step": 1128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.571315057608263e-07,
+      "loss": 1.9131,
+      "step": 1129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.570917759237187e-07,
+      "loss": 1.9052,
+      "step": 1130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.57052046086611e-07,
+      "loss": 1.7931,
+      "step": 1131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.570123162495032e-07,
+      "loss": 1.7573,
+      "step": 1132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.569725864123955e-07,
+      "loss": 1.8494,
+      "step": 1133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.56932856575288e-07,
+      "loss": 1.9687,
+      "step": 1134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.568931267381803e-07,
+      "loss": 2.05,
+      "step": 1135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.568533969010726e-07,
+      "loss": 1.8262,
+      "step": 1136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.56813667063965e-07,
+      "loss": 2.2236,
+      "step": 1137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.567739372268573e-07,
+      "loss": 1.9075,
+      "step": 1138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.567342073897497e-07,
+      "loss": 1.9295,
+      "step": 1139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.56694477552642e-07,
+      "loss": 1.9153,
+      "step": 1140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.566547477155344e-07,
+      "loss": 1.8512,
+      "step": 1141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.566150178784268e-07,
+      "loss": 1.9534,
+      "step": 1142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.565752880413191e-07,
+      "loss": 1.91,
+      "step": 1143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.565355582042113e-07,
+      "loss": 1.9104,
+      "step": 1144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.564958283671036e-07,
+      "loss": 2.043,
+      "step": 1145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.56456098529996e-07,
+      "loss": 1.8834,
+      "step": 1146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.564163686928883e-07,
+      "loss": 1.9998,
+      "step": 1147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.563766388557807e-07,
+      "loss": 1.8065,
+      "step": 1148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.56336909018673e-07,
+      "loss": 1.7649,
+      "step": 1149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.562971791815652e-07,
+      "loss": 1.8437,
+      "step": 1150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.562574493444576e-07,
+      "loss": 2.0685,
+      "step": 1151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.5621771950735e-07,
+      "loss": 1.9222,
+      "step": 1152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.561779896702423e-07,
+      "loss": 1.8121,
+      "step": 1153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.561382598331346e-07,
+      "loss": 1.9686,
+      "step": 1154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.56098529996027e-07,
+      "loss": 1.8592,
+      "step": 1155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.560588001589194e-07,
+      "loss": 1.9854,
+      "step": 1156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.560190703218117e-07,
+      "loss": 1.9267,
+      "step": 1157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.55979340484704e-07,
+      "loss": 2.0174,
+      "step": 1158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.559396106475962e-07,
+      "loss": 1.6786,
+      "step": 1159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.558998808104886e-07,
+      "loss": 2.0443,
+      "step": 1160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.55860150973381e-07,
+      "loss": 1.9786,
+      "step": 1161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.558204211362733e-07,
+      "loss": 2.0102,
+      "step": 1162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.557806912991656e-07,
+      "loss": 1.9309,
+      "step": 1163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.55740961462058e-07,
+      "loss": 1.9727,
+      "step": 1164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.557012316249504e-07,
+      "loss": 1.972,
+      "step": 1165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.556615017878427e-07,
+      "loss": 1.7849,
+      "step": 1166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.55621771950735e-07,
+      "loss": 1.9606,
+      "step": 1167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.555820421136272e-07,
+      "loss": 1.939,
+      "step": 1168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.555423122765196e-07,
+      "loss": 1.8335,
+      "step": 1169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.55502582439412e-07,
+      "loss": 1.974,
+      "step": 1170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.554628526023043e-07,
+      "loss": 2.0498,
+      "step": 1171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.554231227651967e-07,
+      "loss": 1.9903,
+      "step": 1172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.55383392928089e-07,
+      "loss": 1.8674,
+      "step": 1173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.553436630909812e-07,
+      "loss": 1.8891,
+      "step": 1174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.553039332538735e-07,
+      "loss": 1.8717,
+      "step": 1175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.552642034167659e-07,
+      "loss": 1.9106,
+      "step": 1176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.552244735796582e-07,
+      "loss": 1.9792,
+      "step": 1177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.551847437425506e-07,
+      "loss": 2.0071,
+      "step": 1178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.55145013905443e-07,
+      "loss": 1.9997,
+      "step": 1179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.551052840683353e-07,
+      "loss": 1.8509,
+      "step": 1180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.550655542312277e-07,
+      "loss": 1.8536,
+      "step": 1181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.5502582439412e-07,
+      "loss": 1.7978,
+      "step": 1182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.549860945570124e-07,
+      "loss": 1.8127,
+      "step": 1183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.549463647199047e-07,
+      "loss": 1.8307,
+      "step": 1184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.549066348827969e-07,
+      "loss": 1.921,
+      "step": 1185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.548669050456892e-07,
+      "loss": 1.9632,
+      "step": 1186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.548271752085816e-07,
+      "loss": 1.901,
+      "step": 1187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.54787445371474e-07,
+      "loss": 1.9321,
+      "step": 1188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.547477155343663e-07,
+      "loss": 1.9921,
+      "step": 1189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.547079856972587e-07,
+      "loss": 1.85,
+      "step": 1190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.54668255860151e-07,
+      "loss": 1.9381,
+      "step": 1191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.546285260230432e-07,
+      "loss": 1.9183,
+      "step": 1192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.545887961859355e-07,
+      "loss": 2.0278,
+      "step": 1193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.54549066348828e-07,
+      "loss": 1.8782,
+      "step": 1194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.545093365117203e-07,
+      "loss": 2.0152,
+      "step": 1195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.544696066746126e-07,
+      "loss": 1.9499,
+      "step": 1196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.54429876837505e-07,
+      "loss": 1.8445,
+      "step": 1197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.543901470003971e-07,
+      "loss": 1.9272,
+      "step": 1198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.543504171632895e-07,
+      "loss": 1.8872,
+      "step": 1199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.543106873261818e-07,
+      "loss": 2.0201,
+      "step": 1200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.542709574890742e-07,
+      "loss": 1.8879,
+      "step": 1201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.542312276519665e-07,
+      "loss": 2.005,
+      "step": 1202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.54191497814859e-07,
+      "loss": 1.8798,
+      "step": 1203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.541517679777513e-07,
+      "loss": 1.8647,
+      "step": 1204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.541120381406436e-07,
+      "loss": 2.0329,
+      "step": 1205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.54072308303536e-07,
+      "loss": 1.727,
+      "step": 1206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.540325784664283e-07,
+      "loss": 1.992,
+      "step": 1207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.539928486293207e-07,
+      "loss": 1.9672,
+      "step": 1208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.53953118792213e-07,
+      "loss": 1.7548,
+      "step": 1209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.539133889551052e-07,
+      "loss": 1.9533,
+      "step": 1210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.538736591179976e-07,
+      "loss": 1.8639,
+      "step": 1211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.5383392928089e-07,
+      "loss": 2.0032,
+      "step": 1212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.537941994437823e-07,
+      "loss": 1.8309,
+      "step": 1213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.537544696066746e-07,
+      "loss": 1.9436,
+      "step": 1214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.53714739769567e-07,
+      "loss": 1.9415,
+      "step": 1215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.536750099324591e-07,
+      "loss": 2.0072,
+      "step": 1216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.536352800953516e-07,
+      "loss": 1.9527,
+      "step": 1217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.535955502582438e-07,
+      "loss": 1.9685,
+      "step": 1218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.535558204211362e-07,
+      "loss": 1.892,
+      "step": 1219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.535160905840286e-07,
+      "loss": 1.7901,
+      "step": 1220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.534763607469209e-07,
+      "loss": 1.9449,
+      "step": 1221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.534366309098132e-07,
+      "loss": 1.825,
+      "step": 1222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.533969010727055e-07,
+      "loss": 2.0147,
+      "step": 1223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.533571712355979e-07,
+      "loss": 1.8432,
+      "step": 1224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.533174413984902e-07,
+      "loss": 1.9728,
+      "step": 1225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.532777115613826e-07,
+      "loss": 1.8703,
+      "step": 1226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.53237981724275e-07,
+      "loss": 1.9222,
+      "step": 1227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.531982518871672e-07,
+      "loss": 1.8739,
+      "step": 1228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.531585220500596e-07,
+      "loss": 1.9706,
+      "step": 1229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.531187922129519e-07,
+      "loss": 2.0614,
+      "step": 1230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.530790623758442e-07,
+      "loss": 2.0414,
+      "step": 1231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.530393325387365e-07,
+      "loss": 1.9019,
+      "step": 1232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.529996027016289e-07,
+      "loss": 2.0275,
+      "step": 1233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.529598728645212e-07,
+      "loss": 1.9941,
+      "step": 1234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.529201430274135e-07,
+      "loss": 2.0036,
+      "step": 1235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.528804131903059e-07,
+      "loss": 1.8596,
+      "step": 1236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.528406833531982e-07,
+      "loss": 1.8127,
+      "step": 1237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.528009535160906e-07,
+      "loss": 1.9473,
+      "step": 1238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.527612236789829e-07,
+      "loss": 2.0504,
+      "step": 1239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.527214938418752e-07,
+      "loss": 1.9676,
+      "step": 1240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.526817640047675e-07,
+      "loss": 1.8814,
+      "step": 1241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.526420341676599e-07,
+      "loss": 1.9251,
+      "step": 1242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.526023043305523e-07,
+      "loss": 1.9434,
+      "step": 1243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.525625744934446e-07,
+      "loss": 1.9113,
+      "step": 1244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.525228446563369e-07,
+      "loss": 1.9676,
+      "step": 1245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.524831148192291e-07,
+      "loss": 1.8468,
+      "step": 1246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.524433849821215e-07,
+      "loss": 1.8757,
+      "step": 1247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.524036551450138e-07,
+      "loss": 1.8717,
+      "step": 1248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.523639253079062e-07,
+      "loss": 1.9911,
+      "step": 1249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.523241954707986e-07,
+      "loss": 1.84,
+      "step": 1250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.522844656336908e-07,
+      "loss": 1.9262,
+      "step": 1251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.522447357965832e-07,
+      "loss": 1.8459,
+      "step": 1252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.522050059594755e-07,
+      "loss": 1.8686,
+      "step": 1253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.521652761223679e-07,
+      "loss": 2.013,
+      "step": 1254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.521255462852602e-07,
+      "loss": 1.8998,
+      "step": 1255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.520858164481526e-07,
+      "loss": 1.9193,
+      "step": 1256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.520460866110447e-07,
+      "loss": 1.9417,
+      "step": 1257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.520063567739371e-07,
+      "loss": 1.8989,
+      "step": 1258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.519666269368295e-07,
+      "loss": 1.7269,
+      "step": 1259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.519268970997218e-07,
+      "loss": 1.9645,
+      "step": 1260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.518871672626142e-07,
+      "loss": 1.8336,
+      "step": 1261
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.24662965900079303,
+      "eval_loss": 1.9108086824417114,
+      "eval_runtime": 6.3124,
+      "eval_samples_per_second": 399.529,
+      "eval_steps_per_second": 6.337,
+      "step": 1261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.518474374255065e-07,
+      "loss": 1.9488,
+      "step": 1262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.518077075883988e-07,
+      "loss": 1.9616,
+      "step": 1263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.517679777512911e-07,
+      "loss": 1.7328,
+      "step": 1264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.517282479141835e-07,
+      "loss": 1.9264,
+      "step": 1265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.516885180770759e-07,
+      "loss": 1.9606,
+      "step": 1266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.516487882399682e-07,
+      "loss": 1.7967,
+      "step": 1267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.516090584028606e-07,
+      "loss": 1.8167,
+      "step": 1268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.515693285657528e-07,
+      "loss": 1.847,
+      "step": 1269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.515295987286452e-07,
+      "loss": 1.8995,
+      "step": 1270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.514898688915374e-07,
+      "loss": 1.9675,
+      "step": 1271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.514501390544298e-07,
+      "loss": 1.9042,
+      "step": 1272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.514104092173222e-07,
+      "loss": 1.6028,
+      "step": 1273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.513706793802145e-07,
+      "loss": 2.0836,
+      "step": 1274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.513309495431068e-07,
+      "loss": 1.9101,
+      "step": 1275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.512912197059991e-07,
+      "loss": 1.9805,
+      "step": 1276
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.512514898688915e-07,
+      "loss": 1.9603,
+      "step": 1277
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.512117600317838e-07,
+      "loss": 1.8108,
+      "step": 1278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.511720301946762e-07,
+      "loss": 1.8523,
+      "step": 1279
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.511323003575686e-07,
+      "loss": 1.8511,
+      "step": 1280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.510925705204608e-07,
+      "loss": 2.0269,
+      "step": 1281
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.510528406833532e-07,
+      "loss": 1.837,
+      "step": 1282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.510131108462455e-07,
+      "loss": 1.9575,
+      "step": 1283
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.509733810091378e-07,
+      "loss": 1.7776,
+      "step": 1284
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.509336511720301e-07,
+      "loss": 1.9202,
+      "step": 1285
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.508939213349225e-07,
+      "loss": 1.9261,
+      "step": 1286
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.508541914978147e-07,
+      "loss": 1.9305,
+      "step": 1287
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.508144616607071e-07,
+      "loss": 1.8187,
+      "step": 1288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.507747318235995e-07,
+      "loss": 1.7812,
+      "step": 1289
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.507350019864918e-07,
+      "loss": 2.0454,
+      "step": 1290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.506952721493842e-07,
+      "loss": 2.0638,
+      "step": 1291
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.506555423122765e-07,
+      "loss": 1.8238,
+      "step": 1292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.506158124751688e-07,
+      "loss": 1.9233,
+      "step": 1293
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.505760826380611e-07,
+      "loss": 1.8192,
+      "step": 1294
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.505363528009535e-07,
+      "loss": 1.7668,
+      "step": 1295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.504966229638459e-07,
+      "loss": 1.8294,
+      "step": 1296
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.504568931267381e-07,
+      "loss": 2.098,
+      "step": 1297
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.504171632896305e-07,
+      "loss": 2.0532,
+      "step": 1298
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.503774334525227e-07,
+      "loss": 1.8942,
+      "step": 1299
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.503377036154151e-07,
+      "loss": 1.8581,
+      "step": 1300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.502979737783074e-07,
+      "loss": 1.8282,
+      "step": 1301
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.502582439411998e-07,
+      "loss": 2.0216,
+      "step": 1302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.502185141040921e-07,
+      "loss": 1.8161,
+      "step": 1303
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.501787842669845e-07,
+      "loss": 1.931,
+      "step": 1304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.501390544298768e-07,
+      "loss": 1.8697,
+      "step": 1305
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.500993245927691e-07,
+      "loss": 1.8268,
+      "step": 1306
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.500595947556615e-07,
+      "loss": 2.0384,
+      "step": 1307
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.500198649185538e-07,
+      "loss": 1.9866,
+      "step": 1308
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.499801350814462e-07,
+      "loss": 1.8151,
+      "step": 1309
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.499404052443385e-07,
+      "loss": 2.0895,
+      "step": 1310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.499006754072307e-07,
+      "loss": 1.7872,
+      "step": 1311
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.49860945570123e-07,
+      "loss": 1.8439,
+      "step": 1312
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.498212157330154e-07,
+      "loss": 1.9565,
+      "step": 1313
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.497814858959078e-07,
+      "loss": 2.0201,
+      "step": 1314
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.497417560588001e-07,
+      "loss": 1.8531,
+      "step": 1315
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.497020262216925e-07,
+      "loss": 1.8579,
+      "step": 1316
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.496622963845847e-07,
+      "loss": 1.8127,
+      "step": 1317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.496225665474771e-07,
+      "loss": 1.8328,
+      "step": 1318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.495828367103694e-07,
+      "loss": 1.8966,
+      "step": 1319
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.495431068732618e-07,
+      "loss": 1.7368,
+      "step": 1320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.495033770361542e-07,
+      "loss": 2.0702,
+      "step": 1321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.494636471990465e-07,
+      "loss": 1.9202,
+      "step": 1322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.494239173619387e-07,
+      "loss": 1.8319,
+      "step": 1323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.49384187524831e-07,
+      "loss": 1.8051,
+      "step": 1324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.493444576877234e-07,
+      "loss": 1.9053,
+      "step": 1325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.493047278506157e-07,
+      "loss": 1.9869,
+      "step": 1326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.492649980135081e-07,
+      "loss": 1.9629,
+      "step": 1327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.492252681764005e-07,
+      "loss": 1.9712,
+      "step": 1328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.491855383392927e-07,
+      "loss": 1.9727,
+      "step": 1329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.491458085021851e-07,
+      "loss": 1.8757,
+      "step": 1330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.491060786650774e-07,
+      "loss": 1.8639,
+      "step": 1331
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.490663488279698e-07,
+      "loss": 1.9296,
+      "step": 1332
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.490266189908621e-07,
+      "loss": 1.8641,
+      "step": 1333
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489868891537545e-07,
+      "loss": 1.756,
+      "step": 1334
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489471593166468e-07,
+      "loss": 1.9495,
+      "step": 1335
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489074294795391e-07,
+      "loss": 1.9551,
+      "step": 1336
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.488676996424314e-07,
+      "loss": 1.8068,
+      "step": 1337
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.488279698053237e-07,
+      "loss": 1.7894,
+      "step": 1338
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.487882399682161e-07,
+      "loss": 1.8775,
+      "step": 1339
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.487485101311084e-07,
+      "loss": 1.8845,
+      "step": 1340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.487087802940007e-07,
+      "loss": 1.9556,
+      "step": 1341
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.48669050456893e-07,
+      "loss": 1.6874,
+      "step": 1342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.486293206197854e-07,
+      "loss": 1.8438,
+      "step": 1343
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.485895907826778e-07,
+      "loss": 1.799,
+      "step": 1344
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.485498609455701e-07,
+      "loss": 1.9862,
+      "step": 1345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.485101311084625e-07,
+      "loss": 1.8448,
+      "step": 1346
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.484704012713547e-07,
+      "loss": 2.095,
+      "step": 1347
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.484306714342471e-07,
+      "loss": 1.9286,
+      "step": 1348
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.483909415971394e-07,
+      "loss": 1.8694,
+      "step": 1349
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.483512117600317e-07,
+      "loss": 1.8379,
+      "step": 1350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.483114819229241e-07,
+      "loss": 1.8053,
+      "step": 1351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.482717520858164e-07,
+      "loss": 1.9931,
+      "step": 1352
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.482320222487087e-07,
+      "loss": 1.8214,
+      "step": 1353
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.48192292411601e-07,
+      "loss": 2.0343,
+      "step": 1354
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.481525625744934e-07,
+      "loss": 1.8722,
+      "step": 1355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.481128327373857e-07,
+      "loss": 1.9972,
+      "step": 1356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.480731029002781e-07,
+      "loss": 2.0936,
+      "step": 1357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.480333730631705e-07,
+      "loss": 1.8164,
+      "step": 1358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.479936432260627e-07,
+      "loss": 1.9015,
+      "step": 1359
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.479539133889551e-07,
+      "loss": 1.8747,
+      "step": 1360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.479141835518474e-07,
+      "loss": 1.9234,
+      "step": 1361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.478744537147398e-07,
+      "loss": 1.8301,
+      "step": 1362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.47834723877632e-07,
+      "loss": 1.7267,
+      "step": 1363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.477949940405244e-07,
+      "loss": 1.9825,
+      "step": 1364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.477552642034166e-07,
+      "loss": 1.9282,
+      "step": 1365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.47715534366309e-07,
+      "loss": 1.9455,
+      "step": 1366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.476758045292014e-07,
+      "loss": 1.8188,
+      "step": 1367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.476360746920937e-07,
+      "loss": 1.8683,
+      "step": 1368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.475963448549861e-07,
+      "loss": 1.8541,
+      "step": 1369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.475566150178784e-07,
+      "loss": 2.0458,
+      "step": 1370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.475168851807707e-07,
+      "loss": 1.9714,
+      "step": 1371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.47477155343663e-07,
+      "loss": 1.7668,
+      "step": 1372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.474374255065554e-07,
+      "loss": 1.8704,
+      "step": 1373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.473976956694478e-07,
+      "loss": 1.8766,
+      "step": 1374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.473579658323401e-07,
+      "loss": 1.7605,
+      "step": 1375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.473182359952325e-07,
+      "loss": 2.1542,
+      "step": 1376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.472785061581246e-07,
+      "loss": 1.7991,
+      "step": 1377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.47238776321017e-07,
+      "loss": 1.9286,
+      "step": 1378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.471990464839093e-07,
+      "loss": 1.8013,
+      "step": 1379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.471593166468017e-07,
+      "loss": 1.8869,
+      "step": 1380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.47119586809694e-07,
+      "loss": 1.7826,
+      "step": 1381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.470798569725864e-07,
+      "loss": 2.0216,
+      "step": 1382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.470401271354787e-07,
+      "loss": 1.8613,
+      "step": 1383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.47000397298371e-07,
+      "loss": 1.8345,
+      "step": 1384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.469606674612634e-07,
+      "loss": 1.9223,
+      "step": 1385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.469209376241557e-07,
+      "loss": 1.8072,
+      "step": 1386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.468812077870481e-07,
+      "loss": 1.9409,
+      "step": 1387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.468414779499404e-07,
+      "loss": 1.6992,
+      "step": 1388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.468017481128326e-07,
+      "loss": 1.8285,
+      "step": 1389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.46762018275725e-07,
+      "loss": 1.9527,
+      "step": 1390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.467222884386173e-07,
+      "loss": 1.7379,
+      "step": 1391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.466825586015097e-07,
+      "loss": 1.8127,
+      "step": 1392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.46642828764402e-07,
+      "loss": 1.7804,
+      "step": 1393
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.466030989272944e-07,
+      "loss": 1.7785,
+      "step": 1394
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.465633690901866e-07,
+      "loss": 1.9322,
+      "step": 1395
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.46523639253079e-07,
+      "loss": 1.8663,
+      "step": 1396
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.464839094159713e-07,
+      "loss": 1.9046,
+      "step": 1397
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.464441795788637e-07,
+      "loss": 1.9996,
+      "step": 1398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.464044497417561e-07,
+      "loss": 1.8441,
+      "step": 1399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.463647199046484e-07,
+      "loss": 1.7529,
+      "step": 1400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.463249900675407e-07,
+      "loss": 1.8,
+      "step": 1401
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.462852602304329e-07,
+      "loss": 1.855,
+      "step": 1402
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.462455303933253e-07,
+      "loss": 1.7759,
+      "step": 1403
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.462058005562176e-07,
+      "loss": 1.9843,
+      "step": 1404
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.4616607071911e-07,
+      "loss": 1.9527,
+      "step": 1405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.461263408820024e-07,
+      "loss": 1.8275,
+      "step": 1406
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.460866110448946e-07,
+      "loss": 1.7677,
+      "step": 1407
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.46046881207787e-07,
+      "loss": 1.7602,
+      "step": 1408
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.460071513706793e-07,
+      "loss": 1.9252,
+      "step": 1409
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.459674215335717e-07,
+      "loss": 2.1018,
+      "step": 1410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.45927691696464e-07,
+      "loss": 1.888,
+      "step": 1411
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.458879618593564e-07,
+      "loss": 1.966,
+      "step": 1412
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.458482320222487e-07,
+      "loss": 1.8597,
+      "step": 1413
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.45808502185141e-07,
+      "loss": 1.7164,
+      "step": 1414
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.457687723480334e-07,
+      "loss": 1.6022,
+      "step": 1415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.457290425109256e-07,
+      "loss": 2.2244,
+      "step": 1416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.45689312673818e-07,
+      "loss": 1.7715,
+      "step": 1417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.456495828367103e-07,
+      "loss": 1.929,
+      "step": 1418
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.456098529996026e-07,
+      "loss": 1.8494,
+      "step": 1419
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.455701231624949e-07,
+      "loss": 1.9095,
+      "step": 1420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.455303933253873e-07,
+      "loss": 1.8688,
+      "step": 1421
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.454906634882797e-07,
+      "loss": 1.9817,
+      "step": 1422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.45450933651172e-07,
+      "loss": 1.8584,
+      "step": 1423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.454112038140644e-07,
+      "loss": 2.1307,
+      "step": 1424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.453714739769566e-07,
+      "loss": 1.9584,
+      "step": 1425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.45331744139849e-07,
+      "loss": 1.9618,
+      "step": 1426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.452920143027413e-07,
+      "loss": 1.8737,
+      "step": 1427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.452522844656337e-07,
+      "loss": 2.0203,
+      "step": 1428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.45212554628526e-07,
+      "loss": 1.9326,
+      "step": 1429
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.451728247914183e-07,
+      "loss": 1.951,
+      "step": 1430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.451330949543106e-07,
+      "loss": 1.7787,
+      "step": 1431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.450933651172029e-07,
+      "loss": 1.904,
+      "step": 1432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.450536352800953e-07,
+      "loss": 1.9104,
+      "step": 1433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.450139054429876e-07,
+      "loss": 1.8639,
+      "step": 1434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.4497417560588e-07,
+      "loss": 1.74,
+      "step": 1435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.449344457687724e-07,
+      "loss": 2.1736,
+      "step": 1436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.448947159316646e-07,
+      "loss": 2.1075,
+      "step": 1437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.44854986094557e-07,
+      "loss": 1.7147,
+      "step": 1438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.448152562574493e-07,
+      "loss": 1.881,
+      "step": 1439
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.447755264203417e-07,
+      "loss": 1.8041,
+      "step": 1440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.44735796583234e-07,
+      "loss": 1.9059,
+      "step": 1441
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.446960667461263e-07,
+      "loss": 1.9051,
+      "step": 1442
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.446563369090185e-07,
+      "loss": 2.0378,
+      "step": 1443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.446166070719109e-07,
+      "loss": 1.8282,
+      "step": 1444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.445768772348033e-07,
+      "loss": 1.8608,
+      "step": 1445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.445371473976956e-07,
+      "loss": 1.9477,
+      "step": 1446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.44497417560588e-07,
+      "loss": 1.9926,
+      "step": 1447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.444576877234803e-07,
+      "loss": 1.7844,
+      "step": 1448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.444179578863726e-07,
+      "loss": 1.8972,
+      "step": 1449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.443782280492649e-07,
+      "loss": 1.7678,
+      "step": 1450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.443384982121573e-07,
+      "loss": 1.8797,
+      "step": 1451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.442987683750497e-07,
+      "loss": 1.9243,
+      "step": 1452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.44259038537942e-07,
+      "loss": 2.0417,
+      "step": 1453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.442193087008344e-07,
+      "loss": 1.8037,
+      "step": 1454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.441795788637265e-07,
+      "loss": 1.7726,
+      "step": 1455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.441398490266189e-07,
+      "loss": 1.8244,
+      "step": 1456
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.441001191895112e-07,
+      "loss": 2.0224,
+      "step": 1457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.440603893524036e-07,
+      "loss": 1.9429,
+      "step": 1458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.44020659515296e-07,
+      "loss": 1.8366,
+      "step": 1459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.439809296781883e-07,
+      "loss": 2.016,
+      "step": 1460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.439411998410806e-07,
+      "loss": 1.7961,
+      "step": 1461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.439014700039729e-07,
+      "loss": 1.9534,
+      "step": 1462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.438617401668653e-07,
+      "loss": 1.7133,
+      "step": 1463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.438220103297576e-07,
+      "loss": 1.8309,
+      "step": 1464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.4378228049265e-07,
+      "loss": 1.9177,
+      "step": 1465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.437425506555423e-07,
+      "loss": 1.9089,
+      "step": 1466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.437028208184346e-07,
+      "loss": 1.9559,
+      "step": 1467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.436630909813269e-07,
+      "loss": 1.7597,
+      "step": 1468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.436233611442192e-07,
+      "loss": 2.0913,
+      "step": 1469
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.435836313071116e-07,
+      "loss": 1.821,
+      "step": 1470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.435439014700039e-07,
+      "loss": 1.9895,
+      "step": 1471
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.435041716328963e-07,
+      "loss": 1.7462,
+      "step": 1472
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.434644417957885e-07,
+      "loss": 2.0575,
+      "step": 1473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.434247119586809e-07,
+      "loss": 1.8991,
+      "step": 1474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.433849821215733e-07,
+      "loss": 1.8078,
+      "step": 1475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.433452522844656e-07,
+      "loss": 2.0573,
+      "step": 1476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.43305522447358e-07,
+      "loss": 1.8484,
+      "step": 1477
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.432657926102503e-07,
+      "loss": 1.7718,
+      "step": 1478
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.432260627731426e-07,
+      "loss": 1.989,
+      "step": 1479
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.431863329360349e-07,
+      "loss": 1.8027,
+      "step": 1480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.431466030989273e-07,
+      "loss": 1.9911,
+      "step": 1481
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.431068732618195e-07,
+      "loss": 2.0355,
+      "step": 1482
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.430671434247119e-07,
+      "loss": 1.7956,
+      "step": 1483
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.430274135876043e-07,
+      "loss": 2.0303,
+      "step": 1484
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.429876837504965e-07,
+      "loss": 1.7721,
+      "step": 1485
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.429479539133889e-07,
+      "loss": 1.9122,
+      "step": 1486
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.429082240762812e-07,
+      "loss": 1.9398,
+      "step": 1487
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.428684942391736e-07,
+      "loss": 1.7509,
+      "step": 1488
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.428287644020659e-07,
+      "loss": 1.8317,
+      "step": 1489
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.427890345649583e-07,
+      "loss": 1.8698,
+      "step": 1490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.427493047278506e-07,
+      "loss": 1.8624,
+      "step": 1491
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.427095748907429e-07,
+      "loss": 1.9371,
+      "step": 1492
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.426698450536353e-07,
+      "loss": 1.8748,
+      "step": 1493
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.426301152165276e-07,
+      "loss": 1.8774,
+      "step": 1494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.425903853794199e-07,
+      "loss": 1.8327,
+      "step": 1495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.425506555423122e-07,
+      "loss": 1.7383,
+      "step": 1496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.425109257052045e-07,
+      "loss": 2.1276,
+      "step": 1497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.424711958680968e-07,
+      "loss": 1.9773,
+      "step": 1498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.424314660309892e-07,
+      "loss": 1.7587,
+      "step": 1499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.423917361938816e-07,
+      "loss": 2.0042,
+      "step": 1500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.423520063567739e-07,
+      "loss": 1.8042,
+      "step": 1501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.423122765196663e-07,
+      "loss": 1.9387,
+      "step": 1502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.422725466825585e-07,
+      "loss": 1.951,
+      "step": 1503
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.422328168454509e-07,
+      "loss": 2.0503,
+      "step": 1504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.421930870083432e-07,
+      "loss": 2.007,
+      "step": 1505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.421533571712356e-07,
+      "loss": 1.991,
+      "step": 1506
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.42113627334128e-07,
+      "loss": 1.8548,
+      "step": 1507
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.420738974970202e-07,
+      "loss": 1.8979,
+      "step": 1508
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.420341676599125e-07,
+      "loss": 1.8418,
+      "step": 1509
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.419944378228048e-07,
+      "loss": 1.7856,
+      "step": 1510
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.419547079856972e-07,
+      "loss": 1.8694,
+      "step": 1511
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.419149781485895e-07,
+      "loss": 1.8732,
+      "step": 1512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.418752483114819e-07,
+      "loss": 1.8855,
+      "step": 1513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.418355184743743e-07,
+      "loss": 1.9386,
+      "step": 1514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.417957886372665e-07,
+      "loss": 1.9113,
+      "step": 1515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.417560588001589e-07,
+      "loss": 1.9617,
+      "step": 1516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.417163289630512e-07,
+      "loss": 1.865,
+      "step": 1517
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.416765991259436e-07,
+      "loss": 2.009,
+      "step": 1518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.416368692888359e-07,
+      "loss": 1.8945,
+      "step": 1519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.415971394517283e-07,
+      "loss": 1.7404,
+      "step": 1520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.415574096146204e-07,
+      "loss": 1.9701,
+      "step": 1521
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.415176797775128e-07,
+      "loss": 1.7994,
+      "step": 1522
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.414779499404052e-07,
+      "loss": 1.8338,
+      "step": 1523
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.414382201032975e-07,
+      "loss": 1.8921,
+      "step": 1524
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.413984902661899e-07,
+      "loss": 1.7868,
+      "step": 1525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.413587604290822e-07,
+      "loss": 2.0154,
+      "step": 1526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.413190305919745e-07,
+      "loss": 1.8407,
+      "step": 1527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.412793007548668e-07,
+      "loss": 1.7192,
+      "step": 1528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.412395709177592e-07,
+      "loss": 1.8054,
+      "step": 1529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.411998410806516e-07,
+      "loss": 1.7993,
+      "step": 1530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.411601112435439e-07,
+      "loss": 1.9469,
+      "step": 1531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.411203814064363e-07,
+      "loss": 1.7138,
+      "step": 1532
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.410806515693285e-07,
+      "loss": 1.9511,
+      "step": 1533
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.410409217322208e-07,
+      "loss": 2.1358,
+      "step": 1534
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.410011918951131e-07,
+      "loss": 1.857,
+      "step": 1535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.409614620580055e-07,
+      "loss": 1.7832,
+      "step": 1536
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.409217322208978e-07,
+      "loss": 1.8542,
+      "step": 1537
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.408820023837902e-07,
+      "loss": 1.9669,
+      "step": 1538
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.408422725466825e-07,
+      "loss": 1.9553,
+      "step": 1539
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.408025427095748e-07,
+      "loss": 2.1033,
+      "step": 1540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.407628128724672e-07,
+      "loss": 1.8656,
+      "step": 1541
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.407230830353595e-07,
+      "loss": 1.943,
+      "step": 1542
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.406833531982519e-07,
+      "loss": 1.6793,
+      "step": 1543
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.406436233611442e-07,
+      "loss": 1.9078,
+      "step": 1544
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.406038935240365e-07,
+      "loss": 1.9027,
+      "step": 1545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.405641636869289e-07,
+      "loss": 1.7934,
+      "step": 1546
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.405244338498212e-07,
+      "loss": 1.8336,
+      "step": 1547
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.404847040127135e-07,
+      "loss": 1.9682,
+      "step": 1548
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.404449741756058e-07,
+      "loss": 1.9349,
+      "step": 1549
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.404052443384982e-07,
+      "loss": 1.8189,
+      "step": 1550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.403655145013904e-07,
+      "loss": 1.9225,
+      "step": 1551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.403257846642828e-07,
+      "loss": 1.9978,
+      "step": 1552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.402860548271752e-07,
+      "loss": 1.9706,
+      "step": 1553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.402463249900675e-07,
+      "loss": 1.8884,
+      "step": 1554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.402065951529599e-07,
+      "loss": 1.7899,
+      "step": 1555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.401668653158522e-07,
+      "loss": 1.9523,
+      "step": 1556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.401271354787445e-07,
+      "loss": 1.9383,
+      "step": 1557
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.400874056416368e-07,
+      "loss": 1.7792,
+      "step": 1558
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.400476758045292e-07,
+      "loss": 1.9465,
+      "step": 1559
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.400079459674215e-07,
+      "loss": 1.9118,
+      "step": 1560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.399682161303138e-07,
+      "loss": 1.9379,
+      "step": 1561
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.399284862932062e-07,
+      "loss": 2.0116,
+      "step": 1562
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.398887564560984e-07,
+      "loss": 2.0303,
+      "step": 1563
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.398490266189908e-07,
+      "loss": 2.0466,
+      "step": 1564
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.398092967818831e-07,
+      "loss": 1.7492,
+      "step": 1565
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.397695669447755e-07,
+      "loss": 1.7539,
+      "step": 1566
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.397298371076678e-07,
+      "loss": 1.8865,
+      "step": 1567
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.396901072705602e-07,
+      "loss": 2.0273,
+      "step": 1568
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.396503774334525e-07,
+      "loss": 2.1112,
+      "step": 1569
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.396106475963448e-07,
+      "loss": 1.9532,
+      "step": 1570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.395709177592372e-07,
+      "loss": 1.8745,
+      "step": 1571
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.395311879221295e-07,
+      "loss": 1.9341,
+      "step": 1572
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.394914580850219e-07,
+      "loss": 1.8546,
+      "step": 1573
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.394517282479141e-07,
+      "loss": 1.9626,
+      "step": 1574
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.394119984108064e-07,
+      "loss": 1.7667,
+      "step": 1575
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.393722685736987e-07,
+      "loss": 1.7737,
+      "step": 1576
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.393325387365911e-07,
+      "loss": 1.8172,
+      "step": 1577
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.392928088994835e-07,
+      "loss": 1.8398,
+      "step": 1578
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.392530790623758e-07,
+      "loss": 1.7831,
+      "step": 1579
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.392133492252682e-07,
+      "loss": 1.6408,
+      "step": 1580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.391736193881604e-07,
+      "loss": 1.9161,
+      "step": 1581
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.391338895510528e-07,
+      "loss": 1.8828,
+      "step": 1582
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.390941597139451e-07,
+      "loss": 1.9075,
+      "step": 1583
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.390544298768375e-07,
+      "loss": 1.9557,
+      "step": 1584
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.390147000397299e-07,
+      "loss": 1.8504,
+      "step": 1585
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.389749702026222e-07,
+      "loss": 1.9237,
+      "step": 1586
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.389352403655144e-07,
+      "loss": 1.7106,
+      "step": 1587
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.388955105284067e-07,
+      "loss": 1.9594,
+      "step": 1588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.388557806912991e-07,
+      "loss": 1.9575,
+      "step": 1589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.388160508541914e-07,
+      "loss": 1.8479,
+      "step": 1590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.387763210170838e-07,
+      "loss": 1.7399,
+      "step": 1591
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.387365911799762e-07,
+      "loss": 1.8638,
+      "step": 1592
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.386968613428684e-07,
+      "loss": 1.8174,
+      "step": 1593
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.386571315057608e-07,
+      "loss": 1.7369,
+      "step": 1594
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.386174016686531e-07,
+      "loss": 1.9163,
+      "step": 1595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.385776718315455e-07,
+      "loss": 1.7436,
+      "step": 1596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.385379419944378e-07,
+      "loss": 1.9226,
+      "step": 1597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.384982121573302e-07,
+      "loss": 1.9386,
+      "step": 1598
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.384584823202224e-07,
+      "loss": 1.8618,
+      "step": 1599
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.384187524831147e-07,
+      "loss": 1.8426,
+      "step": 1600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.383790226460071e-07,
+      "loss": 1.9803,
+      "step": 1601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.383392928088994e-07,
+      "loss": 1.9365,
+      "step": 1602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.382995629717918e-07,
+      "loss": 1.7842,
+      "step": 1603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.382598331346841e-07,
+      "loss": 1.6708,
+      "step": 1604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.382201032975764e-07,
+      "loss": 1.8455,
+      "step": 1605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.381803734604687e-07,
+      "loss": 1.9729,
+      "step": 1606
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.381406436233611e-07,
+      "loss": 1.8992,
+      "step": 1607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.381009137862535e-07,
+      "loss": 1.9754,
+      "step": 1608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.380611839491458e-07,
+      "loss": 1.975,
+      "step": 1609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.380214541120382e-07,
+      "loss": 1.9896,
+      "step": 1610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.379817242749304e-07,
+      "loss": 1.9668,
+      "step": 1611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.379419944378228e-07,
+      "loss": 1.8652,
+      "step": 1612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.379022646007151e-07,
+      "loss": 1.8887,
+      "step": 1613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.378625347636074e-07,
+      "loss": 1.9933,
+      "step": 1614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.378228049264997e-07,
+      "loss": 1.9421,
+      "step": 1615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.377830750893921e-07,
+      "loss": 1.8675,
+      "step": 1616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.377433452522844e-07,
+      "loss": 1.6455,
+      "step": 1617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.377036154151767e-07,
+      "loss": 2.0771,
+      "step": 1618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.376638855780691e-07,
+      "loss": 1.8247,
+      "step": 1619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.376241557409614e-07,
+      "loss": 1.8864,
+      "step": 1620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.375844259038538e-07,
+      "loss": 2.0087,
+      "step": 1621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.375446960667461e-07,
+      "loss": 1.9015,
+      "step": 1622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.375049662296384e-07,
+      "loss": 1.8673,
+      "step": 1623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.374652363925308e-07,
+      "loss": 1.887,
+      "step": 1624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.374255065554231e-07,
+      "loss": 1.7454,
+      "step": 1625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.373857767183155e-07,
+      "loss": 1.8668,
+      "step": 1626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.373460468812077e-07,
+      "loss": 1.926,
+      "step": 1627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.373063170441001e-07,
+      "loss": 2.1563,
+      "step": 1628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.372665872069923e-07,
+      "loss": 1.9052,
+      "step": 1629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.372268573698847e-07,
+      "loss": 1.8149,
+      "step": 1630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.37187127532777e-07,
+      "loss": 1.8953,
+      "step": 1631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.371473976956694e-07,
+      "loss": 1.8713,
+      "step": 1632
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.371076678585618e-07,
+      "loss": 1.9827,
+      "step": 1633
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.370679380214541e-07,
+      "loss": 1.829,
+      "step": 1634
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.370282081843464e-07,
+      "loss": 1.8173,
+      "step": 1635
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.369884783472387e-07,
+      "loss": 1.8711,
+      "step": 1636
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.369487485101311e-07,
+      "loss": 1.8977,
+      "step": 1637
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.369090186730234e-07,
+      "loss": 1.9796,
+      "step": 1638
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.368692888359158e-07,
+      "loss": 1.8425,
+      "step": 1639
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.368295589988081e-07,
+      "loss": 2.0081,
+      "step": 1640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.367898291617003e-07,
+      "loss": 1.7659,
+      "step": 1641
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.367500993245927e-07,
+      "loss": 1.9844,
+      "step": 1642
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.36710369487485e-07,
+      "loss": 1.8912,
+      "step": 1643
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.366706396503774e-07,
+      "loss": 1.8013,
+      "step": 1644
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.366309098132697e-07,
+      "loss": 2.0088,
+      "step": 1645
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.365911799761621e-07,
+      "loss": 1.7839,
+      "step": 1646
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.365514501390544e-07,
+      "loss": 1.9062,
+      "step": 1647
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.365117203019467e-07,
+      "loss": 1.8911,
+      "step": 1648
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.364719904648391e-07,
+      "loss": 2.0073,
+      "step": 1649
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.364322606277314e-07,
+      "loss": 1.8392,
+      "step": 1650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.363925307906238e-07,
+      "loss": 1.9443,
+      "step": 1651
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.363528009535161e-07,
+      "loss": 1.8739,
+      "step": 1652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.363130711164083e-07,
+      "loss": 1.8493,
+      "step": 1653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.362733412793006e-07,
+      "loss": 1.8783,
+      "step": 1654
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.36233611442193e-07,
+      "loss": 1.7854,
+      "step": 1655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.361938816050854e-07,
+      "loss": 2.0891,
+      "step": 1656
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.361541517679777e-07,
+      "loss": 1.853,
+      "step": 1657
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.361144219308701e-07,
+      "loss": 1.6598,
+      "step": 1658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.360746920937623e-07,
+      "loss": 1.7621,
+      "step": 1659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.360349622566547e-07,
+      "loss": 1.9203,
+      "step": 1660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.35995232419547e-07,
+      "loss": 1.6833,
+      "step": 1661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.359555025824394e-07,
+      "loss": 2.1578,
+      "step": 1662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.359157727453318e-07,
+      "loss": 1.78,
+      "step": 1663
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.358760429082241e-07,
+      "loss": 1.8471,
+      "step": 1664
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.358363130711164e-07,
+      "loss": 1.8486,
+      "step": 1665
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.357965832340086e-07,
+      "loss": 1.8959,
+      "step": 1666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.35756853396901e-07,
+      "loss": 1.9574,
+      "step": 1667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.357171235597933e-07,
+      "loss": 1.9274,
+      "step": 1668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.356773937226857e-07,
+      "loss": 1.9616,
+      "step": 1669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.356376638855781e-07,
+      "loss": 1.8737,
+      "step": 1670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.355979340484703e-07,
+      "loss": 1.8954,
+      "step": 1671
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.355582042113627e-07,
+      "loss": 1.9592,
+      "step": 1672
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.35518474374255e-07,
+      "loss": 1.8378,
+      "step": 1673
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.354787445371474e-07,
+      "loss": 1.8851,
+      "step": 1674
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.354390147000397e-07,
+      "loss": 1.8516,
+      "step": 1675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.353992848629321e-07,
+      "loss": 1.899,
+      "step": 1676
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.353595550258243e-07,
+      "loss": 1.857,
+      "step": 1677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.353198251887167e-07,
+      "loss": 1.6631,
+      "step": 1678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.352800953516091e-07,
+      "loss": 1.8199,
+      "step": 1679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.352403655145013e-07,
+      "loss": 1.8779,
+      "step": 1680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.352006356773937e-07,
+      "loss": 1.8315,
+      "step": 1681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.35160905840286e-07,
+      "loss": 1.7726,
+      "step": 1682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.351211760031783e-07,
+      "loss": 2.023,
+      "step": 1683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.350814461660706e-07,
+      "loss": 1.8984,
+      "step": 1684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.35041716328963e-07,
+      "loss": 1.8639,
+      "step": 1685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.350019864918554e-07,
+      "loss": 1.7974,
+      "step": 1686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.349622566547477e-07,
+      "loss": 1.8058,
+      "step": 1687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.349225268176401e-07,
+      "loss": 2.0053,
+      "step": 1688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.348827969805323e-07,
+      "loss": 1.9751,
+      "step": 1689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.348430671434247e-07,
+      "loss": 1.7422,
+      "step": 1690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.34803337306317e-07,
+      "loss": 1.8542,
+      "step": 1691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.347636074692094e-07,
+      "loss": 1.8782,
+      "step": 1692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.347238776321016e-07,
+      "loss": 1.8872,
+      "step": 1693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.34684147794994e-07,
+      "loss": 1.9138,
+      "step": 1694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.346444179578863e-07,
+      "loss": 1.7643,
+      "step": 1695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.346046881207786e-07,
+      "loss": 2.1127,
+      "step": 1696
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.34564958283671e-07,
+      "loss": 1.8684,
+      "step": 1697
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.345252284465633e-07,
+      "loss": 1.8036,
+      "step": 1698
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.344854986094557e-07,
+      "loss": 1.7935,
+      "step": 1699
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.34445768772348e-07,
+      "loss": 2.1483,
+      "step": 1700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.344060389352403e-07,
+      "loss": 1.8576,
+      "step": 1701
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.343663090981327e-07,
+      "loss": 1.8488,
+      "step": 1702
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.34326579261025e-07,
+      "loss": 1.883,
+      "step": 1703
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.342868494239174e-07,
+      "loss": 1.6922,
+      "step": 1704
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.342471195868097e-07,
+      "loss": 1.846,
+      "step": 1705
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.34207389749702e-07,
+      "loss": 1.9266,
+      "step": 1706
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.341676599125942e-07,
+      "loss": 1.9678,
+      "step": 1707
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.341279300754866e-07,
+      "loss": 2.0118,
+      "step": 1708
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.34088200238379e-07,
+      "loss": 1.7886,
+      "step": 1709
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.340484704012713e-07,
+      "loss": 1.9398,
+      "step": 1710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.340087405641637e-07,
+      "loss": 1.979,
+      "step": 1711
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.33969010727056e-07,
+      "loss": 1.9022,
+      "step": 1712
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.339292808899483e-07,
+      "loss": 1.7205,
+      "step": 1713
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.338895510528406e-07,
+      "loss": 1.8964,
+      "step": 1714
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.33849821215733e-07,
+      "loss": 1.7635,
+      "step": 1715
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.338100913786253e-07,
+      "loss": 1.8035,
+      "step": 1716
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.337703615415177e-07,
+      "loss": 1.9883,
+      "step": 1717
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.337306317044101e-07,
+      "loss": 1.9366,
+      "step": 1718
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.336909018673022e-07,
+      "loss": 2.0303,
+      "step": 1719
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.336511720301946e-07,
+      "loss": 1.9736,
+      "step": 1720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.336114421930869e-07,
+      "loss": 2.0976,
+      "step": 1721
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.335717123559793e-07,
+      "loss": 1.8861,
+      "step": 1722
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.335319825188716e-07,
+      "loss": 2.1389,
+      "step": 1723
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.33492252681764e-07,
+      "loss": 1.904,
+      "step": 1724
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.334525228446563e-07,
+      "loss": 1.8235,
+      "step": 1725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.334127930075486e-07,
+      "loss": 1.9619,
+      "step": 1726
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.33373063170441e-07,
+      "loss": 1.8829,
+      "step": 1727
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.333333333333333e-07,
+      "loss": 1.926,
+      "step": 1728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.332936034962257e-07,
+      "loss": 2.0846,
+      "step": 1729
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.33253873659118e-07,
+      "loss": 1.6051,
+      "step": 1730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.332141438220103e-07,
+      "loss": 1.7896,
+      "step": 1731
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.331744139849025e-07,
+      "loss": 2.0517,
+      "step": 1732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.331346841477949e-07,
+      "loss": 1.7057,
+      "step": 1733
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.330949543106873e-07,
+      "loss": 1.719,
+      "step": 1734
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.330552244735796e-07,
+      "loss": 1.8471,
+      "step": 1735
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.33015494636472e-07,
+      "loss": 1.8726,
+      "step": 1736
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.329757647993642e-07,
+      "loss": 1.7146,
+      "step": 1737
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.329360349622566e-07,
+      "loss": 1.7798,
+      "step": 1738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.328963051251489e-07,
+      "loss": 1.8059,
+      "step": 1739
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.328565752880413e-07,
+      "loss": 1.777,
+      "step": 1740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.328168454509337e-07,
+      "loss": 1.9612,
+      "step": 1741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.32777115613826e-07,
+      "loss": 1.8784,
+      "step": 1742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.327373857767183e-07,
+      "loss": 2.0038,
+      "step": 1743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.326976559396106e-07,
+      "loss": 1.7719,
+      "step": 1744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.32657926102503e-07,
+      "loss": 1.7272,
+      "step": 1745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.326181962653952e-07,
+      "loss": 1.8271,
+      "step": 1746
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.325784664282876e-07,
+      "loss": 1.916,
+      "step": 1747
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.3253873659118e-07,
+      "loss": 1.8337,
+      "step": 1748
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.324990067540722e-07,
+      "loss": 2.3227,
+      "step": 1749
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.324592769169646e-07,
+      "loss": 1.9001,
+      "step": 1750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.324195470798569e-07,
+      "loss": 1.9088,
+      "step": 1751
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.323798172427493e-07,
+      "loss": 2.0123,
+      "step": 1752
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.323400874056416e-07,
+      "loss": 1.9524,
+      "step": 1753
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.32300357568534e-07,
+      "loss": 1.9379,
+      "step": 1754
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.322606277314262e-07,
+      "loss": 1.9,
+      "step": 1755
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.322208978943186e-07,
+      "loss": 1.8636,
+      "step": 1756
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.32181168057211e-07,
+      "loss": 2.283,
+      "step": 1757
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.321414382201033e-07,
+      "loss": 1.8504,
+      "step": 1758
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.321017083829956e-07,
+      "loss": 1.873,
+      "step": 1759
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.320619785458879e-07,
+      "loss": 1.7583,
+      "step": 1760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.320222487087802e-07,
+      "loss": 1.751,
+      "step": 1761
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.319825188716725e-07,
+      "loss": 1.8069,
+      "step": 1762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.319427890345649e-07,
+      "loss": 1.6744,
+      "step": 1763
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.319030591974573e-07,
+      "loss": 1.7604,
+      "step": 1764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.318633293603496e-07,
+      "loss": 1.6618,
+      "step": 1765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.31823599523242e-07,
+      "loss": 1.7387,
+      "step": 1766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.317838696861342e-07,
+      "loss": 1.8767,
+      "step": 1767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.317441398490266e-07,
+      "loss": 1.874,
+      "step": 1768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.317044100119189e-07,
+      "loss": 1.9809,
+      "step": 1769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.316646801748113e-07,
+      "loss": 1.6888,
+      "step": 1770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.316249503377037e-07,
+      "loss": 1.8877,
+      "step": 1771
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.315852205005959e-07,
+      "loss": 1.9185,
+      "step": 1772
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.315454906634882e-07,
+      "loss": 2.173,
+      "step": 1773
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.315057608263805e-07,
+      "loss": 1.7635,
+      "step": 1774
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.314660309892729e-07,
+      "loss": 1.8297,
+      "step": 1775
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.314263011521652e-07,
+      "loss": 1.8624,
+      "step": 1776
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.313865713150576e-07,
+      "loss": 2.1627,
+      "step": 1777
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.3134684147795e-07,
+      "loss": 1.9688,
+      "step": 1778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.313071116408422e-07,
+      "loss": 1.6902,
+      "step": 1779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.312673818037346e-07,
+      "loss": 1.9911,
+      "step": 1780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.312276519666269e-07,
+      "loss": 2.0373,
+      "step": 1781
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.311879221295193e-07,
+      "loss": 1.8728,
+      "step": 1782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.311481922924116e-07,
+      "loss": 1.8268,
+      "step": 1783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.31108462455304e-07,
+      "loss": 2.0235,
+      "step": 1784
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.310687326181961e-07,
+      "loss": 1.8652,
+      "step": 1785
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.310290027810885e-07,
+      "loss": 1.7614,
+      "step": 1786
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.309892729439809e-07,
+      "loss": 1.9394,
+      "step": 1787
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.309495431068732e-07,
+      "loss": 2.0649,
+      "step": 1788
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.309098132697656e-07,
+      "loss": 1.9198,
+      "step": 1789
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.308700834326579e-07,
+      "loss": 1.9826,
+      "step": 1790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.308303535955502e-07,
+      "loss": 2.1059,
+      "step": 1791
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.307906237584425e-07,
+      "loss": 1.6396,
+      "step": 1792
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.307508939213349e-07,
+      "loss": 1.9393,
+      "step": 1793
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.307111640842273e-07,
+      "loss": 2.1312,
+      "step": 1794
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.306714342471196e-07,
+      "loss": 1.7144,
+      "step": 1795
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.30631704410012e-07,
+      "loss": 2.0377,
+      "step": 1796
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.305919745729042e-07,
+      "loss": 2.0489,
+      "step": 1797
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.305522447357965e-07,
+      "loss": 1.8102,
+      "step": 1798
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.305125148986888e-07,
+      "loss": 1.8999,
+      "step": 1799
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.304727850615812e-07,
+      "loss": 1.8934,
+      "step": 1800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.304330552244735e-07,
+      "loss": 1.9802,
+      "step": 1801
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.303933253873659e-07,
+      "loss": 1.8598,
+      "step": 1802
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.303535955502582e-07,
+      "loss": 1.8184,
+      "step": 1803
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.303138657131505e-07,
+      "loss": 1.9592,
+      "step": 1804
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.302741358760429e-07,
+      "loss": 1.8069,
+      "step": 1805
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.302344060389352e-07,
+      "loss": 1.9284,
+      "step": 1806
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.301946762018276e-07,
+      "loss": 1.9716,
+      "step": 1807
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.301549463647199e-07,
+      "loss": 1.7598,
+      "step": 1808
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.301152165276122e-07,
+      "loss": 1.8174,
+      "step": 1809
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.300754866905046e-07,
+      "loss": 1.9736,
+      "step": 1810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.300357568533969e-07,
+      "loss": 2.0083,
+      "step": 1811
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.299960270162892e-07,
+      "loss": 1.7683,
+      "step": 1812
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.299562971791815e-07,
+      "loss": 1.6539,
+      "step": 1813
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.299165673420739e-07,
+      "loss": 1.9144,
+      "step": 1814
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.298768375049661e-07,
+      "loss": 1.9427,
+      "step": 1815
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.298371076678585e-07,
+      "loss": 1.7701,
+      "step": 1816
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.297973778307508e-07,
+      "loss": 1.9803,
+      "step": 1817
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.297576479936432e-07,
+      "loss": 1.7313,
+      "step": 1818
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.297179181565356e-07,
+      "loss": 1.8594,
+      "step": 1819
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.296781883194279e-07,
+      "loss": 1.991,
+      "step": 1820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.296384584823202e-07,
+      "loss": 1.8589,
+      "step": 1821
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.295987286452125e-07,
+      "loss": 1.907,
+      "step": 1822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.295589988081049e-07,
+      "loss": 1.9116,
+      "step": 1823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.295192689709972e-07,
+      "loss": 1.8362,
+      "step": 1824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.294795391338895e-07,
+      "loss": 1.8283,
+      "step": 1825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.294398092967819e-07,
+      "loss": 1.916,
+      "step": 1826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.294000794596741e-07,
+      "loss": 1.939,
+      "step": 1827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.293603496225665e-07,
+      "loss": 2.1165,
+      "step": 1828
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.293206197854588e-07,
+      "loss": 1.9945,
+      "step": 1829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.292808899483512e-07,
+      "loss": 1.8235,
+      "step": 1830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.292411601112435e-07,
+      "loss": 1.7866,
+      "step": 1831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.292014302741359e-07,
+      "loss": 1.8862,
+      "step": 1832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.291617004370281e-07,
+      "loss": 1.7874,
+      "step": 1833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.291219705999205e-07,
+      "loss": 1.9421,
+      "step": 1834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.290822407628129e-07,
+      "loss": 1.9353,
+      "step": 1835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.290425109257052e-07,
+      "loss": 1.8465,
+      "step": 1836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.290027810885976e-07,
+      "loss": 1.9925,
+      "step": 1837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.289630512514898e-07,
+      "loss": 1.833,
+      "step": 1838
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.289233214143821e-07,
+      "loss": 2.0272,
+      "step": 1839
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.288835915772744e-07,
+      "loss": 2.0018,
+      "step": 1840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.288438617401668e-07,
+      "loss": 1.8936,
+      "step": 1841
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.288041319030592e-07,
+      "loss": 1.9495,
+      "step": 1842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.287644020659515e-07,
+      "loss": 1.9557,
+      "step": 1843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.287246722288439e-07,
+      "loss": 1.8058,
+      "step": 1844
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.286849423917361e-07,
+      "loss": 2.1592,
+      "step": 1845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.286452125546285e-07,
+      "loss": 1.8278,
+      "step": 1846
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.286054827175208e-07,
+      "loss": 1.9554,
+      "step": 1847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.285657528804132e-07,
+      "loss": 1.9836,
+      "step": 1848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.285260230433056e-07,
+      "loss": 2.1647,
+      "step": 1849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.284862932061978e-07,
+      "loss": 1.7337,
+      "step": 1850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.284465633690901e-07,
+      "loss": 1.8313,
+      "step": 1851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.284068335319824e-07,
+      "loss": 1.8824,
+      "step": 1852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.283671036948748e-07,
+      "loss": 1.8811,
+      "step": 1853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.283273738577671e-07,
+      "loss": 1.7837,
+      "step": 1854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.282876440206595e-07,
+      "loss": 2.0729,
+      "step": 1855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.282479141835517e-07,
+      "loss": 2.0121,
+      "step": 1856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.282081843464441e-07,
+      "loss": 1.8249,
+      "step": 1857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.281684545093365e-07,
+      "loss": 2.116,
+      "step": 1858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.281287246722288e-07,
+      "loss": 1.6168,
+      "step": 1859
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.280889948351212e-07,
+      "loss": 1.8735,
+      "step": 1860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.280492649980135e-07,
+      "loss": 2.0018,
+      "step": 1861
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.280095351609058e-07,
+      "loss": 1.9869,
+      "step": 1862
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.279698053237981e-07,
+      "loss": 1.8309,
+      "step": 1863
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.279300754866904e-07,
+      "loss": 1.9286,
+      "step": 1864
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.278903456495828e-07,
+      "loss": 2.0787,
+      "step": 1865
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.278506158124751e-07,
+      "loss": 1.9449,
+      "step": 1866
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.278108859753675e-07,
+      "loss": 1.9787,
+      "step": 1867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.277711561382597e-07,
+      "loss": 1.8486,
+      "step": 1868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.277314263011521e-07,
+      "loss": 2.005,
+      "step": 1869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.276916964640444e-07,
+      "loss": 1.957,
+      "step": 1870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.276519666269368e-07,
+      "loss": 1.9242,
+      "step": 1871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.276122367898292e-07,
+      "loss": 1.9765,
+      "step": 1872
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.275725069527215e-07,
+      "loss": 1.8087,
+      "step": 1873
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.275327771156138e-07,
+      "loss": 1.953,
+      "step": 1874
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.274930472785061e-07,
+      "loss": 1.8847,
+      "step": 1875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.274533174413985e-07,
+      "loss": 1.8285,
+      "step": 1876
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.274135876042908e-07,
+      "loss": 1.7905,
+      "step": 1877
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.273738577671831e-07,
+      "loss": 1.8116,
+      "step": 1878
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.273341279300754e-07,
+      "loss": 1.8832,
+      "step": 1879
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.272943980929677e-07,
+      "loss": 1.787,
+      "step": 1880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.272546682558601e-07,
+      "loss": 2.0739,
+      "step": 1881
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.272149384187524e-07,
+      "loss": 1.9139,
+      "step": 1882
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.271752085816448e-07,
+      "loss": 1.9826,
+      "step": 1883
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.271354787445371e-07,
+      "loss": 1.9025,
+      "step": 1884
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.270957489074295e-07,
+      "loss": 2.0218,
+      "step": 1885
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.270560190703217e-07,
+      "loss": 1.9572,
+      "step": 1886
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.270162892332141e-07,
+      "loss": 1.7296,
+      "step": 1887
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.269765593961065e-07,
+      "loss": 1.7346,
+      "step": 1888
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.269368295589988e-07,
+      "loss": 1.7869,
+      "step": 1889
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.268970997218912e-07,
+      "loss": 2.0011,
+      "step": 1890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.268573698847834e-07,
+      "loss": 1.9044,
+      "step": 1891
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.268176400476757e-07,
+      "loss": 1.7965,
+      "step": 1892
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.26777910210568e-07,
+      "loss": 1.7902,
+      "step": 1893
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.267381803734604e-07,
+      "loss": 1.9376,
+      "step": 1894
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.266984505363527e-07,
+      "loss": 1.7673,
+      "step": 1895
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.266587206992451e-07,
+      "loss": 1.947,
+      "step": 1896
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.266189908621375e-07,
+      "loss": 1.9384,
+      "step": 1897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.265792610250297e-07,
+      "loss": 1.9288,
+      "step": 1898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.265395311879221e-07,
+      "loss": 1.9331,
+      "step": 1899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.264998013508144e-07,
+      "loss": 1.8094,
+      "step": 1900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.264600715137068e-07,
+      "loss": 1.7805,
+      "step": 1901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.264203416765991e-07,
+      "loss": 1.7083,
+      "step": 1902
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.263806118394915e-07,
+      "loss": 1.8439,
+      "step": 1903
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.263408820023836e-07,
+      "loss": 1.7821,
+      "step": 1904
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.26301152165276e-07,
+      "loss": 1.9113,
+      "step": 1905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.262614223281684e-07,
+      "loss": 1.6972,
+      "step": 1906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.262216924910607e-07,
+      "loss": 1.8431,
+      "step": 1907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.261819626539531e-07,
+      "loss": 1.7593,
+      "step": 1908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.261422328168454e-07,
+      "loss": 1.7474,
+      "step": 1909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.261025029797377e-07,
+      "loss": 1.8455,
+      "step": 1910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.2606277314263e-07,
+      "loss": 2.0669,
+      "step": 1911
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.260230433055224e-07,
+      "loss": 1.9536,
+      "step": 1912
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.259833134684148e-07,
+      "loss": 1.9174,
+      "step": 1913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.259435836313071e-07,
+      "loss": 1.8251,
+      "step": 1914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.259038537941995e-07,
+      "loss": 1.734,
+      "step": 1915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.258641239570917e-07,
+      "loss": 1.9813,
+      "step": 1916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.25824394119984e-07,
+      "loss": 1.8549,
+      "step": 1917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.257846642828763e-07,
+      "loss": 1.7464,
+      "step": 1918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.257449344457687e-07,
+      "loss": 1.8039,
+      "step": 1919
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.257052046086611e-07,
+      "loss": 1.8863,
+      "step": 1920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.256654747715534e-07,
+      "loss": 1.7935,
+      "step": 1921
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.256257449344457e-07,
+      "loss": 1.949,
+      "step": 1922
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.25586015097338e-07,
+      "loss": 1.8643,
+      "step": 1923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.255462852602304e-07,
+      "loss": 1.6937,
+      "step": 1924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.255065554231227e-07,
+      "loss": 1.9354,
+      "step": 1925
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.254668255860151e-07,
+      "loss": 1.7525,
+      "step": 1926
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.254270957489075e-07,
+      "loss": 2.0547,
+      "step": 1927
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.253873659117997e-07,
+      "loss": 1.7947,
+      "step": 1928
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.253476360746921e-07,
+      "loss": 2.0199,
+      "step": 1929
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.253079062375843e-07,
+      "loss": 2.0431,
+      "step": 1930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.252681764004767e-07,
+      "loss": 1.8873,
+      "step": 1931
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.25228446563369e-07,
+      "loss": 1.9589,
+      "step": 1932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.251887167262614e-07,
+      "loss": 2.1706,
+      "step": 1933
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.251489868891536e-07,
+      "loss": 1.7862,
+      "step": 1934
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.25109257052046e-07,
+      "loss": 1.7622,
+      "step": 1935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.250695272149384e-07,
+      "loss": 1.8987,
+      "step": 1936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.250297973778307e-07,
+      "loss": 1.7645,
+      "step": 1937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.249900675407231e-07,
+      "loss": 1.8755,
+      "step": 1938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.249503377036154e-07,
+      "loss": 1.7408,
+      "step": 1939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.249106078665077e-07,
+      "loss": 1.9528,
+      "step": 1940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.248708780294e-07,
+      "loss": 1.671,
+      "step": 1941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.248311481922924e-07,
+      "loss": 1.9489,
+      "step": 1942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.247914183551848e-07,
+      "loss": 1.6957,
+      "step": 1943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.24751688518077e-07,
+      "loss": 1.7815,
+      "step": 1944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.247119586809694e-07,
+      "loss": 1.8497,
+      "step": 1945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.246722288438616e-07,
+      "loss": 1.8093,
+      "step": 1946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.24632499006754e-07,
+      "loss": 1.9312,
+      "step": 1947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.245927691696463e-07,
+      "loss": 1.8755,
+      "step": 1948
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.245530393325387e-07,
+      "loss": 1.9449,
+      "step": 1949
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.24513309495431e-07,
+      "loss": 1.9041,
+      "step": 1950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.244735796583234e-07,
+      "loss": 1.8984,
+      "step": 1951
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.244338498212157e-07,
+      "loss": 1.689,
+      "step": 1952
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.24394119984108e-07,
+      "loss": 1.8895,
+      "step": 1953
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.243543901470004e-07,
+      "loss": 1.8571,
+      "step": 1954
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.243146603098927e-07,
+      "loss": 1.8389,
+      "step": 1955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.242749304727851e-07,
+      "loss": 1.8485,
+      "step": 1956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.242352006356773e-07,
+      "loss": 1.8963,
+      "step": 1957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.241954707985696e-07,
+      "loss": 1.9615,
+      "step": 1958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.24155740961462e-07,
+      "loss": 1.7126,
+      "step": 1959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.241160111243543e-07,
+      "loss": 1.9303,
+      "step": 1960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.240762812872467e-07,
+      "loss": 2.0118,
+      "step": 1961
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.24036551450139e-07,
+      "loss": 1.7138,
+      "step": 1962
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.239968216130314e-07,
+      "loss": 2.233,
+      "step": 1963
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.239570917759236e-07,
+      "loss": 2.0635,
+      "step": 1964
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.23917361938816e-07,
+      "loss": 1.8521,
+      "step": 1965
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.238776321017084e-07,
+      "loss": 1.7453,
+      "step": 1966
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.238379022646007e-07,
+      "loss": 1.9212,
+      "step": 1967
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.237981724274931e-07,
+      "loss": 1.8786,
+      "step": 1968
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.237584425903854e-07,
+      "loss": 1.9567,
+      "step": 1969
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.237187127532776e-07,
+      "loss": 2.0487,
+      "step": 1970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.236789829161699e-07,
+      "loss": 2.1364,
+      "step": 1971
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.236392530790623e-07,
+      "loss": 1.9668,
+      "step": 1972
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.235995232419546e-07,
+      "loss": 2.1858,
+      "step": 1973
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.23559793404847e-07,
+      "loss": 1.8098,
+      "step": 1974
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.235200635677394e-07,
+      "loss": 1.7598,
+      "step": 1975
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.234803337306316e-07,
+      "loss": 1.7789,
+      "step": 1976
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.23440603893524e-07,
+      "loss": 1.9722,
+      "step": 1977
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.234008740564163e-07,
+      "loss": 1.9132,
+      "step": 1978
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.233611442193087e-07,
+      "loss": 1.8584,
+      "step": 1979
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.23321414382201e-07,
+      "loss": 1.8135,
+      "step": 1980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.232816845450934e-07,
+      "loss": 1.755,
+      "step": 1981
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.232419547079857e-07,
+      "loss": 1.7303,
+      "step": 1982
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.232022248708779e-07,
+      "loss": 1.8011,
+      "step": 1983
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.231624950337703e-07,
+      "loss": 1.948,
+      "step": 1984
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.231227651966626e-07,
+      "loss": 2.034,
+      "step": 1985
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.23083035359555e-07,
+      "loss": 1.9512,
+      "step": 1986
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.230433055224473e-07,
+      "loss": 2.0339,
+      "step": 1987
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.230035756853396e-07,
+      "loss": 1.9381,
+      "step": 1988
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.22963845848232e-07,
+      "loss": 2.0065,
+      "step": 1989
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.229241160111243e-07,
+      "loss": 1.908,
+      "step": 1990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.228843861740167e-07,
+      "loss": 1.8774,
+      "step": 1991
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.22844656336909e-07,
+      "loss": 1.8111,
+      "step": 1992
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.228049264998014e-07,
+      "loss": 1.7912,
+      "step": 1993
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.227651966626936e-07,
+      "loss": 1.8638,
+      "step": 1994
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.22725466825586e-07,
+      "loss": 1.8903,
+      "step": 1995
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.226857369884782e-07,
+      "loss": 1.8296,
+      "step": 1996
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.226460071513706e-07,
+      "loss": 1.8723,
+      "step": 1997
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.22606277314263e-07,
+      "loss": 1.752,
+      "step": 1998
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.225665474771553e-07,
+      "loss": 1.8504,
+      "step": 1999
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.225268176400476e-07,
+      "loss": 1.8619,
+      "step": 2000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.224870878029399e-07,
+      "loss": 1.9827,
+      "step": 2001
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.224473579658323e-07,
+      "loss": 1.8823,
+      "step": 2002
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.224076281287246e-07,
+      "loss": 1.9782,
+      "step": 2003
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.22367898291617e-07,
+      "loss": 1.7782,
+      "step": 2004
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.223281684545094e-07,
+      "loss": 1.8991,
+      "step": 2005
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.222884386174016e-07,
+      "loss": 1.7577,
+      "step": 2006
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.22248708780294e-07,
+      "loss": 1.8362,
+      "step": 2007
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.222089789431863e-07,
+      "loss": 1.8589,
+      "step": 2008
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.221692491060787e-07,
+      "loss": 1.9351,
+      "step": 2009
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.221295192689709e-07,
+      "loss": 1.8241,
+      "step": 2010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.220897894318633e-07,
+      "loss": 2.0449,
+      "step": 2011
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.220500595947555e-07,
+      "loss": 1.9455,
+      "step": 2012
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.220103297576479e-07,
+      "loss": 1.678,
+      "step": 2013
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.219705999205403e-07,
+      "loss": 1.9023,
+      "step": 2014
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.219308700834326e-07,
+      "loss": 1.9371,
+      "step": 2015
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.21891140246325e-07,
+      "loss": 1.8313,
+      "step": 2016
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.218514104092173e-07,
+      "loss": 1.9198,
+      "step": 2017
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.218116805721096e-07,
+      "loss": 1.6567,
+      "step": 2018
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.217719507350019e-07,
+      "loss": 1.8421,
+      "step": 2019
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.217322208978943e-07,
+      "loss": 1.9044,
+      "step": 2020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.216924910607867e-07,
+      "loss": 1.9409,
+      "step": 2021
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.21652761223679e-07,
+      "loss": 1.9075,
+      "step": 2022
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.216130313865713e-07,
+      "loss": 1.8589,
+      "step": 2023
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.215733015494635e-07,
+      "loss": 1.8128,
+      "step": 2024
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.215335717123559e-07,
+      "loss": 1.956,
+      "step": 2025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.214938418752482e-07,
+      "loss": 1.7586,
+      "step": 2026
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.214541120381406e-07,
+      "loss": 1.7734,
+      "step": 2027
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.21414382201033e-07,
+      "loss": 1.8002,
+      "step": 2028
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.213746523639253e-07,
+      "loss": 1.7953,
+      "step": 2029
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.213349225268176e-07,
+      "loss": 1.9909,
+      "step": 2030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.212951926897099e-07,
+      "loss": 1.6872,
+      "step": 2031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.212554628526023e-07,
+      "loss": 1.8797,
+      "step": 2032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.212157330154946e-07,
+      "loss": 2.1024,
+      "step": 2033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.21176003178387e-07,
+      "loss": 1.9357,
+      "step": 2034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.211362733412793e-07,
+      "loss": 1.8908,
+      "step": 2035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.210965435041715e-07,
+      "loss": 1.8634,
+      "step": 2036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210568136670639e-07,
+      "loss": 1.8326,
+      "step": 2037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210170838299562e-07,
+      "loss": 1.7461,
+      "step": 2038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209773539928486e-07,
+      "loss": 1.7455,
+      "step": 2039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209376241557409e-07,
+      "loss": 1.7211,
+      "step": 2040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208978943186333e-07,
+      "loss": 1.7903,
+      "step": 2041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208581644815255e-07,
+      "loss": 1.7244,
+      "step": 2042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208184346444179e-07,
+      "loss": 1.755,
+      "step": 2043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207787048073103e-07,
+      "loss": 1.6925,
+      "step": 2044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207389749702026e-07,
+      "loss": 2.0647,
+      "step": 2045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20699245133095e-07,
+      "loss": 2.0807,
+      "step": 2046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206595152959873e-07,
+      "loss": 1.9295,
+      "step": 2047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206197854588796e-07,
+      "loss": 1.7904,
+      "step": 2048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205800556217718e-07,
+      "loss": 1.9022,
+      "step": 2049
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.205403257846642e-07,
+      "loss": 2.103,
+      "step": 2050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.205005959475565e-07,
+      "loss": 2.0145,
+      "step": 2051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.204608661104489e-07,
+      "loss": 1.7841,
+      "step": 2052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.204211362733413e-07,
+      "loss": 1.9257,
+      "step": 2053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.203814064362335e-07,
+      "loss": 1.9259,
+      "step": 2054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.203416765991259e-07,
+      "loss": 1.864,
+      "step": 2055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.203019467620182e-07,
+      "loss": 1.6693,
+      "step": 2056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.202622169249106e-07,
+      "loss": 1.7625,
+      "step": 2057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.202224870878029e-07,
+      "loss": 2.0909,
+      "step": 2058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.201827572506953e-07,
+      "loss": 2.0146,
+      "step": 2059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.201430274135876e-07,
+      "loss": 1.6975,
+      "step": 2060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.201032975764799e-07,
+      "loss": 1.9201,
+      "step": 2061
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.200635677393722e-07,
+      "loss": 1.8039,
+      "step": 2062
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.200238379022645e-07,
+      "loss": 1.8795,
+      "step": 2063
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.199841080651569e-07,
+      "loss": 1.9203,
+      "step": 2064
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.199443782280492e-07,
+      "loss": 1.9306,
+      "step": 2065
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.199046483909415e-07,
+      "loss": 1.6261,
+      "step": 2066
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.198649185538338e-07,
+      "loss": 1.6894,
+      "step": 2067
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.198251887167262e-07,
+      "loss": 1.7655,
+      "step": 2068
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.197854588796186e-07,
+      "loss": 1.7795,
+      "step": 2069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.197457290425109e-07,
+      "loss": 1.9784,
+      "step": 2070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.197059992054033e-07,
+      "loss": 1.8079,
+      "step": 2071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.196662693682955e-07,
+      "loss": 1.8918,
+      "step": 2072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.196265395311879e-07,
+      "loss": 1.8666,
+      "step": 2073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.195868096940802e-07,
+      "loss": 1.876,
+      "step": 2074
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.195470798569725e-07,
+      "loss": 1.7832,
+      "step": 2075
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.195073500198649e-07,
+      "loss": 1.9842,
+      "step": 2076
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.194676201827572e-07,
+      "loss": 2.2233,
+      "step": 2077
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.194278903456495e-07,
+      "loss": 1.8051,
+      "step": 2078
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.193881605085418e-07,
+      "loss": 1.8256,
+      "step": 2079
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.193484306714342e-07,
+      "loss": 1.786,
+      "step": 2080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.193087008343265e-07,
+      "loss": 1.9534,
+      "step": 2081
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.192689709972189e-07,
+      "loss": 1.6883,
+      "step": 2082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.192292411601113e-07,
+      "loss": 2.0944,
+      "step": 2083
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.191895113230035e-07,
+      "loss": 1.8969,
+      "step": 2084
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.191497814858959e-07,
+      "loss": 1.8229,
+      "step": 2085
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.191100516487882e-07,
+      "loss": 1.7102,
+      "step": 2086
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.190703218116806e-07,
+      "loss": 1.9462,
+      "step": 2087
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.190305919745729e-07,
+      "loss": 2.0288,
+      "step": 2088
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.189908621374652e-07,
+      "loss": 1.8127,
+      "step": 2089
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.189511323003574e-07,
+      "loss": 1.7722,
+      "step": 2090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.189114024632498e-07,
+      "loss": 1.8361,
+      "step": 2091
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.188716726261422e-07,
+      "loss": 1.7048,
+      "step": 2092
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.188319427890345e-07,
+      "loss": 1.8097,
+      "step": 2093
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.187922129519269e-07,
+      "loss": 1.7617,
+      "step": 2094
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.187524831148192e-07,
+      "loss": 2.0074,
+      "step": 2095
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.187127532777115e-07,
+      "loss": 1.75,
+      "step": 2096
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.186730234406038e-07,
+      "loss": 1.9962,
+      "step": 2097
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.186332936034962e-07,
+      "loss": 1.7558,
+      "step": 2098
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.185935637663886e-07,
+      "loss": 1.9119,
+      "step": 2099
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.185538339292809e-07,
+      "loss": 1.9188,
+      "step": 2100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.185141040921733e-07,
+      "loss": 1.9019,
+      "step": 2101
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.184743742550654e-07,
+      "loss": 1.8015,
+      "step": 2102
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.184346444179578e-07,
+      "loss": 1.821,
+      "step": 2103
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.183949145808501e-07,
+      "loss": 1.8589,
+      "step": 2104
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.183551847437425e-07,
+      "loss": 1.9377,
+      "step": 2105
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.183154549066349e-07,
+      "loss": 1.8249,
+      "step": 2106
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.182757250695272e-07,
+      "loss": 1.8332,
+      "step": 2107
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.182359952324195e-07,
+      "loss": 1.84,
+      "step": 2108
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.181962653953118e-07,
+      "loss": 1.8343,
+      "step": 2109
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.181565355582042e-07,
+      "loss": 1.8273,
+      "step": 2110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.181168057210965e-07,
+      "loss": 1.9542,
+      "step": 2111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.180770758839889e-07,
+      "loss": 1.542,
+      "step": 2112
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.180373460468813e-07,
+      "loss": 1.9349,
+      "step": 2113
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.179976162097735e-07,
+      "loss": 1.8765,
+      "step": 2114
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.179578863726658e-07,
+      "loss": 1.78,
+      "step": 2115
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.179181565355581e-07,
+      "loss": 1.8992,
+      "step": 2116
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.178784266984505e-07,
+      "loss": 1.5324,
+      "step": 2117
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.178386968613428e-07,
+      "loss": 1.7544,
+      "step": 2118
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.177989670242352e-07,
+      "loss": 1.8031,
+      "step": 2119
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.177592371871274e-07,
+      "loss": 2.0745,
+      "step": 2120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.177195073500198e-07,
+      "loss": 1.815,
+      "step": 2121
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.176797775129122e-07,
+      "loss": 1.7856,
+      "step": 2122
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.176400476758045e-07,
+      "loss": 1.7165,
+      "step": 2123
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.176003178386969e-07,
+      "loss": 1.8452,
+      "step": 2124
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.175605880015892e-07,
+      "loss": 1.9614,
+      "step": 2125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.175208581644815e-07,
+      "loss": 1.9088,
+      "step": 2126
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.174811283273738e-07,
+      "loss": 2.0634,
+      "step": 2127
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.174413984902661e-07,
+      "loss": 1.9046,
+      "step": 2128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.174016686531584e-07,
+      "loss": 1.9924,
+      "step": 2129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.173619388160508e-07,
+      "loss": 2.0192,
+      "step": 2130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.173222089789432e-07,
+      "loss": 2.0074,
+      "step": 2131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.172824791418354e-07,
+      "loss": 2.0738,
+      "step": 2132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.172427493047278e-07,
+      "loss": 2.0124,
+      "step": 2133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.172030194676201e-07,
+      "loss": 1.9881,
+      "step": 2134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.171632896305125e-07,
+      "loss": 1.9255,
+      "step": 2135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.171235597934048e-07,
+      "loss": 1.7899,
+      "step": 2136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.170838299562972e-07,
+      "loss": 1.7825,
+      "step": 2137
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.170441001191895e-07,
+      "loss": 2.0576,
+      "step": 2138
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.170043702820818e-07,
+      "loss": 1.8212,
+      "step": 2139
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.169646404449742e-07,
+      "loss": 1.9165,
+      "step": 2140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.169249106078664e-07,
+      "loss": 1.8187,
+      "step": 2141
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.168851807707588e-07,
+      "loss": 1.9697,
+      "step": 2142
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.168454509336511e-07,
+      "loss": 2.1476,
+      "step": 2143
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.168057210965434e-07,
+      "loss": 1.7934,
+      "step": 2144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.167659912594357e-07,
+      "loss": 1.7895,
+      "step": 2145
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.167262614223281e-07,
+      "loss": 1.9512,
+      "step": 2146
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.166865315852205e-07,
+      "loss": 1.9268,
+      "step": 2147
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.166468017481128e-07,
+      "loss": 1.8175,
+      "step": 2148
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.166070719110052e-07,
+      "loss": 1.8617,
+      "step": 2149
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.165673420738974e-07,
+      "loss": 2.1599,
+      "step": 2150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.165276122367898e-07,
+      "loss": 1.6683,
+      "step": 2151
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.164878823996821e-07,
+      "loss": 1.9833,
+      "step": 2152
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.164481525625745e-07,
+      "loss": 1.9899,
+      "step": 2153
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.164084227254669e-07,
+      "loss": 2.1488,
+      "step": 2154
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.163686928883591e-07,
+      "loss": 1.7522,
+      "step": 2155
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.163289630512514e-07,
+      "loss": 2.0335,
+      "step": 2156
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.162892332141437e-07,
+      "loss": 1.6535,
+      "step": 2157
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.162495033770361e-07,
+      "loss": 1.8263,
+      "step": 2158
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.162097735399284e-07,
+      "loss": 1.8956,
+      "step": 2159
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.161700437028208e-07,
+      "loss": 1.9555,
+      "step": 2160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.161303138657132e-07,
+      "loss": 1.8879,
+      "step": 2161
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.160905840286054e-07,
+      "loss": 1.7863,
+      "step": 2162
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.160508541914978e-07,
+      "loss": 2.0244,
+      "step": 2163
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.160111243543901e-07,
+      "loss": 1.9889,
+      "step": 2164
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.159713945172825e-07,
+      "loss": 1.7998,
+      "step": 2165
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.159316646801748e-07,
+      "loss": 1.7984,
+      "step": 2166
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.158919348430672e-07,
+      "loss": 1.9758,
+      "step": 2167
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.158522050059593e-07,
+      "loss": 2.0976,
+      "step": 2168
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.158124751688517e-07,
+      "loss": 1.7373,
+      "step": 2169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.157727453317441e-07,
+      "loss": 1.7887,
+      "step": 2170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.157330154946364e-07,
+      "loss": 1.7254,
+      "step": 2171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.156932856575288e-07,
+      "loss": 1.9135,
+      "step": 2172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.156535558204211e-07,
+      "loss": 1.8171,
+      "step": 2173
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.156138259833134e-07,
+      "loss": 1.7536,
+      "step": 2174
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.155740961462057e-07,
+      "loss": 1.9086,
+      "step": 2175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.155343663090981e-07,
+      "loss": 1.8908,
+      "step": 2176
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.154946364719905e-07,
+      "loss": 1.7082,
+      "step": 2177
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.154549066348828e-07,
+      "loss": 1.8203,
+      "step": 2178
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.154151767977752e-07,
+      "loss": 1.8932,
+      "step": 2179
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.153754469606674e-07,
+      "loss": 1.6754,
+      "step": 2180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.153357171235597e-07,
+      "loss": 1.7066,
+      "step": 2181
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.15295987286452e-07,
+      "loss": 2.1376,
+      "step": 2182
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.152562574493444e-07,
+      "loss": 1.759,
+      "step": 2183
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.152165276122368e-07,
+      "loss": 1.8854,
+      "step": 2184
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.151767977751291e-07,
+      "loss": 1.9539,
+      "step": 2185
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.151370679380214e-07,
+      "loss": 1.8034,
+      "step": 2186
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.150973381009137e-07,
+      "loss": 1.8431,
+      "step": 2187
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.150576082638061e-07,
+      "loss": 1.8476,
+      "step": 2188
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.150178784266984e-07,
+      "loss": 1.5484,
+      "step": 2189
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.149781485895908e-07,
+      "loss": 1.9653,
+      "step": 2190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.149384187524832e-07,
+      "loss": 1.8688,
+      "step": 2191
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.148986889153754e-07,
+      "loss": 1.9061,
+      "step": 2192
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.148589590782678e-07,
+      "loss": 1.7757,
+      "step": 2193
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.1481922924116e-07,
+      "loss": 1.9595,
+      "step": 2194
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.147794994040524e-07,
+      "loss": 1.8624,
+      "step": 2195
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.147397695669447e-07,
+      "loss": 1.9138,
+      "step": 2196
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.147000397298371e-07,
+      "loss": 1.755,
+      "step": 2197
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.146603098927293e-07,
+      "loss": 1.6939,
+      "step": 2198
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.146205800556217e-07,
+      "loss": 1.8214,
+      "step": 2199
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.145808502185141e-07,
+      "loss": 1.8209,
+      "step": 2200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.145411203814064e-07,
+      "loss": 2.2252,
+      "step": 2201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.145013905442988e-07,
+      "loss": 2.0575,
+      "step": 2202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.144616607071911e-07,
+      "loss": 1.765,
+      "step": 2203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.144219308700834e-07,
+      "loss": 1.7646,
+      "step": 2204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.143822010329757e-07,
+      "loss": 1.8661,
+      "step": 2205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.143424711958681e-07,
+      "loss": 1.827,
+      "step": 2206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.143027413587603e-07,
+      "loss": 1.9955,
+      "step": 2207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.142630115216527e-07,
+      "loss": 1.9105,
+      "step": 2208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.142232816845451e-07,
+      "loss": 1.8895,
+      "step": 2209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.141835518474373e-07,
+      "loss": 1.9382,
+      "step": 2210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.141438220103297e-07,
+      "loss": 1.8953,
+      "step": 2211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.14104092173222e-07,
+      "loss": 1.7337,
+      "step": 2212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.140643623361144e-07,
+      "loss": 1.5662,
+      "step": 2213
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.140246324990067e-07,
+      "loss": 1.9572,
+      "step": 2214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.139849026618991e-07,
+      "loss": 1.8374,
+      "step": 2215
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.139451728247914e-07,
+      "loss": 1.7891,
+      "step": 2216
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.139054429876837e-07,
+      "loss": 1.733,
+      "step": 2217
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.138657131505761e-07,
+      "loss": 1.9468,
+      "step": 2218
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.138259833134684e-07,
+      "loss": 1.6461,
+      "step": 2219
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.137862534763608e-07,
+      "loss": 2.0242,
+      "step": 2220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.13746523639253e-07,
+      "loss": 2.0235,
+      "step": 2221
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.137067938021453e-07,
+      "loss": 1.931,
+      "step": 2222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.136670639650376e-07,
+      "loss": 2.0208,
+      "step": 2223
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.1362733412793e-07,
+      "loss": 1.7628,
+      "step": 2224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.135876042908224e-07,
+      "loss": 1.5786,
+      "step": 2225
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.135478744537147e-07,
+      "loss": 1.7271,
+      "step": 2226
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.135081446166071e-07,
+      "loss": 1.7161,
+      "step": 2227
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.134684147794993e-07,
+      "loss": 1.8121,
+      "step": 2228
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.134286849423917e-07,
+      "loss": 1.9414,
+      "step": 2229
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.13388955105284e-07,
+      "loss": 1.9732,
+      "step": 2230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.133492252681764e-07,
+      "loss": 1.6867,
+      "step": 2231
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.133094954310688e-07,
+      "loss": 1.8268,
+      "step": 2232
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.132697655939611e-07,
+      "loss": 1.975,
+      "step": 2233
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.132300357568533e-07,
+      "loss": 1.9822,
+      "step": 2234
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.131903059197456e-07,
+      "loss": 1.7972,
+      "step": 2235
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.13150576082638e-07,
+      "loss": 1.8409,
+      "step": 2236
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.131108462455303e-07,
+      "loss": 1.9297,
+      "step": 2237
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.130711164084227e-07,
+      "loss": 2.1404,
+      "step": 2238
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.130313865713151e-07,
+      "loss": 1.9969,
+      "step": 2239
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.129916567342073e-07,
+      "loss": 1.8453,
+      "step": 2240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.129519268970997e-07,
+      "loss": 1.6883,
+      "step": 2241
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.12912197059992e-07,
+      "loss": 1.9163,
+      "step": 2242
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.128724672228844e-07,
+      "loss": 1.8004,
+      "step": 2243
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.128327373857767e-07,
+      "loss": 1.7068,
+      "step": 2244
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.127930075486691e-07,
+      "loss": 1.6632,
+      "step": 2245
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.127532777115613e-07,
+      "loss": 1.9283,
+      "step": 2246
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.127135478744536e-07,
+      "loss": 1.887,
+      "step": 2247
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.12673818037346e-07,
+      "loss": 1.9388,
+      "step": 2248
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.126340882002383e-07,
+      "loss": 1.8832,
+      "step": 2249
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.125943583631307e-07,
+      "loss": 1.8847,
+      "step": 2250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.12554628526023e-07,
+      "loss": 1.8256,
+      "step": 2251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.125148986889153e-07,
+      "loss": 2.0291,
+      "step": 2252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.124751688518076e-07,
+      "loss": 1.7327,
+      "step": 2253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.124354390147e-07,
+      "loss": 2.0231,
+      "step": 2254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.123957091775924e-07,
+      "loss": 1.7939,
+      "step": 2255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.123559793404847e-07,
+      "loss": 1.9383,
+      "step": 2256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.123162495033771e-07,
+      "loss": 1.8857,
+      "step": 2257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.122765196662693e-07,
+      "loss": 1.6749,
+      "step": 2258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.122367898291617e-07,
+      "loss": 1.8049,
+      "step": 2259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.121970599920539e-07,
+      "loss": 2.0218,
+      "step": 2260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.121573301549463e-07,
+      "loss": 1.9172,
+      "step": 2261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.121176003178387e-07,
+      "loss": 1.8528,
+      "step": 2262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.12077870480731e-07,
+      "loss": 1.8648,
+      "step": 2263
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.120381406436233e-07,
+      "loss": 1.9381,
+      "step": 2264
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.119984108065156e-07,
+      "loss": 1.9766,
+      "step": 2265
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.11958680969408e-07,
+      "loss": 1.8641,
+      "step": 2266
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.119189511323003e-07,
+      "loss": 1.8303,
+      "step": 2267
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.118792212951927e-07,
+      "loss": 1.8784,
+      "step": 2268
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.11839491458085e-07,
+      "loss": 1.8166,
+      "step": 2269
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.117997616209773e-07,
+      "loss": 1.7985,
+      "step": 2270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.117600317838697e-07,
+      "loss": 1.9864,
+      "step": 2271
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.11720301946762e-07,
+      "loss": 1.8541,
+      "step": 2272
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.116805721096543e-07,
+      "loss": 1.9212,
+      "step": 2273
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.116408422725466e-07,
+      "loss": 2.0507,
+      "step": 2274
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.11601112435439e-07,
+      "loss": 1.7457,
+      "step": 2275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.115613825983312e-07,
+      "loss": 1.8132,
+      "step": 2276
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.115216527612236e-07,
+      "loss": 1.7669,
+      "step": 2277
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.11481922924116e-07,
+      "loss": 1.8783,
+      "step": 2278
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.114421930870083e-07,
+      "loss": 2.0406,
+      "step": 2279
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.114024632499007e-07,
+      "loss": 1.8277,
+      "step": 2280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.11362733412793e-07,
+      "loss": 1.9606,
+      "step": 2281
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.113230035756853e-07,
+      "loss": 1.8628,
+      "step": 2282
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.112832737385776e-07,
+      "loss": 1.7172,
+      "step": 2283
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.1124354390147e-07,
+      "loss": 1.8777,
+      "step": 2284
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.112038140643624e-07,
+      "loss": 1.9053,
+      "step": 2285
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.111640842272547e-07,
+      "loss": 1.8749,
+      "step": 2286
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.11124354390147e-07,
+      "loss": 1.7713,
+      "step": 2287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.110846245530392e-07,
+      "loss": 1.9919,
+      "step": 2288
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.110448947159316e-07,
+      "loss": 2.0322,
+      "step": 2289
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.110051648788239e-07,
+      "loss": 1.9115,
+      "step": 2290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.109654350417163e-07,
+      "loss": 2.0012,
+      "step": 2291
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.109257052046086e-07,
+      "loss": 1.8934,
+      "step": 2292
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.10885975367501e-07,
+      "loss": 1.8977,
+      "step": 2293
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.108462455303933e-07,
+      "loss": 2.0317,
+      "step": 2294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.108065156932856e-07,
+      "loss": 1.9039,
+      "step": 2295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.10766785856178e-07,
+      "loss": 2.0607,
+      "step": 2296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.107270560190703e-07,
+      "loss": 2.0528,
+      "step": 2297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.106873261819627e-07,
+      "loss": 2.1777,
+      "step": 2298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.10647596344855e-07,
+      "loss": 2.0269,
+      "step": 2299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.106078665077472e-07,
+      "loss": 1.7149,
+      "step": 2300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.105681366706395e-07,
+      "loss": 1.7161,
+      "step": 2301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.105284068335319e-07,
+      "loss": 1.7628,
+      "step": 2302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.104886769964243e-07,
+      "loss": 1.9759,
+      "step": 2303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.104489471593166e-07,
+      "loss": 1.9203,
+      "step": 2304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.10409217322209e-07,
+      "loss": 1.7506,
+      "step": 2305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.103694874851012e-07,
+      "loss": 1.8099,
+      "step": 2306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.103297576479936e-07,
+      "loss": 1.8783,
+      "step": 2307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.10290027810886e-07,
+      "loss": 1.8001,
+      "step": 2308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.102502979737783e-07,
+      "loss": 1.944,
+      "step": 2309
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.102105681366707e-07,
+      "loss": 1.9107,
+      "step": 2310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.10170838299563e-07,
+      "loss": 1.8029,
+      "step": 2311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.101311084624553e-07,
+      "loss": 1.797,
+      "step": 2312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.100913786253475e-07,
+      "loss": 1.7666,
+      "step": 2313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.100516487882399e-07,
+      "loss": 1.8207,
+      "step": 2314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.100119189511322e-07,
+      "loss": 1.9258,
+      "step": 2315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.099721891140246e-07,
+      "loss": 1.9431,
+      "step": 2316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.09932459276917e-07,
+      "loss": 2.0064,
+      "step": 2317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.098927294398092e-07,
+      "loss": 1.8769,
+      "step": 2318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.098529996027016e-07,
+      "loss": 1.7908,
+      "step": 2319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.098132697655939e-07,
+      "loss": 1.9503,
+      "step": 2320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.097735399284863e-07,
+      "loss": 1.7767,
+      "step": 2321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.097338100913786e-07,
+      "loss": 1.7876,
+      "step": 2322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.09694080254271e-07,
+      "loss": 1.8389,
+      "step": 2323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.096543504171632e-07,
+      "loss": 1.9724,
+      "step": 2324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.096146205800556e-07,
+      "loss": 1.698,
+      "step": 2325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.095748907429479e-07,
+      "loss": 1.7523,
+      "step": 2326
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.095351609058402e-07,
+      "loss": 1.8168,
+      "step": 2327
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.094954310687326e-07,
+      "loss": 1.8324,
+      "step": 2328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.094557012316249e-07,
+      "loss": 1.9359,
+      "step": 2329
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.094159713945172e-07,
+      "loss": 1.9964,
+      "step": 2330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.093762415574095e-07,
+      "loss": 1.8772,
+      "step": 2331
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.093365117203019e-07,
+      "loss": 1.9243,
+      "step": 2332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.092967818831943e-07,
+      "loss": 1.8412,
+      "step": 2333
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.092570520460866e-07,
+      "loss": 1.8822,
+      "step": 2334
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.09217322208979e-07,
+      "loss": 1.8535,
+      "step": 2335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.091775923718712e-07,
+      "loss": 1.8845,
+      "step": 2336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.091378625347636e-07,
+      "loss": 1.7473,
+      "step": 2337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.090981326976559e-07,
+      "loss": 1.7826,
+      "step": 2338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.090584028605482e-07,
+      "loss": 1.824,
+      "step": 2339
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.090186730234406e-07,
+      "loss": 1.8069,
+      "step": 2340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.089789431863329e-07,
+      "loss": 1.7678,
+      "step": 2341
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.089392133492252e-07,
+      "loss": 2.0086,
+      "step": 2342
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.088994835121175e-07,
+      "loss": 1.894,
+      "step": 2343
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.088597536750099e-07,
+      "loss": 1.7655,
+      "step": 2344
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.088200238379022e-07,
+      "loss": 2.1305,
+      "step": 2345
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.087802940007946e-07,
+      "loss": 1.8149,
+      "step": 2346
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.08740564163687e-07,
+      "loss": 1.7633,
+      "step": 2347
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.087008343265792e-07,
+      "loss": 1.668,
+      "step": 2348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.086611044894716e-07,
+      "loss": 1.6687,
+      "step": 2349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.086213746523639e-07,
+      "loss": 1.8646,
+      "step": 2350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.085816448152563e-07,
+      "loss": 2.0609,
+      "step": 2351
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.085419149781486e-07,
+      "loss": 1.9098,
+      "step": 2352
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.085021851410409e-07,
+      "loss": 1.9014,
+      "step": 2353
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.084624553039331e-07,
+      "loss": 1.7393,
+      "step": 2354
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.084227254668255e-07,
+      "loss": 1.7941,
+      "step": 2355
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.083829956297179e-07,
+      "loss": 1.7574,
+      "step": 2356
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.083432657926102e-07,
+      "loss": 1.649,
+      "step": 2357
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.083035359555026e-07,
+      "loss": 2.006,
+      "step": 2358
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.082638061183949e-07,
+      "loss": 2.0458,
+      "step": 2359
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.082240762812872e-07,
+      "loss": 1.7613,
+      "step": 2360
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.081843464441795e-07,
+      "loss": 1.9308,
+      "step": 2361
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.081446166070719e-07,
+      "loss": 1.885,
+      "step": 2362
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.081048867699643e-07,
+      "loss": 1.9949,
+      "step": 2363
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.080651569328566e-07,
+      "loss": 1.693,
+      "step": 2364
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.08025427095749e-07,
+      "loss": 1.8297,
+      "step": 2365
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.079856972586411e-07,
+      "loss": 1.8973,
+      "step": 2366
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.079459674215335e-07,
+      "loss": 1.9139,
+      "step": 2367
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.079062375844258e-07,
+      "loss": 1.8701,
+      "step": 2368
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.078665077473182e-07,
+      "loss": 1.7687,
+      "step": 2369
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.078267779102105e-07,
+      "loss": 1.7669,
+      "step": 2370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.077870480731029e-07,
+      "loss": 1.6602,
+      "step": 2371
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.077473182359952e-07,
+      "loss": 1.8649,
+      "step": 2372
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.077075883988875e-07,
+      "loss": 1.7325,
+      "step": 2373
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.076678585617799e-07,
+      "loss": 1.8902,
+      "step": 2374
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.076281287246722e-07,
+      "loss": 1.6568,
+      "step": 2375
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.075883988875646e-07,
+      "loss": 1.8735,
+      "step": 2376
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.075486690504569e-07,
+      "loss": 1.7914,
+      "step": 2377
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.075089392133492e-07,
+      "loss": 1.7381,
+      "step": 2378
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.074692093762414e-07,
+      "loss": 1.9318,
+      "step": 2379
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.074294795391338e-07,
+      "loss": 1.916,
+      "step": 2380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.073897497020262e-07,
+      "loss": 1.7208,
+      "step": 2381
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.073500198649185e-07,
+      "loss": 2.0392,
+      "step": 2382
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.073102900278109e-07,
+      "loss": 1.8655,
+      "step": 2383
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.072705601907031e-07,
+      "loss": 1.7664,
+      "step": 2384
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.072308303535955e-07,
+      "loss": 1.7708,
+      "step": 2385
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.071911005164878e-07,
+      "loss": 1.8545,
+      "step": 2386
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.071513706793802e-07,
+      "loss": 1.8204,
+      "step": 2387
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.071116408422726e-07,
+      "loss": 1.8655,
+      "step": 2388
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.070719110051649e-07,
+      "loss": 1.8537,
+      "step": 2389
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.070321811680572e-07,
+      "loss": 1.7157,
+      "step": 2390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.069924513309495e-07,
+      "loss": 1.9078,
+      "step": 2391
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.069527214938418e-07,
+      "loss": 1.6909,
+      "step": 2392
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.069129916567341e-07,
+      "loss": 1.8769,
+      "step": 2393
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.068732618196265e-07,
+      "loss": 1.8994,
+      "step": 2394
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.068335319825189e-07,
+      "loss": 1.7981,
+      "step": 2395
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.067938021454111e-07,
+      "loss": 1.7736,
+      "step": 2396
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.067540723083035e-07,
+      "loss": 1.853,
+      "step": 2397
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.067143424711958e-07,
+      "loss": 1.7986,
+      "step": 2398
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.066746126340882e-07,
+      "loss": 1.8396,
+      "step": 2399
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.066348827969805e-07,
+      "loss": 1.733,
+      "step": 2400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.065951529598729e-07,
+      "loss": 1.7921,
+      "step": 2401
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.065554231227651e-07,
+      "loss": 2.2619,
+      "step": 2402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.065156932856575e-07,
+      "loss": 1.8429,
+      "step": 2403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.064759634485499e-07,
+      "loss": 1.854,
+      "step": 2404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.064362336114421e-07,
+      "loss": 1.8506,
+      "step": 2405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.063965037743345e-07,
+      "loss": 1.7294,
+      "step": 2406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.063567739372268e-07,
+      "loss": 1.9847,
+      "step": 2407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.063170441001191e-07,
+      "loss": 1.7784,
+      "step": 2408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.062773142630114e-07,
+      "loss": 1.7935,
+      "step": 2409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.062375844259038e-07,
+      "loss": 1.8013,
+      "step": 2410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.061978545887962e-07,
+      "loss": 2.1785,
+      "step": 2411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.061581247516885e-07,
+      "loss": 1.9102,
+      "step": 2412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.061183949145809e-07,
+      "loss": 1.8683,
+      "step": 2413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.060786650774731e-07,
+      "loss": 1.8229,
+      "step": 2414
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.060389352403655e-07,
+      "loss": 2.1953,
+      "step": 2415
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.059992054032578e-07,
+      "loss": 2.2634,
+      "step": 2416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.059594755661502e-07,
+      "loss": 1.697,
+      "step": 2417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.059197457290426e-07,
+      "loss": 1.8442,
+      "step": 2418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.058800158919348e-07,
+      "loss": 2.054,
+      "step": 2419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.058402860548271e-07,
+      "loss": 1.9804,
+      "step": 2420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.058005562177194e-07,
+      "loss": 1.9735,
+      "step": 2421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.057608263806118e-07,
+      "loss": 1.7617,
+      "step": 2422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.057210965435041e-07,
+      "loss": 1.8499,
+      "step": 2423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.056813667063965e-07,
+      "loss": 1.9936,
+      "step": 2424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.056416368692889e-07,
+      "loss": 1.9774,
+      "step": 2425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.056019070321811e-07,
+      "loss": 2.1674,
+      "step": 2426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.055621771950735e-07,
+      "loss": 1.7273,
+      "step": 2427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.055224473579658e-07,
+      "loss": 1.8091,
+      "step": 2428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.054827175208582e-07,
+      "loss": 1.9279,
+      "step": 2429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.054429876837505e-07,
+      "loss": 1.967,
+      "step": 2430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.054032578466429e-07,
+      "loss": 1.8683,
+      "step": 2431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.05363528009535e-07,
+      "loss": 1.9166,
+      "step": 2432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.053237981724274e-07,
+      "loss": 1.8766,
+      "step": 2433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.052840683353198e-07,
+      "loss": 1.7691,
+      "step": 2434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.052443384982121e-07,
+      "loss": 1.852,
+      "step": 2435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.052046086611045e-07,
+      "loss": 1.8292,
+      "step": 2436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.051648788239968e-07,
+      "loss": 1.598,
+      "step": 2437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.051251489868891e-07,
+      "loss": 1.5888,
+      "step": 2438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.050854191497814e-07,
+      "loss": 2.2336,
+      "step": 2439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.050456893126738e-07,
+      "loss": 2.0701,
+      "step": 2440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.050059594755662e-07,
+      "loss": 1.9702,
+      "step": 2441
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.049662296384585e-07,
+      "loss": 1.822,
+      "step": 2442
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.049264998013509e-07,
+      "loss": 1.8639,
+      "step": 2443
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.048867699642431e-07,
+      "loss": 1.7101,
+      "step": 2444
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.048470401271354e-07,
+      "loss": 1.6791,
+      "step": 2445
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.048073102900277e-07,
+      "loss": 1.7307,
+      "step": 2446
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.047675804529201e-07,
+      "loss": 1.8803,
+      "step": 2447
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.047278506158124e-07,
+      "loss": 1.6639,
+      "step": 2448
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.046881207787047e-07,
+      "loss": 1.84,
+      "step": 2449
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.046483909415971e-07,
+      "loss": 1.6348,
+      "step": 2450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.046086611044894e-07,
+      "loss": 2.0536,
+      "step": 2451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.045689312673818e-07,
+      "loss": 2.0939,
+      "step": 2452
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.045292014302741e-07,
+      "loss": 1.9498,
+      "step": 2453
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.044894715931665e-07,
+      "loss": 1.9414,
+      "step": 2454
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.044497417560587e-07,
+      "loss": 1.6377,
+      "step": 2455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.044100119189511e-07,
+      "loss": 1.7526,
+      "step": 2456
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.043702820818435e-07,
+      "loss": 1.7825,
+      "step": 2457
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.043305522447357e-07,
+      "loss": 2.1542,
+      "step": 2458
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.042908224076281e-07,
+      "loss": 1.8917,
+      "step": 2459
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.042510925705204e-07,
+      "loss": 1.7615,
+      "step": 2460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.042113627334127e-07,
+      "loss": 1.9458,
+      "step": 2461
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.04171632896305e-07,
+      "loss": 1.9399,
+      "step": 2462
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.041319030591974e-07,
+      "loss": 1.8447,
+      "step": 2463
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.040921732220897e-07,
+      "loss": 1.616,
+      "step": 2464
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.040524433849821e-07,
+      "loss": 1.6632,
+      "step": 2465
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.040127135478745e-07,
+      "loss": 1.6473,
+      "step": 2466
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.039729837107667e-07,
+      "loss": 2.3128,
+      "step": 2467
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.039332538736591e-07,
+      "loss": 2.0032,
+      "step": 2468
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.038935240365514e-07,
+      "loss": 2.0575,
+      "step": 2469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.038537941994438e-07,
+      "loss": 1.9316,
+      "step": 2470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.03814064362336e-07,
+      "loss": 1.8094,
+      "step": 2471
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.037743345252284e-07,
+      "loss": 1.68,
+      "step": 2472
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.037346046881207e-07,
+      "loss": 2.002,
+      "step": 2473
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.03694874851013e-07,
+      "loss": 1.7483,
+      "step": 2474
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.036551450139054e-07,
+      "loss": 2.0021,
+      "step": 2475
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.036154151767977e-07,
+      "loss": 1.9136,
+      "step": 2476
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.035756853396901e-07,
+      "loss": 1.954,
+      "step": 2477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.035359555025824e-07,
+      "loss": 1.6692,
+      "step": 2478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.034962256654747e-07,
+      "loss": 1.9238,
+      "step": 2479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.03456495828367e-07,
+      "loss": 2.0297,
+      "step": 2480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.034167659912594e-07,
+      "loss": 1.7132,
+      "step": 2481
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.033770361541518e-07,
+      "loss": 2.0188,
+      "step": 2482
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.033373063170441e-07,
+      "loss": 1.8444,
+      "step": 2483
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.032975764799365e-07,
+      "loss": 1.7655,
+      "step": 2484
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.032578466428286e-07,
+      "loss": 1.9138,
+      "step": 2485
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.03218116805721e-07,
+      "loss": 2.0322,
+      "step": 2486
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.031783869686133e-07,
+      "loss": 1.9386,
+      "step": 2487
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.031386571315057e-07,
+      "loss": 1.7414,
+      "step": 2488
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.030989272943981e-07,
+      "loss": 1.6985,
+      "step": 2489
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.030591974572904e-07,
+      "loss": 2.0175,
+      "step": 2490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.030194676201827e-07,
+      "loss": 1.8802,
+      "step": 2491
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.02979737783075e-07,
+      "loss": 2.0779,
+      "step": 2492
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.029400079459674e-07,
+      "loss": 1.9118,
+      "step": 2493
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.029002781088597e-07,
+      "loss": 2.0767,
+      "step": 2494
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.028605482717521e-07,
+      "loss": 1.8545,
+      "step": 2495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.028208184346445e-07,
+      "loss": 1.9641,
+      "step": 2496
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.027810885975366e-07,
+      "loss": 1.8993,
+      "step": 2497
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.02741358760429e-07,
+      "loss": 2.0536,
+      "step": 2498
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.027016289233213e-07,
+      "loss": 1.6222,
+      "step": 2499
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.026618990862137e-07,
+      "loss": 1.8662,
+      "step": 2500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.02622169249106e-07,
+      "loss": 1.6884,
+      "step": 2501
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.025824394119984e-07,
+      "loss": 1.9331,
+      "step": 2502
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.025427095748906e-07,
+      "loss": 1.9553,
+      "step": 2503
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.02502979737783e-07,
+      "loss": 1.7379,
+      "step": 2504
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.024632499006754e-07,
+      "loss": 1.9094,
+      "step": 2505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.024235200635677e-07,
+      "loss": 1.8626,
+      "step": 2506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.023837902264601e-07,
+      "loss": 1.7407,
+      "step": 2507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.023440603893524e-07,
+      "loss": 1.8762,
+      "step": 2508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.023043305522447e-07,
+      "loss": 1.809,
+      "step": 2509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.02264600715137e-07,
+      "loss": 1.9899,
+      "step": 2510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.022248708780293e-07,
+      "loss": 1.7444,
+      "step": 2511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.021851410409217e-07,
+      "loss": 1.7296,
+      "step": 2512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.02145411203814e-07,
+      "loss": 2.0567,
+      "step": 2513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.021056813667064e-07,
+      "loss": 1.731,
+      "step": 2514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.020659515295986e-07,
+      "loss": 2.0631,
+      "step": 2515
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.02026221692491e-07,
+      "loss": 2.096,
+      "step": 2516
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.019864918553833e-07,
+      "loss": 1.8858,
+      "step": 2517
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.019467620182757e-07,
+      "loss": 1.9987,
+      "step": 2518
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.019070321811681e-07,
+      "loss": 1.9697,
+      "step": 2519
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.018673023440604e-07,
+      "loss": 1.8096,
+      "step": 2520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.018275725069527e-07,
+      "loss": 1.9896,
+      "step": 2521
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.01787842669845e-07,
+      "loss": 1.7915,
+      "step": 2522
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.3009516256938937,
+      "eval_loss": 1.8710556030273438,
+      "eval_runtime": 6.432,
+      "eval_samples_per_second": 392.104,
+      "eval_steps_per_second": 6.219,
+      "step": 2522
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.017481128327374e-07,
+      "loss": 1.6319,
+      "step": 2523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.017083829956296e-07,
+      "loss": 1.5688,
+      "step": 2524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.01668653158522e-07,
+      "loss": 2.0627,
+      "step": 2525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.016289233214143e-07,
+      "loss": 1.9113,
+      "step": 2526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.015891934843066e-07,
+      "loss": 1.8403,
+      "step": 2527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.01549463647199e-07,
+      "loss": 1.8399,
+      "step": 2528
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.015097338100913e-07,
+      "loss": 2.0022,
+      "step": 2529
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.014700039729837e-07,
+      "loss": 1.8106,
+      "step": 2530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.01430274135876e-07,
+      "loss": 2.0251,
+      "step": 2531
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.013905442987684e-07,
+      "loss": 1.6793,
+      "step": 2532
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.013508144616606e-07,
+      "loss": 1.8904,
+      "step": 2533
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.01311084624553e-07,
+      "loss": 2.0093,
+      "step": 2534
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.012713547874454e-07,
+      "loss": 2.0205,
+      "step": 2535
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.012316249503377e-07,
+      "loss": 1.8155,
+      "step": 2536
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.0119189511323e-07,
+      "loss": 1.9307,
+      "step": 2537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.011521652761223e-07,
+      "loss": 1.9615,
+      "step": 2538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.011124354390146e-07,
+      "loss": 1.742,
+      "step": 2539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.010727056019069e-07,
+      "loss": 1.6722,
+      "step": 2540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.010329757647993e-07,
+      "loss": 1.9119,
+      "step": 2541
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.009932459276916e-07,
+      "loss": 1.8198,
+      "step": 2542
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.00953516090584e-07,
+      "loss": 1.6106,
+      "step": 2543
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.009137862534764e-07,
+      "loss": 1.8112,
+      "step": 2544
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.008740564163686e-07,
+      "loss": 1.9494,
+      "step": 2545
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.00834326579261e-07,
+      "loss": 1.8298,
+      "step": 2546
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.007945967421533e-07,
+      "loss": 1.811,
+      "step": 2547
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.007548669050457e-07,
+      "loss": 1.8337,
+      "step": 2548
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.00715137067938e-07,
+      "loss": 2.0916,
+      "step": 2549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.006754072308304e-07,
+      "loss": 1.8775,
+      "step": 2550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.006356773937226e-07,
+      "loss": 1.8128,
+      "step": 2551
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.005959475566149e-07,
+      "loss": 1.688,
+      "step": 2552
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.005562177195073e-07,
+      "loss": 1.9753,
+      "step": 2553
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.005164878823996e-07,
+      "loss": 1.7578,
+      "step": 2554
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.00476758045292e-07,
+      "loss": 1.7666,
+      "step": 2555
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.004370282081843e-07,
+      "loss": 2.0119,
+      "step": 2556
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.003972983710766e-07,
+      "loss": 2.1757,
+      "step": 2557
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.00357568533969e-07,
+      "loss": 1.8315,
+      "step": 2558
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.003178386968613e-07,
+      "loss": 2.0139,
+      "step": 2559
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.002781088597537e-07,
+      "loss": 1.7504,
+      "step": 2560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.00238379022646e-07,
+      "loss": 1.8144,
+      "step": 2561
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.001986491855384e-07,
+      "loss": 1.8984,
+      "step": 2562
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.001589193484305e-07,
+      "loss": 1.8249,
+      "step": 2563
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.001191895113229e-07,
+      "loss": 1.8639,
+      "step": 2564
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.000794596742152e-07,
+      "loss": 2.1846,
+      "step": 2565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.000397298371076e-07,
+      "loss": 1.7457,
+      "step": 2566
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9e-07,
+      "loss": 2.0382,
+      "step": 2567
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.999602701628923e-07,
+      "loss": 1.9633,
+      "step": 2568
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.999205403257846e-07,
+      "loss": 1.9552,
+      "step": 2569
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.998808104886769e-07,
+      "loss": 1.766,
+      "step": 2570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.998410806515693e-07,
+      "loss": 1.886,
+      "step": 2571
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.998013508144616e-07,
+      "loss": 1.9339,
+      "step": 2572
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.99761620977354e-07,
+      "loss": 1.5328,
+      "step": 2573
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.997218911402464e-07,
+      "loss": 1.9132,
+      "step": 2574
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.996821613031386e-07,
+      "loss": 1.8758,
+      "step": 2575
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.99642431466031e-07,
+      "loss": 1.5938,
+      "step": 2576
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.996027016289232e-07,
+      "loss": 1.8741,
+      "step": 2577
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.995629717918156e-07,
+      "loss": 1.6864,
+      "step": 2578
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.995232419547079e-07,
+      "loss": 2.0258,
+      "step": 2579
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.994835121176003e-07,
+      "loss": 1.9586,
+      "step": 2580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.994437822804925e-07,
+      "loss": 1.8255,
+      "step": 2581
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.994040524433849e-07,
+      "loss": 1.8682,
+      "step": 2582
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.993643226062773e-07,
+      "loss": 2.0492,
+      "step": 2583
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.993245927691696e-07,
+      "loss": 1.9342,
+      "step": 2584
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.99284862932062e-07,
+      "loss": 1.634,
+      "step": 2585
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.992451330949543e-07,
+      "loss": 2.0577,
+      "step": 2586
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.992054032578466e-07,
+      "loss": 1.7077,
+      "step": 2587
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.991656734207389e-07,
+      "loss": 1.6379,
+      "step": 2588
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.991259435836313e-07,
+      "loss": 1.7635,
+      "step": 2589
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.990862137465236e-07,
+      "loss": 1.7766,
+      "step": 2590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.990464839094159e-07,
+      "loss": 1.7921,
+      "step": 2591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.990067540723083e-07,
+      "loss": 1.7139,
+      "step": 2592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.989670242352005e-07,
+      "loss": 1.6942,
+      "step": 2593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.989272943980929e-07,
+      "loss": 2.037,
+      "step": 2594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.988875645609852e-07,
+      "loss": 2.0177,
+      "step": 2595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.988478347238776e-07,
+      "loss": 1.7462,
+      "step": 2596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.9880810488677e-07,
+      "loss": 1.7547,
+      "step": 2597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.987683750496623e-07,
+      "loss": 1.9355,
+      "step": 2598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.987286452125546e-07,
+      "loss": 1.8779,
+      "step": 2599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.986889153754469e-07,
+      "loss": 1.8662,
+      "step": 2600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.986491855383393e-07,
+      "loss": 1.5456,
+      "step": 2601
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.986094557012316e-07,
+      "loss": 1.8546,
+      "step": 2602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.985697258641239e-07,
+      "loss": 1.853,
+      "step": 2603
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.985299960270162e-07,
+      "loss": 2.0637,
+      "step": 2604
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.984902661899085e-07,
+      "loss": 1.9385,
+      "step": 2605
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.984505363528009e-07,
+      "loss": 1.6295,
+      "step": 2606
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.984108065156932e-07,
+      "loss": 1.7576,
+      "step": 2607
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.983710766785856e-07,
+      "loss": 1.9891,
+      "step": 2608
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.983313468414779e-07,
+      "loss": 1.7151,
+      "step": 2609
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.982916170043703e-07,
+      "loss": 1.6505,
+      "step": 2610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.982518871672625e-07,
+      "loss": 1.808,
+      "step": 2611
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.982121573301549e-07,
+      "loss": 1.7486,
+      "step": 2612
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.981724274930473e-07,
+      "loss": 1.9942,
+      "step": 2613
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.981326976559396e-07,
+      "loss": 1.7163,
+      "step": 2614
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.98092967818832e-07,
+      "loss": 1.8867,
+      "step": 2615
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.980532379817243e-07,
+      "loss": 1.9754,
+      "step": 2616
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.980135081446165e-07,
+      "loss": 1.6914,
+      "step": 2617
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.979737783075088e-07,
+      "loss": 1.9133,
+      "step": 2618
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.979340484704012e-07,
+      "loss": 1.8452,
+      "step": 2619
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.978943186332935e-07,
+      "loss": 1.8009,
+      "step": 2620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.978545887961859e-07,
+      "loss": 1.582,
+      "step": 2621
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.978148589590783e-07,
+      "loss": 1.8237,
+      "step": 2622
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.977751291219705e-07,
+      "loss": 1.7371,
+      "step": 2623
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.977353992848629e-07,
+      "loss": 1.693,
+      "step": 2624
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.976956694477552e-07,
+      "loss": 1.8779,
+      "step": 2625
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.976559396106476e-07,
+      "loss": 1.8526,
+      "step": 2626
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.9761620977354e-07,
+      "loss": 2.1389,
+      "step": 2627
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.975764799364323e-07,
+      "loss": 1.9693,
+      "step": 2628
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.975367500993245e-07,
+      "loss": 1.8653,
+      "step": 2629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.974970202622168e-07,
+      "loss": 1.6277,
+      "step": 2630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.974572904251092e-07,
+      "loss": 1.8229,
+      "step": 2631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.974175605880015e-07,
+      "loss": 1.8813,
+      "step": 2632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.973778307508939e-07,
+      "loss": 1.7088,
+      "step": 2633
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.973381009137862e-07,
+      "loss": 1.842,
+      "step": 2634
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.972983710766785e-07,
+      "loss": 1.7763,
+      "step": 2635
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.972586412395709e-07,
+      "loss": 1.708,
+      "step": 2636
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.972189114024632e-07,
+      "loss": 1.5998,
+      "step": 2637
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.971791815653556e-07,
+      "loss": 1.9157,
+      "step": 2638
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.971394517282479e-07,
+      "loss": 1.9168,
+      "step": 2639
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.970997218911403e-07,
+      "loss": 1.9973,
+      "step": 2640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.970599920540325e-07,
+      "loss": 1.7274,
+      "step": 2641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.970202622169249e-07,
+      "loss": 1.9295,
+      "step": 2642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.969805323798171e-07,
+      "loss": 2.0523,
+      "step": 2643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.969408025427095e-07,
+      "loss": 1.6749,
+      "step": 2644
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.969010727056019e-07,
+      "loss": 2.205,
+      "step": 2645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.968613428684942e-07,
+      "loss": 1.8297,
+      "step": 2646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.968216130313865e-07,
+      "loss": 2.2123,
+      "step": 2647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.967818831942788e-07,
+      "loss": 1.7611,
+      "step": 2648
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.967421533571712e-07,
+      "loss": 2.0106,
+      "step": 2649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.967024235200635e-07,
+      "loss": 1.8359,
+      "step": 2650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.966626936829559e-07,
+      "loss": 1.8311,
+      "step": 2651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.966229638458483e-07,
+      "loss": 1.8514,
+      "step": 2652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.965832340087405e-07,
+      "loss": 1.8297,
+      "step": 2653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.965435041716329e-07,
+      "loss": 1.6914,
+      "step": 2654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.965037743345252e-07,
+      "loss": 1.8292,
+      "step": 2655
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.964640444974175e-07,
+      "loss": 1.68,
+      "step": 2656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.964243146603098e-07,
+      "loss": 1.9492,
+      "step": 2657
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.963845848232022e-07,
+      "loss": 1.976,
+      "step": 2658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.963448549860944e-07,
+      "loss": 1.8836,
+      "step": 2659
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.963051251489868e-07,
+      "loss": 1.9419,
+      "step": 2660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.962653953118792e-07,
+      "loss": 1.8242,
+      "step": 2661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.962256654747715e-07,
+      "loss": 1.9411,
+      "step": 2662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.961859356376639e-07,
+      "loss": 1.7416,
+      "step": 2663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.961462058005562e-07,
+      "loss": 1.6406,
+      "step": 2664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.961064759634485e-07,
+      "loss": 1.7059,
+      "step": 2665
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.960667461263408e-07,
+      "loss": 1.8926,
+      "step": 2666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.960270162892332e-07,
+      "loss": 1.9929,
+      "step": 2667
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.959872864521256e-07,
+      "loss": 1.966,
+      "step": 2668
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.959475566150178e-07,
+      "loss": 1.8723,
+      "step": 2669
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.959078267779102e-07,
+      "loss": 1.9563,
+      "step": 2670
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.958680969408024e-07,
+      "loss": 1.8248,
+      "step": 2671
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.958283671036948e-07,
+      "loss": 2.0858,
+      "step": 2672
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.957886372665871e-07,
+      "loss": 1.7235,
+      "step": 2673
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.957489074294795e-07,
+      "loss": 1.8061,
+      "step": 2674
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.957091775923719e-07,
+      "loss": 1.793,
+      "step": 2675
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.956694477552642e-07,
+      "loss": 1.7396,
+      "step": 2676
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.956297179181565e-07,
+      "loss": 1.7163,
+      "step": 2677
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.955899880810488e-07,
+      "loss": 1.7661,
+      "step": 2678
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.955502582439412e-07,
+      "loss": 1.8422,
+      "step": 2679
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.955105284068335e-07,
+      "loss": 1.8028,
+      "step": 2680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.954707985697259e-07,
+      "loss": 1.7268,
+      "step": 2681
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.954310687326183e-07,
+      "loss": 2.0463,
+      "step": 2682
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.953913388955104e-07,
+      "loss": 1.762,
+      "step": 2683
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.953516090584028e-07,
+      "loss": 1.7541,
+      "step": 2684
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.953118792212951e-07,
+      "loss": 2.0086,
+      "step": 2685
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.952721493841875e-07,
+      "loss": 1.8379,
+      "step": 2686
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.952324195470798e-07,
+      "loss": 1.9495,
+      "step": 2687
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.951926897099722e-07,
+      "loss": 1.8116,
+      "step": 2688
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.951529598728644e-07,
+      "loss": 1.7648,
+      "step": 2689
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.951132300357568e-07,
+      "loss": 1.8644,
+      "step": 2690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.950735001986492e-07,
+      "loss": 1.754,
+      "step": 2691
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.950337703615415e-07,
+      "loss": 1.8629,
+      "step": 2692
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.949940405244339e-07,
+      "loss": 1.8529,
+      "step": 2693
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.949543106873262e-07,
+      "loss": 1.8586,
+      "step": 2694
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.949145808502184e-07,
+      "loss": 1.6419,
+      "step": 2695
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.948748510131107e-07,
+      "loss": 1.7083,
+      "step": 2696
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.948351211760031e-07,
+      "loss": 1.8429,
+      "step": 2697
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.947953913388954e-07,
+      "loss": 1.6907,
+      "step": 2698
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.947556615017878e-07,
+      "loss": 1.7907,
+      "step": 2699
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.947159316646802e-07,
+      "loss": 1.6828,
+      "step": 2700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.946762018275724e-07,
+      "loss": 2.0293,
+      "step": 2701
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.946364719904648e-07,
+      "loss": 2.0208,
+      "step": 2702
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.945967421533571e-07,
+      "loss": 1.7172,
+      "step": 2703
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.945570123162495e-07,
+      "loss": 1.7297,
+      "step": 2704
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.945172824791418e-07,
+      "loss": 2.0005,
+      "step": 2705
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.944775526420342e-07,
+      "loss": 1.8157,
+      "step": 2706
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.944378228049265e-07,
+      "loss": 1.867,
+      "step": 2707
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.943980929678187e-07,
+      "loss": 1.8464,
+      "step": 2708
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.943583631307111e-07,
+      "loss": 1.8758,
+      "step": 2709
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.943186332936034e-07,
+      "loss": 1.9573,
+      "step": 2710
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.942789034564958e-07,
+      "loss": 1.9087,
+      "step": 2711
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.942391736193881e-07,
+      "loss": 1.6407,
+      "step": 2712
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.941994437822804e-07,
+      "loss": 1.8597,
+      "step": 2713
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.941597139451728e-07,
+      "loss": 1.8094,
+      "step": 2714
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.941199841080651e-07,
+      "loss": 1.9358,
+      "step": 2715
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.940802542709575e-07,
+      "loss": 2.0501,
+      "step": 2716
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.940405244338498e-07,
+      "loss": 1.7737,
+      "step": 2717
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.940007945967422e-07,
+      "loss": 1.6949,
+      "step": 2718
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.939610647596344e-07,
+      "loss": 1.83,
+      "step": 2719
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.939213349225268e-07,
+      "loss": 1.7944,
+      "step": 2720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.938816050854191e-07,
+      "loss": 1.7768,
+      "step": 2721
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.938418752483114e-07,
+      "loss": 1.82,
+      "step": 2722
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.938021454112038e-07,
+      "loss": 2.0836,
+      "step": 2723
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.937624155740961e-07,
+      "loss": 1.8779,
+      "step": 2724
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.937226857369884e-07,
+      "loss": 1.7705,
+      "step": 2725
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.936829558998807e-07,
+      "loss": 1.8498,
+      "step": 2726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.936432260627731e-07,
+      "loss": 1.9004,
+      "step": 2727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.936034962256654e-07,
+      "loss": 2.2712,
+      "step": 2728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.935637663885578e-07,
+      "loss": 1.9741,
+      "step": 2729
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.935240365514502e-07,
+      "loss": 1.9091,
+      "step": 2730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.934843067143424e-07,
+      "loss": 1.9099,
+      "step": 2731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.934445768772348e-07,
+      "loss": 1.7846,
+      "step": 2732
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.934048470401271e-07,
+      "loss": 1.9311,
+      "step": 2733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.933651172030195e-07,
+      "loss": 1.8907,
+      "step": 2734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.933253873659117e-07,
+      "loss": 1.6548,
+      "step": 2735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.932856575288041e-07,
+      "loss": 1.9074,
+      "step": 2736
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.932459276916963e-07,
+      "loss": 1.7371,
+      "step": 2737
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.932061978545887e-07,
+      "loss": 1.9852,
+      "step": 2738
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.931664680174811e-07,
+      "loss": 2.1141,
+      "step": 2739
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.931267381803734e-07,
+      "loss": 1.8146,
+      "step": 2740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.930870083432658e-07,
+      "loss": 1.9089,
+      "step": 2741
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.930472785061581e-07,
+      "loss": 1.7902,
+      "step": 2742
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.930075486690504e-07,
+      "loss": 1.7311,
+      "step": 2743
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.929678188319427e-07,
+      "loss": 2.0622,
+      "step": 2744
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.929280889948351e-07,
+      "loss": 2.0909,
+      "step": 2745
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.928883591577275e-07,
+      "loss": 1.793,
+      "step": 2746
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.928486293206198e-07,
+      "loss": 1.7435,
+      "step": 2747
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.928088994835121e-07,
+      "loss": 1.7635,
+      "step": 2748
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.927691696464043e-07,
+      "loss": 1.7091,
+      "step": 2749
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.927294398092967e-07,
+      "loss": 1.9883,
+      "step": 2750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.92689709972189e-07,
+      "loss": 1.869,
+      "step": 2751
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.926499801350814e-07,
+      "loss": 1.6284,
+      "step": 2752
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.926102502979738e-07,
+      "loss": 1.6405,
+      "step": 2753
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.925705204608661e-07,
+      "loss": 2.0029,
+      "step": 2754
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.925307906237584e-07,
+      "loss": 1.8985,
+      "step": 2755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.924910607866507e-07,
+      "loss": 1.9805,
+      "step": 2756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.924513309495431e-07,
+      "loss": 1.8267,
+      "step": 2757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.924116011124354e-07,
+      "loss": 1.7846,
+      "step": 2758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.923718712753278e-07,
+      "loss": 1.77,
+      "step": 2759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.923321414382202e-07,
+      "loss": 1.7756,
+      "step": 2760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.922924116011123e-07,
+      "loss": 1.7576,
+      "step": 2761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.922526817640047e-07,
+      "loss": 1.6461,
+      "step": 2762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.92212951926897e-07,
+      "loss": 1.7916,
+      "step": 2763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.921732220897894e-07,
+      "loss": 1.7116,
+      "step": 2764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.921334922526817e-07,
+      "loss": 1.983,
+      "step": 2765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.920937624155741e-07,
+      "loss": 1.9867,
+      "step": 2766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.920540325784663e-07,
+      "loss": 1.8458,
+      "step": 2767
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.920143027413587e-07,
+      "loss": 1.7133,
+      "step": 2768
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.919745729042511e-07,
+      "loss": 2.1576,
+      "step": 2769
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.919348430671434e-07,
+      "loss": 2.1294,
+      "step": 2770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.918951132300358e-07,
+      "loss": 1.7633,
+      "step": 2771
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.918553833929281e-07,
+      "loss": 1.6844,
+      "step": 2772
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.918156535558204e-07,
+      "loss": 1.9626,
+      "step": 2773
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.917759237187126e-07,
+      "loss": 1.8471,
+      "step": 2774
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.91736193881605e-07,
+      "loss": 2.06,
+      "step": 2775
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.916964640444973e-07,
+      "loss": 2.0721,
+      "step": 2776
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.916567342073897e-07,
+      "loss": 1.7788,
+      "step": 2777
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.916170043702821e-07,
+      "loss": 1.6722,
+      "step": 2778
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.915772745331743e-07,
+      "loss": 1.9903,
+      "step": 2779
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.915375446960667e-07,
+      "loss": 2.1456,
+      "step": 2780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.91497814858959e-07,
+      "loss": 1.7274,
+      "step": 2781
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.914580850218514e-07,
+      "loss": 1.8213,
+      "step": 2782
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.914183551847437e-07,
+      "loss": 2.015,
+      "step": 2783
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.913786253476361e-07,
+      "loss": 1.7622,
+      "step": 2784
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.913388955105284e-07,
+      "loss": 1.6407,
+      "step": 2785
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.912991656734207e-07,
+      "loss": 1.8215,
+      "step": 2786
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.912594358363131e-07,
+      "loss": 1.902,
+      "step": 2787
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.912197059992053e-07,
+      "loss": 2.0004,
+      "step": 2788
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.911799761620977e-07,
+      "loss": 1.8808,
+      "step": 2789
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.9114024632499e-07,
+      "loss": 1.8515,
+      "step": 2790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.911005164878823e-07,
+      "loss": 1.8104,
+      "step": 2791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.910607866507747e-07,
+      "loss": 1.7545,
+      "step": 2792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.91021056813667e-07,
+      "loss": 1.9183,
+      "step": 2793
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.909813269765594e-07,
+      "loss": 1.8131,
+      "step": 2794
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.909415971394517e-07,
+      "loss": 1.8787,
+      "step": 2795
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.909018673023441e-07,
+      "loss": 1.7005,
+      "step": 2796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.908621374652363e-07,
+      "loss": 1.7256,
+      "step": 2797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.908224076281287e-07,
+      "loss": 2.043,
+      "step": 2798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.90782677791021e-07,
+      "loss": 1.7543,
+      "step": 2799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.907429479539134e-07,
+      "loss": 1.9043,
+      "step": 2800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.907032181168057e-07,
+      "loss": 1.8636,
+      "step": 2801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.90663488279698e-07,
+      "loss": 1.68,
+      "step": 2802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.906237584425903e-07,
+      "loss": 1.8597,
+      "step": 2803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.905840286054826e-07,
+      "loss": 1.731,
+      "step": 2804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.90544298768375e-07,
+      "loss": 1.943,
+      "step": 2805
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.905045689312673e-07,
+      "loss": 1.8907,
+      "step": 2806
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.904648390941597e-07,
+      "loss": 1.8714,
+      "step": 2807
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.904251092570521e-07,
+      "loss": 1.7502,
+      "step": 2808
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.903853794199443e-07,
+      "loss": 1.7519,
+      "step": 2809
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.903456495828367e-07,
+      "loss": 1.8342,
+      "step": 2810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.90305919745729e-07,
+      "loss": 1.8428,
+      "step": 2811
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.902661899086214e-07,
+      "loss": 1.8635,
+      "step": 2812
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.902264600715137e-07,
+      "loss": 2.1046,
+      "step": 2813
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.90186730234406e-07,
+      "loss": 1.6454,
+      "step": 2814
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.901470003972982e-07,
+      "loss": 1.8607,
+      "step": 2815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.901072705601906e-07,
+      "loss": 1.8408,
+      "step": 2816
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.90067540723083e-07,
+      "loss": 1.734,
+      "step": 2817
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.900278108859753e-07,
+      "loss": 1.8321,
+      "step": 2818
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.899880810488677e-07,
+      "loss": 1.7874,
+      "step": 2819
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.8994835121176e-07,
+      "loss": 1.9728,
+      "step": 2820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.899086213746523e-07,
+      "loss": 1.7126,
+      "step": 2821
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.898688915375446e-07,
+      "loss": 1.66,
+      "step": 2822
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.89829161700437e-07,
+      "loss": 2.1024,
+      "step": 2823
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.897894318633294e-07,
+      "loss": 1.9468,
+      "step": 2824
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.897497020262217e-07,
+      "loss": 1.8038,
+      "step": 2825
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.897099721891141e-07,
+      "loss": 1.792,
+      "step": 2826
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.896702423520062e-07,
+      "loss": 1.873,
+      "step": 2827
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.896305125148986e-07,
+      "loss": 1.8358,
+      "step": 2828
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.895907826777909e-07,
+      "loss": 1.8517,
+      "step": 2829
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.895510528406833e-07,
+      "loss": 1.856,
+      "step": 2830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.895113230035757e-07,
+      "loss": 1.7945,
+      "step": 2831
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.89471593166468e-07,
+      "loss": 1.9464,
+      "step": 2832
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.894318633293603e-07,
+      "loss": 2.1114,
+      "step": 2833
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.893921334922526e-07,
+      "loss": 1.6103,
+      "step": 2834
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.89352403655145e-07,
+      "loss": 1.9874,
+      "step": 2835
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.893126738180373e-07,
+      "loss": 1.6372,
+      "step": 2836
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.892729439809297e-07,
+      "loss": 1.7255,
+      "step": 2837
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.892332141438221e-07,
+      "loss": 2.2292,
+      "step": 2838
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.891934843067143e-07,
+      "loss": 1.7299,
+      "step": 2839
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.891537544696066e-07,
+      "loss": 1.7054,
+      "step": 2840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.891140246324989e-07,
+      "loss": 1.7499,
+      "step": 2841
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.890742947953913e-07,
+      "loss": 1.9707,
+      "step": 2842
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.890345649582836e-07,
+      "loss": 1.9437,
+      "step": 2843
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.88994835121176e-07,
+      "loss": 1.6166,
+      "step": 2844
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.889551052840682e-07,
+      "loss": 1.9241,
+      "step": 2845
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.889153754469606e-07,
+      "loss": 1.7255,
+      "step": 2846
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.88875645609853e-07,
+      "loss": 1.8578,
+      "step": 2847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.888359157727453e-07,
+      "loss": 1.7441,
+      "step": 2848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.887961859356377e-07,
+      "loss": 1.7808,
+      "step": 2849
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.8875645609853e-07,
+      "loss": 1.7585,
+      "step": 2850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.887167262614223e-07,
+      "loss": 2.0269,
+      "step": 2851
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.886769964243146e-07,
+      "loss": 1.8818,
+      "step": 2852
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.88637266587207e-07,
+      "loss": 1.8177,
+      "step": 2853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.885975367500992e-07,
+      "loss": 1.8824,
+      "step": 2854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.885578069129916e-07,
+      "loss": 1.9736,
+      "step": 2855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.88518077075884e-07,
+      "loss": 1.4915,
+      "step": 2856
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.884783472387762e-07,
+      "loss": 1.9586,
+      "step": 2857
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.884386174016686e-07,
+      "loss": 1.6719,
+      "step": 2858
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.883988875645609e-07,
+      "loss": 1.7229,
+      "step": 2859
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.883591577274533e-07,
+      "loss": 2.0408,
+      "step": 2860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.883194278903456e-07,
+      "loss": 1.9492,
+      "step": 2861
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.88279698053238e-07,
+      "loss": 1.8923,
+      "step": 2862
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.882399682161303e-07,
+      "loss": 1.7423,
+      "step": 2863
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.882002383790226e-07,
+      "loss": 2.121,
+      "step": 2864
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.88160508541915e-07,
+      "loss": 1.8489,
+      "step": 2865
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.881207787048073e-07,
+      "loss": 1.7371,
+      "step": 2866
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.880810488676996e-07,
+      "loss": 1.8063,
+      "step": 2867
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.880413190305919e-07,
+      "loss": 1.8158,
+      "step": 2868
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.880015891934842e-07,
+      "loss": 1.7215,
+      "step": 2869
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.879618593563766e-07,
+      "loss": 1.636,
+      "step": 2870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.879221295192689e-07,
+      "loss": 1.8283,
+      "step": 2871
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.878823996821613e-07,
+      "loss": 1.6095,
+      "step": 2872
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.878426698450536e-07,
+      "loss": 1.727,
+      "step": 2873
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.87802940007946e-07,
+      "loss": 1.758,
+      "step": 2874
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.877632101708382e-07,
+      "loss": 1.8321,
+      "step": 2875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.877234803337306e-07,
+      "loss": 1.981,
+      "step": 2876
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.87683750496623e-07,
+      "loss": 1.9301,
+      "step": 2877
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.876440206595153e-07,
+      "loss": 1.9425,
+      "step": 2878
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.876042908224077e-07,
+      "loss": 1.6855,
+      "step": 2879
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.875645609852999e-07,
+      "loss": 2.0365,
+      "step": 2880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.875248311481922e-07,
+      "loss": 1.9344,
+      "step": 2881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.874851013110845e-07,
+      "loss": 1.7347,
+      "step": 2882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.874453714739769e-07,
+      "loss": 1.8268,
+      "step": 2883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.874056416368692e-07,
+      "loss": 1.5795,
+      "step": 2884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.873659117997616e-07,
+      "loss": 1.836,
+      "step": 2885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.87326181962654e-07,
+      "loss": 1.7886,
+      "step": 2886
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.872864521255462e-07,
+      "loss": 1.8647,
+      "step": 2887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.872467222884386e-07,
+      "loss": 1.824,
+      "step": 2888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.872069924513309e-07,
+      "loss": 1.9007,
+      "step": 2889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.871672626142233e-07,
+      "loss": 1.8593,
+      "step": 2890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.871275327771156e-07,
+      "loss": 1.9534,
+      "step": 2891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.87087802940008e-07,
+      "loss": 1.7865,
+      "step": 2892
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.870480731029001e-07,
+      "loss": 1.9878,
+      "step": 2893
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.870083432657925e-07,
+      "loss": 1.8222,
+      "step": 2894
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.869686134286849e-07,
+      "loss": 1.831,
+      "step": 2895
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.869288835915772e-07,
+      "loss": 1.8979,
+      "step": 2896
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.868891537544696e-07,
+      "loss": 2.1147,
+      "step": 2897
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.868494239173619e-07,
+      "loss": 1.7965,
+      "step": 2898
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.868096940802542e-07,
+      "loss": 1.9572,
+      "step": 2899
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.867699642431465e-07,
+      "loss": 1.7997,
+      "step": 2900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.867302344060389e-07,
+      "loss": 1.9882,
+      "step": 2901
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.866905045689313e-07,
+      "loss": 1.7315,
+      "step": 2902
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.866507747318236e-07,
+      "loss": 1.8111,
+      "step": 2903
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.86611044894716e-07,
+      "loss": 1.6695,
+      "step": 2904
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.865713150576082e-07,
+      "loss": 1.7984,
+      "step": 2905
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.865315852205005e-07,
+      "loss": 1.6572,
+      "step": 2906
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.864918553833928e-07,
+      "loss": 1.9937,
+      "step": 2907
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.864521255462852e-07,
+      "loss": 1.8033,
+      "step": 2908
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.864123957091776e-07,
+      "loss": 1.9307,
+      "step": 2909
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.863726658720699e-07,
+      "loss": 1.6663,
+      "step": 2910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.863329360349622e-07,
+      "loss": 1.8003,
+      "step": 2911
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.862932061978545e-07,
+      "loss": 1.7018,
+      "step": 2912
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.862534763607469e-07,
+      "loss": 1.8751,
+      "step": 2913
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.862137465236392e-07,
+      "loss": 1.8152,
+      "step": 2914
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.861740166865316e-07,
+      "loss": 1.884,
+      "step": 2915
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.86134286849424e-07,
+      "loss": 1.7428,
+      "step": 2916
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.860945570123162e-07,
+      "loss": 1.6821,
+      "step": 2917
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.860548271752086e-07,
+      "loss": 1.869,
+      "step": 2918
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.860150973381009e-07,
+      "loss": 1.7024,
+      "step": 2919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.859753675009932e-07,
+      "loss": 2.007,
+      "step": 2920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.859356376638855e-07,
+      "loss": 1.9235,
+      "step": 2921
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.858959078267779e-07,
+      "loss": 1.7599,
+      "step": 2922
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.858561779896701e-07,
+      "loss": 1.73,
+      "step": 2923
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.858164481525625e-07,
+      "loss": 1.8367,
+      "step": 2924
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.857767183154549e-07,
+      "loss": 1.784,
+      "step": 2925
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.857369884783472e-07,
+      "loss": 1.549,
+      "step": 2926
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.856972586412396e-07,
+      "loss": 1.5569,
+      "step": 2927
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.856575288041319e-07,
+      "loss": 1.89,
+      "step": 2928
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.856177989670242e-07,
+      "loss": 1.9841,
+      "step": 2929
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.855780691299165e-07,
+      "loss": 1.9569,
+      "step": 2930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.855383392928089e-07,
+      "loss": 1.8972,
+      "step": 2931
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.854986094557013e-07,
+      "loss": 1.7054,
+      "step": 2932
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.854588796185935e-07,
+      "loss": 1.7807,
+      "step": 2933
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.854191497814859e-07,
+      "loss": 1.9051,
+      "step": 2934
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.853794199443781e-07,
+      "loss": 2.0171,
+      "step": 2935
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.853396901072705e-07,
+      "loss": 1.8267,
+      "step": 2936
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.852999602701628e-07,
+      "loss": 1.7562,
+      "step": 2937
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.852602304330552e-07,
+      "loss": 1.8517,
+      "step": 2938
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.852205005959475e-07,
+      "loss": 1.7861,
+      "step": 2939
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.851807707588399e-07,
+      "loss": 2.0192,
+      "step": 2940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.851410409217322e-07,
+      "loss": 1.7278,
+      "step": 2941
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.851013110846245e-07,
+      "loss": 1.9552,
+      "step": 2942
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.850615812475169e-07,
+      "loss": 1.7644,
+      "step": 2943
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.850218514104092e-07,
+      "loss": 1.7875,
+      "step": 2944
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.849821215733016e-07,
+      "loss": 1.653,
+      "step": 2945
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.849423917361938e-07,
+      "loss": 1.8719,
+      "step": 2946
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.849026618990861e-07,
+      "loss": 1.9158,
+      "step": 2947
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.848629320619785e-07,
+      "loss": 1.8096,
+      "step": 2948
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.848232022248708e-07,
+      "loss": 2.0426,
+      "step": 2949
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.847834723877632e-07,
+      "loss": 1.808,
+      "step": 2950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.847437425506555e-07,
+      "loss": 1.9066,
+      "step": 2951
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.847040127135479e-07,
+      "loss": 2.0249,
+      "step": 2952
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.846642828764401e-07,
+      "loss": 1.8607,
+      "step": 2953
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.846245530393325e-07,
+      "loss": 1.8466,
+      "step": 2954
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.845848232022249e-07,
+      "loss": 2.0838,
+      "step": 2955
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.845450933651172e-07,
+      "loss": 1.6588,
+      "step": 2956
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.845053635280096e-07,
+      "loss": 1.6487,
+      "step": 2957
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.844656336909019e-07,
+      "loss": 1.8323,
+      "step": 2958
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.844259038537941e-07,
+      "loss": 1.8442,
+      "step": 2959
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.843861740166864e-07,
+      "loss": 1.9306,
+      "step": 2960
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.843464441795788e-07,
+      "loss": 1.6816,
+      "step": 2961
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.843067143424711e-07,
+      "loss": 1.8714,
+      "step": 2962
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.842669845053635e-07,
+      "loss": 1.9252,
+      "step": 2963
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.842272546682559e-07,
+      "loss": 1.9441,
+      "step": 2964
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.841875248311481e-07,
+      "loss": 1.8007,
+      "step": 2965
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.841477949940405e-07,
+      "loss": 1.6617,
+      "step": 2966
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.841080651569328e-07,
+      "loss": 1.8168,
+      "step": 2967
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.840683353198252e-07,
+      "loss": 1.6689,
+      "step": 2968
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.840286054827175e-07,
+      "loss": 1.8044,
+      "step": 2969
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.839888756456099e-07,
+      "loss": 1.7183,
+      "step": 2970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.839491458085022e-07,
+      "loss": 1.7613,
+      "step": 2971
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.839094159713944e-07,
+      "loss": 1.9557,
+      "step": 2972
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.838696861342868e-07,
+      "loss": 2.0037,
+      "step": 2973
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.838299562971791e-07,
+      "loss": 1.846,
+      "step": 2974
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.837902264600715e-07,
+      "loss": 1.7556,
+      "step": 2975
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.837504966229638e-07,
+      "loss": 1.7733,
+      "step": 2976
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.837107667858561e-07,
+      "loss": 1.7018,
+      "step": 2977
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.836710369487484e-07,
+      "loss": 2.0426,
+      "step": 2978
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.836313071116408e-07,
+      "loss": 1.7182,
+      "step": 2979
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.835915772745332e-07,
+      "loss": 1.8427,
+      "step": 2980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.835518474374255e-07,
+      "loss": 1.9444,
+      "step": 2981
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.835121176003179e-07,
+      "loss": 1.9164,
+      "step": 2982
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.834723877632101e-07,
+      "loss": 1.6915,
+      "step": 2983
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.834326579261025e-07,
+      "loss": 1.9825,
+      "step": 2984
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.833929280889948e-07,
+      "loss": 2.0525,
+      "step": 2985
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.833531982518871e-07,
+      "loss": 1.6585,
+      "step": 2986
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.833134684147795e-07,
+      "loss": 1.8587,
+      "step": 2987
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.832737385776718e-07,
+      "loss": 1.7827,
+      "step": 2988
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.832340087405641e-07,
+      "loss": 1.7967,
+      "step": 2989
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.831942789034564e-07,
+      "loss": 1.8756,
+      "step": 2990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.831545490663488e-07,
+      "loss": 1.9103,
+      "step": 2991
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.831148192292411e-07,
+      "loss": 1.9613,
+      "step": 2992
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.830750893921335e-07,
+      "loss": 1.7321,
+      "step": 2993
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.830353595550259e-07,
+      "loss": 1.681,
+      "step": 2994
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.829956297179181e-07,
+      "loss": 2.0301,
+      "step": 2995
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.829558998808105e-07,
+      "loss": 1.7727,
+      "step": 2996
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.829161700437028e-07,
+      "loss": 1.7181,
+      "step": 2997
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.828764402065952e-07,
+      "loss": 1.9377,
+      "step": 2998
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.828367103694874e-07,
+      "loss": 1.8425,
+      "step": 2999
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.827969805323798e-07,
+      "loss": 1.6915,
+      "step": 3000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.82757250695272e-07,
+      "loss": 1.8138,
+      "step": 3001
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.827175208581644e-07,
+      "loss": 1.6435,
+      "step": 3002
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.826777910210568e-07,
+      "loss": 1.6741,
+      "step": 3003
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.826380611839491e-07,
+      "loss": 1.8337,
+      "step": 3004
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.825983313468415e-07,
+      "loss": 1.7434,
+      "step": 3005
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.825586015097338e-07,
+      "loss": 1.7465,
+      "step": 3006
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.825188716726261e-07,
+      "loss": 1.7238,
+      "step": 3007
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.824791418355184e-07,
+      "loss": 1.8703,
+      "step": 3008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.824394119984108e-07,
+      "loss": 1.8783,
+      "step": 3009
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.823996821613032e-07,
+      "loss": 2.0759,
+      "step": 3010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.823599523241955e-07,
+      "loss": 2.0508,
+      "step": 3011
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.823202224870878e-07,
+      "loss": 1.5516,
+      "step": 3012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.8228049264998e-07,
+      "loss": 1.6647,
+      "step": 3013
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.822407628128724e-07,
+      "loss": 1.6334,
+      "step": 3014
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.822010329757647e-07,
+      "loss": 2.1876,
+      "step": 3015
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.821613031386571e-07,
+      "loss": 2.2275,
+      "step": 3016
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.821215733015494e-07,
+      "loss": 1.7953,
+      "step": 3017
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.820818434644418e-07,
+      "loss": 2.0657,
+      "step": 3018
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.820421136273341e-07,
+      "loss": 1.9178,
+      "step": 3019
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.820023837902264e-07,
+      "loss": 1.9465,
+      "step": 3020
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.819626539531188e-07,
+      "loss": 1.8972,
+      "step": 3021
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.819229241160111e-07,
+      "loss": 1.8641,
+      "step": 3022
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.818831942789035e-07,
+      "loss": 2.1986,
+      "step": 3023
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.818434644417958e-07,
+      "loss": 1.7323,
+      "step": 3024
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.81803734604688e-07,
+      "loss": 1.5906,
+      "step": 3025
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.817640047675804e-07,
+      "loss": 1.9312,
+      "step": 3026
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.817242749304727e-07,
+      "loss": 2.0262,
+      "step": 3027
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.816845450933651e-07,
+      "loss": 1.9061,
+      "step": 3028
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.816448152562574e-07,
+      "loss": 1.7539,
+      "step": 3029
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.816050854191498e-07,
+      "loss": 1.8481,
+      "step": 3030
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.81565355582042e-07,
+      "loss": 1.7304,
+      "step": 3031
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.815256257449344e-07,
+      "loss": 1.7919,
+      "step": 3032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.814858959078268e-07,
+      "loss": 1.7288,
+      "step": 3033
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.814461660707191e-07,
+      "loss": 1.7342,
+      "step": 3034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.814064362336115e-07,
+      "loss": 1.8542,
+      "step": 3035
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.813667063965038e-07,
+      "loss": 1.8066,
+      "step": 3036
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.813269765593961e-07,
+      "loss": 1.8331,
+      "step": 3037
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.812872467222883e-07,
+      "loss": 1.6283,
+      "step": 3038
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.812475168851807e-07,
+      "loss": 1.7576,
+      "step": 3039
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.81207787048073e-07,
+      "loss": 1.6418,
+      "step": 3040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.811680572109654e-07,
+      "loss": 1.9182,
+      "step": 3041
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.811283273738578e-07,
+      "loss": 1.9006,
+      "step": 3042
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.8108859753675e-07,
+      "loss": 1.8463,
+      "step": 3043
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.810488676996424e-07,
+      "loss": 1.8392,
+      "step": 3044
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.810091378625347e-07,
+      "loss": 1.8443,
+      "step": 3045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.809694080254271e-07,
+      "loss": 1.7094,
+      "step": 3046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.809296781883194e-07,
+      "loss": 1.8684,
+      "step": 3047
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.808899483512117e-07,
+      "loss": 1.9273,
+      "step": 3048
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.80850218514104e-07,
+      "loss": 1.7647,
+      "step": 3049
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.808104886769964e-07,
+      "loss": 2.0433,
+      "step": 3050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.807707588398888e-07,
+      "loss": 1.7305,
+      "step": 3051
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.80731029002781e-07,
+      "loss": 1.7288,
+      "step": 3052
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.806912991656734e-07,
+      "loss": 1.6726,
+      "step": 3053
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.806515693285656e-07,
+      "loss": 1.746,
+      "step": 3054
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.80611839491458e-07,
+      "loss": 1.9873,
+      "step": 3055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.805721096543503e-07,
+      "loss": 1.9351,
+      "step": 3056
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.805323798172427e-07,
+      "loss": 1.7326,
+      "step": 3057
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.804926499801351e-07,
+      "loss": 1.9854,
+      "step": 3058
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.804529201430274e-07,
+      "loss": 1.8752,
+      "step": 3059
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.804131903059197e-07,
+      "loss": 2.112,
+      "step": 3060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.80373460468812e-07,
+      "loss": 1.7279,
+      "step": 3061
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.803337306317044e-07,
+      "loss": 1.9615,
+      "step": 3062
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.802940007945967e-07,
+      "loss": 1.7133,
+      "step": 3063
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.802542709574891e-07,
+      "loss": 1.7994,
+      "step": 3064
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.802145411203814e-07,
+      "loss": 2.0037,
+      "step": 3065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.801748112832736e-07,
+      "loss": 1.974,
+      "step": 3066
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.80135081446166e-07,
+      "loss": 2.0745,
+      "step": 3067
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.800953516090583e-07,
+      "loss": 1.7843,
+      "step": 3068
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.800556217719507e-07,
+      "loss": 1.8786,
+      "step": 3069
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 8.80015891934843e-07,
+      "loss": 2.1317,
+      "step": 3070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.799761620977354e-07,
+      "loss": 1.7829,
+      "step": 3071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.799364322606276e-07,
+      "loss": 1.7062,
+      "step": 3072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.7989670242352e-07,
+      "loss": 1.8292,
+      "step": 3073
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.798569725864124e-07,
+      "loss": 1.8141,
+      "step": 3074
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.798172427493047e-07,
+      "loss": 1.9881,
+      "step": 3075
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.797775129121971e-07,
+      "loss": 1.9499,
+      "step": 3076
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.797377830750894e-07,
+      "loss": 1.7956,
+      "step": 3077
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.796980532379816e-07,
+      "loss": 1.9698,
+      "step": 3078
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.796583234008739e-07,
+      "loss": 1.8018,
+      "step": 3079
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.796185935637663e-07,
+      "loss": 1.8499,
+      "step": 3080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.795788637266587e-07,
+      "loss": 1.6111,
+      "step": 3081
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.79539133889551e-07,
+      "loss": 1.7874,
+      "step": 3082
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.794994040524434e-07,
+      "loss": 1.8192,
+      "step": 3083
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.794596742153356e-07,
+      "loss": 1.7608,
+      "step": 3084
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.79419944378228e-07,
+      "loss": 1.8776,
+      "step": 3085
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.793802145411203e-07,
+      "loss": 1.7417,
+      "step": 3086
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.793404847040127e-07,
+      "loss": 1.947,
+      "step": 3087
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.793007548669051e-07,
+      "loss": 1.739,
+      "step": 3088
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.792610250297974e-07,
+      "loss": 1.8878,
+      "step": 3089
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.792212951926897e-07,
+      "loss": 1.895,
+      "step": 3090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.791815653555819e-07,
+      "loss": 1.6171,
+      "step": 3091
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.791418355184743e-07,
+      "loss": 1.9317,
+      "step": 3092
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.791021056813666e-07,
+      "loss": 1.7531,
+      "step": 3093
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.79062375844259e-07,
+      "loss": 1.821,
+      "step": 3094
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.790226460071513e-07,
+      "loss": 1.8572,
+      "step": 3095
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.789829161700436e-07,
+      "loss": 1.9179,
+      "step": 3096
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.78943186332936e-07,
+      "loss": 1.8965,
+      "step": 3097
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.789034564958283e-07,
+      "loss": 1.7774,
+      "step": 3098
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.788637266587207e-07,
+      "loss": 1.8758,
+      "step": 3099
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.78823996821613e-07,
+      "loss": 1.7197,
+      "step": 3100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.787842669845054e-07,
+      "loss": 1.6961,
+      "step": 3101
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.787445371473976e-07,
+      "loss": 1.8029,
+      "step": 3102
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.7870480731029e-07,
+      "loss": 1.7684,
+      "step": 3103
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.786650774731823e-07,
+      "loss": 1.7423,
+      "step": 3104
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.786253476360746e-07,
+      "loss": 1.7563,
+      "step": 3105
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.78585617798967e-07,
+      "loss": 1.8512,
+      "step": 3106
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.785458879618593e-07,
+      "loss": 1.9115,
+      "step": 3107
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.785061581247516e-07,
+      "loss": 1.8701,
+      "step": 3108
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.784664282876439e-07,
+      "loss": 1.6993,
+      "step": 3109
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.784266984505363e-07,
+      "loss": 1.8248,
+      "step": 3110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.783869686134287e-07,
+      "loss": 1.9936,
+      "step": 3111
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.78347238776321e-07,
+      "loss": 1.9455,
+      "step": 3112
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.783075089392134e-07,
+      "loss": 2.0673,
+      "step": 3113
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.782677791021056e-07,
+      "loss": 1.547,
+      "step": 3114
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.78228049264998e-07,
+      "loss": 1.9326,
+      "step": 3115
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.781883194278903e-07,
+      "loss": 2.1217,
+      "step": 3116
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.781485895907827e-07,
+      "loss": 1.4793,
+      "step": 3117
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.781088597536749e-07,
+      "loss": 1.8794,
+      "step": 3118
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.780691299165673e-07,
+      "loss": 1.9025,
+      "step": 3119
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.780294000794596e-07,
+      "loss": 1.9034,
+      "step": 3120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.779896702423519e-07,
+      "loss": 2.0134,
+      "step": 3121
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.779499404052443e-07,
+      "loss": 1.8615,
+      "step": 3122
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.779102105681366e-07,
+      "loss": 1.6747,
+      "step": 3123
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.77870480731029e-07,
+      "loss": 1.8945,
+      "step": 3124
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.778307508939213e-07,
+      "loss": 1.898,
+      "step": 3125
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.777910210568136e-07,
+      "loss": 1.6323,
+      "step": 3126
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.77751291219706e-07,
+      "loss": 1.8631,
+      "step": 3127
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.777115613825983e-07,
+      "loss": 1.6886,
+      "step": 3128
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.776718315454907e-07,
+      "loss": 1.7284,
+      "step": 3129
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.77632101708383e-07,
+      "loss": 1.7966,
+      "step": 3130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.775923718712753e-07,
+      "loss": 2.0615,
+      "step": 3131
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.775526420341675e-07,
+      "loss": 1.7434,
+      "step": 3132
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.775129121970599e-07,
+      "loss": 1.927,
+      "step": 3133
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.774731823599522e-07,
+      "loss": 1.8364,
+      "step": 3134
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.774334525228446e-07,
+      "loss": 1.7853,
+      "step": 3135
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.77393722685737e-07,
+      "loss": 1.8784,
+      "step": 3136
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.773539928486293e-07,
+      "loss": 1.7675,
+      "step": 3137
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.773142630115216e-07,
+      "loss": 1.5502,
+      "step": 3138
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.772745331744139e-07,
+      "loss": 1.7459,
+      "step": 3139
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.772348033373063e-07,
+      "loss": 1.8227,
+      "step": 3140
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.771950735001986e-07,
+      "loss": 1.7091,
+      "step": 3141
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.77155343663091e-07,
+      "loss": 2.1187,
+      "step": 3142
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.771156138259834e-07,
+      "loss": 1.9521,
+      "step": 3143
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.770758839888755e-07,
+      "loss": 1.8608,
+      "step": 3144
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.770361541517679e-07,
+      "loss": 2.0757,
+      "step": 3145
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.769964243146602e-07,
+      "loss": 1.7473,
+      "step": 3146
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.769566944775526e-07,
+      "loss": 1.7166,
+      "step": 3147
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.769169646404449e-07,
+      "loss": 1.8191,
+      "step": 3148
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.768772348033373e-07,
+      "loss": 1.9323,
+      "step": 3149
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.768375049662295e-07,
+      "loss": 1.9184,
+      "step": 3150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.767977751291219e-07,
+      "loss": 1.9206,
+      "step": 3151
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.767580452920143e-07,
+      "loss": 1.6593,
+      "step": 3152
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.767183154549066e-07,
+      "loss": 1.6717,
+      "step": 3153
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.76678585617799e-07,
+      "loss": 1.7183,
+      "step": 3154
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.766388557806913e-07,
+      "loss": 1.787,
+      "step": 3155
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.765991259435836e-07,
+      "loss": 1.994,
+      "step": 3156
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.765593961064758e-07,
+      "loss": 1.8915,
+      "step": 3157
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.765196662693682e-07,
+      "loss": 1.6788,
+      "step": 3158
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.764799364322606e-07,
+      "loss": 2.0215,
+      "step": 3159
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.764402065951529e-07,
+      "loss": 1.6319,
+      "step": 3160
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.764004767580453e-07,
+      "loss": 2.1709,
+      "step": 3161
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.763607469209375e-07,
+      "loss": 1.686,
+      "step": 3162
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.763210170838299e-07,
+      "loss": 1.9606,
+      "step": 3163
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.762812872467222e-07,
+      "loss": 1.6924,
+      "step": 3164
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.762415574096146e-07,
+      "loss": 1.6838,
+      "step": 3165
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.76201827572507e-07,
+      "loss": 1.6904,
+      "step": 3166
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.761620977353993e-07,
+      "loss": 1.8338,
+      "step": 3167
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.761223678982916e-07,
+      "loss": 1.8452,
+      "step": 3168
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.760826380611839e-07,
+      "loss": 2.0807,
+      "step": 3169
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.760429082240762e-07,
+      "loss": 2.0109,
+      "step": 3170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.760031783869685e-07,
+      "loss": 1.8514,
+      "step": 3171
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.759634485498609e-07,
+      "loss": 1.8139,
+      "step": 3172
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.759237187127532e-07,
+      "loss": 1.7213,
+      "step": 3173
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.758839888756455e-07,
+      "loss": 1.7328,
+      "step": 3174
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.758442590385379e-07,
+      "loss": 2.109,
+      "step": 3175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.758045292014302e-07,
+      "loss": 1.972,
+      "step": 3176
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.757647993643226e-07,
+      "loss": 1.9572,
+      "step": 3177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.757250695272149e-07,
+      "loss": 1.8953,
+      "step": 3178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.756853396901073e-07,
+      "loss": 1.7838,
+      "step": 3179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.756456098529995e-07,
+      "loss": 1.645,
+      "step": 3180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.756058800158919e-07,
+      "loss": 1.5309,
+      "step": 3181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.755661501787843e-07,
+      "loss": 1.642,
+      "step": 3182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.755264203416766e-07,
+      "loss": 2.028,
+      "step": 3183
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.754866905045689e-07,
+      "loss": 1.899,
+      "step": 3184
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.754469606674612e-07,
+      "loss": 1.7813,
+      "step": 3185
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.754072308303535e-07,
+      "loss": 1.6962,
+      "step": 3186
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.753675009932458e-07,
+      "loss": 1.7656,
+      "step": 3187
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.753277711561382e-07,
+      "loss": 1.7806,
+      "step": 3188
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.752880413190306e-07,
+      "loss": 1.7933,
+      "step": 3189
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.752483114819229e-07,
+      "loss": 2.0776,
+      "step": 3190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.752085816448153e-07,
+      "loss": 1.9373,
+      "step": 3191
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.751688518077075e-07,
+      "loss": 1.8286,
+      "step": 3192
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.751291219705999e-07,
+      "loss": 1.7907,
+      "step": 3193
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.750893921334922e-07,
+      "loss": 1.9611,
+      "step": 3194
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.750496622963846e-07,
+      "loss": 1.9001,
+      "step": 3195
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.75009932459277e-07,
+      "loss": 1.9337,
+      "step": 3196
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.749702026221692e-07,
+      "loss": 1.7756,
+      "step": 3197
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.749304727850615e-07,
+      "loss": 1.8215,
+      "step": 3198
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.748907429479538e-07,
+      "loss": 1.5398,
+      "step": 3199
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.748510131108462e-07,
+      "loss": 1.7718,
+      "step": 3200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.748112832737385e-07,
+      "loss": 1.8053,
+      "step": 3201
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.747715534366309e-07,
+      "loss": 1.6683,
+      "step": 3202
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.747318235995232e-07,
+      "loss": 1.7841,
+      "step": 3203
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.746920937624155e-07,
+      "loss": 1.7568,
+      "step": 3204
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.746523639253079e-07,
+      "loss": 1.969,
+      "step": 3205
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.746126340882002e-07,
+      "loss": 1.7197,
+      "step": 3206
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.745729042510926e-07,
+      "loss": 1.9801,
+      "step": 3207
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.745331744139849e-07,
+      "loss": 2.0111,
+      "step": 3208
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.744934445768773e-07,
+      "loss": 1.7698,
+      "step": 3209
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.744537147397694e-07,
+      "loss": 1.8818,
+      "step": 3210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.744139849026618e-07,
+      "loss": 1.9363,
+      "step": 3211
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.743742550655541e-07,
+      "loss": 1.7527,
+      "step": 3212
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.743345252284465e-07,
+      "loss": 1.7423,
+      "step": 3213
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.742947953913389e-07,
+      "loss": 1.91,
+      "step": 3214
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.742550655542312e-07,
+      "loss": 1.8031,
+      "step": 3215
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.742153357171235e-07,
+      "loss": 1.952,
+      "step": 3216
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.741756058800158e-07,
+      "loss": 1.8214,
+      "step": 3217
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.741358760429082e-07,
+      "loss": 1.8253,
+      "step": 3218
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.740961462058005e-07,
+      "loss": 1.8255,
+      "step": 3219
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.740564163686929e-07,
+      "loss": 1.9754,
+      "step": 3220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.740166865315853e-07,
+      "loss": 1.812,
+      "step": 3221
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.739769566944775e-07,
+      "loss": 1.7409,
+      "step": 3222
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.739372268573698e-07,
+      "loss": 1.7337,
+      "step": 3223
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.738974970202621e-07,
+      "loss": 1.7511,
+      "step": 3224
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.738577671831545e-07,
+      "loss": 1.9686,
+      "step": 3225
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.738180373460468e-07,
+      "loss": 1.8123,
+      "step": 3226
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.737783075089392e-07,
+      "loss": 1.6074,
+      "step": 3227
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.737385776718314e-07,
+      "loss": 1.901,
+      "step": 3228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.736988478347238e-07,
+      "loss": 1.9232,
+      "step": 3229
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.736591179976162e-07,
+      "loss": 2.0365,
+      "step": 3230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.736193881605085e-07,
+      "loss": 1.7291,
+      "step": 3231
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.735796583234009e-07,
+      "loss": 1.9739,
+      "step": 3232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.735399284862932e-07,
+      "loss": 1.8044,
+      "step": 3233
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.735001986491855e-07,
+      "loss": 1.9035,
+      "step": 3234
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.734604688120778e-07,
+      "loss": 1.7075,
+      "step": 3235
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.734207389749701e-07,
+      "loss": 2.0379,
+      "step": 3236
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.733810091378625e-07,
+      "loss": 1.7195,
+      "step": 3237
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.733412793007548e-07,
+      "loss": 1.8353,
+      "step": 3238
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.733015494636472e-07,
+      "loss": 2.0461,
+      "step": 3239
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.732618196265394e-07,
+      "loss": 1.8509,
+      "step": 3240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.732220897894318e-07,
+      "loss": 1.8114,
+      "step": 3241
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.731823599523241e-07,
+      "loss": 1.4863,
+      "step": 3242
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.731426301152165e-07,
+      "loss": 2.3269,
+      "step": 3243
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.731029002781089e-07,
+      "loss": 1.7513,
+      "step": 3244
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.730631704410012e-07,
+      "loss": 1.8585,
+      "step": 3245
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.730234406038935e-07,
+      "loss": 2.064,
+      "step": 3246
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.729837107667858e-07,
+      "loss": 1.6067,
+      "step": 3247
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.729439809296782e-07,
+      "loss": 2.0013,
+      "step": 3248
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.729042510925705e-07,
+      "loss": 1.9547,
+      "step": 3249
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.728645212554628e-07,
+      "loss": 1.5429,
+      "step": 3250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.728247914183551e-07,
+      "loss": 1.8628,
+      "step": 3251
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.727850615812474e-07,
+      "loss": 1.6706,
+      "step": 3252
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.727453317441398e-07,
+      "loss": 1.6935,
+      "step": 3253
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.727056019070321e-07,
+      "loss": 1.7776,
+      "step": 3254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.726658720699245e-07,
+      "loss": 1.7242,
+      "step": 3255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.726261422328168e-07,
+      "loss": 1.7282,
+      "step": 3256
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.725864123957092e-07,
+      "loss": 1.7474,
+      "step": 3257
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.725466825586014e-07,
+      "loss": 1.878,
+      "step": 3258
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.725069527214938e-07,
+      "loss": 1.7906,
+      "step": 3259
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.724672228843862e-07,
+      "loss": 1.8215,
+      "step": 3260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.724274930472785e-07,
+      "loss": 1.5837,
+      "step": 3261
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.723877632101709e-07,
+      "loss": 2.1529,
+      "step": 3262
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.723480333730631e-07,
+      "loss": 1.6685,
+      "step": 3263
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.723083035359554e-07,
+      "loss": 1.7215,
+      "step": 3264
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.722685736988477e-07,
+      "loss": 1.845,
+      "step": 3265
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.722288438617401e-07,
+      "loss": 1.775,
+      "step": 3266
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.721891140246325e-07,
+      "loss": 1.9229,
+      "step": 3267
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.721493841875248e-07,
+      "loss": 1.7501,
+      "step": 3268
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.721096543504172e-07,
+      "loss": 2.1379,
+      "step": 3269
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.720699245133094e-07,
+      "loss": 1.8283,
+      "step": 3270
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.720301946762018e-07,
+      "loss": 1.7547,
+      "step": 3271
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.719904648390941e-07,
+      "loss": 1.915,
+      "step": 3272
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.719507350019865e-07,
+      "loss": 1.9769,
+      "step": 3273
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.719110051648789e-07,
+      "loss": 1.8313,
+      "step": 3274
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.718712753277712e-07,
+      "loss": 1.6861,
+      "step": 3275
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.718315454906634e-07,
+      "loss": 1.6722,
+      "step": 3276
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.717918156535557e-07,
+      "loss": 2.025,
+      "step": 3277
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.717520858164481e-07,
+      "loss": 1.741,
+      "step": 3278
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.717123559793404e-07,
+      "loss": 1.9888,
+      "step": 3279
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.716726261422328e-07,
+      "loss": 2.0673,
+      "step": 3280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.716328963051251e-07,
+      "loss": 1.6815,
+      "step": 3281
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.715931664680174e-07,
+      "loss": 1.7529,
+      "step": 3282
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.715534366309098e-07,
+      "loss": 1.9729,
+      "step": 3283
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.715137067938021e-07,
+      "loss": 1.7811,
+      "step": 3284
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714739769566945e-07,
+      "loss": 1.7365,
+      "step": 3285
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714342471195868e-07,
+      "loss": 1.7954,
+      "step": 3286
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.713945172824792e-07,
+      "loss": 1.8383,
+      "step": 3287
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.713547874453714e-07,
+      "loss": 1.8462,
+      "step": 3288
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.713150576082637e-07,
+      "loss": 1.7794,
+      "step": 3289
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.71275327771156e-07,
+      "loss": 1.7729,
+      "step": 3290
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.712355979340484e-07,
+      "loss": 1.9678,
+      "step": 3291
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.711958680969408e-07,
+      "loss": 1.5718,
+      "step": 3292
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.711561382598331e-07,
+      "loss": 1.9784,
+      "step": 3293
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.711164084227254e-07,
+      "loss": 1.8184,
+      "step": 3294
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.710766785856177e-07,
+      "loss": 1.911,
+      "step": 3295
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.710369487485101e-07,
+      "loss": 1.9737,
+      "step": 3296
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.709972189114024e-07,
+      "loss": 2.0104,
+      "step": 3297
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.709574890742948e-07,
+      "loss": 1.7791,
+      "step": 3298
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.709177592371872e-07,
+      "loss": 1.7967,
+      "step": 3299
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.708780294000794e-07,
+      "loss": 1.7701,
+      "step": 3300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.708382995629718e-07,
+      "loss": 1.8965,
+      "step": 3301
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.70798569725864e-07,
+      "loss": 2.0171,
+      "step": 3302
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.707588398887564e-07,
+      "loss": 1.7579,
+      "step": 3303
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.707191100516487e-07,
+      "loss": 1.8104,
+      "step": 3304
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.706793802145411e-07,
+      "loss": 1.8879,
+      "step": 3305
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.706396503774333e-07,
+      "loss": 1.7961,
+      "step": 3306
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.705999205403257e-07,
+      "loss": 2.0575,
+      "step": 3307
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.705601907032181e-07,
+      "loss": 1.924,
+      "step": 3308
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.705204608661104e-07,
+      "loss": 1.7389,
+      "step": 3309
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.704807310290028e-07,
+      "loss": 1.6864,
+      "step": 3310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.704410011918951e-07,
+      "loss": 1.7936,
+      "step": 3311
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.704012713547874e-07,
+      "loss": 1.8484,
+      "step": 3312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.703615415176797e-07,
+      "loss": 1.948,
+      "step": 3313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.703218116805721e-07,
+      "loss": 1.9359,
+      "step": 3314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.702820818434645e-07,
+      "loss": 1.8995,
+      "step": 3315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.702423520063567e-07,
+      "loss": 2.0909,
+      "step": 3316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.702026221692491e-07,
+      "loss": 1.8196,
+      "step": 3317
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.701628923321413e-07,
+      "loss": 1.5476,
+      "step": 3318
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.701231624950337e-07,
+      "loss": 1.4952,
+      "step": 3319
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.70083432657926e-07,
+      "loss": 1.8046,
+      "step": 3320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.700437028208184e-07,
+      "loss": 1.9505,
+      "step": 3321
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.700039729837108e-07,
+      "loss": 1.6816,
+      "step": 3322
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.699642431466031e-07,
+      "loss": 1.8955,
+      "step": 3323
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.699245133094954e-07,
+      "loss": 1.7884,
+      "step": 3324
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.698847834723877e-07,
+      "loss": 1.8358,
+      "step": 3325
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.698450536352801e-07,
+      "loss": 1.8425,
+      "step": 3326
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.698053237981724e-07,
+      "loss": 1.6731,
+      "step": 3327
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.697655939610648e-07,
+      "loss": 1.7822,
+      "step": 3328
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.69725864123957e-07,
+      "loss": 1.8589,
+      "step": 3329
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.696861342868493e-07,
+      "loss": 1.7064,
+      "step": 3330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.696464044497417e-07,
+      "loss": 1.9448,
+      "step": 3331
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.69606674612634e-07,
+      "loss": 1.8305,
+      "step": 3332
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.695669447755264e-07,
+      "loss": 1.8923,
+      "step": 3333
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.695272149384187e-07,
+      "loss": 1.6495,
+      "step": 3334
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.694874851013111e-07,
+      "loss": 2.0179,
+      "step": 3335
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.694477552642033e-07,
+      "loss": 1.8697,
+      "step": 3336
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.694080254270957e-07,
+      "loss": 1.7369,
+      "step": 3337
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.693682955899881e-07,
+      "loss": 1.7976,
+      "step": 3338
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.693285657528804e-07,
+      "loss": 1.7287,
+      "step": 3339
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.692888359157728e-07,
+      "loss": 1.7959,
+      "step": 3340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.692491060786651e-07,
+      "loss": 1.7381,
+      "step": 3341
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.692093762415573e-07,
+      "loss": 1.9344,
+      "step": 3342
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.691696464044496e-07,
+      "loss": 1.9551,
+      "step": 3343
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.69129916567342e-07,
+      "loss": 1.5893,
+      "step": 3344
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.690901867302344e-07,
+      "loss": 1.708,
+      "step": 3345
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.690504568931267e-07,
+      "loss": 1.6734,
+      "step": 3346
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.690107270560191e-07,
+      "loss": 1.8064,
+      "step": 3347
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.689709972189113e-07,
+      "loss": 1.8454,
+      "step": 3348
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.689312673818037e-07,
+      "loss": 1.6946,
+      "step": 3349
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.68891537544696e-07,
+      "loss": 2.0171,
+      "step": 3350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.688518077075884e-07,
+      "loss": 1.895,
+      "step": 3351
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.688120778704808e-07,
+      "loss": 1.9615,
+      "step": 3352
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.687723480333731e-07,
+      "loss": 1.7643,
+      "step": 3353
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.687326181962654e-07,
+      "loss": 1.6806,
+      "step": 3354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.686928883591576e-07,
+      "loss": 1.8142,
+      "step": 3355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.6865315852205e-07,
+      "loss": 2.0355,
+      "step": 3356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.686134286849423e-07,
+      "loss": 2.0691,
+      "step": 3357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.685736988478347e-07,
+      "loss": 1.7505,
+      "step": 3358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.68533969010727e-07,
+      "loss": 1.7843,
+      "step": 3359
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.684942391736193e-07,
+      "loss": 1.8543,
+      "step": 3360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.684545093365117e-07,
+      "loss": 2.0304,
+      "step": 3361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.68414779499404e-07,
+      "loss": 1.7064,
+      "step": 3362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.683750496622964e-07,
+      "loss": 1.9053,
+      "step": 3363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.683353198251887e-07,
+      "loss": 2.0851,
+      "step": 3364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.682955899880811e-07,
+      "loss": 1.4711,
+      "step": 3365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.682558601509733e-07,
+      "loss": 1.8242,
+      "step": 3366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.682161303138657e-07,
+      "loss": 1.9643,
+      "step": 3367
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.681764004767579e-07,
+      "loss": 1.6952,
+      "step": 3368
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.681366706396503e-07,
+      "loss": 1.7051,
+      "step": 3369
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.680969408025427e-07,
+      "loss": 1.7699,
+      "step": 3370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.68057210965435e-07,
+      "loss": 1.4842,
+      "step": 3371
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.680174811283273e-07,
+      "loss": 1.8224,
+      "step": 3372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.679777512912196e-07,
+      "loss": 1.7114,
+      "step": 3373
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.67938021454112e-07,
+      "loss": 1.7508,
+      "step": 3374
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.678982916170043e-07,
+      "loss": 1.8635,
+      "step": 3375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.678585617798967e-07,
+      "loss": 1.8985,
+      "step": 3376
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.678188319427891e-07,
+      "loss": 1.9732,
+      "step": 3377
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.677791021056813e-07,
+      "loss": 1.7583,
+      "step": 3378
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.677393722685737e-07,
+      "loss": 2.0091,
+      "step": 3379
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.67699642431466e-07,
+      "loss": 1.9763,
+      "step": 3380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.676599125943583e-07,
+      "loss": 2.0293,
+      "step": 3381
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.676201827572506e-07,
+      "loss": 2.0732,
+      "step": 3382
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.67580452920143e-07,
+      "loss": 2.0971,
+      "step": 3383
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.675407230830352e-07,
+      "loss": 2.1458,
+      "step": 3384
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 8.675009932459276e-07,
+      "loss": 1.6805,
+      "step": 3385
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.6746126340882e-07,
+      "loss": 1.7744,
+      "step": 3386
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.674215335717123e-07,
+      "loss": 1.9406,
+      "step": 3387
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.673818037346047e-07,
+      "loss": 2.2003,
+      "step": 3388
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.67342073897497e-07,
+      "loss": 1.8379,
+      "step": 3389
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.673023440603893e-07,
+      "loss": 1.722,
+      "step": 3390
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.672626142232816e-07,
+      "loss": 1.9752,
+      "step": 3391
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.67222884386174e-07,
+      "loss": 1.6492,
+      "step": 3392
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.671831545490664e-07,
+      "loss": 1.8965,
+      "step": 3393
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.671434247119587e-07,
+      "loss": 1.8244,
+      "step": 3394
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.67103694874851e-07,
+      "loss": 1.6879,
+      "step": 3395
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.670639650377432e-07,
+      "loss": 1.9969,
+      "step": 3396
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.670242352006356e-07,
+      "loss": 1.8024,
+      "step": 3397
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.669845053635279e-07,
+      "loss": 1.889,
+      "step": 3398
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.669447755264203e-07,
+      "loss": 1.8187,
+      "step": 3399
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.669050456893127e-07,
+      "loss": 2.0969,
+      "step": 3400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.66865315852205e-07,
+      "loss": 1.5726,
+      "step": 3401
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.668255860150973e-07,
+      "loss": 1.5479,
+      "step": 3402
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.667858561779896e-07,
+      "loss": 1.645,
+      "step": 3403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.66746126340882e-07,
+      "loss": 2.2479,
+      "step": 3404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.667063965037743e-07,
+      "loss": 1.8314,
+      "step": 3405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.666666666666667e-07,
+      "loss": 1.8647,
+      "step": 3406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.666269368295591e-07,
+      "loss": 2.0012,
+      "step": 3407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.665872069924512e-07,
+      "loss": 1.6943,
+      "step": 3408
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.665474771553436e-07,
+      "loss": 1.8113,
+      "step": 3409
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.665077473182359e-07,
+      "loss": 1.7676,
+      "step": 3410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.664680174811283e-07,
+      "loss": 2.0263,
+      "step": 3411
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.664282876440206e-07,
+      "loss": 2.0237,
+      "step": 3412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.66388557806913e-07,
+      "loss": 1.7156,
+      "step": 3413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.663488279698052e-07,
+      "loss": 1.9447,
+      "step": 3414
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.663090981326976e-07,
+      "loss": 1.6749,
+      "step": 3415
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.6626936829559e-07,
+      "loss": 1.8546,
+      "step": 3416
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.662296384584823e-07,
+      "loss": 1.9833,
+      "step": 3417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.661899086213747e-07,
+      "loss": 1.6684,
+      "step": 3418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.66150178784267e-07,
+      "loss": 1.9123,
+      "step": 3419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.661104489471593e-07,
+      "loss": 1.7025,
+      "step": 3420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.660707191100515e-07,
+      "loss": 1.9121,
+      "step": 3421
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.660309892729439e-07,
+      "loss": 1.8882,
+      "step": 3422
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.659912594358363e-07,
+      "loss": 1.8262,
+      "step": 3423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.659515295987286e-07,
+      "loss": 1.6837,
+      "step": 3424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.65911799761621e-07,
+      "loss": 1.969,
+      "step": 3425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.658720699245132e-07,
+      "loss": 1.779,
+      "step": 3426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.658323400874056e-07,
+      "loss": 1.7062,
+      "step": 3427
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.657926102502979e-07,
+      "loss": 1.9079,
+      "step": 3428
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.657528804131903e-07,
+      "loss": 1.6363,
+      "step": 3429
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.657131505760827e-07,
+      "loss": 1.7325,
+      "step": 3430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.65673420738975e-07,
+      "loss": 1.8497,
+      "step": 3431
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.656336909018673e-07,
+      "loss": 1.85,
+      "step": 3432
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.655939610647596e-07,
+      "loss": 1.6822,
+      "step": 3433
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.655542312276519e-07,
+      "loss": 1.8853,
+      "step": 3434
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.655145013905442e-07,
+      "loss": 1.9122,
+      "step": 3435
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.654747715534366e-07,
+      "loss": 1.7611,
+      "step": 3436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.654350417163289e-07,
+      "loss": 1.7666,
+      "step": 3437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.653953118792212e-07,
+      "loss": 1.9759,
+      "step": 3438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.653555820421136e-07,
+      "loss": 1.4641,
+      "step": 3439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.653158522050059e-07,
+      "loss": 1.6692,
+      "step": 3440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.652761223678983e-07,
+      "loss": 1.7442,
+      "step": 3441
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.652363925307906e-07,
+      "loss": 1.6863,
+      "step": 3442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.65196662693683e-07,
+      "loss": 1.5676,
+      "step": 3443
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.651569328565752e-07,
+      "loss": 1.8898,
+      "step": 3444
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.651172030194676e-07,
+      "loss": 1.9827,
+      "step": 3445
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.6507747318236e-07,
+      "loss": 1.7634,
+      "step": 3446
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.650377433452522e-07,
+      "loss": 1.8065,
+      "step": 3447
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.649980135081446e-07,
+      "loss": 1.8567,
+      "step": 3448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.649582836710369e-07,
+      "loss": 1.8328,
+      "step": 3449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.649185538339292e-07,
+      "loss": 1.8986,
+      "step": 3450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.648788239968215e-07,
+      "loss": 1.6841,
+      "step": 3451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.648390941597139e-07,
+      "loss": 1.7577,
+      "step": 3452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.647993643226062e-07,
+      "loss": 2.0125,
+      "step": 3453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.647596344854986e-07,
+      "loss": 1.7121,
+      "step": 3454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.64719904648391e-07,
+      "loss": 1.8186,
+      "step": 3455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.646801748112832e-07,
+      "loss": 1.7204,
+      "step": 3456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.646404449741756e-07,
+      "loss": 1.9906,
+      "step": 3457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.646007151370679e-07,
+      "loss": 1.6635,
+      "step": 3458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.645609852999603e-07,
+      "loss": 1.9647,
+      "step": 3459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.645212554628526e-07,
+      "loss": 1.8201,
+      "step": 3460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.644815256257449e-07,
+      "loss": 1.8299,
+      "step": 3461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.644417957886371e-07,
+      "loss": 2.0413,
+      "step": 3462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.644020659515295e-07,
+      "loss": 1.6466,
+      "step": 3463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.643623361144219e-07,
+      "loss": 1.8018,
+      "step": 3464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.643226062773142e-07,
+      "loss": 1.8731,
+      "step": 3465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.642828764402066e-07,
+      "loss": 1.8616,
+      "step": 3466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.642431466030989e-07,
+      "loss": 1.9106,
+      "step": 3467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.642034167659912e-07,
+      "loss": 1.6355,
+      "step": 3468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.641636869288835e-07,
+      "loss": 1.7989,
+      "step": 3469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.641239570917759e-07,
+      "loss": 1.914,
+      "step": 3470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.640842272546683e-07,
+      "loss": 1.9232,
+      "step": 3471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.640444974175606e-07,
+      "loss": 1.6821,
+      "step": 3472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.64004767580453e-07,
+      "loss": 2.0336,
+      "step": 3473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.639650377433451e-07,
+      "loss": 1.921,
+      "step": 3474
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.639253079062375e-07,
+      "loss": 1.6461,
+      "step": 3475
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.638855780691298e-07,
+      "loss": 1.5019,
+      "step": 3476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.638458482320222e-07,
+      "loss": 1.719,
+      "step": 3477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.638061183949146e-07,
+      "loss": 1.7903,
+      "step": 3478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.637663885578069e-07,
+      "loss": 1.7343,
+      "step": 3479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.637266587206992e-07,
+      "loss": 1.9568,
+      "step": 3480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.636869288835915e-07,
+      "loss": 1.6498,
+      "step": 3481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.636471990464839e-07,
+      "loss": 1.7625,
+      "step": 3482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.636074692093762e-07,
+      "loss": 1.6544,
+      "step": 3483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.635677393722686e-07,
+      "loss": 1.7616,
+      "step": 3484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.63528009535161e-07,
+      "loss": 1.6755,
+      "step": 3485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.634882796980532e-07,
+      "loss": 1.8786,
+      "step": 3486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.634485498609455e-07,
+      "loss": 1.666,
+      "step": 3487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.634088200238378e-07,
+      "loss": 2.2054,
+      "step": 3488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.633690901867302e-07,
+      "loss": 2.0145,
+      "step": 3489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.633293603496225e-07,
+      "loss": 1.6581,
+      "step": 3490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.632896305125149e-07,
+      "loss": 1.8536,
+      "step": 3491
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.632499006754071e-07,
+      "loss": 1.6135,
+      "step": 3492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.632101708382995e-07,
+      "loss": 1.5165,
+      "step": 3493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.631704410011919e-07,
+      "loss": 1.6877,
+      "step": 3494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.631307111640842e-07,
+      "loss": 1.7497,
+      "step": 3495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.630909813269766e-07,
+      "loss": 1.8038,
+      "step": 3496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.630512514898689e-07,
+      "loss": 2.0002,
+      "step": 3497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.630115216527612e-07,
+      "loss": 1.6547,
+      "step": 3498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.629717918156535e-07,
+      "loss": 1.8025,
+      "step": 3499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.629320619785458e-07,
+      "loss": 1.6133,
+      "step": 3500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.628923321414382e-07,
+      "loss": 1.8913,
+      "step": 3501
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.628526023043305e-07,
+      "loss": 1.9032,
+      "step": 3502
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.628128724672229e-07,
+      "loss": 1.9301,
+      "step": 3503
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.627731426301151e-07,
+      "loss": 1.9136,
+      "step": 3504
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.627334127930075e-07,
+      "loss": 1.9656,
+      "step": 3505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.626936829558998e-07,
+      "loss": 1.8365,
+      "step": 3506
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.626539531187922e-07,
+      "loss": 1.8656,
+      "step": 3507
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.626142232816846e-07,
+      "loss": 1.8091,
+      "step": 3508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.625744934445769e-07,
+      "loss": 1.7585,
+      "step": 3509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.625347636074692e-07,
+      "loss": 1.8302,
+      "step": 3510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.624950337703615e-07,
+      "loss": 1.9444,
+      "step": 3511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624553039332539e-07,
+      "loss": 1.9099,
+      "step": 3512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624155740961461e-07,
+      "loss": 1.8614,
+      "step": 3513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.623758442590385e-07,
+      "loss": 1.6595,
+      "step": 3514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.623361144219308e-07,
+      "loss": 2.0015,
+      "step": 3515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622963845848231e-07,
+      "loss": 1.8957,
+      "step": 3516
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622566547477155e-07,
+      "loss": 1.9057,
+      "step": 3517
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622169249106078e-07,
+      "loss": 1.8028,
+      "step": 3518
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.621771950735002e-07,
+      "loss": 1.8681,
+      "step": 3519
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.621374652363925e-07,
+      "loss": 1.7589,
+      "step": 3520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620977353992849e-07,
+      "loss": 1.5584,
+      "step": 3521
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620580055621771e-07,
+      "loss": 1.6564,
+      "step": 3522
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620182757250695e-07,
+      "loss": 1.4871,
+      "step": 3523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619785458879619e-07,
+      "loss": 1.4378,
+      "step": 3524
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.619388160508542e-07,
+      "loss": 1.7009,
+      "step": 3525
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.618990862137466e-07,
+      "loss": 1.8407,
+      "step": 3526
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.618593563766388e-07,
+      "loss": 1.9057,
+      "step": 3527
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.618196265395311e-07,
+      "loss": 2.0089,
+      "step": 3528
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.617798967024234e-07,
+      "loss": 1.9887,
+      "step": 3529
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.617401668653158e-07,
+      "loss": 1.8306,
+      "step": 3530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.617004370282081e-07,
+      "loss": 1.9252,
+      "step": 3531
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.616607071911005e-07,
+      "loss": 1.6362,
+      "step": 3532
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.616209773539929e-07,
+      "loss": 2.2112,
+      "step": 3533
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.615812475168851e-07,
+      "loss": 1.9428,
+      "step": 3534
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.615415176797775e-07,
+      "loss": 1.7201,
+      "step": 3535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.615017878426698e-07,
+      "loss": 1.6167,
+      "step": 3536
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.614620580055622e-07,
+      "loss": 1.5191,
+      "step": 3537
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.614223281684545e-07,
+      "loss": 1.7591,
+      "step": 3538
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.613825983313469e-07,
+      "loss": 2.3054,
+      "step": 3539
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.61342868494239e-07,
+      "loss": 1.9093,
+      "step": 3540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.613031386571314e-07,
+      "loss": 1.8534,
+      "step": 3541
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.612634088200238e-07,
+      "loss": 1.8153,
+      "step": 3542
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.612236789829161e-07,
+      "loss": 1.9001,
+      "step": 3543
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.611839491458085e-07,
+      "loss": 1.8162,
+      "step": 3544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.611442193087008e-07,
+      "loss": 1.6999,
+      "step": 3545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.611044894715931e-07,
+      "loss": 1.8546,
+      "step": 3546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.610647596344854e-07,
+      "loss": 1.8749,
+      "step": 3547
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.610250297973778e-07,
+      "loss": 2.0933,
+      "step": 3548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.609852999602702e-07,
+      "loss": 1.5584,
+      "step": 3549
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.609455701231625e-07,
+      "loss": 1.8242,
+      "step": 3550
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.609058402860549e-07,
+      "loss": 1.85,
+      "step": 3551
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.608661104489471e-07,
+      "loss": 1.8403,
+      "step": 3552
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.608263806118394e-07,
+      "loss": 1.8219,
+      "step": 3553
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.607866507747317e-07,
+      "loss": 2.0632,
+      "step": 3554
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.607469209376241e-07,
+      "loss": 1.7754,
+      "step": 3555
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.607071911005165e-07,
+      "loss": 1.8426,
+      "step": 3556
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.606674612634088e-07,
+      "loss": 1.5754,
+      "step": 3557
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.606277314263011e-07,
+      "loss": 1.9709,
+      "step": 3558
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.605880015891934e-07,
+      "loss": 1.8987,
+      "step": 3559
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.605482717520858e-07,
+      "loss": 2.1731,
+      "step": 3560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.605085419149781e-07,
+      "loss": 1.7613,
+      "step": 3561
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.604688120778705e-07,
+      "loss": 1.5324,
+      "step": 3562
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.604290822407629e-07,
+      "loss": 1.9712,
+      "step": 3563
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.603893524036551e-07,
+      "loss": 1.9504,
+      "step": 3564
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.603496225665475e-07,
+      "loss": 1.6818,
+      "step": 3565
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.603098927294397e-07,
+      "loss": 1.7194,
+      "step": 3566
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.602701628923321e-07,
+      "loss": 1.6447,
+      "step": 3567
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.602304330552244e-07,
+      "loss": 1.8533,
+      "step": 3568
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.601907032181168e-07,
+      "loss": 1.6616,
+      "step": 3569
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.60150973381009e-07,
+      "loss": 1.6944,
+      "step": 3570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.601112435439014e-07,
+      "loss": 1.7206,
+      "step": 3571
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.600715137067938e-07,
+      "loss": 1.8239,
+      "step": 3572
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.600317838696861e-07,
+      "loss": 1.8404,
+      "step": 3573
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.599920540325785e-07,
+      "loss": 2.0502,
+      "step": 3574
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.599523241954708e-07,
+      "loss": 1.8089,
+      "step": 3575
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.599125943583631e-07,
+      "loss": 1.7732,
+      "step": 3576
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.598728645212554e-07,
+      "loss": 1.6011,
+      "step": 3577
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.598331346841478e-07,
+      "loss": 1.9086,
+      "step": 3578
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.5979340484704e-07,
+      "loss": 1.7027,
+      "step": 3579
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.597536750099324e-07,
+      "loss": 1.7481,
+      "step": 3580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.597139451728248e-07,
+      "loss": 1.9555,
+      "step": 3581
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.59674215335717e-07,
+      "loss": 1.8757,
+      "step": 3582
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.596344854986094e-07,
+      "loss": 1.8313,
+      "step": 3583
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.595947556615017e-07,
+      "loss": 1.8661,
+      "step": 3584
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.595550258243941e-07,
+      "loss": 1.6492,
+      "step": 3585
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.595152959872865e-07,
+      "loss": 1.9406,
+      "step": 3586
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.594755661501788e-07,
+      "loss": 1.6287,
+      "step": 3587
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.594358363130711e-07,
+      "loss": 1.8274,
+      "step": 3588
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.593961064759634e-07,
+      "loss": 1.8738,
+      "step": 3589
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.593563766388558e-07,
+      "loss": 1.907,
+      "step": 3590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.593166468017481e-07,
+      "loss": 1.868,
+      "step": 3591
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.592769169646405e-07,
+      "loss": 1.8558,
+      "step": 3592
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.592371871275327e-07,
+      "loss": 1.6136,
+      "step": 3593
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.59197457290425e-07,
+      "loss": 1.6569,
+      "step": 3594
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.591577274533174e-07,
+      "loss": 1.7829,
+      "step": 3595
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.591179976162097e-07,
+      "loss": 1.7656,
+      "step": 3596
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.590782677791021e-07,
+      "loss": 2.0636,
+      "step": 3597
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.590385379419944e-07,
+      "loss": 1.7,
+      "step": 3598
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.589988081048868e-07,
+      "loss": 1.8007,
+      "step": 3599
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.58959078267779e-07,
+      "loss": 1.4043,
+      "step": 3600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.589193484306714e-07,
+      "loss": 1.8415,
+      "step": 3601
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.588796185935638e-07,
+      "loss": 1.5776,
+      "step": 3602
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.588398887564561e-07,
+      "loss": 2.004,
+      "step": 3603
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.588001589193485e-07,
+      "loss": 1.8265,
+      "step": 3604
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.587604290822408e-07,
+      "loss": 1.5178,
+      "step": 3605
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.58720699245133e-07,
+      "loss": 1.7594,
+      "step": 3606
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.586809694080253e-07,
+      "loss": 1.5611,
+      "step": 3607
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.586412395709177e-07,
+      "loss": 1.7561,
+      "step": 3608
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.5860150973381e-07,
+      "loss": 1.9067,
+      "step": 3609
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.585617798967024e-07,
+      "loss": 1.8299,
+      "step": 3610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.585220500595948e-07,
+      "loss": 2.1157,
+      "step": 3611
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.58482320222487e-07,
+      "loss": 1.7629,
+      "step": 3612
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.584425903853794e-07,
+      "loss": 1.8098,
+      "step": 3613
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.584028605482717e-07,
+      "loss": 1.6971,
+      "step": 3614
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.583631307111641e-07,
+      "loss": 1.9895,
+      "step": 3615
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.583234008740564e-07,
+      "loss": 1.9433,
+      "step": 3616
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.582836710369488e-07,
+      "loss": 1.8457,
+      "step": 3617
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.582439411998411e-07,
+      "loss": 1.4562,
+      "step": 3618
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.582042113627333e-07,
+      "loss": 1.9531,
+      "step": 3619
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.581644815256257e-07,
+      "loss": 1.7556,
+      "step": 3620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.58124751688518e-07,
+      "loss": 2.0958,
+      "step": 3621
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.580850218514104e-07,
+      "loss": 1.868,
+      "step": 3622
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.580452920143027e-07,
+      "loss": 1.7714,
+      "step": 3623
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.58005562177195e-07,
+      "loss": 1.7877,
+      "step": 3624
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.579658323400873e-07,
+      "loss": 1.8906,
+      "step": 3625
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.579261025029797e-07,
+      "loss": 1.7861,
+      "step": 3626
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.578863726658721e-07,
+      "loss": 1.9251,
+      "step": 3627
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.578466428287644e-07,
+      "loss": 1.7458,
+      "step": 3628
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.578069129916568e-07,
+      "loss": 1.5379,
+      "step": 3629
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.57767183154549e-07,
+      "loss": 1.8755,
+      "step": 3630
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.577274533174414e-07,
+      "loss": 1.6232,
+      "step": 3631
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.576877234803336e-07,
+      "loss": 1.8898,
+      "step": 3632
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.57647993643226e-07,
+      "loss": 1.8964,
+      "step": 3633
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.576082638061184e-07,
+      "loss": 1.8601,
+      "step": 3634
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.575685339690107e-07,
+      "loss": 1.6127,
+      "step": 3635
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.57528804131903e-07,
+      "loss": 1.5389,
+      "step": 3636
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.574890742947953e-07,
+      "loss": 1.5973,
+      "step": 3637
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.574493444576877e-07,
+      "loss": 1.8261,
+      "step": 3638
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.5740961462058e-07,
+      "loss": 1.8081,
+      "step": 3639
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.573698847834724e-07,
+      "loss": 1.9072,
+      "step": 3640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.573301549463648e-07,
+      "loss": 1.7489,
+      "step": 3641
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.57290425109257e-07,
+      "loss": 1.8178,
+      "step": 3642
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.572506952721494e-07,
+      "loss": 2.011,
+      "step": 3643
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.572109654350417e-07,
+      "loss": 1.8168,
+      "step": 3644
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.57171235597934e-07,
+      "loss": 1.6791,
+      "step": 3645
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.571315057608263e-07,
+      "loss": 1.9135,
+      "step": 3646
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.570917759237186e-07,
+      "loss": 1.7954,
+      "step": 3647
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.570520460866109e-07,
+      "loss": 1.9951,
+      "step": 3648
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.570123162495033e-07,
+      "loss": 1.7452,
+      "step": 3649
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.569725864123957e-07,
+      "loss": 1.7744,
+      "step": 3650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.56932856575288e-07,
+      "loss": 1.6521,
+      "step": 3651
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.568931267381804e-07,
+      "loss": 1.8426,
+      "step": 3652
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.568533969010726e-07,
+      "loss": 1.8125,
+      "step": 3653
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.56813667063965e-07,
+      "loss": 1.9433,
+      "step": 3654
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.567739372268573e-07,
+      "loss": 1.6878,
+      "step": 3655
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.567342073897497e-07,
+      "loss": 1.8162,
+      "step": 3656
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.566944775526421e-07,
+      "loss": 1.8181,
+      "step": 3657
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.566547477155344e-07,
+      "loss": 1.9197,
+      "step": 3658
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.566150178784266e-07,
+      "loss": 1.8225,
+      "step": 3659
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.565752880413189e-07,
+      "loss": 2.0987,
+      "step": 3660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.565355582042113e-07,
+      "loss": 1.8109,
+      "step": 3661
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.564958283671036e-07,
+      "loss": 2.0316,
+      "step": 3662
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.56456098529996e-07,
+      "loss": 1.8242,
+      "step": 3663
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.564163686928884e-07,
+      "loss": 1.9436,
+      "step": 3664
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.563766388557806e-07,
+      "loss": 1.7869,
+      "step": 3665
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.56336909018673e-07,
+      "loss": 1.7637,
+      "step": 3666
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.562971791815653e-07,
+      "loss": 2.0613,
+      "step": 3667
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.562574493444577e-07,
+      "loss": 2.0663,
+      "step": 3668
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.5621771950735e-07,
+      "loss": 1.7615,
+      "step": 3669
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.561779896702424e-07,
+      "loss": 1.6158,
+      "step": 3670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.561382598331345e-07,
+      "loss": 1.8245,
+      "step": 3671
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.560985299960269e-07,
+      "loss": 1.6937,
+      "step": 3672
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.560588001589193e-07,
+      "loss": 1.7361,
+      "step": 3673
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.560190703218116e-07,
+      "loss": 1.7165,
+      "step": 3674
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.55979340484704e-07,
+      "loss": 1.969,
+      "step": 3675
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.559396106475963e-07,
+      "loss": 1.8744,
+      "step": 3676
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.558998808104886e-07,
+      "loss": 1.7019,
+      "step": 3677
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.558601509733809e-07,
+      "loss": 1.8954,
+      "step": 3678
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.558204211362733e-07,
+      "loss": 1.6461,
+      "step": 3679
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.557806912991657e-07,
+      "loss": 1.8096,
+      "step": 3680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.55740961462058e-07,
+      "loss": 1.64,
+      "step": 3681
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.557012316249504e-07,
+      "loss": 1.7443,
+      "step": 3682
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.556615017878426e-07,
+      "loss": 1.8023,
+      "step": 3683
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.55621771950735e-07,
+      "loss": 2.1617,
+      "step": 3684
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.555820421136272e-07,
+      "loss": 1.5374,
+      "step": 3685
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.555423122765196e-07,
+      "loss": 2.0839,
+      "step": 3686
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.555025824394119e-07,
+      "loss": 1.6608,
+      "step": 3687
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.554628526023043e-07,
+      "loss": 1.6628,
+      "step": 3688
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.554231227651966e-07,
+      "loss": 1.9772,
+      "step": 3689
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.553833929280889e-07,
+      "loss": 1.8438,
+      "step": 3690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.553436630909813e-07,
+      "loss": 1.8894,
+      "step": 3691
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.553039332538736e-07,
+      "loss": 1.8399,
+      "step": 3692
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.55264203416766e-07,
+      "loss": 2.1048,
+      "step": 3693
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.552244735796583e-07,
+      "loss": 1.7885,
+      "step": 3694
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.551847437425506e-07,
+      "loss": 1.6107,
+      "step": 3695
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.55145013905443e-07,
+      "loss": 1.6809,
+      "step": 3696
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.551052840683353e-07,
+      "loss": 2.0344,
+      "step": 3697
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.550655542312276e-07,
+      "loss": 2.1628,
+      "step": 3698
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.550258243941199e-07,
+      "loss": 1.7589,
+      "step": 3699
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.549860945570123e-07,
+      "loss": 1.8848,
+      "step": 3700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.549463647199045e-07,
+      "loss": 1.8817,
+      "step": 3701
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.549066348827969e-07,
+      "loss": 1.8109,
+      "step": 3702
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.548669050456892e-07,
+      "loss": 1.9461,
+      "step": 3703
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.548271752085816e-07,
+      "loss": 1.459,
+      "step": 3704
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.54787445371474e-07,
+      "loss": 1.7714,
+      "step": 3705
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.547477155343663e-07,
+      "loss": 1.8873,
+      "step": 3706
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.547079856972586e-07,
+      "loss": 1.6161,
+      "step": 3707
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.546682558601509e-07,
+      "loss": 1.8357,
+      "step": 3708
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.546285260230433e-07,
+      "loss": 1.7494,
+      "step": 3709
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.545887961859356e-07,
+      "loss": 1.7772,
+      "step": 3710
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.545490663488279e-07,
+      "loss": 1.7326,
+      "step": 3711
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.545093365117203e-07,
+      "loss": 1.9679,
+      "step": 3712
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.544696066746125e-07,
+      "loss": 1.9871,
+      "step": 3713
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.544298768375049e-07,
+      "loss": 1.9075,
+      "step": 3714
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.543901470003972e-07,
+      "loss": 2.0834,
+      "step": 3715
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.543504171632896e-07,
+      "loss": 1.7994,
+      "step": 3716
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.543106873261819e-07,
+      "loss": 2.1229,
+      "step": 3717
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.542709574890743e-07,
+      "loss": 1.7484,
+      "step": 3718
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.542312276519666e-07,
+      "loss": 1.9248,
+      "step": 3719
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.541914978148589e-07,
+      "loss": 2.1184,
+      "step": 3720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.541517679777513e-07,
+      "loss": 1.822,
+      "step": 3721
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.541120381406436e-07,
+      "loss": 1.9918,
+      "step": 3722
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.54072308303536e-07,
+      "loss": 2.0231,
+      "step": 3723
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.540325784664283e-07,
+      "loss": 1.7236,
+      "step": 3724
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.539928486293205e-07,
+      "loss": 1.7063,
+      "step": 3725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.539531187922128e-07,
+      "loss": 1.6591,
+      "step": 3726
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.539133889551052e-07,
+      "loss": 1.7044,
+      "step": 3727
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.538736591179976e-07,
+      "loss": 1.5503,
+      "step": 3728
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.538339292808899e-07,
+      "loss": 1.8809,
+      "step": 3729
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.537941994437823e-07,
+      "loss": 1.8834,
+      "step": 3730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.537544696066745e-07,
+      "loss": 1.8475,
+      "step": 3731
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.537147397695669e-07,
+      "loss": 1.9862,
+      "step": 3732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.536750099324592e-07,
+      "loss": 1.5211,
+      "step": 3733
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.536352800953516e-07,
+      "loss": 1.98,
+      "step": 3734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.53595550258244e-07,
+      "loss": 1.9745,
+      "step": 3735
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.535558204211363e-07,
+      "loss": 1.7174,
+      "step": 3736
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.535160905840285e-07,
+      "loss": 1.8465,
+      "step": 3737
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.534763607469208e-07,
+      "loss": 1.782,
+      "step": 3738
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.534366309098132e-07,
+      "loss": 1.8447,
+      "step": 3739
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.533969010727055e-07,
+      "loss": 1.8809,
+      "step": 3740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.533571712355979e-07,
+      "loss": 1.673,
+      "step": 3741
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.533174413984903e-07,
+      "loss": 1.8544,
+      "step": 3742
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.532777115613825e-07,
+      "loss": 1.8684,
+      "step": 3743
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.532379817242749e-07,
+      "loss": 1.823,
+      "step": 3744
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.531982518871672e-07,
+      "loss": 1.7419,
+      "step": 3745
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.531585220500596e-07,
+      "loss": 1.7567,
+      "step": 3746
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.531187922129519e-07,
+      "loss": 1.8378,
+      "step": 3747
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.530790623758443e-07,
+      "loss": 2.056,
+      "step": 3748
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.530393325387365e-07,
+      "loss": 2.0068,
+      "step": 3749
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.529996027016289e-07,
+      "loss": 1.9282,
+      "step": 3750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.529598728645212e-07,
+      "loss": 2.0341,
+      "step": 3751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.529201430274135e-07,
+      "loss": 1.9042,
+      "step": 3752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.528804131903059e-07,
+      "loss": 1.7096,
+      "step": 3753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.528406833531982e-07,
+      "loss": 1.8974,
+      "step": 3754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.528009535160905e-07,
+      "loss": 1.9807,
+      "step": 3755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.527612236789828e-07,
+      "loss": 1.7751,
+      "step": 3756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.527214938418752e-07,
+      "loss": 2.3495,
+      "step": 3757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.526817640047676e-07,
+      "loss": 1.6779,
+      "step": 3758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.526420341676599e-07,
+      "loss": 1.7568,
+      "step": 3759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.526023043305523e-07,
+      "loss": 1.7763,
+      "step": 3760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.525625744934445e-07,
+      "loss": 1.8027,
+      "step": 3761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.525228446563369e-07,
+      "loss": 1.9287,
+      "step": 3762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.524831148192292e-07,
+      "loss": 1.9005,
+      "step": 3763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.524433849821215e-07,
+      "loss": 1.8486,
+      "step": 3764
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.524036551450138e-07,
+      "loss": 1.7217,
+      "step": 3765
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.523639253079062e-07,
+      "loss": 1.9501,
+      "step": 3766
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.523241954707985e-07,
+      "loss": 1.609,
+      "step": 3767
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.522844656336908e-07,
+      "loss": 1.6453,
+      "step": 3768
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.522447357965832e-07,
+      "loss": 1.7367,
+      "step": 3769
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.522050059594755e-07,
+      "loss": 1.8475,
+      "step": 3770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.521652761223679e-07,
+      "loss": 1.7849,
+      "step": 3771
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.521255462852602e-07,
+      "loss": 1.796,
+      "step": 3772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.520858164481525e-07,
+      "loss": 1.5104,
+      "step": 3773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.520460866110449e-07,
+      "loss": 2.1082,
+      "step": 3774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.520063567739372e-07,
+      "loss": 1.9226,
+      "step": 3775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.519666269368296e-07,
+      "loss": 1.7297,
+      "step": 3776
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.519268970997218e-07,
+      "loss": 1.7354,
+      "step": 3777
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.518871672626142e-07,
+      "loss": 2.0192,
+      "step": 3778
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.518474374255064e-07,
+      "loss": 1.8894,
+      "step": 3779
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.518077075883988e-07,
+      "loss": 1.7988,
+      "step": 3780
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.517679777512911e-07,
+      "loss": 1.9647,
+      "step": 3781
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.517282479141835e-07,
+      "loss": 1.6359,
+      "step": 3782
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.516885180770759e-07,
+      "loss": 1.8981,
+      "step": 3783
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.33822363203806505,
+      "eval_loss": 1.8228262662887573,
+      "eval_runtime": 6.4587,
+      "eval_samples_per_second": 390.483,
+      "eval_steps_per_second": 6.193,
+      "step": 3783
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.516487882399682e-07,
+      "loss": 1.6629,
+      "step": 3784
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.516090584028605e-07,
+      "loss": 1.7115,
+      "step": 3785
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.515693285657528e-07,
+      "loss": 1.6797,
+      "step": 3786
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.515295987286452e-07,
+      "loss": 1.8043,
+      "step": 3787
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.514898688915375e-07,
+      "loss": 1.9168,
+      "step": 3788
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.514501390544299e-07,
+      "loss": 1.5299,
+      "step": 3789
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.514104092173223e-07,
+      "loss": 1.8508,
+      "step": 3790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.513706793802144e-07,
+      "loss": 1.6931,
+      "step": 3791
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.513309495431068e-07,
+      "loss": 1.6914,
+      "step": 3792
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.512912197059991e-07,
+      "loss": 1.7992,
+      "step": 3793
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.512514898688915e-07,
+      "loss": 1.6608,
+      "step": 3794
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.512117600317838e-07,
+      "loss": 1.7817,
+      "step": 3795
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.511720301946762e-07,
+      "loss": 2.06,
+      "step": 3796
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.511323003575685e-07,
+      "loss": 1.7962,
+      "step": 3797
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.510925705204608e-07,
+      "loss": 1.8218,
+      "step": 3798
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.510528406833532e-07,
+      "loss": 1.8486,
+      "step": 3799
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.510131108462455e-07,
+      "loss": 1.7468,
+      "step": 3800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.509733810091379e-07,
+      "loss": 1.9876,
+      "step": 3801
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.509336511720302e-07,
+      "loss": 1.8903,
+      "step": 3802
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.508939213349224e-07,
+      "loss": 1.8858,
+      "step": 3803
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.508541914978147e-07,
+      "loss": 1.6942,
+      "step": 3804
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.508144616607071e-07,
+      "loss": 1.9789,
+      "step": 3805
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.507747318235995e-07,
+      "loss": 1.783,
+      "step": 3806
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.507350019864918e-07,
+      "loss": 2.0242,
+      "step": 3807
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.506952721493842e-07,
+      "loss": 1.7653,
+      "step": 3808
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.506555423122764e-07,
+      "loss": 1.9095,
+      "step": 3809
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.506158124751688e-07,
+      "loss": 1.9244,
+      "step": 3810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.505760826380611e-07,
+      "loss": 1.4882,
+      "step": 3811
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.505363528009535e-07,
+      "loss": 1.9016,
+      "step": 3812
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.504966229638459e-07,
+      "loss": 1.7929,
+      "step": 3813
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.504568931267382e-07,
+      "loss": 1.8156,
+      "step": 3814
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.504171632896305e-07,
+      "loss": 1.792,
+      "step": 3815
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.503774334525228e-07,
+      "loss": 1.8427,
+      "step": 3816
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.503377036154151e-07,
+      "loss": 1.7948,
+      "step": 3817
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.502979737783074e-07,
+      "loss": 1.8883,
+      "step": 3818
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.502582439411998e-07,
+      "loss": 1.8071,
+      "step": 3819
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.502185141040922e-07,
+      "loss": 1.7584,
+      "step": 3820
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.501787842669844e-07,
+      "loss": 1.848,
+      "step": 3821
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.501390544298768e-07,
+      "loss": 2.0526,
+      "step": 3822
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.500993245927691e-07,
+      "loss": 1.7695,
+      "step": 3823
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.500595947556615e-07,
+      "loss": 1.7802,
+      "step": 3824
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.500198649185538e-07,
+      "loss": 1.7622,
+      "step": 3825
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.499801350814462e-07,
+      "loss": 1.8611,
+      "step": 3826
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.499404052443384e-07,
+      "loss": 1.8006,
+      "step": 3827
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.499006754072308e-07,
+      "loss": 1.9292,
+      "step": 3828
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.498609455701232e-07,
+      "loss": 1.8235,
+      "step": 3829
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.498212157330154e-07,
+      "loss": 1.6695,
+      "step": 3830
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.497814858959078e-07,
+      "loss": 1.6557,
+      "step": 3831
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.497417560588001e-07,
+      "loss": 1.7314,
+      "step": 3832
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.497020262216924e-07,
+      "loss": 1.6585,
+      "step": 3833
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.496622963845847e-07,
+      "loss": 1.4736,
+      "step": 3834
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.496225665474771e-07,
+      "loss": 1.8949,
+      "step": 3835
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.495828367103695e-07,
+      "loss": 1.802,
+      "step": 3836
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.495431068732618e-07,
+      "loss": 1.7512,
+      "step": 3837
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.495033770361542e-07,
+      "loss": 1.8783,
+      "step": 3838
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.494636471990464e-07,
+      "loss": 1.868,
+      "step": 3839
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.494239173619388e-07,
+      "loss": 1.8264,
+      "step": 3840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.493841875248311e-07,
+      "loss": 1.46,
+      "step": 3841
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.493444576877235e-07,
+      "loss": 1.7249,
+      "step": 3842
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.493047278506157e-07,
+      "loss": 1.7991,
+      "step": 3843
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.492649980135081e-07,
+      "loss": 1.9109,
+      "step": 3844
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.492252681764004e-07,
+      "loss": 1.7626,
+      "step": 3845
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.491855383392927e-07,
+      "loss": 1.6492,
+      "step": 3846
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.491458085021851e-07,
+      "loss": 1.7348,
+      "step": 3847
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.491060786650774e-07,
+      "loss": 1.7295,
+      "step": 3848
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.490663488279698e-07,
+      "loss": 1.8568,
+      "step": 3849
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.490266189908621e-07,
+      "loss": 1.7614,
+      "step": 3850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.489868891537544e-07,
+      "loss": 1.6177,
+      "step": 3851
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.489471593166468e-07,
+      "loss": 1.7231,
+      "step": 3852
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.489074294795391e-07,
+      "loss": 1.7292,
+      "step": 3853
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.488676996424315e-07,
+      "loss": 1.9204,
+      "step": 3854
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.488279698053238e-07,
+      "loss": 1.7877,
+      "step": 3855
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.487882399682162e-07,
+      "loss": 1.8693,
+      "step": 3856
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.487485101311083e-07,
+      "loss": 1.744,
+      "step": 3857
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.487087802940007e-07,
+      "loss": 1.7231,
+      "step": 3858
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.48669050456893e-07,
+      "loss": 1.7173,
+      "step": 3859
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.486293206197854e-07,
+      "loss": 1.7419,
+      "step": 3860
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.485895907826778e-07,
+      "loss": 1.8775,
+      "step": 3861
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.485498609455701e-07,
+      "loss": 1.8843,
+      "step": 3862
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.485101311084624e-07,
+      "loss": 1.8498,
+      "step": 3863
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.484704012713547e-07,
+      "loss": 1.5179,
+      "step": 3864
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.484306714342471e-07,
+      "loss": 1.697,
+      "step": 3865
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.483909415971394e-07,
+      "loss": 1.7978,
+      "step": 3866
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.483512117600318e-07,
+      "loss": 1.5991,
+      "step": 3867
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.483114819229242e-07,
+      "loss": 1.7895,
+      "step": 3868
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.482717520858163e-07,
+      "loss": 1.9836,
+      "step": 3869
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.482320222487087e-07,
+      "loss": 1.5215,
+      "step": 3870
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.48192292411601e-07,
+      "loss": 2.0689,
+      "step": 3871
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.481525625744934e-07,
+      "loss": 1.8131,
+      "step": 3872
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.481128327373857e-07,
+      "loss": 1.8271,
+      "step": 3873
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.480731029002781e-07,
+      "loss": 1.8301,
+      "step": 3874
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.480333730631704e-07,
+      "loss": 1.9879,
+      "step": 3875
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.479936432260627e-07,
+      "loss": 1.9139,
+      "step": 3876
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.479539133889551e-07,
+      "loss": 1.6082,
+      "step": 3877
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.479141835518474e-07,
+      "loss": 1.781,
+      "step": 3878
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.478744537147398e-07,
+      "loss": 1.6336,
+      "step": 3879
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.478347238776321e-07,
+      "loss": 1.7603,
+      "step": 3880
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.477949940405244e-07,
+      "loss": 2.0638,
+      "step": 3881
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.477552642034168e-07,
+      "loss": 2.0384,
+      "step": 3882
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.47715534366309e-07,
+      "loss": 1.8279,
+      "step": 3883
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.476758045292014e-07,
+      "loss": 1.8668,
+      "step": 3884
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.476360746920937e-07,
+      "loss": 2.2204,
+      "step": 3885
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.475963448549861e-07,
+      "loss": 1.973,
+      "step": 3886
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.475566150178783e-07,
+      "loss": 1.8939,
+      "step": 3887
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.475168851807707e-07,
+      "loss": 1.9278,
+      "step": 3888
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.47477155343663e-07,
+      "loss": 1.6104,
+      "step": 3889
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.474374255065554e-07,
+      "loss": 1.9808,
+      "step": 3890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.473976956694478e-07,
+      "loss": 1.8274,
+      "step": 3891
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.473579658323401e-07,
+      "loss": 1.6677,
+      "step": 3892
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.473182359952324e-07,
+      "loss": 1.8726,
+      "step": 3893
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.472785061581247e-07,
+      "loss": 2.1607,
+      "step": 3894
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.472387763210171e-07,
+      "loss": 1.9377,
+      "step": 3895
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.471990464839093e-07,
+      "loss": 1.8799,
+      "step": 3896
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.471593166468017e-07,
+      "loss": 1.6331,
+      "step": 3897
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.47119586809694e-07,
+      "loss": 1.6188,
+      "step": 3898
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.470798569725863e-07,
+      "loss": 1.7969,
+      "step": 3899
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.470401271354787e-07,
+      "loss": 1.599,
+      "step": 3900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.47000397298371e-07,
+      "loss": 1.8649,
+      "step": 3901
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.469606674612634e-07,
+      "loss": 1.8338,
+      "step": 3902
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.469209376241557e-07,
+      "loss": 2.0379,
+      "step": 3903
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.468812077870481e-07,
+      "loss": 2.0494,
+      "step": 3904
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.468414779499403e-07,
+      "loss": 1.967,
+      "step": 3905
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.468017481128327e-07,
+      "loss": 1.5345,
+      "step": 3906
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.467620182757251e-07,
+      "loss": 2.098,
+      "step": 3907
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.467222884386174e-07,
+      "loss": 1.8444,
+      "step": 3908
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.466825586015097e-07,
+      "loss": 1.6134,
+      "step": 3909
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.46642828764402e-07,
+      "loss": 1.7105,
+      "step": 3910
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.466030989272943e-07,
+      "loss": 1.6284,
+      "step": 3911
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.465633690901866e-07,
+      "loss": 1.7413,
+      "step": 3912
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.46523639253079e-07,
+      "loss": 1.4739,
+      "step": 3913
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.464839094159714e-07,
+      "loss": 1.9049,
+      "step": 3914
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.464441795788637e-07,
+      "loss": 1.7745,
+      "step": 3915
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.464044497417561e-07,
+      "loss": 1.8508,
+      "step": 3916
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.463647199046483e-07,
+      "loss": 1.8861,
+      "step": 3917
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.463249900675407e-07,
+      "loss": 1.8866,
+      "step": 3918
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.46285260230433e-07,
+      "loss": 1.9534,
+      "step": 3919
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.462455303933254e-07,
+      "loss": 1.7785,
+      "step": 3920
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.462058005562178e-07,
+      "loss": 1.7766,
+      "step": 3921
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.461660707191101e-07,
+      "loss": 1.4918,
+      "step": 3922
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.461263408820023e-07,
+      "loss": 1.6646,
+      "step": 3923
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.460866110448946e-07,
+      "loss": 1.6783,
+      "step": 3924
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.46046881207787e-07,
+      "loss": 1.7466,
+      "step": 3925
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.460071513706793e-07,
+      "loss": 1.8635,
+      "step": 3926
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.459674215335717e-07,
+      "loss": 1.8502,
+      "step": 3927
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.45927691696464e-07,
+      "loss": 1.8319,
+      "step": 3928
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.458879618593563e-07,
+      "loss": 1.767,
+      "step": 3929
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.458482320222487e-07,
+      "loss": 1.8741,
+      "step": 3930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.45808502185141e-07,
+      "loss": 1.575,
+      "step": 3931
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.457687723480334e-07,
+      "loss": 1.6905,
+      "step": 3932
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.457290425109257e-07,
+      "loss": 2.0069,
+      "step": 3933
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.456893126738181e-07,
+      "loss": 1.4916,
+      "step": 3934
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.456495828367102e-07,
+      "loss": 1.9644,
+      "step": 3935
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.456098529996026e-07,
+      "loss": 1.5186,
+      "step": 3936
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.45570123162495e-07,
+      "loss": 1.5231,
+      "step": 3937
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.455303933253873e-07,
+      "loss": 1.658,
+      "step": 3938
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.454906634882797e-07,
+      "loss": 1.336,
+      "step": 3939
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.45450933651172e-07,
+      "loss": 2.065,
+      "step": 3940
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.454112038140643e-07,
+      "loss": 2.0025,
+      "step": 3941
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.453714739769566e-07,
+      "loss": 1.913,
+      "step": 3942
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.45331744139849e-07,
+      "loss": 1.6801,
+      "step": 3943
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.452920143027413e-07,
+      "loss": 1.7993,
+      "step": 3944
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.452522844656337e-07,
+      "loss": 1.7445,
+      "step": 3945
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.452125546285261e-07,
+      "loss": 1.6651,
+      "step": 3946
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.451728247914183e-07,
+      "loss": 1.802,
+      "step": 3947
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.451330949543107e-07,
+      "loss": 1.705,
+      "step": 3948
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.450933651172029e-07,
+      "loss": 1.8387,
+      "step": 3949
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.450536352800953e-07,
+      "loss": 1.7426,
+      "step": 3950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.450139054429876e-07,
+      "loss": 1.8145,
+      "step": 3951
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.4497417560588e-07,
+      "loss": 1.6776,
+      "step": 3952
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.449344457687723e-07,
+      "loss": 1.6737,
+      "step": 3953
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.448947159316646e-07,
+      "loss": 1.596,
+      "step": 3954
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.44854986094557e-07,
+      "loss": 1.7076,
+      "step": 3955
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.448152562574493e-07,
+      "loss": 1.8254,
+      "step": 3956
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.447755264203417e-07,
+      "loss": 1.8658,
+      "step": 3957
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.44735796583234e-07,
+      "loss": 1.8923,
+      "step": 3958
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.446960667461263e-07,
+      "loss": 1.7117,
+      "step": 3959
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.446563369090187e-07,
+      "loss": 1.6851,
+      "step": 3960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.44616607071911e-07,
+      "loss": 1.9105,
+      "step": 3961
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.445768772348033e-07,
+      "loss": 1.9622,
+      "step": 3962
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.445371473976956e-07,
+      "loss": 1.8237,
+      "step": 3963
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.44497417560588e-07,
+      "loss": 1.917,
+      "step": 3964
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.444576877234802e-07,
+      "loss": 1.7779,
+      "step": 3965
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.444179578863726e-07,
+      "loss": 1.7696,
+      "step": 3966
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.443782280492649e-07,
+      "loss": 2.0449,
+      "step": 3967
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.443384982121573e-07,
+      "loss": 1.6972,
+      "step": 3968
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.442987683750497e-07,
+      "loss": 1.8005,
+      "step": 3969
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.44259038537942e-07,
+      "loss": 1.6081,
+      "step": 3970
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.442193087008343e-07,
+      "loss": 1.8573,
+      "step": 3971
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.441795788637266e-07,
+      "loss": 1.8407,
+      "step": 3972
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.44139849026619e-07,
+      "loss": 1.6093,
+      "step": 3973
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.441001191895113e-07,
+      "loss": 2.0483,
+      "step": 3974
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.440603893524036e-07,
+      "loss": 1.6503,
+      "step": 3975
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.44020659515296e-07,
+      "loss": 1.6775,
+      "step": 3976
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.439809296781882e-07,
+      "loss": 1.7896,
+      "step": 3977
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.439411998410806e-07,
+      "loss": 1.8429,
+      "step": 3978
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.439014700039729e-07,
+      "loss": 1.8806,
+      "step": 3979
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.438617401668653e-07,
+      "loss": 1.7961,
+      "step": 3980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.438220103297576e-07,
+      "loss": 1.8967,
+      "step": 3981
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.4378228049265e-07,
+      "loss": 1.8328,
+      "step": 3982
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.437425506555422e-07,
+      "loss": 1.9956,
+      "step": 3983
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.437028208184346e-07,
+      "loss": 1.8305,
+      "step": 3984
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.43663090981327e-07,
+      "loss": 1.7999,
+      "step": 3985
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.436233611442193e-07,
+      "loss": 2.1956,
+      "step": 3986
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.435836313071117e-07,
+      "loss": 1.929,
+      "step": 3987
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.43543901470004e-07,
+      "loss": 2.0805,
+      "step": 3988
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.435041716328962e-07,
+      "loss": 1.8051,
+      "step": 3989
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.434644417957885e-07,
+      "loss": 1.8252,
+      "step": 3990
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.434247119586809e-07,
+      "loss": 1.8472,
+      "step": 3991
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.433849821215733e-07,
+      "loss": 1.981,
+      "step": 3992
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.433452522844656e-07,
+      "loss": 1.788,
+      "step": 3993
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.43305522447358e-07,
+      "loss": 2.1013,
+      "step": 3994
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.432657926102502e-07,
+      "loss": 1.7827,
+      "step": 3995
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.432260627731426e-07,
+      "loss": 1.6451,
+      "step": 3996
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.431863329360349e-07,
+      "loss": 1.6681,
+      "step": 3997
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.431466030989273e-07,
+      "loss": 1.5235,
+      "step": 3998
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.431068732618197e-07,
+      "loss": 1.7592,
+      "step": 3999
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.43067143424712e-07,
+      "loss": 1.6112,
+      "step": 4000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.430274135876042e-07,
+      "loss": 1.8097,
+      "step": 4001
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.429876837504965e-07,
+      "loss": 2.106,
+      "step": 4002
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.429479539133889e-07,
+      "loss": 1.7452,
+      "step": 4003
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.429082240762812e-07,
+      "loss": 1.8361,
+      "step": 4004
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.428684942391736e-07,
+      "loss": 1.9009,
+      "step": 4005
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.428287644020659e-07,
+      "loss": 1.6548,
+      "step": 4006
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.427890345649582e-07,
+      "loss": 1.8139,
+      "step": 4007
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.427493047278506e-07,
+      "loss": 1.7139,
+      "step": 4008
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.427095748907429e-07,
+      "loss": 1.8363,
+      "step": 4009
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.426698450536353e-07,
+      "loss": 1.9401,
+      "step": 4010
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.426301152165276e-07,
+      "loss": 1.7834,
+      "step": 4011
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.4259038537942e-07,
+      "loss": 1.6903,
+      "step": 4012
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.425506555423122e-07,
+      "loss": 1.9678,
+      "step": 4013
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.425109257052045e-07,
+      "loss": 1.7301,
+      "step": 4014
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.424711958680968e-07,
+      "loss": 1.813,
+      "step": 4015
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.424314660309892e-07,
+      "loss": 2.1145,
+      "step": 4016
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.423917361938816e-07,
+      "loss": 1.9275,
+      "step": 4017
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.423520063567739e-07,
+      "loss": 2.1734,
+      "step": 4018
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.423122765196662e-07,
+      "loss": 1.4987,
+      "step": 4019
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.422725466825585e-07,
+      "loss": 1.8364,
+      "step": 4020
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.422328168454509e-07,
+      "loss": 1.6411,
+      "step": 4021
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.421930870083432e-07,
+      "loss": 1.7428,
+      "step": 4022
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.421533571712356e-07,
+      "loss": 1.3675,
+      "step": 4023
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.42113627334128e-07,
+      "loss": 2.1243,
+      "step": 4024
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.420738974970202e-07,
+      "loss": 1.9001,
+      "step": 4025
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.420341676599126e-07,
+      "loss": 1.9319,
+      "step": 4026
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.419944378228049e-07,
+      "loss": 1.6866,
+      "step": 4027
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.419547079856972e-07,
+      "loss": 1.705,
+      "step": 4028
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.419149781485895e-07,
+      "loss": 1.6821,
+      "step": 4029
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.418752483114819e-07,
+      "loss": 2.044,
+      "step": 4030
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.418355184743742e-07,
+      "loss": 1.6612,
+      "step": 4031
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.417957886372665e-07,
+      "loss": 1.8965,
+      "step": 4032
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.417560588001589e-07,
+      "loss": 1.8435,
+      "step": 4033
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.417163289630512e-07,
+      "loss": 1.6446,
+      "step": 4034
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.416765991259436e-07,
+      "loss": 1.6793,
+      "step": 4035
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.416368692888359e-07,
+      "loss": 2.1135,
+      "step": 4036
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.415971394517282e-07,
+      "loss": 1.8305,
+      "step": 4037
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.415574096146206e-07,
+      "loss": 1.8039,
+      "step": 4038
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.415176797775129e-07,
+      "loss": 1.9345,
+      "step": 4039
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.414779499404053e-07,
+      "loss": 1.6379,
+      "step": 4040
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.414382201032975e-07,
+      "loss": 1.8853,
+      "step": 4041
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.413984902661899e-07,
+      "loss": 1.707,
+      "step": 4042
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.413587604290821e-07,
+      "loss": 1.5397,
+      "step": 4043
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.413190305919745e-07,
+      "loss": 1.8307,
+      "step": 4044
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.412793007548668e-07,
+      "loss": 1.9106,
+      "step": 4045
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.412395709177592e-07,
+      "loss": 1.5344,
+      "step": 4046
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.411998410806516e-07,
+      "loss": 1.9554,
+      "step": 4047
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.411601112435439e-07,
+      "loss": 1.9164,
+      "step": 4048
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.411203814064362e-07,
+      "loss": 1.5232,
+      "step": 4049
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.410806515693285e-07,
+      "loss": 1.7992,
+      "step": 4050
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.410409217322209e-07,
+      "loss": 1.6238,
+      "step": 4051
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.410011918951132e-07,
+      "loss": 1.7372,
+      "step": 4052
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.409614620580056e-07,
+      "loss": 1.8478,
+      "step": 4053
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.40921732220898e-07,
+      "loss": 1.6826,
+      "step": 4054
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.408820023837901e-07,
+      "loss": 1.9651,
+      "step": 4055
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.408422725466825e-07,
+      "loss": 1.5151,
+      "step": 4056
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.408025427095748e-07,
+      "loss": 1.9442,
+      "step": 4057
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.407628128724672e-07,
+      "loss": 1.7945,
+      "step": 4058
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.407230830353595e-07,
+      "loss": 1.5328,
+      "step": 4059
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.406833531982519e-07,
+      "loss": 1.6175,
+      "step": 4060
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.406436233611441e-07,
+      "loss": 1.7113,
+      "step": 4061
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.406038935240365e-07,
+      "loss": 1.9063,
+      "step": 4062
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.405641636869289e-07,
+      "loss": 1.854,
+      "step": 4063
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.405244338498212e-07,
+      "loss": 1.9029,
+      "step": 4064
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.404847040127136e-07,
+      "loss": 1.7813,
+      "step": 4065
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.404449741756059e-07,
+      "loss": 1.819,
+      "step": 4066
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.404052443384981e-07,
+      "loss": 1.7509,
+      "step": 4067
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.403655145013904e-07,
+      "loss": 2.0836,
+      "step": 4068
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.403257846642828e-07,
+      "loss": 1.7594,
+      "step": 4069
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.402860548271752e-07,
+      "loss": 1.5132,
+      "step": 4070
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.402463249900675e-07,
+      "loss": 1.6175,
+      "step": 4071
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.402065951529599e-07,
+      "loss": 1.9083,
+      "step": 4072
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.401668653158521e-07,
+      "loss": 1.7837,
+      "step": 4073
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.401271354787445e-07,
+      "loss": 1.6591,
+      "step": 4074
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.400874056416368e-07,
+      "loss": 2.1657,
+      "step": 4075
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.400476758045292e-07,
+      "loss": 1.9647,
+      "step": 4076
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.400079459674216e-07,
+      "loss": 1.7079,
+      "step": 4077
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.399682161303139e-07,
+      "loss": 1.6143,
+      "step": 4078
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.399284862932062e-07,
+      "loss": 1.8883,
+      "step": 4079
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.398887564560984e-07,
+      "loss": 1.6518,
+      "step": 4080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.398490266189908e-07,
+      "loss": 1.7594,
+      "step": 4081
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.398092967818831e-07,
+      "loss": 1.8211,
+      "step": 4082
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.397695669447755e-07,
+      "loss": 1.9567,
+      "step": 4083
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.397298371076678e-07,
+      "loss": 1.4755,
+      "step": 4084
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.396901072705601e-07,
+      "loss": 1.6715,
+      "step": 4085
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.396503774334525e-07,
+      "loss": 1.5647,
+      "step": 4086
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.396106475963448e-07,
+      "loss": 1.9052,
+      "step": 4087
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.395709177592372e-07,
+      "loss": 1.6574,
+      "step": 4088
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.395311879221295e-07,
+      "loss": 1.7833,
+      "step": 4089
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.394914580850219e-07,
+      "loss": 2.033,
+      "step": 4090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.394517282479141e-07,
+      "loss": 1.757,
+      "step": 4091
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.394119984108065e-07,
+      "loss": 1.8531,
+      "step": 4092
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.393722685736989e-07,
+      "loss": 1.6533,
+      "step": 4093
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.393325387365911e-07,
+      "loss": 1.6932,
+      "step": 4094
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.392928088994835e-07,
+      "loss": 1.7005,
+      "step": 4095
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.392530790623758e-07,
+      "loss": 1.6727,
+      "step": 4096
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.392133492252681e-07,
+      "loss": 1.7979,
+      "step": 4097
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.391736193881604e-07,
+      "loss": 1.882,
+      "step": 4098
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.391338895510528e-07,
+      "loss": 1.6209,
+      "step": 4099
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.390941597139451e-07,
+      "loss": 1.8032,
+      "step": 4100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.390544298768375e-07,
+      "loss": 1.7279,
+      "step": 4101
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.390147000397299e-07,
+      "loss": 1.6859,
+      "step": 4102
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.389749702026221e-07,
+      "loss": 1.8312,
+      "step": 4103
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.389352403655145e-07,
+      "loss": 1.8085,
+      "step": 4104
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.388955105284068e-07,
+      "loss": 1.4311,
+      "step": 4105
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.388557806912992e-07,
+      "loss": 1.9779,
+      "step": 4106
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.388160508541914e-07,
+      "loss": 1.4903,
+      "step": 4107
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.387763210170838e-07,
+      "loss": 1.7502,
+      "step": 4108
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.38736591179976e-07,
+      "loss": 1.7054,
+      "step": 4109
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.386968613428684e-07,
+      "loss": 1.8519,
+      "step": 4110
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.386571315057608e-07,
+      "loss": 1.8235,
+      "step": 4111
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.386174016686531e-07,
+      "loss": 1.8928,
+      "step": 4112
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.385776718315455e-07,
+      "loss": 1.8155,
+      "step": 4113
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.385379419944378e-07,
+      "loss": 1.8572,
+      "step": 4114
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.384982121573301e-07,
+      "loss": 1.6067,
+      "step": 4115
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.384584823202225e-07,
+      "loss": 1.5969,
+      "step": 4116
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.384187524831148e-07,
+      "loss": 1.8322,
+      "step": 4117
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.383790226460072e-07,
+      "loss": 1.849,
+      "step": 4118
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.383392928088995e-07,
+      "loss": 2.1398,
+      "step": 4119
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.382995629717918e-07,
+      "loss": 1.7586,
+      "step": 4120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.38259833134684e-07,
+      "loss": 1.7747,
+      "step": 4121
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.382201032975764e-07,
+      "loss": 1.6191,
+      "step": 4122
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.381803734604687e-07,
+      "loss": 1.7452,
+      "step": 4123
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.381406436233611e-07,
+      "loss": 1.751,
+      "step": 4124
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.381009137862535e-07,
+      "loss": 1.8514,
+      "step": 4125
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.380611839491458e-07,
+      "loss": 2.0296,
+      "step": 4126
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.380214541120381e-07,
+      "loss": 1.9496,
+      "step": 4127
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.379817242749304e-07,
+      "loss": 1.7795,
+      "step": 4128
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.379419944378228e-07,
+      "loss": 1.8926,
+      "step": 4129
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.379022646007151e-07,
+      "loss": 1.8353,
+      "step": 4130
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.378625347636075e-07,
+      "loss": 1.9877,
+      "step": 4131
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.378228049264999e-07,
+      "loss": 1.5994,
+      "step": 4132
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.37783075089392e-07,
+      "loss": 1.8595,
+      "step": 4133
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.377433452522844e-07,
+      "loss": 1.9146,
+      "step": 4134
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.377036154151767e-07,
+      "loss": 1.5503,
+      "step": 4135
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.376638855780691e-07,
+      "loss": 2.061,
+      "step": 4136
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.376241557409614e-07,
+      "loss": 1.7394,
+      "step": 4137
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.375844259038538e-07,
+      "loss": 1.7304,
+      "step": 4138
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.37544696066746e-07,
+      "loss": 1.8037,
+      "step": 4139
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.375049662296384e-07,
+      "loss": 1.4976,
+      "step": 4140
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.374652363925308e-07,
+      "loss": 1.4798,
+      "step": 4141
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.374255065554231e-07,
+      "loss": 1.6327,
+      "step": 4142
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.373857767183155e-07,
+      "loss": 1.8873,
+      "step": 4143
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.373460468812078e-07,
+      "loss": 1.7703,
+      "step": 4144
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.373063170441001e-07,
+      "loss": 1.6858,
+      "step": 4145
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.372665872069923e-07,
+      "loss": 1.7572,
+      "step": 4146
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.372268573698847e-07,
+      "loss": 1.7331,
+      "step": 4147
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.371871275327771e-07,
+      "loss": 1.5288,
+      "step": 4148
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.371473976956694e-07,
+      "loss": 1.7632,
+      "step": 4149
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.371076678585618e-07,
+      "loss": 1.5793,
+      "step": 4150
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.37067938021454e-07,
+      "loss": 1.7844,
+      "step": 4151
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.370282081843464e-07,
+      "loss": 1.8283,
+      "step": 4152
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.369884783472387e-07,
+      "loss": 1.5833,
+      "step": 4153
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.369487485101311e-07,
+      "loss": 1.6196,
+      "step": 4154
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.369090186730235e-07,
+      "loss": 1.5725,
+      "step": 4155
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.368692888359158e-07,
+      "loss": 1.5506,
+      "step": 4156
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.368295589988081e-07,
+      "loss": 1.8621,
+      "step": 4157
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.367898291617004e-07,
+      "loss": 1.7503,
+      "step": 4158
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.367500993245928e-07,
+      "loss": 1.7977,
+      "step": 4159
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.36710369487485e-07,
+      "loss": 1.8951,
+      "step": 4160
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.366706396503774e-07,
+      "loss": 1.7245,
+      "step": 4161
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.366309098132697e-07,
+      "loss": 1.661,
+      "step": 4162
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.36591179976162e-07,
+      "loss": 1.8265,
+      "step": 4163
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.365514501390544e-07,
+      "loss": 1.6292,
+      "step": 4164
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.365117203019467e-07,
+      "loss": 1.615,
+      "step": 4165
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.364719904648391e-07,
+      "loss": 2.1791,
+      "step": 4166
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.364322606277314e-07,
+      "loss": 1.5529,
+      "step": 4167
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.363925307906238e-07,
+      "loss": 2.0029,
+      "step": 4168
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.36352800953516e-07,
+      "loss": 1.7933,
+      "step": 4169
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.363130711164084e-07,
+      "loss": 1.5413,
+      "step": 4170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.362733412793008e-07,
+      "loss": 1.7042,
+      "step": 4171
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.362336114421931e-07,
+      "loss": 1.811,
+      "step": 4172
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.361938816050854e-07,
+      "loss": 1.8121,
+      "step": 4173
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.361541517679777e-07,
+      "loss": 1.8592,
+      "step": 4174
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.3611442193087e-07,
+      "loss": 1.6655,
+      "step": 4175
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.360746920937623e-07,
+      "loss": 1.7553,
+      "step": 4176
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.360349622566547e-07,
+      "loss": 2.1086,
+      "step": 4177
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.35995232419547e-07,
+      "loss": 1.6021,
+      "step": 4178
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.359555025824394e-07,
+      "loss": 1.871,
+      "step": 4179
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.359157727453318e-07,
+      "loss": 1.7691,
+      "step": 4180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.35876042908224e-07,
+      "loss": 2.1273,
+      "step": 4181
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.358363130711164e-07,
+      "loss": 1.7564,
+      "step": 4182
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.357965832340087e-07,
+      "loss": 1.8663,
+      "step": 4183
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.357568533969011e-07,
+      "loss": 1.8695,
+      "step": 4184
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.357171235597934e-07,
+      "loss": 1.6618,
+      "step": 4185
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.356773937226857e-07,
+      "loss": 1.4847,
+      "step": 4186
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.35637663885578e-07,
+      "loss": 1.8061,
+      "step": 4187
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.355979340484703e-07,
+      "loss": 1.7319,
+      "step": 4188
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.355582042113627e-07,
+      "loss": 1.6127,
+      "step": 4189
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.35518474374255e-07,
+      "loss": 2.0192,
+      "step": 4190
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.354787445371474e-07,
+      "loss": 1.9313,
+      "step": 4191
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.354390147000397e-07,
+      "loss": 1.7362,
+      "step": 4192
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.35399284862932e-07,
+      "loss": 1.6556,
+      "step": 4193
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.353595550258244e-07,
+      "loss": 1.6726,
+      "step": 4194
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.353198251887167e-07,
+      "loss": 1.8367,
+      "step": 4195
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.352800953516091e-07,
+      "loss": 1.762,
+      "step": 4196
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.352403655145014e-07,
+      "loss": 1.9053,
+      "step": 4197
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.352006356773938e-07,
+      "loss": 1.8999,
+      "step": 4198
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.351609058402859e-07,
+      "loss": 1.6212,
+      "step": 4199
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.351211760031783e-07,
+      "loss": 1.8443,
+      "step": 4200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.350814461660706e-07,
+      "loss": 1.7738,
+      "step": 4201
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.35041716328963e-07,
+      "loss": 1.8336,
+      "step": 4202
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.350019864918554e-07,
+      "loss": 1.9988,
+      "step": 4203
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.349622566547477e-07,
+      "loss": 1.451,
+      "step": 4204
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.3492252681764e-07,
+      "loss": 1.7339,
+      "step": 4205
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.348827969805323e-07,
+      "loss": 1.9293,
+      "step": 4206
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.348430671434247e-07,
+      "loss": 2.1458,
+      "step": 4207
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.34803337306317e-07,
+      "loss": 1.5677,
+      "step": 4208
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.347636074692094e-07,
+      "loss": 1.6193,
+      "step": 4209
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.347238776321018e-07,
+      "loss": 1.8774,
+      "step": 4210
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.34684147794994e-07,
+      "loss": 2.0356,
+      "step": 4211
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.346444179578863e-07,
+      "loss": 1.6965,
+      "step": 4212
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.346046881207786e-07,
+      "loss": 2.14,
+      "step": 4213
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.34564958283671e-07,
+      "loss": 1.8361,
+      "step": 4214
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.345252284465633e-07,
+      "loss": 1.8424,
+      "step": 4215
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.344854986094557e-07,
+      "loss": 1.779,
+      "step": 4216
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.344457687723479e-07,
+      "loss": 1.6991,
+      "step": 4217
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.344060389352403e-07,
+      "loss": 1.6386,
+      "step": 4218
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.343663090981327e-07,
+      "loss": 1.935,
+      "step": 4219
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.34326579261025e-07,
+      "loss": 2.3385,
+      "step": 4220
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.342868494239174e-07,
+      "loss": 1.7625,
+      "step": 4221
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.342471195868097e-07,
+      "loss": 1.7526,
+      "step": 4222
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.34207389749702e-07,
+      "loss": 1.8436,
+      "step": 4223
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.341676599125943e-07,
+      "loss": 1.5826,
+      "step": 4224
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.341279300754867e-07,
+      "loss": 1.6825,
+      "step": 4225
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.34088200238379e-07,
+      "loss": 1.7628,
+      "step": 4226
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.340484704012713e-07,
+      "loss": 1.6669,
+      "step": 4227
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.340087405641637e-07,
+      "loss": 1.7921,
+      "step": 4228
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.339690107270559e-07,
+      "loss": 1.7333,
+      "step": 4229
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.339292808899483e-07,
+      "loss": 1.9079,
+      "step": 4230
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.338895510528406e-07,
+      "loss": 1.6282,
+      "step": 4231
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.33849821215733e-07,
+      "loss": 1.8789,
+      "step": 4232
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.338100913786254e-07,
+      "loss": 1.6979,
+      "step": 4233
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.337703615415177e-07,
+      "loss": 1.9276,
+      "step": 4234
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.3373063170441e-07,
+      "loss": 2.176,
+      "step": 4235
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.336909018673023e-07,
+      "loss": 1.6476,
+      "step": 4236
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.336511720301947e-07,
+      "loss": 1.6335,
+      "step": 4237
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.33611442193087e-07,
+      "loss": 1.8397,
+      "step": 4238
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.335717123559793e-07,
+      "loss": 2.0043,
+      "step": 4239
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.335319825188716e-07,
+      "loss": 1.6677,
+      "step": 4240
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.334922526817639e-07,
+      "loss": 1.861,
+      "step": 4241
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.334525228446563e-07,
+      "loss": 1.8924,
+      "step": 4242
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.334127930075486e-07,
+      "loss": 1.8497,
+      "step": 4243
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.33373063170441e-07,
+      "loss": 1.9292,
+      "step": 4244
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.8257,
+      "step": 4245
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.332936034962256e-07,
+      "loss": 2.1096,
+      "step": 4246
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.332538736591179e-07,
+      "loss": 2.1235,
+      "step": 4247
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.332141438220103e-07,
+      "loss": 1.845,
+      "step": 4248
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.331744139849027e-07,
+      "loss": 2.074,
+      "step": 4249
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.33134684147795e-07,
+      "loss": 1.7327,
+      "step": 4250
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.330949543106874e-07,
+      "loss": 1.769,
+      "step": 4251
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.330552244735795e-07,
+      "loss": 1.8501,
+      "step": 4252
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.330154946364719e-07,
+      "loss": 1.8579,
+      "step": 4253
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.329757647993642e-07,
+      "loss": 1.613,
+      "step": 4254
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.329360349622566e-07,
+      "loss": 1.6408,
+      "step": 4255
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.32896305125149e-07,
+      "loss": 2.2469,
+      "step": 4256
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.328565752880413e-07,
+      "loss": 2.0044,
+      "step": 4257
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.328168454509336e-07,
+      "loss": 1.9539,
+      "step": 4258
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.327771156138259e-07,
+      "loss": 1.9097,
+      "step": 4259
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.327373857767183e-07,
+      "loss": 1.6494,
+      "step": 4260
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.326976559396106e-07,
+      "loss": 1.5077,
+      "step": 4261
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.32657926102503e-07,
+      "loss": 1.6729,
+      "step": 4262
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.326181962653953e-07,
+      "loss": 2.1104,
+      "step": 4263
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.325784664282876e-07,
+      "loss": 1.9534,
+      "step": 4264
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.325387365911799e-07,
+      "loss": 1.6929,
+      "step": 4265
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.324990067540722e-07,
+      "loss": 2.1984,
+      "step": 4266
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.324592769169646e-07,
+      "loss": 1.8084,
+      "step": 4267
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.324195470798569e-07,
+      "loss": 1.9004,
+      "step": 4268
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.323798172427493e-07,
+      "loss": 1.9195,
+      "step": 4269
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.323400874056415e-07,
+      "loss": 1.573,
+      "step": 4270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.323003575685339e-07,
+      "loss": 1.8323,
+      "step": 4271
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.322606277314263e-07,
+      "loss": 1.6205,
+      "step": 4272
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.322208978943186e-07,
+      "loss": 1.6351,
+      "step": 4273
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.32181168057211e-07,
+      "loss": 1.6819,
+      "step": 4274
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.321414382201033e-07,
+      "loss": 1.8584,
+      "step": 4275
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.321017083829956e-07,
+      "loss": 1.7595,
+      "step": 4276
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.320619785458879e-07,
+      "loss": 1.9458,
+      "step": 4277
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.320222487087802e-07,
+      "loss": 1.7946,
+      "step": 4278
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.319825188716725e-07,
+      "loss": 1.9217,
+      "step": 4279
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.319427890345649e-07,
+      "loss": 1.7194,
+      "step": 4280
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.319030591974573e-07,
+      "loss": 1.5856,
+      "step": 4281
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.318633293603495e-07,
+      "loss": 1.6947,
+      "step": 4282
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.318235995232419e-07,
+      "loss": 1.8152,
+      "step": 4283
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.317838696861342e-07,
+      "loss": 1.5074,
+      "step": 4284
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.317441398490266e-07,
+      "loss": 1.8976,
+      "step": 4285
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.317044100119189e-07,
+      "loss": 1.8819,
+      "step": 4286
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.316646801748113e-07,
+      "loss": 2.1017,
+      "step": 4287
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.316249503377036e-07,
+      "loss": 1.8218,
+      "step": 4288
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.315852205005959e-07,
+      "loss": 1.5708,
+      "step": 4289
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.315454906634883e-07,
+      "loss": 1.8538,
+      "step": 4290
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.315057608263806e-07,
+      "loss": 1.9772,
+      "step": 4291
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.314660309892729e-07,
+      "loss": 1.8315,
+      "step": 4292
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.314263011521652e-07,
+      "loss": 1.6863,
+      "step": 4293
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.313865713150575e-07,
+      "loss": 1.6295,
+      "step": 4294
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.313468414779498e-07,
+      "loss": 1.9708,
+      "step": 4295
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.313071116408422e-07,
+      "loss": 1.6012,
+      "step": 4296
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.312673818037346e-07,
+      "loss": 1.792,
+      "step": 4297
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.312276519666269e-07,
+      "loss": 1.8798,
+      "step": 4298
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.311879221295193e-07,
+      "loss": 1.7987,
+      "step": 4299
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.311481922924115e-07,
+      "loss": 1.6314,
+      "step": 4300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.311084624553039e-07,
+      "loss": 1.7355,
+      "step": 4301
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.310687326181962e-07,
+      "loss": 1.6861,
+      "step": 4302
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.310290027810886e-07,
+      "loss": 1.5265,
+      "step": 4303
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.30989272943981e-07,
+      "loss": 1.7823,
+      "step": 4304
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.309495431068732e-07,
+      "loss": 1.7691,
+      "step": 4305
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.309098132697655e-07,
+      "loss": 1.5542,
+      "step": 4306
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.308700834326578e-07,
+      "loss": 1.8584,
+      "step": 4307
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.308303535955502e-07,
+      "loss": 1.555,
+      "step": 4308
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.307906237584425e-07,
+      "loss": 1.7567,
+      "step": 4309
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.307508939213349e-07,
+      "loss": 1.64,
+      "step": 4310
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.307111640842273e-07,
+      "loss": 1.8052,
+      "step": 4311
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.306714342471195e-07,
+      "loss": 2.431,
+      "step": 4312
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.306317044100119e-07,
+      "loss": 1.8837,
+      "step": 4313
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.305919745729042e-07,
+      "loss": 1.5635,
+      "step": 4314
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.305522447357966e-07,
+      "loss": 2.0435,
+      "step": 4315
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.305125148986889e-07,
+      "loss": 1.7044,
+      "step": 4316
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.304727850615813e-07,
+      "loss": 1.8122,
+      "step": 4317
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.304330552244734e-07,
+      "loss": 1.8593,
+      "step": 4318
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.303933253873658e-07,
+      "loss": 1.65,
+      "step": 4319
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.303535955502582e-07,
+      "loss": 1.6928,
+      "step": 4320
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.303138657131505e-07,
+      "loss": 1.5457,
+      "step": 4321
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.302741358760429e-07,
+      "loss": 1.8328,
+      "step": 4322
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.302344060389352e-07,
+      "loss": 1.7125,
+      "step": 4323
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.301946762018275e-07,
+      "loss": 1.6906,
+      "step": 4324
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.301549463647198e-07,
+      "loss": 1.6171,
+      "step": 4325
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.301152165276122e-07,
+      "loss": 2.0497,
+      "step": 4326
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.300754866905046e-07,
+      "loss": 1.6359,
+      "step": 4327
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.300357568533969e-07,
+      "loss": 1.928,
+      "step": 4328
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.299960270162893e-07,
+      "loss": 1.6801,
+      "step": 4329
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.299562971791815e-07,
+      "loss": 1.8043,
+      "step": 4330
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.299165673420738e-07,
+      "loss": 2.0094,
+      "step": 4331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.298768375049661e-07,
+      "loss": 1.8842,
+      "step": 4332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.298371076678585e-07,
+      "loss": 1.9561,
+      "step": 4333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.297973778307508e-07,
+      "loss": 1.8144,
+      "step": 4334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.297576479936432e-07,
+      "loss": 1.5902,
+      "step": 4335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.297179181565355e-07,
+      "loss": 1.7107,
+      "step": 4336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.296781883194278e-07,
+      "loss": 1.7166,
+      "step": 4337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.296384584823202e-07,
+      "loss": 2.042,
+      "step": 4338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.295987286452125e-07,
+      "loss": 1.6748,
+      "step": 4339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.295589988081049e-07,
+      "loss": 1.5164,
+      "step": 4340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.295192689709972e-07,
+      "loss": 2.0373,
+      "step": 4341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.294795391338895e-07,
+      "loss": 1.8512,
+      "step": 4342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.294398092967819e-07,
+      "loss": 1.5483,
+      "step": 4343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.294000794596741e-07,
+      "loss": 1.8185,
+      "step": 4344
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.293603496225665e-07,
+      "loss": 1.9096,
+      "step": 4345
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.293206197854588e-07,
+      "loss": 1.807,
+      "step": 4346
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.292808899483512e-07,
+      "loss": 1.7495,
+      "step": 4347
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.292411601112434e-07,
+      "loss": 1.574,
+      "step": 4348
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.292014302741358e-07,
+      "loss": 1.7776,
+      "step": 4349
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.291617004370282e-07,
+      "loss": 1.853,
+      "step": 4350
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.291219705999205e-07,
+      "loss": 1.7168,
+      "step": 4351
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.290822407628129e-07,
+      "loss": 1.7669,
+      "step": 4352
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.290425109257052e-07,
+      "loss": 1.8384,
+      "step": 4353
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.290027810885975e-07,
+      "loss": 1.7786,
+      "step": 4354
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.289630512514898e-07,
+      "loss": 1.8363,
+      "step": 4355
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.289233214143822e-07,
+      "loss": 1.9916,
+      "step": 4356
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.288835915772746e-07,
+      "loss": 1.8312,
+      "step": 4357
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.288438617401668e-07,
+      "loss": 1.7466,
+      "step": 4358
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.288041319030592e-07,
+      "loss": 1.5479,
+      "step": 4359
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.287644020659514e-07,
+      "loss": 1.8349,
+      "step": 4360
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.287246722288438e-07,
+      "loss": 1.8256,
+      "step": 4361
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.286849423917361e-07,
+      "loss": 1.9075,
+      "step": 4362
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.286452125546285e-07,
+      "loss": 1.9343,
+      "step": 4363
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.286054827175208e-07,
+      "loss": 2.0321,
+      "step": 4364
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.285657528804132e-07,
+      "loss": 1.7173,
+      "step": 4365
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.285260230433055e-07,
+      "loss": 1.7621,
+      "step": 4366
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.284862932061978e-07,
+      "loss": 1.6737,
+      "step": 4367
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.284465633690902e-07,
+      "loss": 1.819,
+      "step": 4368
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.284068335319825e-07,
+      "loss": 1.5464,
+      "step": 4369
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.283671036948749e-07,
+      "loss": 1.4677,
+      "step": 4370
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.283273738577671e-07,
+      "loss": 1.7891,
+      "step": 4371
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.282876440206594e-07,
+      "loss": 1.7621,
+      "step": 4372
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.282479141835517e-07,
+      "loss": 1.5718,
+      "step": 4373
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.282081843464441e-07,
+      "loss": 1.7525,
+      "step": 4374
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.281684545093365e-07,
+      "loss": 1.9143,
+      "step": 4375
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.281287246722288e-07,
+      "loss": 1.8045,
+      "step": 4376
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.280889948351212e-07,
+      "loss": 1.8123,
+      "step": 4377
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.280492649980134e-07,
+      "loss": 1.5164,
+      "step": 4378
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.280095351609058e-07,
+      "loss": 1.7733,
+      "step": 4379
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.279698053237981e-07,
+      "loss": 1.4102,
+      "step": 4380
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.279300754866905e-07,
+      "loss": 1.9591,
+      "step": 4381
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.278903456495829e-07,
+      "loss": 1.6922,
+      "step": 4382
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.278506158124752e-07,
+      "loss": 1.9774,
+      "step": 4383
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.278108859753674e-07,
+      "loss": 1.9438,
+      "step": 4384
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.277711561382597e-07,
+      "loss": 2.0181,
+      "step": 4385
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.277314263011521e-07,
+      "loss": 1.732,
+      "step": 4386
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.276916964640444e-07,
+      "loss": 1.8066,
+      "step": 4387
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.276519666269368e-07,
+      "loss": 1.6929,
+      "step": 4388
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.276122367898292e-07,
+      "loss": 1.7271,
+      "step": 4389
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.275725069527214e-07,
+      "loss": 1.8349,
+      "step": 4390
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.275327771156138e-07,
+      "loss": 1.5767,
+      "step": 4391
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.274930472785061e-07,
+      "loss": 1.6515,
+      "step": 4392
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.274533174413985e-07,
+      "loss": 1.7232,
+      "step": 4393
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.274135876042908e-07,
+      "loss": 1.8646,
+      "step": 4394
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.273738577671832e-07,
+      "loss": 1.6922,
+      "step": 4395
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.273341279300754e-07,
+      "loss": 1.7654,
+      "step": 4396
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.272943980929677e-07,
+      "loss": 1.5905,
+      "step": 4397
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.272546682558601e-07,
+      "loss": 1.9224,
+      "step": 4398
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.272149384187524e-07,
+      "loss": 2.1214,
+      "step": 4399
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.271752085816448e-07,
+      "loss": 1.8512,
+      "step": 4400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.271354787445371e-07,
+      "loss": 1.7199,
+      "step": 4401
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.270957489074294e-07,
+      "loss": 1.9283,
+      "step": 4402
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.270560190703217e-07,
+      "loss": 1.7218,
+      "step": 4403
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.270162892332141e-07,
+      "loss": 1.4776,
+      "step": 4404
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.269765593961065e-07,
+      "loss": 1.8417,
+      "step": 4405
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.269368295589988e-07,
+      "loss": 1.5531,
+      "step": 4406
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.268970997218912e-07,
+      "loss": 1.9454,
+      "step": 4407
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.268573698847834e-07,
+      "loss": 1.8815,
+      "step": 4408
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.268176400476758e-07,
+      "loss": 1.5293,
+      "step": 4409
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.26777910210568e-07,
+      "loss": 1.8481,
+      "step": 4410
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.267381803734604e-07,
+      "loss": 1.7208,
+      "step": 4411
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.266984505363528e-07,
+      "loss": 1.7815,
+      "step": 4412
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.266587206992451e-07,
+      "loss": 1.8547,
+      "step": 4413
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.266189908621374e-07,
+      "loss": 1.7099,
+      "step": 4414
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.265792610250297e-07,
+      "loss": 1.9852,
+      "step": 4415
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.265395311879221e-07,
+      "loss": 1.3413,
+      "step": 4416
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.264998013508144e-07,
+      "loss": 1.5769,
+      "step": 4417
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.264600715137068e-07,
+      "loss": 1.8351,
+      "step": 4418
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.264203416765991e-07,
+      "loss": 1.6132,
+      "step": 4419
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.263806118394914e-07,
+      "loss": 2.1698,
+      "step": 4420
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.263408820023838e-07,
+      "loss": 2.0282,
+      "step": 4421
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.263011521652761e-07,
+      "loss": 1.6191,
+      "step": 4422
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.262614223281685e-07,
+      "loss": 1.4037,
+      "step": 4423
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.262216924910607e-07,
+      "loss": 1.9491,
+      "step": 4424
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.261819626539531e-07,
+      "loss": 1.8793,
+      "step": 4425
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.261422328168453e-07,
+      "loss": 1.7428,
+      "step": 4426
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.261025029797377e-07,
+      "loss": 1.4003,
+      "step": 4427
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.2606277314263e-07,
+      "loss": 1.8483,
+      "step": 4428
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.260230433055224e-07,
+      "loss": 1.8903,
+      "step": 4429
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.259833134684148e-07,
+      "loss": 1.7469,
+      "step": 4430
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.259435836313071e-07,
+      "loss": 2.0062,
+      "step": 4431
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.259038537941994e-07,
+      "loss": 1.7985,
+      "step": 4432
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.258641239570917e-07,
+      "loss": 1.5006,
+      "step": 4433
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.258243941199841e-07,
+      "loss": 1.7632,
+      "step": 4434
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.257846642828765e-07,
+      "loss": 1.7871,
+      "step": 4435
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.257449344457688e-07,
+      "loss": 1.871,
+      "step": 4436
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.257052046086611e-07,
+      "loss": 1.7068,
+      "step": 4437
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.256654747715533e-07,
+      "loss": 1.752,
+      "step": 4438
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.256257449344457e-07,
+      "loss": 1.6813,
+      "step": 4439
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.25586015097338e-07,
+      "loss": 1.9632,
+      "step": 4440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.255462852602304e-07,
+      "loss": 1.6729,
+      "step": 4441
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.255065554231227e-07,
+      "loss": 1.9702,
+      "step": 4442
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.254668255860151e-07,
+      "loss": 1.8215,
+      "step": 4443
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.254270957489074e-07,
+      "loss": 1.4442,
+      "step": 4444
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.253873659117997e-07,
+      "loss": 1.9337,
+      "step": 4445
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.253476360746921e-07,
+      "loss": 2.1255,
+      "step": 4446
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.253079062375844e-07,
+      "loss": 1.8157,
+      "step": 4447
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.252681764004768e-07,
+      "loss": 1.8445,
+      "step": 4448
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.252284465633691e-07,
+      "loss": 2.106,
+      "step": 4449
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.251887167262613e-07,
+      "loss": 1.5741,
+      "step": 4450
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.251489868891536e-07,
+      "loss": 2.0961,
+      "step": 4451
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.25109257052046e-07,
+      "loss": 1.7,
+      "step": 4452
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.250695272149384e-07,
+      "loss": 2.0062,
+      "step": 4453
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.250297973778307e-07,
+      "loss": 1.8217,
+      "step": 4454
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.249900675407231e-07,
+      "loss": 1.7052,
+      "step": 4455
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.249503377036153e-07,
+      "loss": 1.8829,
+      "step": 4456
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.249106078665077e-07,
+      "loss": 1.7827,
+      "step": 4457
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.248708780294e-07,
+      "loss": 1.506,
+      "step": 4458
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.248311481922924e-07,
+      "loss": 1.7337,
+      "step": 4459
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.247914183551848e-07,
+      "loss": 1.967,
+      "step": 4460
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.247516885180771e-07,
+      "loss": 1.8558,
+      "step": 4461
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.247119586809694e-07,
+      "loss": 1.6465,
+      "step": 4462
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.246722288438616e-07,
+      "loss": 1.9346,
+      "step": 4463
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.24632499006754e-07,
+      "loss": 2.0655,
+      "step": 4464
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.245927691696463e-07,
+      "loss": 1.7404,
+      "step": 4465
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.245530393325387e-07,
+      "loss": 1.7952,
+      "step": 4466
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.245133094954311e-07,
+      "loss": 1.8293,
+      "step": 4467
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.244735796583233e-07,
+      "loss": 1.6898,
+      "step": 4468
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.244338498212157e-07,
+      "loss": 1.6738,
+      "step": 4469
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.24394119984108e-07,
+      "loss": 1.8279,
+      "step": 4470
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.243543901470004e-07,
+      "loss": 1.6249,
+      "step": 4471
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.243146603098927e-07,
+      "loss": 1.6918,
+      "step": 4472
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.242749304727851e-07,
+      "loss": 1.732,
+      "step": 4473
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.242352006356773e-07,
+      "loss": 2.0094,
+      "step": 4474
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.241954707985697e-07,
+      "loss": 1.9505,
+      "step": 4475
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.24155740961462e-07,
+      "loss": 1.8935,
+      "step": 4476
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.241160111243543e-07,
+      "loss": 1.7948,
+      "step": 4477
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.240762812872467e-07,
+      "loss": 1.7049,
+      "step": 4478
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.24036551450139e-07,
+      "loss": 1.8004,
+      "step": 4479
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.239968216130313e-07,
+      "loss": 1.7554,
+      "step": 4480
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.239570917759236e-07,
+      "loss": 1.7331,
+      "step": 4481
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.23917361938816e-07,
+      "loss": 1.8455,
+      "step": 4482
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.238776321017084e-07,
+      "loss": 1.886,
+      "step": 4483
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.238379022646007e-07,
+      "loss": 1.7823,
+      "step": 4484
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.237981724274931e-07,
+      "loss": 1.7155,
+      "step": 4485
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.237584425903853e-07,
+      "loss": 2.0443,
+      "step": 4486
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.237187127532777e-07,
+      "loss": 1.5648,
+      "step": 4487
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.2367898291617e-07,
+      "loss": 1.8996,
+      "step": 4488
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.236392530790624e-07,
+      "loss": 2.0174,
+      "step": 4489
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.235995232419547e-07,
+      "loss": 1.3476,
+      "step": 4490
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.23559793404847e-07,
+      "loss": 1.6077,
+      "step": 4491
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.235200635677393e-07,
+      "loss": 2.1602,
+      "step": 4492
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.234803337306316e-07,
+      "loss": 1.8603,
+      "step": 4493
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.23440603893524e-07,
+      "loss": 1.6322,
+      "step": 4494
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.234008740564163e-07,
+      "loss": 1.4173,
+      "step": 4495
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.233611442193087e-07,
+      "loss": 1.8042,
+      "step": 4496
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.23321414382201e-07,
+      "loss": 1.6712,
+      "step": 4497
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.232816845450933e-07,
+      "loss": 1.7415,
+      "step": 4498
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.232419547079857e-07,
+      "loss": 1.8985,
+      "step": 4499
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.23202224870878e-07,
+      "loss": 1.5657,
+      "step": 4500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.231624950337704e-07,
+      "loss": 2.0834,
+      "step": 4501
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.231227651966627e-07,
+      "loss": 1.7729,
+      "step": 4502
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.23083035359555e-07,
+      "loss": 1.5893,
+      "step": 4503
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.230433055224472e-07,
+      "loss": 1.9559,
+      "step": 4504
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.230035756853396e-07,
+      "loss": 1.9105,
+      "step": 4505
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.22963845848232e-07,
+      "loss": 1.8912,
+      "step": 4506
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.229241160111243e-07,
+      "loss": 1.871,
+      "step": 4507
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.228843861740167e-07,
+      "loss": 1.9593,
+      "step": 4508
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.22844656336909e-07,
+      "loss": 1.8193,
+      "step": 4509
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.228049264998013e-07,
+      "loss": 2.0525,
+      "step": 4510
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.227651966626936e-07,
+      "loss": 1.6685,
+      "step": 4511
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.22725466825586e-07,
+      "loss": 1.6617,
+      "step": 4512
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.226857369884784e-07,
+      "loss": 1.5999,
+      "step": 4513
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.226460071513707e-07,
+      "loss": 1.8562,
+      "step": 4514
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.226062773142631e-07,
+      "loss": 1.6827,
+      "step": 4515
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.225665474771552e-07,
+      "loss": 1.6603,
+      "step": 4516
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.225268176400476e-07,
+      "loss": 1.7084,
+      "step": 4517
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.224870878029399e-07,
+      "loss": 1.7259,
+      "step": 4518
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.224473579658323e-07,
+      "loss": 1.4096,
+      "step": 4519
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.224076281287246e-07,
+      "loss": 1.8451,
+      "step": 4520
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.22367898291617e-07,
+      "loss": 1.7513,
+      "step": 4521
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.223281684545093e-07,
+      "loss": 1.9558,
+      "step": 4522
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.222884386174016e-07,
+      "loss": 1.6815,
+      "step": 4523
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.22248708780294e-07,
+      "loss": 1.7951,
+      "step": 4524
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.222089789431863e-07,
+      "loss": 1.7359,
+      "step": 4525
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.221692491060787e-07,
+      "loss": 1.6605,
+      "step": 4526
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.22129519268971e-07,
+      "loss": 1.5527,
+      "step": 4527
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.220897894318633e-07,
+      "loss": 1.7616,
+      "step": 4528
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.220500595947555e-07,
+      "loss": 1.794,
+      "step": 4529
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.220103297576479e-07,
+      "loss": 1.7325,
+      "step": 4530
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.219705999205403e-07,
+      "loss": 2.1162,
+      "step": 4531
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.219308700834326e-07,
+      "loss": 2.0436,
+      "step": 4532
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.21891140246325e-07,
+      "loss": 1.8096,
+      "step": 4533
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.218514104092172e-07,
+      "loss": 1.5692,
+      "step": 4534
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.218116805721096e-07,
+      "loss": 1.4559,
+      "step": 4535
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.217719507350019e-07,
+      "loss": 1.6401,
+      "step": 4536
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.217322208978943e-07,
+      "loss": 1.814,
+      "step": 4537
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.216924910607867e-07,
+      "loss": 1.7168,
+      "step": 4538
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.21652761223679e-07,
+      "loss": 1.6296,
+      "step": 4539
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.216130313865713e-07,
+      "loss": 1.8323,
+      "step": 4540
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.215733015494636e-07,
+      "loss": 1.7446,
+      "step": 4541
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.215335717123559e-07,
+      "loss": 2.0645,
+      "step": 4542
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.214938418752482e-07,
+      "loss": 1.7181,
+      "step": 4543
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.214541120381406e-07,
+      "loss": 1.9687,
+      "step": 4544
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.21414382201033e-07,
+      "loss": 1.765,
+      "step": 4545
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.213746523639252e-07,
+      "loss": 1.6019,
+      "step": 4546
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.213349225268176e-07,
+      "loss": 1.5339,
+      "step": 4547
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.212951926897099e-07,
+      "loss": 1.7587,
+      "step": 4548
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.212554628526023e-07,
+      "loss": 1.9168,
+      "step": 4549
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.212157330154946e-07,
+      "loss": 1.8752,
+      "step": 4550
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.21176003178387e-07,
+      "loss": 1.8901,
+      "step": 4551
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.211362733412792e-07,
+      "loss": 1.9047,
+      "step": 4552
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.210965435041716e-07,
+      "loss": 1.7791,
+      "step": 4553
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.21056813667064e-07,
+      "loss": 1.4373,
+      "step": 4554
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.210170838299563e-07,
+      "loss": 1.642,
+      "step": 4555
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.209773539928486e-07,
+      "loss": 1.6818,
+      "step": 4556
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.209376241557409e-07,
+      "loss": 2.0214,
+      "step": 4557
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.208978943186332e-07,
+      "loss": 1.7335,
+      "step": 4558
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.208581644815255e-07,
+      "loss": 1.9025,
+      "step": 4559
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.208184346444179e-07,
+      "loss": 1.5646,
+      "step": 4560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.207787048073103e-07,
+      "loss": 1.6672,
+      "step": 4561
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.207389749702026e-07,
+      "loss": 1.8365,
+      "step": 4562
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.20699245133095e-07,
+      "loss": 1.7327,
+      "step": 4563
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.206595152959872e-07,
+      "loss": 1.9031,
+      "step": 4564
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.206197854588796e-07,
+      "loss": 1.9934,
+      "step": 4565
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.205800556217719e-07,
+      "loss": 1.6773,
+      "step": 4566
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.205403257846643e-07,
+      "loss": 1.8015,
+      "step": 4567
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.205005959475567e-07,
+      "loss": 1.6552,
+      "step": 4568
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.204608661104489e-07,
+      "loss": 1.8631,
+      "step": 4569
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.204211362733412e-07,
+      "loss": 1.7384,
+      "step": 4570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.203814064362335e-07,
+      "loss": 1.7259,
+      "step": 4571
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.203416765991259e-07,
+      "loss": 1.9018,
+      "step": 4572
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.203019467620182e-07,
+      "loss": 1.7256,
+      "step": 4573
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.202622169249106e-07,
+      "loss": 1.6517,
+      "step": 4574
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.20222487087803e-07,
+      "loss": 1.3843,
+      "step": 4575
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.201827572506952e-07,
+      "loss": 1.6348,
+      "step": 4576
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.201430274135876e-07,
+      "loss": 1.6376,
+      "step": 4577
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.201032975764799e-07,
+      "loss": 1.5448,
+      "step": 4578
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.200635677393723e-07,
+      "loss": 1.6991,
+      "step": 4579
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.200238379022646e-07,
+      "loss": 1.8223,
+      "step": 4580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.19984108065157e-07,
+      "loss": 1.9447,
+      "step": 4581
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.199443782280491e-07,
+      "loss": 1.7633,
+      "step": 4582
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.199046483909415e-07,
+      "loss": 1.6112,
+      "step": 4583
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.198649185538339e-07,
+      "loss": 1.5234,
+      "step": 4584
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.198251887167262e-07,
+      "loss": 1.8738,
+      "step": 4585
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.197854588796186e-07,
+      "loss": 2.1925,
+      "step": 4586
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.197457290425109e-07,
+      "loss": 2.0866,
+      "step": 4587
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.197059992054032e-07,
+      "loss": 1.5107,
+      "step": 4588
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.196662693682955e-07,
+      "loss": 1.6369,
+      "step": 4589
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.196265395311879e-07,
+      "loss": 1.5968,
+      "step": 4590
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.195868096940803e-07,
+      "loss": 1.8913,
+      "step": 4591
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.195470798569726e-07,
+      "loss": 1.8056,
+      "step": 4592
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.19507350019865e-07,
+      "loss": 1.6747,
+      "step": 4593
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.194676201827572e-07,
+      "loss": 1.8432,
+      "step": 4594
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.194278903456495e-07,
+      "loss": 1.2916,
+      "step": 4595
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.193881605085418e-07,
+      "loss": 2.0655,
+      "step": 4596
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.193484306714342e-07,
+      "loss": 1.6183,
+      "step": 4597
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.193087008343265e-07,
+      "loss": 1.7601,
+      "step": 4598
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.192689709972189e-07,
+      "loss": 1.5539,
+      "step": 4599
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.192292411601112e-07,
+      "loss": 1.6089,
+      "step": 4600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.191895113230035e-07,
+      "loss": 1.3203,
+      "step": 4601
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.191497814858959e-07,
+      "loss": 1.5933,
+      "step": 4602
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.191100516487882e-07,
+      "loss": 1.7513,
+      "step": 4603
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.190703218116806e-07,
+      "loss": 1.9129,
+      "step": 4604
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.190305919745729e-07,
+      "loss": 1.736,
+      "step": 4605
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.189908621374652e-07,
+      "loss": 1.6887,
+      "step": 4606
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.189511323003576e-07,
+      "loss": 2.0619,
+      "step": 4607
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.189114024632498e-07,
+      "loss": 1.812,
+      "step": 4608
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.188716726261422e-07,
+      "loss": 1.7705,
+      "step": 4609
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.188319427890345e-07,
+      "loss": 1.7621,
+      "step": 4610
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.187922129519269e-07,
+      "loss": 1.4326,
+      "step": 4611
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.187524831148191e-07,
+      "loss": 1.7246,
+      "step": 4612
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.187127532777115e-07,
+      "loss": 1.5394,
+      "step": 4613
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.186730234406038e-07,
+      "loss": 1.8637,
+      "step": 4614
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.186332936034962e-07,
+      "loss": 1.7753,
+      "step": 4615
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.185935637663886e-07,
+      "loss": 1.6568,
+      "step": 4616
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.185538339292809e-07,
+      "loss": 1.7054,
+      "step": 4617
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.185141040921732e-07,
+      "loss": 1.756,
+      "step": 4618
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.184743742550655e-07,
+      "loss": 1.8568,
+      "step": 4619
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.184346444179579e-07,
+      "loss": 1.5907,
+      "step": 4620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.183949145808502e-07,
+      "loss": 1.9083,
+      "step": 4621
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.183551847437425e-07,
+      "loss": 1.6396,
+      "step": 4622
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.183154549066349e-07,
+      "loss": 1.8076,
+      "step": 4623
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.182757250695271e-07,
+      "loss": 1.6068,
+      "step": 4624
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.182359952324195e-07,
+      "loss": 1.9072,
+      "step": 4625
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.181962653953118e-07,
+      "loss": 1.4571,
+      "step": 4626
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.181565355582042e-07,
+      "loss": 1.9287,
+      "step": 4627
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.181168057210965e-07,
+      "loss": 1.7379,
+      "step": 4628
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.180770758839889e-07,
+      "loss": 1.9038,
+      "step": 4629
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.180373460468811e-07,
+      "loss": 1.7845,
+      "step": 4630
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.179976162097735e-07,
+      "loss": 1.756,
+      "step": 4631
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.179578863726659e-07,
+      "loss": 1.8888,
+      "step": 4632
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.179181565355582e-07,
+      "loss": 1.8591,
+      "step": 4633
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.178784266984506e-07,
+      "loss": 1.6099,
+      "step": 4634
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.178386968613428e-07,
+      "loss": 1.8601,
+      "step": 4635
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.177989670242351e-07,
+      "loss": 1.7167,
+      "step": 4636
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.177592371871274e-07,
+      "loss": 1.6634,
+      "step": 4637
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.177195073500198e-07,
+      "loss": 1.8326,
+      "step": 4638
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.176797775129122e-07,
+      "loss": 1.6482,
+      "step": 4639
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.176400476758045e-07,
+      "loss": 1.7939,
+      "step": 4640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.176003178386969e-07,
+      "loss": 1.7855,
+      "step": 4641
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.175605880015891e-07,
+      "loss": 1.654,
+      "step": 4642
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.175208581644815e-07,
+      "loss": 1.6864,
+      "step": 4643
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.174811283273738e-07,
+      "loss": 1.7831,
+      "step": 4644
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.174413984902662e-07,
+      "loss": 1.6684,
+      "step": 4645
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.174016686531586e-07,
+      "loss": 1.724,
+      "step": 4646
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.173619388160509e-07,
+      "loss": 2.039,
+      "step": 4647
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.173222089789431e-07,
+      "loss": 1.8209,
+      "step": 4648
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.172824791418354e-07,
+      "loss": 1.7714,
+      "step": 4649
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.172427493047278e-07,
+      "loss": 1.7126,
+      "step": 4650
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.172030194676201e-07,
+      "loss": 1.6825,
+      "step": 4651
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.171632896305125e-07,
+      "loss": 1.599,
+      "step": 4652
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.171235597934048e-07,
+      "loss": 1.6177,
+      "step": 4653
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.170838299562971e-07,
+      "loss": 1.8492,
+      "step": 4654
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.170441001191895e-07,
+      "loss": 1.8105,
+      "step": 4655
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.170043702820818e-07,
+      "loss": 1.7665,
+      "step": 4656
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.169646404449742e-07,
+      "loss": 1.8055,
+      "step": 4657
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.169249106078665e-07,
+      "loss": 1.5741,
+      "step": 4658
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.168851807707589e-07,
+      "loss": 1.662,
+      "step": 4659
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.168454509336511e-07,
+      "loss": 1.6116,
+      "step": 4660
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.168057210965434e-07,
+      "loss": 1.655,
+      "step": 4661
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.167659912594358e-07,
+      "loss": 1.684,
+      "step": 4662
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.167262614223281e-07,
+      "loss": 1.9326,
+      "step": 4663
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.166865315852205e-07,
+      "loss": 1.686,
+      "step": 4664
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.166468017481128e-07,
+      "loss": 1.7404,
+      "step": 4665
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.166070719110051e-07,
+      "loss": 1.7826,
+      "step": 4666
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.165673420738974e-07,
+      "loss": 1.8477,
+      "step": 4667
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.165276122367898e-07,
+      "loss": 1.713,
+      "step": 4668
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.164878823996822e-07,
+      "loss": 1.8273,
+      "step": 4669
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.164481525625745e-07,
+      "loss": 1.6694,
+      "step": 4670
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.164084227254669e-07,
+      "loss": 1.9466,
+      "step": 4671
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.163686928883591e-07,
+      "loss": 1.799,
+      "step": 4672
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.163289630512515e-07,
+      "loss": 1.8887,
+      "step": 4673
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.162892332141437e-07,
+      "loss": 1.9002,
+      "step": 4674
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.162495033770361e-07,
+      "loss": 1.6493,
+      "step": 4675
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.162097735399284e-07,
+      "loss": 2.0853,
+      "step": 4676
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.161700437028208e-07,
+      "loss": 1.7623,
+      "step": 4677
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.161303138657131e-07,
+      "loss": 1.7992,
+      "step": 4678
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.160905840286054e-07,
+      "loss": 1.7362,
+      "step": 4679
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.160508541914978e-07,
+      "loss": 1.8046,
+      "step": 4680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.160111243543901e-07,
+      "loss": 1.7738,
+      "step": 4681
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.159713945172825e-07,
+      "loss": 1.571,
+      "step": 4682
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.159316646801748e-07,
+      "loss": 1.4921,
+      "step": 4683
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.158919348430671e-07,
+      "loss": 1.5926,
+      "step": 4684
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.158522050059595e-07,
+      "loss": 1.8345,
+      "step": 4685
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.158124751688518e-07,
+      "loss": 1.7986,
+      "step": 4686
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.157727453317442e-07,
+      "loss": 1.8511,
+      "step": 4687
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.157330154946364e-07,
+      "loss": 1.6999,
+      "step": 4688
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.156932856575288e-07,
+      "loss": 1.9885,
+      "step": 4689
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.15653555820421e-07,
+      "loss": 1.8735,
+      "step": 4690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.156138259833134e-07,
+      "loss": 1.7087,
+      "step": 4691
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.155740961462057e-07,
+      "loss": 2.107,
+      "step": 4692
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.155343663090981e-07,
+      "loss": 1.807,
+      "step": 4693
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.154946364719905e-07,
+      "loss": 1.5974,
+      "step": 4694
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.154549066348828e-07,
+      "loss": 1.6517,
+      "step": 4695
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.154151767977751e-07,
+      "loss": 1.9168,
+      "step": 4696
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.153754469606674e-07,
+      "loss": 1.8207,
+      "step": 4697
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.153357171235598e-07,
+      "loss": 1.7864,
+      "step": 4698
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.152959872864521e-07,
+      "loss": 1.8304,
+      "step": 4699
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.152562574493445e-07,
+      "loss": 1.6716,
+      "step": 4700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.152165276122368e-07,
+      "loss": 1.758,
+      "step": 4701
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.15176797775129e-07,
+      "loss": 1.8761,
+      "step": 4702
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.151370679380214e-07,
+      "loss": 1.5784,
+      "step": 4703
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.150973381009137e-07,
+      "loss": 1.8578,
+      "step": 4704
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.150576082638061e-07,
+      "loss": 2.025,
+      "step": 4705
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.150178784266984e-07,
+      "loss": 1.6644,
+      "step": 4706
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.149781485895908e-07,
+      "loss": 1.7849,
+      "step": 4707
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.14938418752483e-07,
+      "loss": 1.6982,
+      "step": 4708
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.148986889153754e-07,
+      "loss": 1.9521,
+      "step": 4709
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.148589590782678e-07,
+      "loss": 1.7074,
+      "step": 4710
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.148192292411601e-07,
+      "loss": 1.6846,
+      "step": 4711
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.147794994040525e-07,
+      "loss": 1.6621,
+      "step": 4712
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.147397695669448e-07,
+      "loss": 1.686,
+      "step": 4713
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.14700039729837e-07,
+      "loss": 1.7232,
+      "step": 4714
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.146603098927293e-07,
+      "loss": 1.8551,
+      "step": 4715
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.146205800556217e-07,
+      "loss": 1.7246,
+      "step": 4716
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.145808502185141e-07,
+      "loss": 1.7059,
+      "step": 4717
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.145411203814064e-07,
+      "loss": 1.6522,
+      "step": 4718
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.145013905442988e-07,
+      "loss": 1.8548,
+      "step": 4719
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.14461660707191e-07,
+      "loss": 1.9614,
+      "step": 4720
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.144219308700834e-07,
+      "loss": 1.8507,
+      "step": 4721
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.143822010329757e-07,
+      "loss": 1.8166,
+      "step": 4722
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.143424711958681e-07,
+      "loss": 2.0083,
+      "step": 4723
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.143027413587605e-07,
+      "loss": 1.64,
+      "step": 4724
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.142630115216528e-07,
+      "loss": 2.0549,
+      "step": 4725
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.142232816845451e-07,
+      "loss": 1.9079,
+      "step": 4726
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.141835518474373e-07,
+      "loss": 1.931,
+      "step": 4727
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.141438220103297e-07,
+      "loss": 1.638,
+      "step": 4728
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.14104092173222e-07,
+      "loss": 2.0336,
+      "step": 4729
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.140643623361144e-07,
+      "loss": 1.5833,
+      "step": 4730
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.140246324990068e-07,
+      "loss": 1.3824,
+      "step": 4731
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.13984902661899e-07,
+      "loss": 1.8088,
+      "step": 4732
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.139451728247914e-07,
+      "loss": 1.9117,
+      "step": 4733
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.139054429876837e-07,
+      "loss": 1.876,
+      "step": 4734
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.138657131505761e-07,
+      "loss": 1.6951,
+      "step": 4735
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.138259833134684e-07,
+      "loss": 1.5127,
+      "step": 4736
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.137862534763608e-07,
+      "loss": 1.8278,
+      "step": 4737
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.13746523639253e-07,
+      "loss": 1.5959,
+      "step": 4738
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.137067938021454e-07,
+      "loss": 1.5659,
+      "step": 4739
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.136670639650377e-07,
+      "loss": 1.6558,
+      "step": 4740
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.1362733412793e-07,
+      "loss": 1.8625,
+      "step": 4741
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.135876042908224e-07,
+      "loss": 1.6864,
+      "step": 4742
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.135478744537147e-07,
+      "loss": 1.69,
+      "step": 4743
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.13508144616607e-07,
+      "loss": 1.5546,
+      "step": 4744
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.134684147794993e-07,
+      "loss": 2.0675,
+      "step": 4745
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.134286849423917e-07,
+      "loss": 2.0304,
+      "step": 4746
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.13388955105284e-07,
+      "loss": 1.9615,
+      "step": 4747
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.133492252681764e-07,
+      "loss": 1.7746,
+      "step": 4748
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.133094954310688e-07,
+      "loss": 1.7388,
+      "step": 4749
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.13269765593961e-07,
+      "loss": 1.7679,
+      "step": 4750
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.132300357568534e-07,
+      "loss": 1.8734,
+      "step": 4751
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.131903059197457e-07,
+      "loss": 2.0024,
+      "step": 4752
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.13150576082638e-07,
+      "loss": 1.6368,
+      "step": 4753
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.131108462455303e-07,
+      "loss": 1.8493,
+      "step": 4754
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.130711164084227e-07,
+      "loss": 1.5579,
+      "step": 4755
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.13031386571315e-07,
+      "loss": 1.6771,
+      "step": 4756
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.129916567342073e-07,
+      "loss": 1.9115,
+      "step": 4757
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.129519268970997e-07,
+      "loss": 1.6329,
+      "step": 4758
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.12912197059992e-07,
+      "loss": 1.7406,
+      "step": 4759
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.128724672228844e-07,
+      "loss": 1.8618,
+      "step": 4760
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.128327373857767e-07,
+      "loss": 2.1436,
+      "step": 4761
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.12793007548669e-07,
+      "loss": 2.0008,
+      "step": 4762
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.127532777115614e-07,
+      "loss": 2.0125,
+      "step": 4763
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.127135478744537e-07,
+      "loss": 1.4594,
+      "step": 4764
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.126738180373461e-07,
+      "loss": 1.7519,
+      "step": 4765
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.126340882002384e-07,
+      "loss": 1.7354,
+      "step": 4766
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.125943583631307e-07,
+      "loss": 1.4764,
+      "step": 4767
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.125546285260229e-07,
+      "loss": 2.1397,
+      "step": 4768
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.125148986889153e-07,
+      "loss": 1.6135,
+      "step": 4769
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.124751688518076e-07,
+      "loss": 1.9839,
+      "step": 4770
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.124354390147e-07,
+      "loss": 1.5576,
+      "step": 4771
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.123957091775924e-07,
+      "loss": 1.5875,
+      "step": 4772
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.123559793404847e-07,
+      "loss": 1.768,
+      "step": 4773
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.12316249503377e-07,
+      "loss": 1.9478,
+      "step": 4774
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.122765196662693e-07,
+      "loss": 1.7255,
+      "step": 4775
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.122367898291617e-07,
+      "loss": 1.9448,
+      "step": 4776
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.12197059992054e-07,
+      "loss": 1.7808,
+      "step": 4777
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.121573301549464e-07,
+      "loss": 1.5399,
+      "step": 4778
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.121176003178388e-07,
+      "loss": 1.9478,
+      "step": 4779
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.120778704807309e-07,
+      "loss": 1.7495,
+      "step": 4780
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.120381406436233e-07,
+      "loss": 1.8819,
+      "step": 4781
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.119984108065156e-07,
+      "loss": 1.8045,
+      "step": 4782
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.11958680969408e-07,
+      "loss": 1.3192,
+      "step": 4783
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.119189511323003e-07,
+      "loss": 1.4484,
+      "step": 4784
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.118792212951927e-07,
+      "loss": 2.1104,
+      "step": 4785
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.11839491458085e-07,
+      "loss": 1.6772,
+      "step": 4786
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.117997616209773e-07,
+      "loss": 1.8532,
+      "step": 4787
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.117600317838697e-07,
+      "loss": 1.667,
+      "step": 4788
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.11720301946762e-07,
+      "loss": 2.1302,
+      "step": 4789
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.116805721096544e-07,
+      "loss": 1.9584,
+      "step": 4790
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.116408422725467e-07,
+      "loss": 1.5674,
+      "step": 4791
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.11601112435439e-07,
+      "loss": 2.1706,
+      "step": 4792
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.115613825983312e-07,
+      "loss": 1.6788,
+      "step": 4793
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.115216527612236e-07,
+      "loss": 1.5773,
+      "step": 4794
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.11481922924116e-07,
+      "loss": 1.5815,
+      "step": 4795
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.114421930870083e-07,
+      "loss": 1.8189,
+      "step": 4796
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.114024632499007e-07,
+      "loss": 1.6068,
+      "step": 4797
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.113627334127929e-07,
+      "loss": 1.786,
+      "step": 4798
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.113230035756853e-07,
+      "loss": 1.7584,
+      "step": 4799
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.112832737385776e-07,
+      "loss": 1.696,
+      "step": 4800
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.1124354390147e-07,
+      "loss": 1.4548,
+      "step": 4801
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.112038140643624e-07,
+      "loss": 1.813,
+      "step": 4802
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.111640842272547e-07,
+      "loss": 1.6885,
+      "step": 4803
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.11124354390147e-07,
+      "loss": 1.7638,
+      "step": 4804
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.110846245530393e-07,
+      "loss": 1.6395,
+      "step": 4805
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.110448947159316e-07,
+      "loss": 1.5617,
+      "step": 4806
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.110051648788239e-07,
+      "loss": 1.6539,
+      "step": 4807
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.109654350417163e-07,
+      "loss": 1.5337,
+      "step": 4808
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.109257052046087e-07,
+      "loss": 1.6655,
+      "step": 4809
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.108859753675009e-07,
+      "loss": 1.6327,
+      "step": 4810
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.108462455303933e-07,
+      "loss": 1.7581,
+      "step": 4811
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.108065156932856e-07,
+      "loss": 1.7738,
+      "step": 4812
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.10766785856178e-07,
+      "loss": 2.074,
+      "step": 4813
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.107270560190703e-07,
+      "loss": 1.9442,
+      "step": 4814
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.106873261819627e-07,
+      "loss": 1.818,
+      "step": 4815
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.106475963448549e-07,
+      "loss": 1.5915,
+      "step": 4816
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.106078665077473e-07,
+      "loss": 1.8511,
+      "step": 4817
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.105681366706397e-07,
+      "loss": 1.8746,
+      "step": 4818
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.105284068335319e-07,
+      "loss": 1.4715,
+      "step": 4819
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.104886769964243e-07,
+      "loss": 1.7168,
+      "step": 4820
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.104489471593166e-07,
+      "loss": 1.5687,
+      "step": 4821
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.104092173222089e-07,
+      "loss": 1.717,
+      "step": 4822
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.103694874851012e-07,
+      "loss": 1.6673,
+      "step": 4823
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.103297576479936e-07,
+      "loss": 1.786,
+      "step": 4824
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.10290027810886e-07,
+      "loss": 1.6946,
+      "step": 4825
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.102502979737783e-07,
+      "loss": 1.7814,
+      "step": 4826
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.102105681366707e-07,
+      "loss": 2.1691,
+      "step": 4827
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.101708382995629e-07,
+      "loss": 1.5918,
+      "step": 4828
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.101311084624553e-07,
+      "loss": 1.6326,
+      "step": 4829
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.100913786253476e-07,
+      "loss": 1.8203,
+      "step": 4830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.1005164878824e-07,
+      "loss": 1.8917,
+      "step": 4831
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.100119189511324e-07,
+      "loss": 1.3582,
+      "step": 4832
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.099721891140246e-07,
+      "loss": 1.7446,
+      "step": 4833
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.099324592769169e-07,
+      "loss": 1.598,
+      "step": 4834
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.098927294398092e-07,
+      "loss": 1.9153,
+      "step": 4835
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.098529996027016e-07,
+      "loss": 1.675,
+      "step": 4836
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.098132697655939e-07,
+      "loss": 2.1788,
+      "step": 4837
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.097735399284863e-07,
+      "loss": 1.9894,
+      "step": 4838
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.097338100913786e-07,
+      "loss": 2.1242,
+      "step": 4839
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.096940802542709e-07,
+      "loss": 1.6726,
+      "step": 4840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.096543504171633e-07,
+      "loss": 1.8228,
+      "step": 4841
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.096146205800556e-07,
+      "loss": 2.1822,
+      "step": 4842
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.09574890742948e-07,
+      "loss": 1.4261,
+      "step": 4843
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.095351609058403e-07,
+      "loss": 1.7393,
+      "step": 4844
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.094954310687325e-07,
+      "loss": 2.0824,
+      "step": 4845
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.094557012316248e-07,
+      "loss": 1.7724,
+      "step": 4846
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.094159713945172e-07,
+      "loss": 1.9568,
+      "step": 4847
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.093762415574095e-07,
+      "loss": 1.5794,
+      "step": 4848
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.093365117203019e-07,
+      "loss": 1.7738,
+      "step": 4849
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.092967818831943e-07,
+      "loss": 1.6328,
+      "step": 4850
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.092570520460865e-07,
+      "loss": 1.6855,
+      "step": 4851
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.092173222089789e-07,
+      "loss": 1.8141,
+      "step": 4852
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.091775923718712e-07,
+      "loss": 1.8512,
+      "step": 4853
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.091378625347636e-07,
+      "loss": 2.0513,
+      "step": 4854
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.090981326976559e-07,
+      "loss": 1.7929,
+      "step": 4855
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.090584028605483e-07,
+      "loss": 1.6107,
+      "step": 4856
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.090186730234406e-07,
+      "loss": 1.8944,
+      "step": 4857
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.089789431863329e-07,
+      "loss": 1.8974,
+      "step": 4858
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.089392133492252e-07,
+      "loss": 1.6856,
+      "step": 4859
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.088994835121175e-07,
+      "loss": 1.8181,
+      "step": 4860
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.088597536750099e-07,
+      "loss": 1.9426,
+      "step": 4861
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.088200238379022e-07,
+      "loss": 1.6475,
+      "step": 4862
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.087802940007945e-07,
+      "loss": 1.6964,
+      "step": 4863
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.087405641636868e-07,
+      "loss": 2.0576,
+      "step": 4864
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.087008343265792e-07,
+      "loss": 1.7333,
+      "step": 4865
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.086611044894716e-07,
+      "loss": 1.6939,
+      "step": 4866
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.086213746523639e-07,
+      "loss": 1.9065,
+      "step": 4867
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.085816448152563e-07,
+      "loss": 1.9164,
+      "step": 4868
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.085419149781485e-07,
+      "loss": 1.9098,
+      "step": 4869
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.085021851410409e-07,
+      "loss": 1.8204,
+      "step": 4870
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.084624553039332e-07,
+      "loss": 1.898,
+      "step": 4871
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.084227254668255e-07,
+      "loss": 1.8217,
+      "step": 4872
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.083829956297179e-07,
+      "loss": 1.8445,
+      "step": 4873
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.083432657926102e-07,
+      "loss": 2.2421,
+      "step": 4874
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.083035359555025e-07,
+      "loss": 1.7418,
+      "step": 4875
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.082638061183948e-07,
+      "loss": 1.7857,
+      "step": 4876
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.082240762812872e-07,
+      "loss": 1.7319,
+      "step": 4877
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.081843464441795e-07,
+      "loss": 1.7575,
+      "step": 4878
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.081446166070719e-07,
+      "loss": 1.9395,
+      "step": 4879
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.081048867699643e-07,
+      "loss": 2.0369,
+      "step": 4880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.080651569328565e-07,
+      "loss": 1.7003,
+      "step": 4881
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.080254270957489e-07,
+      "loss": 1.6688,
+      "step": 4882
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.079856972586412e-07,
+      "loss": 2.0975,
+      "step": 4883
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.079459674215336e-07,
+      "loss": 1.6216,
+      "step": 4884
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.079062375844258e-07,
+      "loss": 1.9244,
+      "step": 4885
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.078665077473182e-07,
+      "loss": 1.8736,
+      "step": 4886
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.078267779102104e-07,
+      "loss": 1.4908,
+      "step": 4887
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.077870480731028e-07,
+      "loss": 1.615,
+      "step": 4888
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.077473182359952e-07,
+      "loss": 1.7905,
+      "step": 4889
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.077075883988875e-07,
+      "loss": 2.0023,
+      "step": 4890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.076678585617799e-07,
+      "loss": 1.6479,
+      "step": 4891
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.076281287246722e-07,
+      "loss": 1.7667,
+      "step": 4892
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.075883988875645e-07,
+      "loss": 1.7366,
+      "step": 4893
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.075486690504568e-07,
+      "loss": 1.615,
+      "step": 4894
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.075089392133492e-07,
+      "loss": 2.1142,
+      "step": 4895
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.074692093762416e-07,
+      "loss": 2.0336,
+      "step": 4896
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.074294795391339e-07,
+      "loss": 1.7197,
+      "step": 4897
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.073897497020263e-07,
+      "loss": 1.5775,
+      "step": 4898
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.073500198649184e-07,
+      "loss": 1.6978,
+      "step": 4899
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.073102900278108e-07,
+      "loss": 1.6702,
+      "step": 4900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.072705601907031e-07,
+      "loss": 1.691,
+      "step": 4901
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.072308303535955e-07,
+      "loss": 1.7153,
+      "step": 4902
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.071911005164879e-07,
+      "loss": 1.4815,
+      "step": 4903
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.071513706793802e-07,
+      "loss": 1.8465,
+      "step": 4904
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.071116408422725e-07,
+      "loss": 2.1282,
+      "step": 4905
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.070719110051648e-07,
+      "loss": 1.6402,
+      "step": 4906
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.070321811680572e-07,
+      "loss": 1.5474,
+      "step": 4907
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.069924513309495e-07,
+      "loss": 1.8557,
+      "step": 4908
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.069527214938419e-07,
+      "loss": 1.8523,
+      "step": 4909
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.069129916567343e-07,
+      "loss": 1.5779,
+      "step": 4910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.068732618196264e-07,
+      "loss": 1.736,
+      "step": 4911
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.068335319825188e-07,
+      "loss": 1.6198,
+      "step": 4912
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.067938021454111e-07,
+      "loss": 1.7295,
+      "step": 4913
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.067540723083035e-07,
+      "loss": 1.59,
+      "step": 4914
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.067143424711958e-07,
+      "loss": 1.873,
+      "step": 4915
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.066746126340882e-07,
+      "loss": 1.8098,
+      "step": 4916
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.066348827969804e-07,
+      "loss": 1.7893,
+      "step": 4917
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.065951529598728e-07,
+      "loss": 1.5076,
+      "step": 4918
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.065554231227652e-07,
+      "loss": 1.4918,
+      "step": 4919
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.065156932856575e-07,
+      "loss": 1.8671,
+      "step": 4920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.064759634485499e-07,
+      "loss": 1.4963,
+      "step": 4921
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.064362336114422e-07,
+      "loss": 1.6961,
+      "step": 4922
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.063965037743345e-07,
+      "loss": 2.0877,
+      "step": 4923
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.063567739372268e-07,
+      "loss": 1.4581,
+      "step": 4924
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.063170441001191e-07,
+      "loss": 1.8497,
+      "step": 4925
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.062773142630114e-07,
+      "loss": 1.586,
+      "step": 4926
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.062375844259038e-07,
+      "loss": 1.7354,
+      "step": 4927
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.061978545887962e-07,
+      "loss": 2.0292,
+      "step": 4928
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.061581247516884e-07,
+      "loss": 1.9762,
+      "step": 4929
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.061183949145808e-07,
+      "loss": 1.9264,
+      "step": 4930
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.060786650774731e-07,
+      "loss": 1.9402,
+      "step": 4931
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.060389352403655e-07,
+      "loss": 1.8019,
+      "step": 4932
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.059992054032578e-07,
+      "loss": 1.869,
+      "step": 4933
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.059594755661502e-07,
+      "loss": 1.7257,
+      "step": 4934
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.059197457290425e-07,
+      "loss": 1.8087,
+      "step": 4935
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.058800158919348e-07,
+      "loss": 1.6855,
+      "step": 4936
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.058402860548272e-07,
+      "loss": 1.8218,
+      "step": 4937
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.058005562177194e-07,
+      "loss": 1.8396,
+      "step": 4938
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.057608263806118e-07,
+      "loss": 1.65,
+      "step": 4939
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.057210965435041e-07,
+      "loss": 1.6164,
+      "step": 4940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.056813667063964e-07,
+      "loss": 1.4809,
+      "step": 4941
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.056416368692887e-07,
+      "loss": 1.7471,
+      "step": 4942
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.056019070321811e-07,
+      "loss": 1.9562,
+      "step": 4943
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.055621771950735e-07,
+      "loss": 1.7399,
+      "step": 4944
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.055224473579658e-07,
+      "loss": 1.662,
+      "step": 4945
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.054827175208582e-07,
+      "loss": 1.7788,
+      "step": 4946
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.054429876837504e-07,
+      "loss": 1.7806,
+      "step": 4947
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.054032578466428e-07,
+      "loss": 2.37,
+      "step": 4948
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.053635280095351e-07,
+      "loss": 1.6834,
+      "step": 4949
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.053237981724275e-07,
+      "loss": 1.8372,
+      "step": 4950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.052840683353198e-07,
+      "loss": 1.8452,
+      "step": 4951
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.052443384982121e-07,
+      "loss": 1.4096,
+      "step": 4952
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.052046086611044e-07,
+      "loss": 1.8775,
+      "step": 4953
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.051648788239967e-07,
+      "loss": 2.0372,
+      "step": 4954
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.051251489868891e-07,
+      "loss": 1.5835,
+      "step": 4955
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.050854191497814e-07,
+      "loss": 1.4369,
+      "step": 4956
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.050456893126738e-07,
+      "loss": 1.7197,
+      "step": 4957
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.050059594755662e-07,
+      "loss": 1.5143,
+      "step": 4958
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.049662296384584e-07,
+      "loss": 1.6732,
+      "step": 4959
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.049264998013508e-07,
+      "loss": 1.6347,
+      "step": 4960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.048867699642431e-07,
+      "loss": 1.7375,
+      "step": 4961
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.048470401271355e-07,
+      "loss": 1.5473,
+      "step": 4962
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.048073102900278e-07,
+      "loss": 1.6467,
+      "step": 4963
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.047675804529202e-07,
+      "loss": 1.5432,
+      "step": 4964
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.047278506158123e-07,
+      "loss": 1.9396,
+      "step": 4965
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.046881207787047e-07,
+      "loss": 1.6094,
+      "step": 4966
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.046483909415971e-07,
+      "loss": 1.6083,
+      "step": 4967
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.046086611044894e-07,
+      "loss": 2.1696,
+      "step": 4968
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.045689312673818e-07,
+      "loss": 2.225,
+      "step": 4969
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.045292014302741e-07,
+      "loss": 1.6708,
+      "step": 4970
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.044894715931664e-07,
+      "loss": 1.7469,
+      "step": 4971
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.044497417560587e-07,
+      "loss": 2.1565,
+      "step": 4972
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.044100119189511e-07,
+      "loss": 1.6841,
+      "step": 4973
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.043702820818435e-07,
+      "loss": 1.9574,
+      "step": 4974
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.043305522447358e-07,
+      "loss": 1.8212,
+      "step": 4975
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.042908224076282e-07,
+      "loss": 1.4106,
+      "step": 4976
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.042510925705203e-07,
+      "loss": 1.5991,
+      "step": 4977
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.042113627334127e-07,
+      "loss": 1.6285,
+      "step": 4978
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.04171632896305e-07,
+      "loss": 2.0736,
+      "step": 4979
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.041319030591974e-07,
+      "loss": 1.5845,
+      "step": 4980
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.040921732220898e-07,
+      "loss": 1.4167,
+      "step": 4981
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.040524433849821e-07,
+      "loss": 1.7437,
+      "step": 4982
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.040127135478744e-07,
+      "loss": 2.0147,
+      "step": 4983
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.039729837107667e-07,
+      "loss": 1.6574,
+      "step": 4984
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.039332538736591e-07,
+      "loss": 1.9141,
+      "step": 4985
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.038935240365514e-07,
+      "loss": 1.8083,
+      "step": 4986
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.038537941994438e-07,
+      "loss": 2.0182,
+      "step": 4987
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.038140643623362e-07,
+      "loss": 2.0124,
+      "step": 4988
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.037743345252284e-07,
+      "loss": 1.6784,
+      "step": 4989
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.037346046881208e-07,
+      "loss": 1.3711,
+      "step": 4990
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.03694874851013e-07,
+      "loss": 1.9845,
+      "step": 4991
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.036551450139054e-07,
+      "loss": 2.1943,
+      "step": 4992
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.036154151767977e-07,
+      "loss": 1.8249,
+      "step": 4993
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.035756853396901e-07,
+      "loss": 2.0672,
+      "step": 4994
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.035359555025823e-07,
+      "loss": 1.7929,
+      "step": 4995
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.034962256654747e-07,
+      "loss": 1.8573,
+      "step": 4996
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.034564958283671e-07,
+      "loss": 1.8001,
+      "step": 4997
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.034167659912594e-07,
+      "loss": 1.9215,
+      "step": 4998
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.033770361541518e-07,
+      "loss": 2.0732,
+      "step": 4999
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.033373063170441e-07,
+      "loss": 2.02,
+      "step": 5000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.032975764799364e-07,
+      "loss": 1.7207,
+      "step": 5001
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.032578466428287e-07,
+      "loss": 2.0534,
+      "step": 5002
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.032181168057211e-07,
+      "loss": 2.036,
+      "step": 5003
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.031783869686133e-07,
+      "loss": 1.4758,
+      "step": 5004
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.031386571315057e-07,
+      "loss": 1.919,
+      "step": 5005
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.030989272943981e-07,
+      "loss": 1.6298,
+      "step": 5006
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.030591974572903e-07,
+      "loss": 1.8095,
+      "step": 5007
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.030194676201827e-07,
+      "loss": 1.6259,
+      "step": 5008
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.02979737783075e-07,
+      "loss": 1.7994,
+      "step": 5009
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.029400079459674e-07,
+      "loss": 1.7818,
+      "step": 5010
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.029002781088597e-07,
+      "loss": 2.0446,
+      "step": 5011
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.028605482717521e-07,
+      "loss": 1.9678,
+      "step": 5012
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.028208184346444e-07,
+      "loss": 1.9519,
+      "step": 5013
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.027810885975367e-07,
+      "loss": 1.7627,
+      "step": 5014
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.027413587604291e-07,
+      "loss": 1.793,
+      "step": 5015
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.027016289233214e-07,
+      "loss": 1.7374,
+      "step": 5016
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.026618990862137e-07,
+      "loss": 1.7035,
+      "step": 5017
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.02622169249106e-07,
+      "loss": 1.567,
+      "step": 5018
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.025824394119983e-07,
+      "loss": 2.2437,
+      "step": 5019
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.025427095748906e-07,
+      "loss": 1.7206,
+      "step": 5020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.02502979737783e-07,
+      "loss": 1.9926,
+      "step": 5021
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.024632499006754e-07,
+      "loss": 1.7441,
+      "step": 5022
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.024235200635677e-07,
+      "loss": 2.0434,
+      "step": 5023
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.023837902264601e-07,
+      "loss": 1.8384,
+      "step": 5024
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.023440603893523e-07,
+      "loss": 1.9668,
+      "step": 5025
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.023043305522447e-07,
+      "loss": 2.001,
+      "step": 5026
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.02264600715137e-07,
+      "loss": 1.9726,
+      "step": 5027
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.022248708780294e-07,
+      "loss": 1.701,
+      "step": 5028
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.021851410409218e-07,
+      "loss": 1.7145,
+      "step": 5029
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.021454112038141e-07,
+      "loss": 1.7827,
+      "step": 5030
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.021056813667063e-07,
+      "loss": 1.7052,
+      "step": 5031
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.020659515295986e-07,
+      "loss": 1.9116,
+      "step": 5032
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.02026221692491e-07,
+      "loss": 1.7298,
+      "step": 5033
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.019864918553833e-07,
+      "loss": 1.4834,
+      "step": 5034
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.019467620182757e-07,
+      "loss": 1.6916,
+      "step": 5035
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.019070321811681e-07,
+      "loss": 1.9626,
+      "step": 5036
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.018673023440603e-07,
+      "loss": 1.589,
+      "step": 5037
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.018275725069527e-07,
+      "loss": 1.9268,
+      "step": 5038
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.01787842669845e-07,
+      "loss": 1.636,
+      "step": 5039
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.017481128327374e-07,
+      "loss": 1.9206,
+      "step": 5040
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.017083829956297e-07,
+      "loss": 1.7558,
+      "step": 5041
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.016686531585221e-07,
+      "loss": 1.4778,
+      "step": 5042
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.016289233214142e-07,
+      "loss": 1.9276,
+      "step": 5043
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.015891934843066e-07,
+      "loss": 1.1538,
+      "step": 5044
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.35448057097541635,
+      "eval_loss": 1.7883225679397583,
+      "eval_runtime": 6.4396,
+      "eval_samples_per_second": 391.637,
+      "eval_steps_per_second": 6.212,
+      "step": 5044
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.01549463647199e-07,
+      "loss": 1.5194,
+      "step": 5045
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.015097338100913e-07,
+      "loss": 1.7954,
+      "step": 5046
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.014700039729837e-07,
+      "loss": 1.824,
+      "step": 5047
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.01430274135876e-07,
+      "loss": 1.6693,
+      "step": 5048
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.013905442987683e-07,
+      "loss": 1.4517,
+      "step": 5049
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.013508144616606e-07,
+      "loss": 1.9729,
+      "step": 5050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.01311084624553e-07,
+      "loss": 1.7269,
+      "step": 5051
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.012713547874454e-07,
+      "loss": 1.6884,
+      "step": 5052
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.012316249503377e-07,
+      "loss": 1.8518,
+      "step": 5053
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.011918951132301e-07,
+      "loss": 1.508,
+      "step": 5054
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.011521652761223e-07,
+      "loss": 1.4053,
+      "step": 5055
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.011124354390147e-07,
+      "loss": 1.7884,
+      "step": 5056
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.010727056019069e-07,
+      "loss": 1.3653,
+      "step": 5057
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.010329757647993e-07,
+      "loss": 1.8998,
+      "step": 5058
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.009932459276917e-07,
+      "loss": 1.6071,
+      "step": 5059
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.00953516090584e-07,
+      "loss": 1.7261,
+      "step": 5060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.009137862534763e-07,
+      "loss": 1.6066,
+      "step": 5061
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.008740564163686e-07,
+      "loss": 1.6944,
+      "step": 5062
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.00834326579261e-07,
+      "loss": 1.6313,
+      "step": 5063
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.007945967421533e-07,
+      "loss": 1.744,
+      "step": 5064
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.007548669050457e-07,
+      "loss": 1.8037,
+      "step": 5065
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.00715137067938e-07,
+      "loss": 1.6903,
+      "step": 5066
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.006754072308303e-07,
+      "loss": 1.785,
+      "step": 5067
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.006356773937227e-07,
+      "loss": 1.5073,
+      "step": 5068
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.00595947556615e-07,
+      "loss": 1.7164,
+      "step": 5069
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.005562177195073e-07,
+      "loss": 1.7291,
+      "step": 5070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.005164878823996e-07,
+      "loss": 1.7537,
+      "step": 5071
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.00476758045292e-07,
+      "loss": 1.5035,
+      "step": 5072
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.004370282081842e-07,
+      "loss": 1.7119,
+      "step": 5073
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.003972983710766e-07,
+      "loss": 1.9759,
+      "step": 5074
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.00357568533969e-07,
+      "loss": 1.7782,
+      "step": 5075
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.003178386968613e-07,
+      "loss": 1.9638,
+      "step": 5076
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.002781088597537e-07,
+      "loss": 1.5533,
+      "step": 5077
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.00238379022646e-07,
+      "loss": 1.9756,
+      "step": 5078
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.001986491855383e-07,
+      "loss": 1.4564,
+      "step": 5079
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.001589193484306e-07,
+      "loss": 1.711,
+      "step": 5080
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.00119189511323e-07,
+      "loss": 1.7602,
+      "step": 5081
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.000794596742154e-07,
+      "loss": 1.7786,
+      "step": 5082
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.000397298371076e-07,
+      "loss": 1.8936,
+      "step": 5083
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8e-07,
+      "loss": 1.6157,
+      "step": 5084
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.999602701628922e-07,
+      "loss": 1.8438,
+      "step": 5085
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.999205403257846e-07,
+      "loss": 1.8334,
+      "step": 5086
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.998808104886769e-07,
+      "loss": 1.6247,
+      "step": 5087
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.998410806515693e-07,
+      "loss": 1.7625,
+      "step": 5088
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.998013508144616e-07,
+      "loss": 1.4686,
+      "step": 5089
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.99761620977354e-07,
+      "loss": 1.8804,
+      "step": 5090
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.997218911402463e-07,
+      "loss": 1.868,
+      "step": 5091
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.996821613031386e-07,
+      "loss": 1.9224,
+      "step": 5092
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.99642431466031e-07,
+      "loss": 1.7111,
+      "step": 5093
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.996027016289233e-07,
+      "loss": 1.839,
+      "step": 5094
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.995629717918157e-07,
+      "loss": 1.8742,
+      "step": 5095
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.99523241954708e-07,
+      "loss": 2.0328,
+      "step": 5096
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.994835121176002e-07,
+      "loss": 1.6969,
+      "step": 5097
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.994437822804926e-07,
+      "loss": 1.7711,
+      "step": 5098
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.994040524433849e-07,
+      "loss": 2.1212,
+      "step": 5099
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.993643226062773e-07,
+      "loss": 1.6729,
+      "step": 5100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.993245927691696e-07,
+      "loss": 1.5135,
+      "step": 5101
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.99284862932062e-07,
+      "loss": 1.6989,
+      "step": 5102
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.992451330949542e-07,
+      "loss": 1.6277,
+      "step": 5103
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.992054032578466e-07,
+      "loss": 1.6207,
+      "step": 5104
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.99165673420739e-07,
+      "loss": 1.7597,
+      "step": 5105
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.991259435836313e-07,
+      "loss": 1.8727,
+      "step": 5106
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.990862137465237e-07,
+      "loss": 1.8719,
+      "step": 5107
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.99046483909416e-07,
+      "loss": 1.9241,
+      "step": 5108
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.990067540723082e-07,
+      "loss": 1.7309,
+      "step": 5109
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.989670242352005e-07,
+      "loss": 1.6139,
+      "step": 5110
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.989272943980929e-07,
+      "loss": 1.5543,
+      "step": 5111
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.988875645609852e-07,
+      "loss": 1.9166,
+      "step": 5112
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.988478347238776e-07,
+      "loss": 1.8092,
+      "step": 5113
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.9880810488677e-07,
+      "loss": 1.749,
+      "step": 5114
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.987683750496622e-07,
+      "loss": 1.6,
+      "step": 5115
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.987286452125546e-07,
+      "loss": 1.6143,
+      "step": 5116
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.986889153754469e-07,
+      "loss": 1.453,
+      "step": 5117
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.986491855383393e-07,
+      "loss": 1.6603,
+      "step": 5118
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.986094557012316e-07,
+      "loss": 1.5033,
+      "step": 5119
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.98569725864124e-07,
+      "loss": 1.6878,
+      "step": 5120
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.985299960270163e-07,
+      "loss": 1.6797,
+      "step": 5121
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.984902661899086e-07,
+      "loss": 1.7326,
+      "step": 5122
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.984505363528009e-07,
+      "loss": 1.4309,
+      "step": 5123
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.984108065156932e-07,
+      "loss": 1.767,
+      "step": 5124
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.983710766785856e-07,
+      "loss": 1.5822,
+      "step": 5125
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.983313468414779e-07,
+      "loss": 1.7805,
+      "step": 5126
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.982916170043702e-07,
+      "loss": 2.2058,
+      "step": 5127
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.982518871672625e-07,
+      "loss": 1.6544,
+      "step": 5128
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.982121573301549e-07,
+      "loss": 1.7941,
+      "step": 5129
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.981724274930473e-07,
+      "loss": 2.0204,
+      "step": 5130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.981326976559396e-07,
+      "loss": 1.8005,
+      "step": 5131
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.98092967818832e-07,
+      "loss": 1.867,
+      "step": 5132
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.980532379817242e-07,
+      "loss": 2.1576,
+      "step": 5133
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.980135081446166e-07,
+      "loss": 1.7804,
+      "step": 5134
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.979737783075089e-07,
+      "loss": 1.7395,
+      "step": 5135
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.979340484704012e-07,
+      "loss": 1.6198,
+      "step": 5136
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.978943186332936e-07,
+      "loss": 1.7469,
+      "step": 5137
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.978545887961859e-07,
+      "loss": 1.7314,
+      "step": 5138
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.978148589590782e-07,
+      "loss": 1.7034,
+      "step": 5139
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.977751291219705e-07,
+      "loss": 1.5526,
+      "step": 5140
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.977353992848629e-07,
+      "loss": 1.6966,
+      "step": 5141
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.976956694477552e-07,
+      "loss": 1.8326,
+      "step": 5142
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.976559396106476e-07,
+      "loss": 1.8296,
+      "step": 5143
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.9761620977354e-07,
+      "loss": 2.0832,
+      "step": 5144
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.975764799364322e-07,
+      "loss": 2.2287,
+      "step": 5145
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.975367500993246e-07,
+      "loss": 1.7519,
+      "step": 5146
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.974970202622169e-07,
+      "loss": 1.6923,
+      "step": 5147
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.974572904251093e-07,
+      "loss": 1.4753,
+      "step": 5148
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.974175605880015e-07,
+      "loss": 1.6243,
+      "step": 5149
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.973778307508939e-07,
+      "loss": 1.8881,
+      "step": 5150
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.973381009137861e-07,
+      "loss": 1.6082,
+      "step": 5151
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.972983710766785e-07,
+      "loss": 1.8573,
+      "step": 5152
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.972586412395709e-07,
+      "loss": 1.4408,
+      "step": 5153
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.972189114024632e-07,
+      "loss": 1.6356,
+      "step": 5154
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.971791815653556e-07,
+      "loss": 1.7819,
+      "step": 5155
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.971394517282479e-07,
+      "loss": 1.8028,
+      "step": 5156
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.970997218911402e-07,
+      "loss": 1.6707,
+      "step": 5157
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.970599920540325e-07,
+      "loss": 1.8758,
+      "step": 5158
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.970202622169249e-07,
+      "loss": 1.8514,
+      "step": 5159
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.969805323798173e-07,
+      "loss": 1.5998,
+      "step": 5160
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.969408025427096e-07,
+      "loss": 1.9211,
+      "step": 5161
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.96901072705602e-07,
+      "loss": 1.6785,
+      "step": 5162
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.968613428684941e-07,
+      "loss": 1.2889,
+      "step": 5163
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.968216130313865e-07,
+      "loss": 2.0672,
+      "step": 5164
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.967818831942788e-07,
+      "loss": 1.533,
+      "step": 5165
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.967421533571712e-07,
+      "loss": 1.8663,
+      "step": 5166
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.967024235200635e-07,
+      "loss": 1.886,
+      "step": 5167
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.966626936829559e-07,
+      "loss": 1.6002,
+      "step": 5168
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.966229638458482e-07,
+      "loss": 1.5489,
+      "step": 5169
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.965832340087405e-07,
+      "loss": 1.6211,
+      "step": 5170
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.965435041716329e-07,
+      "loss": 1.962,
+      "step": 5171
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.965037743345252e-07,
+      "loss": 1.7484,
+      "step": 5172
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.964640444974176e-07,
+      "loss": 1.9825,
+      "step": 5173
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.964243146603099e-07,
+      "loss": 1.518,
+      "step": 5174
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.963845848232021e-07,
+      "loss": 1.9867,
+      "step": 5175
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.963448549860945e-07,
+      "loss": 1.5881,
+      "step": 5176
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.963051251489868e-07,
+      "loss": 1.6076,
+      "step": 5177
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.962653953118792e-07,
+      "loss": 2.0617,
+      "step": 5178
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.962256654747715e-07,
+      "loss": 1.8802,
+      "step": 5179
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.961859356376639e-07,
+      "loss": 1.7362,
+      "step": 5180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.961462058005561e-07,
+      "loss": 1.8083,
+      "step": 5181
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.961064759634485e-07,
+      "loss": 2.292,
+      "step": 5182
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.960667461263408e-07,
+      "loss": 1.7353,
+      "step": 5183
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.960270162892332e-07,
+      "loss": 1.9308,
+      "step": 5184
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.959872864521256e-07,
+      "loss": 1.5666,
+      "step": 5185
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.959475566150179e-07,
+      "loss": 1.8545,
+      "step": 5186
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.959078267779102e-07,
+      "loss": 1.7032,
+      "step": 5187
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.958680969408025e-07,
+      "loss": 1.8392,
+      "step": 5188
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.958283671036948e-07,
+      "loss": 1.7611,
+      "step": 5189
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.957886372665871e-07,
+      "loss": 1.886,
+      "step": 5190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.957489074294795e-07,
+      "loss": 1.6233,
+      "step": 5191
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.957091775923719e-07,
+      "loss": 1.4485,
+      "step": 5192
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.956694477552641e-07,
+      "loss": 1.8675,
+      "step": 5193
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.956297179181565e-07,
+      "loss": 1.9952,
+      "step": 5194
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.955899880810488e-07,
+      "loss": 1.8614,
+      "step": 5195
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.955502582439412e-07,
+      "loss": 1.7047,
+      "step": 5196
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.955105284068335e-07,
+      "loss": 1.7552,
+      "step": 5197
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.954707985697259e-07,
+      "loss": 1.7879,
+      "step": 5198
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.954310687326182e-07,
+      "loss": 1.8156,
+      "step": 5199
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.953913388955105e-07,
+      "loss": 1.9488,
+      "step": 5200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.953516090584029e-07,
+      "loss": 2.0379,
+      "step": 5201
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.953118792212951e-07,
+      "loss": 1.5142,
+      "step": 5202
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.952721493841875e-07,
+      "loss": 1.7414,
+      "step": 5203
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.952324195470798e-07,
+      "loss": 1.7461,
+      "step": 5204
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.951926897099721e-07,
+      "loss": 1.9247,
+      "step": 5205
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.951529598728644e-07,
+      "loss": 1.7022,
+      "step": 5206
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.951132300357568e-07,
+      "loss": 1.7273,
+      "step": 5207
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.950735001986492e-07,
+      "loss": 1.8774,
+      "step": 5208
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.950337703615415e-07,
+      "loss": 2.0006,
+      "step": 5209
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.949940405244339e-07,
+      "loss": 1.9029,
+      "step": 5210
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.949543106873261e-07,
+      "loss": 1.3873,
+      "step": 5211
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.949145808502185e-07,
+      "loss": 1.7776,
+      "step": 5212
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.948748510131108e-07,
+      "loss": 1.6401,
+      "step": 5213
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.948351211760032e-07,
+      "loss": 1.5597,
+      "step": 5214
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.947953913388955e-07,
+      "loss": 1.7331,
+      "step": 5215
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.947556615017878e-07,
+      "loss": 1.6468,
+      "step": 5216
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.947159316646801e-07,
+      "loss": 1.8465,
+      "step": 5217
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.946762018275724e-07,
+      "loss": 1.7343,
+      "step": 5218
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.946364719904648e-07,
+      "loss": 1.47,
+      "step": 5219
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.945967421533571e-07,
+      "loss": 1.9368,
+      "step": 5220
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.945570123162495e-07,
+      "loss": 1.6446,
+      "step": 5221
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.945172824791419e-07,
+      "loss": 1.8229,
+      "step": 5222
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.944775526420341e-07,
+      "loss": 1.9819,
+      "step": 5223
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.944378228049265e-07,
+      "loss": 1.6399,
+      "step": 5224
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.943980929678188e-07,
+      "loss": 1.5843,
+      "step": 5225
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.943583631307112e-07,
+      "loss": 1.6532,
+      "step": 5226
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.943186332936035e-07,
+      "loss": 1.6458,
+      "step": 5227
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.942789034564959e-07,
+      "loss": 1.6092,
+      "step": 5228
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.94239173619388e-07,
+      "loss": 1.6208,
+      "step": 5229
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.941994437822804e-07,
+      "loss": 1.8394,
+      "step": 5230
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.941597139451728e-07,
+      "loss": 1.8793,
+      "step": 5231
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.941199841080651e-07,
+      "loss": 1.7428,
+      "step": 5232
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.940802542709575e-07,
+      "loss": 1.3127,
+      "step": 5233
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.940405244338498e-07,
+      "loss": 1.7325,
+      "step": 5234
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.940007945967421e-07,
+      "loss": 1.4457,
+      "step": 5235
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.939610647596344e-07,
+      "loss": 1.6285,
+      "step": 5236
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.939213349225268e-07,
+      "loss": 1.3521,
+      "step": 5237
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.938816050854192e-07,
+      "loss": 1.8791,
+      "step": 5238
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.938418752483115e-07,
+      "loss": 1.845,
+      "step": 5239
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.938021454112039e-07,
+      "loss": 1.6941,
+      "step": 5240
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.93762415574096e-07,
+      "loss": 1.7192,
+      "step": 5241
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.937226857369884e-07,
+      "loss": 1.8291,
+      "step": 5242
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.936829558998807e-07,
+      "loss": 1.8894,
+      "step": 5243
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.936432260627731e-07,
+      "loss": 1.7224,
+      "step": 5244
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.936034962256654e-07,
+      "loss": 1.6504,
+      "step": 5245
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.935637663885578e-07,
+      "loss": 1.7716,
+      "step": 5246
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.935240365514501e-07,
+      "loss": 1.7029,
+      "step": 5247
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.934843067143424e-07,
+      "loss": 1.7815,
+      "step": 5248
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.934445768772348e-07,
+      "loss": 1.4586,
+      "step": 5249
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.934048470401271e-07,
+      "loss": 1.5988,
+      "step": 5250
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.933651172030195e-07,
+      "loss": 1.8053,
+      "step": 5251
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.933253873659118e-07,
+      "loss": 1.5384,
+      "step": 5252
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.932856575288041e-07,
+      "loss": 1.7075,
+      "step": 5253
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.932459276916965e-07,
+      "loss": 1.6488,
+      "step": 5254
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.932061978545887e-07,
+      "loss": 1.9863,
+      "step": 5255
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.931664680174811e-07,
+      "loss": 1.4413,
+      "step": 5256
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.931267381803734e-07,
+      "loss": 1.8483,
+      "step": 5257
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.930870083432658e-07,
+      "loss": 1.8545,
+      "step": 5258
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.93047278506158e-07,
+      "loss": 1.8057,
+      "step": 5259
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.930075486690504e-07,
+      "loss": 1.8406,
+      "step": 5260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.929678188319427e-07,
+      "loss": 1.9784,
+      "step": 5261
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.929280889948351e-07,
+      "loss": 1.4894,
+      "step": 5262
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.928883591577275e-07,
+      "loss": 1.8538,
+      "step": 5263
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.928486293206198e-07,
+      "loss": 1.7513,
+      "step": 5264
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.928088994835121e-07,
+      "loss": 1.807,
+      "step": 5265
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.927691696464044e-07,
+      "loss": 1.7056,
+      "step": 5266
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.927294398092968e-07,
+      "loss": 1.947,
+      "step": 5267
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.92689709972189e-07,
+      "loss": 1.6919,
+      "step": 5268
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.926499801350814e-07,
+      "loss": 1.6886,
+      "step": 5269
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.926102502979738e-07,
+      "loss": 1.2121,
+      "step": 5270
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.92570520460866e-07,
+      "loss": 1.7184,
+      "step": 5271
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.925307906237584e-07,
+      "loss": 1.7431,
+      "step": 5272
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.924910607866507e-07,
+      "loss": 1.7672,
+      "step": 5273
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.924513309495431e-07,
+      "loss": 1.4527,
+      "step": 5274
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.924116011124354e-07,
+      "loss": 1.7913,
+      "step": 5275
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.923718712753278e-07,
+      "loss": 1.4126,
+      "step": 5276
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.9233214143822e-07,
+      "loss": 2.0101,
+      "step": 5277
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.922924116011124e-07,
+      "loss": 1.5988,
+      "step": 5278
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.922526817640048e-07,
+      "loss": 1.803,
+      "step": 5279
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.922129519268971e-07,
+      "loss": 2.0257,
+      "step": 5280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.921732220897894e-07,
+      "loss": 1.5712,
+      "step": 5281
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.921334922526817e-07,
+      "loss": 1.8921,
+      "step": 5282
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.92093762415574e-07,
+      "loss": 1.6418,
+      "step": 5283
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.920540325784663e-07,
+      "loss": 1.7867,
+      "step": 5284
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.920143027413587e-07,
+      "loss": 1.8101,
+      "step": 5285
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.919745729042511e-07,
+      "loss": 1.6652,
+      "step": 5286
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.919348430671434e-07,
+      "loss": 1.4505,
+      "step": 5287
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.918951132300358e-07,
+      "loss": 1.59,
+      "step": 5288
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.91855383392928e-07,
+      "loss": 1.4078,
+      "step": 5289
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.918156535558204e-07,
+      "loss": 1.5838,
+      "step": 5290
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.917759237187127e-07,
+      "loss": 1.7088,
+      "step": 5291
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.917361938816051e-07,
+      "loss": 1.406,
+      "step": 5292
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.916964640444975e-07,
+      "loss": 1.5855,
+      "step": 5293
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.916567342073898e-07,
+      "loss": 1.4948,
+      "step": 5294
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.91617004370282e-07,
+      "loss": 1.978,
+      "step": 5295
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.915772745331743e-07,
+      "loss": 1.7641,
+      "step": 5296
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.915375446960667e-07,
+      "loss": 1.8895,
+      "step": 5297
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.91497814858959e-07,
+      "loss": 1.7658,
+      "step": 5298
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.914580850218514e-07,
+      "loss": 1.6395,
+      "step": 5299
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.914183551847438e-07,
+      "loss": 1.5612,
+      "step": 5300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.91378625347636e-07,
+      "loss": 2.1115,
+      "step": 5301
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.913388955105284e-07,
+      "loss": 1.7744,
+      "step": 5302
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.912991656734207e-07,
+      "loss": 1.7274,
+      "step": 5303
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.912594358363131e-07,
+      "loss": 1.578,
+      "step": 5304
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.912197059992054e-07,
+      "loss": 1.7384,
+      "step": 5305
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.911799761620978e-07,
+      "loss": 1.6557,
+      "step": 5306
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.911402463249899e-07,
+      "loss": 1.497,
+      "step": 5307
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.911005164878823e-07,
+      "loss": 1.7096,
+      "step": 5308
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.910607866507747e-07,
+      "loss": 1.6755,
+      "step": 5309
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.91021056813667e-07,
+      "loss": 1.6382,
+      "step": 5310
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.909813269765594e-07,
+      "loss": 1.9147,
+      "step": 5311
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.909415971394517e-07,
+      "loss": 1.7047,
+      "step": 5312
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.90901867302344e-07,
+      "loss": 1.8231,
+      "step": 5313
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.908621374652363e-07,
+      "loss": 1.9426,
+      "step": 5314
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.908224076281287e-07,
+      "loss": 1.7248,
+      "step": 5315
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.907826777910211e-07,
+      "loss": 2.2154,
+      "step": 5316
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.907429479539134e-07,
+      "loss": 1.7178,
+      "step": 5317
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.907032181168058e-07,
+      "loss": 1.7948,
+      "step": 5318
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.90663488279698e-07,
+      "loss": 1.5289,
+      "step": 5319
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.906237584425904e-07,
+      "loss": 1.9985,
+      "step": 5320
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.905840286054826e-07,
+      "loss": 1.186,
+      "step": 5321
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.90544298768375e-07,
+      "loss": 1.7193,
+      "step": 5322
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.905045689312673e-07,
+      "loss": 1.4889,
+      "step": 5323
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.904648390941597e-07,
+      "loss": 1.6809,
+      "step": 5324
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.90425109257052e-07,
+      "loss": 1.8509,
+      "step": 5325
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.903853794199443e-07,
+      "loss": 1.7153,
+      "step": 5326
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.903456495828367e-07,
+      "loss": 1.7545,
+      "step": 5327
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.90305919745729e-07,
+      "loss": 1.7524,
+      "step": 5328
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.902661899086214e-07,
+      "loss": 1.6441,
+      "step": 5329
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.902264600715137e-07,
+      "loss": 1.646,
+      "step": 5330
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.90186730234406e-07,
+      "loss": 1.4902,
+      "step": 5331
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.901470003972984e-07,
+      "loss": 1.6037,
+      "step": 5332
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.901072705601907e-07,
+      "loss": 2.1069,
+      "step": 5333
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.90067540723083e-07,
+      "loss": 1.7935,
+      "step": 5334
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.900278108859753e-07,
+      "loss": 1.7706,
+      "step": 5335
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.899880810488677e-07,
+      "loss": 1.889,
+      "step": 5336
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.899483512117599e-07,
+      "loss": 1.5949,
+      "step": 5337
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.899086213746523e-07,
+      "loss": 1.8602,
+      "step": 5338
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.898688915375446e-07,
+      "loss": 1.865,
+      "step": 5339
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.89829161700437e-07,
+      "loss": 1.9208,
+      "step": 5340
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.897894318633294e-07,
+      "loss": 1.8015,
+      "step": 5341
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.897497020262217e-07,
+      "loss": 1.7938,
+      "step": 5342
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.89709972189114e-07,
+      "loss": 1.8114,
+      "step": 5343
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.896702423520063e-07,
+      "loss": 1.6263,
+      "step": 5344
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.896305125148987e-07,
+      "loss": 1.3205,
+      "step": 5345
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.89590782677791e-07,
+      "loss": 1.6141,
+      "step": 5346
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.895510528406833e-07,
+      "loss": 1.9754,
+      "step": 5347
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.895113230035757e-07,
+      "loss": 1.5535,
+      "step": 5348
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.894715931664679e-07,
+      "loss": 1.95,
+      "step": 5349
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.894318633293603e-07,
+      "loss": 1.6445,
+      "step": 5350
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.893921334922526e-07,
+      "loss": 1.7005,
+      "step": 5351
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.89352403655145e-07,
+      "loss": 1.7542,
+      "step": 5352
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.893126738180373e-07,
+      "loss": 1.9402,
+      "step": 5353
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.892729439809297e-07,
+      "loss": 1.6444,
+      "step": 5354
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.89233214143822e-07,
+      "loss": 1.8436,
+      "step": 5355
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.891934843067143e-07,
+      "loss": 1.6844,
+      "step": 5356
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.891537544696067e-07,
+      "loss": 2.1721,
+      "step": 5357
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.89114024632499e-07,
+      "loss": 1.939,
+      "step": 5358
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.890742947953914e-07,
+      "loss": 1.7762,
+      "step": 5359
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.890345649582837e-07,
+      "loss": 1.5927,
+      "step": 5360
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.889948351211759e-07,
+      "loss": 1.5784,
+      "step": 5361
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.889551052840682e-07,
+      "loss": 1.8335,
+      "step": 5362
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.889153754469606e-07,
+      "loss": 1.7798,
+      "step": 5363
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.88875645609853e-07,
+      "loss": 1.713,
+      "step": 5364
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.888359157727453e-07,
+      "loss": 1.565,
+      "step": 5365
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.887961859356377e-07,
+      "loss": 1.6323,
+      "step": 5366
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.887564560985299e-07,
+      "loss": 1.8907,
+      "step": 5367
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.887167262614223e-07,
+      "loss": 1.7896,
+      "step": 5368
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.886769964243146e-07,
+      "loss": 1.5031,
+      "step": 5369
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.88637266587207e-07,
+      "loss": 1.4994,
+      "step": 5370
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.885975367500994e-07,
+      "loss": 1.4742,
+      "step": 5371
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.885578069129917e-07,
+      "loss": 1.8802,
+      "step": 5372
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.885180770758839e-07,
+      "loss": 1.3427,
+      "step": 5373
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.884783472387762e-07,
+      "loss": 1.8239,
+      "step": 5374
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.884386174016686e-07,
+      "loss": 1.618,
+      "step": 5375
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.883988875645609e-07,
+      "loss": 1.6845,
+      "step": 5376
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.883591577274533e-07,
+      "loss": 1.9681,
+      "step": 5377
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.883194278903457e-07,
+      "loss": 1.8685,
+      "step": 5378
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.882796980532379e-07,
+      "loss": 1.6645,
+      "step": 5379
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.882399682161303e-07,
+      "loss": 1.8722,
+      "step": 5380
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.882002383790226e-07,
+      "loss": 1.8242,
+      "step": 5381
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.88160508541915e-07,
+      "loss": 1.8234,
+      "step": 5382
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.881207787048073e-07,
+      "loss": 1.8432,
+      "step": 5383
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.880810488676997e-07,
+      "loss": 1.9092,
+      "step": 5384
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.880413190305919e-07,
+      "loss": 1.8415,
+      "step": 5385
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.880015891934842e-07,
+      "loss": 1.6302,
+      "step": 5386
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.879618593563766e-07,
+      "loss": 1.804,
+      "step": 5387
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.879221295192689e-07,
+      "loss": 1.7714,
+      "step": 5388
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.878823996821613e-07,
+      "loss": 1.7186,
+      "step": 5389
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.878426698450536e-07,
+      "loss": 1.529,
+      "step": 5390
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.878029400079459e-07,
+      "loss": 1.5821,
+      "step": 5391
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.877632101708382e-07,
+      "loss": 1.6283,
+      "step": 5392
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.877234803337306e-07,
+      "loss": 1.7015,
+      "step": 5393
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.87683750496623e-07,
+      "loss": 2.0974,
+      "step": 5394
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.876440206595153e-07,
+      "loss": 1.681,
+      "step": 5395
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.876042908224077e-07,
+      "loss": 1.7822,
+      "step": 5396
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.875645609852999e-07,
+      "loss": 1.4885,
+      "step": 5397
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.875248311481923e-07,
+      "loss": 1.9007,
+      "step": 5398
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.874851013110846e-07,
+      "loss": 1.5973,
+      "step": 5399
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.874453714739769e-07,
+      "loss": 1.6765,
+      "step": 5400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.874056416368692e-07,
+      "loss": 1.757,
+      "step": 5401
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.873659117997616e-07,
+      "loss": 1.7159,
+      "step": 5402
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.873261819626539e-07,
+      "loss": 1.7016,
+      "step": 5403
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.872864521255462e-07,
+      "loss": 1.5758,
+      "step": 5404
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.872467222884386e-07,
+      "loss": 1.5557,
+      "step": 5405
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.872069924513309e-07,
+      "loss": 1.4283,
+      "step": 5406
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.871672626142233e-07,
+      "loss": 1.7436,
+      "step": 5407
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.871275327771156e-07,
+      "loss": 1.7096,
+      "step": 5408
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.870878029400079e-07,
+      "loss": 1.6857,
+      "step": 5409
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.870480731029003e-07,
+      "loss": 1.6553,
+      "step": 5410
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.870083432657926e-07,
+      "loss": 2.2357,
+      "step": 5411
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.86968613428685e-07,
+      "loss": 1.8125,
+      "step": 5412
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.869288835915772e-07,
+      "loss": 1.4557,
+      "step": 5413
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.868891537544696e-07,
+      "loss": 1.6483,
+      "step": 5414
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.868494239173618e-07,
+      "loss": 2.1779,
+      "step": 5415
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.868096940802542e-07,
+      "loss": 1.6964,
+      "step": 5416
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.867699642431466e-07,
+      "loss": 1.7456,
+      "step": 5417
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.867302344060389e-07,
+      "loss": 1.6592,
+      "step": 5418
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.866905045689313e-07,
+      "loss": 1.9019,
+      "step": 5419
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.866507747318236e-07,
+      "loss": 1.6216,
+      "step": 5420
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.866110448947159e-07,
+      "loss": 1.8383,
+      "step": 5421
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.865713150576082e-07,
+      "loss": 1.8089,
+      "step": 5422
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.865315852205006e-07,
+      "loss": 1.6167,
+      "step": 5423
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.86491855383393e-07,
+      "loss": 1.8631,
+      "step": 5424
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.864521255462853e-07,
+      "loss": 1.5521,
+      "step": 5425
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.864123957091776e-07,
+      "loss": 1.6496,
+      "step": 5426
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.863726658720698e-07,
+      "loss": 1.9399,
+      "step": 5427
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.863329360349622e-07,
+      "loss": 1.6793,
+      "step": 5428
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.862932061978545e-07,
+      "loss": 1.778,
+      "step": 5429
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.862534763607469e-07,
+      "loss": 1.9231,
+      "step": 5430
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.862137465236392e-07,
+      "loss": 1.9485,
+      "step": 5431
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.861740166865316e-07,
+      "loss": 1.9805,
+      "step": 5432
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.861342868494239e-07,
+      "loss": 1.4811,
+      "step": 5433
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.860945570123162e-07,
+      "loss": 1.9394,
+      "step": 5434
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.860548271752086e-07,
+      "loss": 1.4132,
+      "step": 5435
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.860150973381009e-07,
+      "loss": 2.0518,
+      "step": 5436
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.859753675009933e-07,
+      "loss": 1.8142,
+      "step": 5437
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.859356376638856e-07,
+      "loss": 1.9977,
+      "step": 5438
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.858959078267778e-07,
+      "loss": 1.9022,
+      "step": 5439
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.858561779896701e-07,
+      "loss": 1.7413,
+      "step": 5440
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.858164481525625e-07,
+      "loss": 1.4959,
+      "step": 5441
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.857767183154549e-07,
+      "loss": 1.7769,
+      "step": 5442
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.857369884783472e-07,
+      "loss": 1.7021,
+      "step": 5443
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.856972586412395e-07,
+      "loss": 1.7842,
+      "step": 5444
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.856575288041318e-07,
+      "loss": 1.5541,
+      "step": 5445
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.856177989670242e-07,
+      "loss": 1.927,
+      "step": 5446
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.855780691299165e-07,
+      "loss": 1.8621,
+      "step": 5447
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.855383392928089e-07,
+      "loss": 1.8075,
+      "step": 5448
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.854986094557013e-07,
+      "loss": 1.8167,
+      "step": 5449
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.854588796185935e-07,
+      "loss": 2.148,
+      "step": 5450
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.854191497814859e-07,
+      "loss": 2.1382,
+      "step": 5451
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.853794199443781e-07,
+      "loss": 1.6034,
+      "step": 5452
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.853396901072705e-07,
+      "loss": 2.2628,
+      "step": 5453
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.852999602701628e-07,
+      "loss": 1.6926,
+      "step": 5454
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.852602304330552e-07,
+      "loss": 1.8058,
+      "step": 5455
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.852205005959474e-07,
+      "loss": 1.6545,
+      "step": 5456
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.851807707588398e-07,
+      "loss": 2.0783,
+      "step": 5457
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.851410409217322e-07,
+      "loss": 1.703,
+      "step": 5458
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.851013110846245e-07,
+      "loss": 1.5565,
+      "step": 5459
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.850615812475169e-07,
+      "loss": 1.657,
+      "step": 5460
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.850218514104092e-07,
+      "loss": 1.2623,
+      "step": 5461
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.849821215733015e-07,
+      "loss": 1.725,
+      "step": 5462
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.849423917361938e-07,
+      "loss": 1.4654,
+      "step": 5463
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.849026618990862e-07,
+      "loss": 1.7498,
+      "step": 5464
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.848629320619786e-07,
+      "loss": 1.6962,
+      "step": 5465
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.848232022248708e-07,
+      "loss": 2.0559,
+      "step": 5466
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.847834723877632e-07,
+      "loss": 1.4154,
+      "step": 5467
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.847437425506554e-07,
+      "loss": 1.9678,
+      "step": 5468
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.847040127135478e-07,
+      "loss": 2.0598,
+      "step": 5469
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.846642828764401e-07,
+      "loss": 1.6223,
+      "step": 5470
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.846245530393325e-07,
+      "loss": 1.5018,
+      "step": 5471
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.845848232022249e-07,
+      "loss": 1.624,
+      "step": 5472
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.845450933651172e-07,
+      "loss": 1.767,
+      "step": 5473
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.845053635280095e-07,
+      "loss": 1.5786,
+      "step": 5474
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.844656336909018e-07,
+      "loss": 1.8082,
+      "step": 5475
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.844259038537942e-07,
+      "loss": 1.9878,
+      "step": 5476
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.843861740166865e-07,
+      "loss": 1.5781,
+      "step": 5477
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.843464441795789e-07,
+      "loss": 1.5617,
+      "step": 5478
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.843067143424711e-07,
+      "loss": 1.5667,
+      "step": 5479
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.842669845053634e-07,
+      "loss": 2.0319,
+      "step": 5480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.842272546682558e-07,
+      "loss": 1.5854,
+      "step": 5481
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.841875248311481e-07,
+      "loss": 1.9424,
+      "step": 5482
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.841477949940405e-07,
+      "loss": 1.6602,
+      "step": 5483
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.841080651569328e-07,
+      "loss": 1.785,
+      "step": 5484
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.840683353198252e-07,
+      "loss": 1.624,
+      "step": 5485
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.840286054827174e-07,
+      "loss": 1.6931,
+      "step": 5486
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.839888756456098e-07,
+      "loss": 1.6746,
+      "step": 5487
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.839491458085022e-07,
+      "loss": 1.6949,
+      "step": 5488
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.839094159713945e-07,
+      "loss": 1.9108,
+      "step": 5489
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.838696861342869e-07,
+      "loss": 1.6169,
+      "step": 5490
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.838299562971792e-07,
+      "loss": 1.7176,
+      "step": 5491
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.837902264600714e-07,
+      "loss": 1.7508,
+      "step": 5492
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.837504966229637e-07,
+      "loss": 1.6875,
+      "step": 5493
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.837107667858561e-07,
+      "loss": 1.88,
+      "step": 5494
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.836710369487485e-07,
+      "loss": 1.9022,
+      "step": 5495
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.836313071116408e-07,
+      "loss": 1.9185,
+      "step": 5496
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.835915772745332e-07,
+      "loss": 1.6117,
+      "step": 5497
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.835518474374254e-07,
+      "loss": 2.137,
+      "step": 5498
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.835121176003178e-07,
+      "loss": 1.7675,
+      "step": 5499
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.834723877632101e-07,
+      "loss": 1.5907,
+      "step": 5500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.834326579261025e-07,
+      "loss": 1.9562,
+      "step": 5501
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.833929280889948e-07,
+      "loss": 1.8992,
+      "step": 5502
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.833531982518872e-07,
+      "loss": 1.9583,
+      "step": 5503
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.833134684147795e-07,
+      "loss": 1.8979,
+      "step": 5504
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.832737385776717e-07,
+      "loss": 2.0501,
+      "step": 5505
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.832340087405641e-07,
+      "loss": 1.8622,
+      "step": 5506
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.831942789034564e-07,
+      "loss": 1.8676,
+      "step": 5507
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.831545490663488e-07,
+      "loss": 1.6027,
+      "step": 5508
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.831148192292411e-07,
+      "loss": 1.5858,
+      "step": 5509
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.830750893921334e-07,
+      "loss": 2.2492,
+      "step": 5510
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.830353595550258e-07,
+      "loss": 1.5877,
+      "step": 5511
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.829956297179181e-07,
+      "loss": 1.871,
+      "step": 5512
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.829558998808105e-07,
+      "loss": 1.6129,
+      "step": 5513
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.829161700437028e-07,
+      "loss": 1.7603,
+      "step": 5514
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.828764402065952e-07,
+      "loss": 1.6604,
+      "step": 5515
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.828367103694874e-07,
+      "loss": 1.7759,
+      "step": 5516
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.827969805323798e-07,
+      "loss": 1.6559,
+      "step": 5517
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.82757250695272e-07,
+      "loss": 1.423,
+      "step": 5518
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.827175208581644e-07,
+      "loss": 1.9902,
+      "step": 5519
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.826777910210568e-07,
+      "loss": 1.861,
+      "step": 5520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.826380611839491e-07,
+      "loss": 1.7435,
+      "step": 5521
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.825983313468414e-07,
+      "loss": 1.5636,
+      "step": 5522
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.825586015097337e-07,
+      "loss": 1.9567,
+      "step": 5523
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.825188716726261e-07,
+      "loss": 1.8208,
+      "step": 5524
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.824791418355184e-07,
+      "loss": 1.7983,
+      "step": 5525
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.824394119984108e-07,
+      "loss": 1.8118,
+      "step": 5526
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.823996821613032e-07,
+      "loss": 1.4758,
+      "step": 5527
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.823599523241954e-07,
+      "loss": 1.8048,
+      "step": 5528
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.823202224870878e-07,
+      "loss": 1.7531,
+      "step": 5529
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.822804926499801e-07,
+      "loss": 1.6903,
+      "step": 5530
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.822407628128725e-07,
+      "loss": 1.611,
+      "step": 5531
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.822010329757647e-07,
+      "loss": 1.5291,
+      "step": 5532
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.821613031386571e-07,
+      "loss": 1.4285,
+      "step": 5533
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.821215733015493e-07,
+      "loss": 1.6616,
+      "step": 5534
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.820818434644417e-07,
+      "loss": 1.897,
+      "step": 5535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.820421136273341e-07,
+      "loss": 1.4391,
+      "step": 5536
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.820023837902264e-07,
+      "loss": 1.8301,
+      "step": 5537
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819626539531188e-07,
+      "loss": 1.7762,
+      "step": 5538
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819229241160111e-07,
+      "loss": 1.9355,
+      "step": 5539
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.818831942789034e-07,
+      "loss": 1.8246,
+      "step": 5540
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.818434644417957e-07,
+      "loss": 1.8553,
+      "step": 5541
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.818037346046881e-07,
+      "loss": 1.5639,
+      "step": 5542
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.817640047675805e-07,
+      "loss": 1.9936,
+      "step": 5543
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.817242749304728e-07,
+      "loss": 1.7285,
+      "step": 5544
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.816845450933651e-07,
+      "loss": 1.5478,
+      "step": 5545
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.816448152562573e-07,
+      "loss": 1.6969,
+      "step": 5546
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.816050854191497e-07,
+      "loss": 1.9026,
+      "step": 5547
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.81565355582042e-07,
+      "loss": 1.4006,
+      "step": 5548
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.815256257449344e-07,
+      "loss": 1.7295,
+      "step": 5549
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.814858959078268e-07,
+      "loss": 2.0476,
+      "step": 5550
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.814461660707191e-07,
+      "loss": 1.4367,
+      "step": 5551
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.814064362336114e-07,
+      "loss": 1.8046,
+      "step": 5552
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.813667063965037e-07,
+      "loss": 1.8319,
+      "step": 5553
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.813269765593961e-07,
+      "loss": 2.0349,
+      "step": 5554
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.812872467222884e-07,
+      "loss": 1.8268,
+      "step": 5555
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.812475168851808e-07,
+      "loss": 1.8206,
+      "step": 5556
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.812077870480732e-07,
+      "loss": 1.5912,
+      "step": 5557
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.811680572109653e-07,
+      "loss": 1.7316,
+      "step": 5558
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.811283273738577e-07,
+      "loss": 1.6805,
+      "step": 5559
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.8108859753675e-07,
+      "loss": 1.6801,
+      "step": 5560
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.810488676996424e-07,
+      "loss": 1.4958,
+      "step": 5561
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.810091378625347e-07,
+      "loss": 1.8145,
+      "step": 5562
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.809694080254271e-07,
+      "loss": 2.2022,
+      "step": 5563
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.809296781883193e-07,
+      "loss": 1.9568,
+      "step": 5564
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.808899483512117e-07,
+      "loss": 1.6911,
+      "step": 5565
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.808502185141041e-07,
+      "loss": 1.6676,
+      "step": 5566
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.808104886769964e-07,
+      "loss": 1.4276,
+      "step": 5567
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.807707588398888e-07,
+      "loss": 1.7413,
+      "step": 5568
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.807310290027811e-07,
+      "loss": 1.6479,
+      "step": 5569
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.806912991656734e-07,
+      "loss": 1.6904,
+      "step": 5570
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.806515693285656e-07,
+      "loss": 1.8894,
+      "step": 5571
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.80611839491458e-07,
+      "loss": 1.8459,
+      "step": 5572
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.805721096543504e-07,
+      "loss": 1.7058,
+      "step": 5573
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.805323798172427e-07,
+      "loss": 1.4792,
+      "step": 5574
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.804926499801351e-07,
+      "loss": 1.9499,
+      "step": 5575
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.804529201430273e-07,
+      "loss": 1.6342,
+      "step": 5576
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.804131903059197e-07,
+      "loss": 2.0586,
+      "step": 5577
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.80373460468812e-07,
+      "loss": 1.9151,
+      "step": 5578
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.803337306317044e-07,
+      "loss": 1.5873,
+      "step": 5579
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.802940007945967e-07,
+      "loss": 1.8229,
+      "step": 5580
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.802542709574891e-07,
+      "loss": 1.594,
+      "step": 5581
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.802145411203814e-07,
+      "loss": 1.8243,
+      "step": 5582
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.801748112832737e-07,
+      "loss": 1.6765,
+      "step": 5583
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.80135081446166e-07,
+      "loss": 1.7885,
+      "step": 5584
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.800953516090583e-07,
+      "loss": 1.8046,
+      "step": 5585
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.800556217719507e-07,
+      "loss": 1.6278,
+      "step": 5586
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.80015891934843e-07,
+      "loss": 2.1374,
+      "step": 5587
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.799761620977353e-07,
+      "loss": 1.8003,
+      "step": 5588
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.799364322606277e-07,
+      "loss": 1.7715,
+      "step": 5589
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.7989670242352e-07,
+      "loss": 1.4885,
+      "step": 5590
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.798569725864124e-07,
+      "loss": 1.6794,
+      "step": 5591
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.798172427493047e-07,
+      "loss": 1.7067,
+      "step": 5592
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.797775129121971e-07,
+      "loss": 2.2002,
+      "step": 5593
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.797377830750893e-07,
+      "loss": 1.4887,
+      "step": 5594
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.796980532379817e-07,
+      "loss": 1.8394,
+      "step": 5595
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.79658323400874e-07,
+      "loss": 1.7722,
+      "step": 5596
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.796185935637664e-07,
+      "loss": 1.8622,
+      "step": 5597
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.795788637266587e-07,
+      "loss": 1.7451,
+      "step": 5598
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.79539133889551e-07,
+      "loss": 1.7102,
+      "step": 5599
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.794994040524433e-07,
+      "loss": 1.6583,
+      "step": 5600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.794596742153356e-07,
+      "loss": 1.6454,
+      "step": 5601
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.79419944378228e-07,
+      "loss": 1.5626,
+      "step": 5602
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.793802145411203e-07,
+      "loss": 1.7351,
+      "step": 5603
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.793404847040127e-07,
+      "loss": 1.6509,
+      "step": 5604
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.793007548669051e-07,
+      "loss": 1.9218,
+      "step": 5605
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.792610250297973e-07,
+      "loss": 1.7236,
+      "step": 5606
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.792212951926897e-07,
+      "loss": 1.9694,
+      "step": 5607
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.79181565355582e-07,
+      "loss": 1.8944,
+      "step": 5608
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.791418355184744e-07,
+      "loss": 1.9229,
+      "step": 5609
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.791021056813667e-07,
+      "loss": 1.9587,
+      "step": 5610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.79062375844259e-07,
+      "loss": 1.5077,
+      "step": 5611
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.790226460071512e-07,
+      "loss": 1.9564,
+      "step": 5612
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.789829161700436e-07,
+      "loss": 1.89,
+      "step": 5613
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.78943186332936e-07,
+      "loss": 1.9413,
+      "step": 5614
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.789034564958283e-07,
+      "loss": 1.6295,
+      "step": 5615
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.788637266587207e-07,
+      "loss": 1.7052,
+      "step": 5616
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.78823996821613e-07,
+      "loss": 1.7426,
+      "step": 5617
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.787842669845053e-07,
+      "loss": 1.4988,
+      "step": 5618
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.787445371473976e-07,
+      "loss": 1.8993,
+      "step": 5619
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.7870480731029e-07,
+      "loss": 2.0233,
+      "step": 5620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.786650774731824e-07,
+      "loss": 1.6174,
+      "step": 5621
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.786253476360747e-07,
+      "loss": 2.0211,
+      "step": 5622
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.785856177989671e-07,
+      "loss": 1.936,
+      "step": 5623
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.785458879618592e-07,
+      "loss": 1.8551,
+      "step": 5624
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.785061581247516e-07,
+      "loss": 1.6075,
+      "step": 5625
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.784664282876439e-07,
+      "loss": 1.5255,
+      "step": 5626
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.784266984505363e-07,
+      "loss": 1.7282,
+      "step": 5627
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.783869686134287e-07,
+      "loss": 1.8992,
+      "step": 5628
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.78347238776321e-07,
+      "loss": 1.5635,
+      "step": 5629
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.783075089392133e-07,
+      "loss": 1.4435,
+      "step": 5630
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.782677791021056e-07,
+      "loss": 2.1338,
+      "step": 5631
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.78228049264998e-07,
+      "loss": 1.8402,
+      "step": 5632
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.781883194278903e-07,
+      "loss": 1.74,
+      "step": 5633
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.781485895907827e-07,
+      "loss": 1.5727,
+      "step": 5634
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.781088597536751e-07,
+      "loss": 1.9179,
+      "step": 5635
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.780691299165673e-07,
+      "loss": 1.7902,
+      "step": 5636
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.780294000794596e-07,
+      "loss": 1.8085,
+      "step": 5637
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.779896702423519e-07,
+      "loss": 1.5056,
+      "step": 5638
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.779499404052443e-07,
+      "loss": 1.6613,
+      "step": 5639
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.779102105681366e-07,
+      "loss": 1.9514,
+      "step": 5640
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.77870480731029e-07,
+      "loss": 1.8943,
+      "step": 5641
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.778307508939212e-07,
+      "loss": 1.8118,
+      "step": 5642
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.777910210568136e-07,
+      "loss": 2.0043,
+      "step": 5643
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.77751291219706e-07,
+      "loss": 1.5895,
+      "step": 5644
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.777115613825983e-07,
+      "loss": 1.7034,
+      "step": 5645
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.776718315454907e-07,
+      "loss": 1.6333,
+      "step": 5646
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.77632101708383e-07,
+      "loss": 1.5478,
+      "step": 5647
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.775923718712753e-07,
+      "loss": 1.8833,
+      "step": 5648
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.775526420341676e-07,
+      "loss": 1.5774,
+      "step": 5649
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.775129121970599e-07,
+      "loss": 1.462,
+      "step": 5650
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.774731823599523e-07,
+      "loss": 1.8822,
+      "step": 5651
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.774334525228446e-07,
+      "loss": 1.8511,
+      "step": 5652
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.77393722685737e-07,
+      "loss": 1.6964,
+      "step": 5653
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.773539928486292e-07,
+      "loss": 1.6952,
+      "step": 5654
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.773142630115216e-07,
+      "loss": 1.8491,
+      "step": 5655
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.772745331744139e-07,
+      "loss": 1.6448,
+      "step": 5656
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.772348033373063e-07,
+      "loss": 2.0424,
+      "step": 5657
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.771950735001986e-07,
+      "loss": 2.0266,
+      "step": 5658
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.77155343663091e-07,
+      "loss": 1.9088,
+      "step": 5659
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.771156138259833e-07,
+      "loss": 1.8191,
+      "step": 5660
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.770758839888756e-07,
+      "loss": 1.9334,
+      "step": 5661
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.77036154151768e-07,
+      "loss": 1.8952,
+      "step": 5662
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.769964243146603e-07,
+      "loss": 1.6604,
+      "step": 5663
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.769566944775526e-07,
+      "loss": 1.6343,
+      "step": 5664
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.769169646404449e-07,
+      "loss": 1.5363,
+      "step": 5665
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.768772348033372e-07,
+      "loss": 1.7292,
+      "step": 5666
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.768375049662296e-07,
+      "loss": 1.578,
+      "step": 5667
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.767977751291219e-07,
+      "loss": 1.7467,
+      "step": 5668
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.767580452920143e-07,
+      "loss": 1.701,
+      "step": 5669
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.767183154549066e-07,
+      "loss": 1.648,
+      "step": 5670
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.76678585617799e-07,
+      "loss": 1.8133,
+      "step": 5671
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.766388557806912e-07,
+      "loss": 2.0515,
+      "step": 5672
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.765991259435836e-07,
+      "loss": 1.7398,
+      "step": 5673
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.76559396106476e-07,
+      "loss": 1.8031,
+      "step": 5674
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.765196662693683e-07,
+      "loss": 1.7219,
+      "step": 5675
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.764799364322607e-07,
+      "loss": 1.5603,
+      "step": 5676
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.764402065951529e-07,
+      "loss": 1.6461,
+      "step": 5677
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.764004767580452e-07,
+      "loss": 1.901,
+      "step": 5678
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.763607469209375e-07,
+      "loss": 1.5628,
+      "step": 5679
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.763210170838299e-07,
+      "loss": 1.6835,
+      "step": 5680
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.762812872467222e-07,
+      "loss": 1.8655,
+      "step": 5681
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.762415574096146e-07,
+      "loss": 2.0045,
+      "step": 5682
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.76201827572507e-07,
+      "loss": 1.6264,
+      "step": 5683
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.761620977353992e-07,
+      "loss": 2.2089,
+      "step": 5684
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.761223678982916e-07,
+      "loss": 1.7596,
+      "step": 5685
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.760826380611839e-07,
+      "loss": 1.6836,
+      "step": 5686
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.760429082240763e-07,
+      "loss": 1.7703,
+      "step": 5687
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.760031783869686e-07,
+      "loss": 1.9481,
+      "step": 5688
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.75963448549861e-07,
+      "loss": 1.8333,
+      "step": 5689
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.759237187127531e-07,
+      "loss": 1.7943,
+      "step": 5690
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.758839888756455e-07,
+      "loss": 1.88,
+      "step": 5691
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.758442590385379e-07,
+      "loss": 1.6543,
+      "step": 5692
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.758045292014302e-07,
+      "loss": 1.5208,
+      "step": 5693
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.757647993643226e-07,
+      "loss": 1.4552,
+      "step": 5694
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.757250695272149e-07,
+      "loss": 1.7345,
+      "step": 5695
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.756853396901072e-07,
+      "loss": 2.2104,
+      "step": 5696
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.756456098529995e-07,
+      "loss": 1.849,
+      "step": 5697
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.756058800158919e-07,
+      "loss": 2.0532,
+      "step": 5698
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.755661501787843e-07,
+      "loss": 1.916,
+      "step": 5699
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.755264203416766e-07,
+      "loss": 1.5505,
+      "step": 5700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.75486690504569e-07,
+      "loss": 1.7448,
+      "step": 5701
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.754469606674612e-07,
+      "loss": 1.5704,
+      "step": 5702
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.754072308303535e-07,
+      "loss": 1.7277,
+      "step": 5703
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.753675009932458e-07,
+      "loss": 1.7301,
+      "step": 5704
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.753277711561382e-07,
+      "loss": 1.4524,
+      "step": 5705
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.752880413190306e-07,
+      "loss": 1.8037,
+      "step": 5706
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.752483114819229e-07,
+      "loss": 1.7992,
+      "step": 5707
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.752085816448152e-07,
+      "loss": 1.7319,
+      "step": 5708
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.751688518077075e-07,
+      "loss": 1.9306,
+      "step": 5709
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.751291219705999e-07,
+      "loss": 1.3679,
+      "step": 5710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.750893921334922e-07,
+      "loss": 1.9444,
+      "step": 5711
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.750496622963846e-07,
+      "loss": 1.8432,
+      "step": 5712
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.75009932459277e-07,
+      "loss": 1.6866,
+      "step": 5713
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.749702026221692e-07,
+      "loss": 1.5085,
+      "step": 5714
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.749304727850616e-07,
+      "loss": 1.8466,
+      "step": 5715
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.748907429479538e-07,
+      "loss": 1.5317,
+      "step": 5716
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.748510131108462e-07,
+      "loss": 1.924,
+      "step": 5717
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.748112832737385e-07,
+      "loss": 1.8489,
+      "step": 5718
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.747715534366309e-07,
+      "loss": 1.829,
+      "step": 5719
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.747318235995231e-07,
+      "loss": 1.7665,
+      "step": 5720
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.746920937624155e-07,
+      "loss": 1.694,
+      "step": 5721
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.746523639253079e-07,
+      "loss": 1.6494,
+      "step": 5722
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.746126340882002e-07,
+      "loss": 1.611,
+      "step": 5723
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.745729042510926e-07,
+      "loss": 1.9485,
+      "step": 5724
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.745331744139849e-07,
+      "loss": 1.7934,
+      "step": 5725
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.744934445768772e-07,
+      "loss": 1.8736,
+      "step": 5726
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.744537147397695e-07,
+      "loss": 1.7,
+      "step": 5727
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.744139849026619e-07,
+      "loss": 1.7266,
+      "step": 5728
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.743742550655543e-07,
+      "loss": 1.4278,
+      "step": 5729
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.743345252284465e-07,
+      "loss": 1.6796,
+      "step": 5730
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.742947953913389e-07,
+      "loss": 1.1396,
+      "step": 5731
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.742550655542311e-07,
+      "loss": 1.7416,
+      "step": 5732
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.742153357171235e-07,
+      "loss": 1.7441,
+      "step": 5733
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.741756058800158e-07,
+      "loss": 1.8338,
+      "step": 5734
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.741358760429082e-07,
+      "loss": 1.9439,
+      "step": 5735
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.740961462058006e-07,
+      "loss": 1.5627,
+      "step": 5736
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.740564163686929e-07,
+      "loss": 2.0143,
+      "step": 5737
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.740166865315852e-07,
+      "loss": 1.6949,
+      "step": 5738
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.739769566944775e-07,
+      "loss": 1.6923,
+      "step": 5739
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.739372268573699e-07,
+      "loss": 2.0423,
+      "step": 5740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.738974970202622e-07,
+      "loss": 2.0088,
+      "step": 5741
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.738577671831546e-07,
+      "loss": 1.7082,
+      "step": 5742
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.738180373460468e-07,
+      "loss": 1.7318,
+      "step": 5743
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.737783075089391e-07,
+      "loss": 1.9391,
+      "step": 5744
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.737385776718315e-07,
+      "loss": 2.0204,
+      "step": 5745
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.736988478347238e-07,
+      "loss": 1.5827,
+      "step": 5746
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.736591179976162e-07,
+      "loss": 1.9721,
+      "step": 5747
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.736193881605085e-07,
+      "loss": 1.6694,
+      "step": 5748
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.735796583234009e-07,
+      "loss": 1.8235,
+      "step": 5749
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.735399284862931e-07,
+      "loss": 1.9622,
+      "step": 5750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.735001986491855e-07,
+      "loss": 1.5261,
+      "step": 5751
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.734604688120779e-07,
+      "loss": 1.7383,
+      "step": 5752
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.734207389749702e-07,
+      "loss": 1.985,
+      "step": 5753
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.733810091378626e-07,
+      "loss": 1.9317,
+      "step": 5754
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.733412793007549e-07,
+      "loss": 1.6913,
+      "step": 5755
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.733015494636471e-07,
+      "loss": 1.8008,
+      "step": 5756
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.732618196265394e-07,
+      "loss": 1.5264,
+      "step": 5757
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.732220897894318e-07,
+      "loss": 1.9827,
+      "step": 5758
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.731823599523241e-07,
+      "loss": 1.888,
+      "step": 5759
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.731426301152165e-07,
+      "loss": 1.4037,
+      "step": 5760
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.731029002781089e-07,
+      "loss": 1.9514,
+      "step": 5761
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.730631704410011e-07,
+      "loss": 1.8484,
+      "step": 5762
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.730234406038935e-07,
+      "loss": 1.9935,
+      "step": 5763
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.729837107667858e-07,
+      "loss": 1.4354,
+      "step": 5764
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.729439809296782e-07,
+      "loss": 1.8299,
+      "step": 5765
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.729042510925705e-07,
+      "loss": 2.1728,
+      "step": 5766
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.728645212554629e-07,
+      "loss": 1.6718,
+      "step": 5767
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.728247914183552e-07,
+      "loss": 1.7803,
+      "step": 5768
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.727850615812474e-07,
+      "loss": 1.5664,
+      "step": 5769
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.727453317441398e-07,
+      "loss": 1.764,
+      "step": 5770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.727056019070321e-07,
+      "loss": 1.6465,
+      "step": 5771
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.726658720699245e-07,
+      "loss": 1.7899,
+      "step": 5772
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.726261422328168e-07,
+      "loss": 1.9837,
+      "step": 5773
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.725864123957091e-07,
+      "loss": 1.9775,
+      "step": 5774
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.725466825586014e-07,
+      "loss": 1.7754,
+      "step": 5775
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.725069527214938e-07,
+      "loss": 1.6052,
+      "step": 5776
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.724672228843862e-07,
+      "loss": 1.7109,
+      "step": 5777
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.724274930472785e-07,
+      "loss": 1.7562,
+      "step": 5778
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.723877632101709e-07,
+      "loss": 1.5696,
+      "step": 5779
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.723480333730631e-07,
+      "loss": 1.7164,
+      "step": 5780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.723083035359555e-07,
+      "loss": 1.9149,
+      "step": 5781
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.722685736988477e-07,
+      "loss": 1.5229,
+      "step": 5782
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.722288438617401e-07,
+      "loss": 1.9021,
+      "step": 5783
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.721891140246325e-07,
+      "loss": 1.9239,
+      "step": 5784
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.721493841875248e-07,
+      "loss": 1.6505,
+      "step": 5785
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.721096543504171e-07,
+      "loss": 1.7678,
+      "step": 5786
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.720699245133094e-07,
+      "loss": 1.7578,
+      "step": 5787
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.720301946762018e-07,
+      "loss": 1.983,
+      "step": 5788
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.719904648390941e-07,
+      "loss": 1.6095,
+      "step": 5789
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.719507350019865e-07,
+      "loss": 1.8753,
+      "step": 5790
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.719110051648789e-07,
+      "loss": 1.7107,
+      "step": 5791
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.718712753277711e-07,
+      "loss": 1.8931,
+      "step": 5792
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.718315454906635e-07,
+      "loss": 1.7077,
+      "step": 5793
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.717918156535558e-07,
+      "loss": 1.7221,
+      "step": 5794
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.717520858164482e-07,
+      "loss": 1.3905,
+      "step": 5795
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.717123559793404e-07,
+      "loss": 1.4395,
+      "step": 5796
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.716726261422328e-07,
+      "loss": 1.7614,
+      "step": 5797
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.71632896305125e-07,
+      "loss": 1.6191,
+      "step": 5798
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.715931664680174e-07,
+      "loss": 1.5108,
+      "step": 5799
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.715534366309098e-07,
+      "loss": 1.6236,
+      "step": 5800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.715137067938021e-07,
+      "loss": 1.7944,
+      "step": 5801
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.714739769566945e-07,
+      "loss": 1.3005,
+      "step": 5802
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.714342471195868e-07,
+      "loss": 1.5843,
+      "step": 5803
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.713945172824791e-07,
+      "loss": 1.6131,
+      "step": 5804
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.713547874453714e-07,
+      "loss": 1.6062,
+      "step": 5805
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.713150576082638e-07,
+      "loss": 1.7784,
+      "step": 5806
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.712753277711562e-07,
+      "loss": 1.4413,
+      "step": 5807
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.712355979340485e-07,
+      "loss": 1.7132,
+      "step": 5808
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.711958680969408e-07,
+      "loss": 1.7089,
+      "step": 5809
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.71156138259833e-07,
+      "loss": 1.719,
+      "step": 5810
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.711164084227254e-07,
+      "loss": 1.6907,
+      "step": 5811
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.710766785856177e-07,
+      "loss": 1.6626,
+      "step": 5812
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.710369487485101e-07,
+      "loss": 1.6494,
+      "step": 5813
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.709972189114025e-07,
+      "loss": 1.4798,
+      "step": 5814
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.709574890742948e-07,
+      "loss": 1.584,
+      "step": 5815
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.709177592371871e-07,
+      "loss": 1.7173,
+      "step": 5816
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.708780294000794e-07,
+      "loss": 1.7126,
+      "step": 5817
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.708382995629718e-07,
+      "loss": 1.9637,
+      "step": 5818
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.707985697258641e-07,
+      "loss": 1.761,
+      "step": 5819
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.707588398887565e-07,
+      "loss": 1.6854,
+      "step": 5820
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.707191100516488e-07,
+      "loss": 1.8236,
+      "step": 5821
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.70679380214541e-07,
+      "loss": 1.6259,
+      "step": 5822
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.706396503774334e-07,
+      "loss": 1.8426,
+      "step": 5823
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.705999205403257e-07,
+      "loss": 1.6997,
+      "step": 5824
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.705601907032181e-07,
+      "loss": 1.7664,
+      "step": 5825
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.705204608661104e-07,
+      "loss": 1.3226,
+      "step": 5826
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.704807310290028e-07,
+      "loss": 1.7745,
+      "step": 5827
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.70441001191895e-07,
+      "loss": 2.1815,
+      "step": 5828
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.704012713547874e-07,
+      "loss": 2.1522,
+      "step": 5829
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.703615415176798e-07,
+      "loss": 1.9801,
+      "step": 5830
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.703218116805721e-07,
+      "loss": 1.9219,
+      "step": 5831
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.702820818434645e-07,
+      "loss": 1.6762,
+      "step": 5832
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.702423520063568e-07,
+      "loss": 1.7014,
+      "step": 5833
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.702026221692491e-07,
+      "loss": 2.2314,
+      "step": 5834
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.701628923321413e-07,
+      "loss": 1.7563,
+      "step": 5835
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.701231624950337e-07,
+      "loss": 1.446,
+      "step": 5836
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.70083432657926e-07,
+      "loss": 1.4869,
+      "step": 5837
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.700437028208184e-07,
+      "loss": 1.9069,
+      "step": 5838
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.700039729837108e-07,
+      "loss": 1.909,
+      "step": 5839
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.69964243146603e-07,
+      "loss": 1.782,
+      "step": 5840
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.699245133094954e-07,
+      "loss": 1.6962,
+      "step": 5841
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.698847834723877e-07,
+      "loss": 1.6838,
+      "step": 5842
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.698450536352801e-07,
+      "loss": 1.783,
+      "step": 5843
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.698053237981724e-07,
+      "loss": 1.4477,
+      "step": 5844
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.697655939610648e-07,
+      "loss": 1.7632,
+      "step": 5845
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.697258641239571e-07,
+      "loss": 1.5319,
+      "step": 5846
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.696861342868494e-07,
+      "loss": 1.8673,
+      "step": 5847
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.696464044497417e-07,
+      "loss": 1.8339,
+      "step": 5848
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.69606674612634e-07,
+      "loss": 1.879,
+      "step": 5849
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.695669447755264e-07,
+      "loss": 1.9356,
+      "step": 5850
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.695272149384187e-07,
+      "loss": 1.6972,
+      "step": 5851
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.69487485101311e-07,
+      "loss": 1.7559,
+      "step": 5852
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.694477552642033e-07,
+      "loss": 1.9934,
+      "step": 5853
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.694080254270957e-07,
+      "loss": 1.692,
+      "step": 5854
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.693682955899881e-07,
+      "loss": 1.5628,
+      "step": 5855
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.693285657528804e-07,
+      "loss": 1.6448,
+      "step": 5856
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.692888359157728e-07,
+      "loss": 1.773,
+      "step": 5857
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.69249106078665e-07,
+      "loss": 1.9612,
+      "step": 5858
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.692093762415574e-07,
+      "loss": 1.6138,
+      "step": 5859
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.691696464044497e-07,
+      "loss": 1.6949,
+      "step": 5860
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.691299165673421e-07,
+      "loss": 2.0657,
+      "step": 5861
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.690901867302344e-07,
+      "loss": 1.8606,
+      "step": 5862
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.690504568931267e-07,
+      "loss": 1.5913,
+      "step": 5863
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.69010727056019e-07,
+      "loss": 1.4716,
+      "step": 5864
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.689709972189113e-07,
+      "loss": 1.3924,
+      "step": 5865
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.689312673818037e-07,
+      "loss": 2.0333,
+      "step": 5866
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.68891537544696e-07,
+      "loss": 1.6877,
+      "step": 5867
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.688518077075884e-07,
+      "loss": 1.6215,
+      "step": 5868
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.688120778704808e-07,
+      "loss": 1.7267,
+      "step": 5869
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.68772348033373e-07,
+      "loss": 1.7116,
+      "step": 5870
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.687326181962654e-07,
+      "loss": 1.5012,
+      "step": 5871
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.686928883591577e-07,
+      "loss": 1.6377,
+      "step": 5872
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.686531585220501e-07,
+      "loss": 1.8253,
+      "step": 5873
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.686134286849424e-07,
+      "loss": 1.8196,
+      "step": 5874
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.685736988478347e-07,
+      "loss": 1.7416,
+      "step": 5875
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.685339690107269e-07,
+      "loss": 1.6624,
+      "step": 5876
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.684942391736193e-07,
+      "loss": 1.7677,
+      "step": 5877
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.684545093365117e-07,
+      "loss": 1.9112,
+      "step": 5878
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.68414779499404e-07,
+      "loss": 1.5387,
+      "step": 5879
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.683750496622964e-07,
+      "loss": 1.7533,
+      "step": 5880
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.683353198251887e-07,
+      "loss": 1.5712,
+      "step": 5881
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.68295589988081e-07,
+      "loss": 1.8361,
+      "step": 5882
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.682558601509733e-07,
+      "loss": 1.7501,
+      "step": 5883
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.682161303138657e-07,
+      "loss": 1.5885,
+      "step": 5884
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.681764004767581e-07,
+      "loss": 1.5116,
+      "step": 5885
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.681366706396504e-07,
+      "loss": 1.6506,
+      "step": 5886
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.680969408025428e-07,
+      "loss": 1.7206,
+      "step": 5887
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.680572109654349e-07,
+      "loss": 1.6257,
+      "step": 5888
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.680174811283273e-07,
+      "loss": 1.4546,
+      "step": 5889
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.679777512912196e-07,
+      "loss": 1.7012,
+      "step": 5890
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.67938021454112e-07,
+      "loss": 1.7665,
+      "step": 5891
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.678982916170044e-07,
+      "loss": 1.7928,
+      "step": 5892
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.678585617798967e-07,
+      "loss": 1.6442,
+      "step": 5893
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.67818831942789e-07,
+      "loss": 1.9252,
+      "step": 5894
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.677791021056813e-07,
+      "loss": 1.5176,
+      "step": 5895
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.677393722685737e-07,
+      "loss": 1.4765,
+      "step": 5896
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.67699642431466e-07,
+      "loss": 1.6264,
+      "step": 5897
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.676599125943584e-07,
+      "loss": 2.2619,
+      "step": 5898
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.676201827572507e-07,
+      "loss": 1.5287,
+      "step": 5899
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.67580452920143e-07,
+      "loss": 2.1262,
+      "step": 5900
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.675407230830353e-07,
+      "loss": 1.296,
+      "step": 5901
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.675009932459276e-07,
+      "loss": 1.7076,
+      "step": 5902
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.6746126340882e-07,
+      "loss": 1.5625,
+      "step": 5903
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.674215335717123e-07,
+      "loss": 1.3386,
+      "step": 5904
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.673818037346047e-07,
+      "loss": 2.1328,
+      "step": 5905
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.673420738974969e-07,
+      "loss": 1.7634,
+      "step": 5906
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.673023440603893e-07,
+      "loss": 1.7398,
+      "step": 5907
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.672626142232817e-07,
+      "loss": 1.8218,
+      "step": 5908
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.67222884386174e-07,
+      "loss": 1.5154,
+      "step": 5909
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.671831545490664e-07,
+      "loss": 1.5683,
+      "step": 5910
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.671434247119587e-07,
+      "loss": 1.9855,
+      "step": 5911
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.67103694874851e-07,
+      "loss": 1.8368,
+      "step": 5912
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.670639650377433e-07,
+      "loss": 1.8679,
+      "step": 5913
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.670242352006356e-07,
+      "loss": 1.3222,
+      "step": 5914
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.669845053635279e-07,
+      "loss": 1.5365,
+      "step": 5915
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.669447755264203e-07,
+      "loss": 1.7904,
+      "step": 5916
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.669050456893127e-07,
+      "loss": 2.2087,
+      "step": 5917
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.668653158522049e-07,
+      "loss": 1.638,
+      "step": 5918
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.668255860150973e-07,
+      "loss": 1.6118,
+      "step": 5919
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.667858561779896e-07,
+      "loss": 2.0621,
+      "step": 5920
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.66746126340882e-07,
+      "loss": 1.5602,
+      "step": 5921
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.667063965037743e-07,
+      "loss": 1.3388,
+      "step": 5922
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.666666666666667e-07,
+      "loss": 1.956,
+      "step": 5923
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.66626936829559e-07,
+      "loss": 1.5818,
+      "step": 5924
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.665872069924513e-07,
+      "loss": 1.8482,
+      "step": 5925
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.665474771553437e-07,
+      "loss": 1.4161,
+      "step": 5926
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.66507747318236e-07,
+      "loss": 1.6712,
+      "step": 5927
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.664680174811283e-07,
+      "loss": 1.7373,
+      "step": 5928
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.664282876440206e-07,
+      "loss": 1.82,
+      "step": 5929
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.663885578069129e-07,
+      "loss": 1.6114,
+      "step": 5930
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.663488279698052e-07,
+      "loss": 2.2613,
+      "step": 5931
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.663090981326976e-07,
+      "loss": 2.1866,
+      "step": 5932
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.6626936829559e-07,
+      "loss": 1.7381,
+      "step": 5933
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.662296384584823e-07,
+      "loss": 1.8241,
+      "step": 5934
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.661899086213747e-07,
+      "loss": 1.7944,
+      "step": 5935
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.661501787842669e-07,
+      "loss": 1.7585,
+      "step": 5936
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.661104489471593e-07,
+      "loss": 1.7643,
+      "step": 5937
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.660707191100516e-07,
+      "loss": 1.8081,
+      "step": 5938
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.66030989272944e-07,
+      "loss": 1.6369,
+      "step": 5939
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.659912594358364e-07,
+      "loss": 1.6869,
+      "step": 5940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.659515295987286e-07,
+      "loss": 1.506,
+      "step": 5941
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.659117997616209e-07,
+      "loss": 1.4672,
+      "step": 5942
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.658720699245132e-07,
+      "loss": 1.9081,
+      "step": 5943
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.658323400874056e-07,
+      "loss": 1.4407,
+      "step": 5944
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.657926102502979e-07,
+      "loss": 1.5888,
+      "step": 5945
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.657528804131903e-07,
+      "loss": 1.7674,
+      "step": 5946
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.657131505760827e-07,
+      "loss": 1.5407,
+      "step": 5947
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.656734207389749e-07,
+      "loss": 1.6616,
+      "step": 5948
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.656336909018673e-07,
+      "loss": 1.7184,
+      "step": 5949
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.655939610647596e-07,
+      "loss": 1.8806,
+      "step": 5950
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.65554231227652e-07,
+      "loss": 1.4838,
+      "step": 5951
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.655145013905443e-07,
+      "loss": 1.6085,
+      "step": 5952
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.654747715534367e-07,
+      "loss": 1.7546,
+      "step": 5953
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.654350417163288e-07,
+      "loss": 1.6891,
+      "step": 5954
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.653953118792212e-07,
+      "loss": 1.4385,
+      "step": 5955
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.653555820421136e-07,
+      "loss": 2.0771,
+      "step": 5956
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.653158522050059e-07,
+      "loss": 1.8805,
+      "step": 5957
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.652761223678983e-07,
+      "loss": 1.5317,
+      "step": 5958
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.652363925307906e-07,
+      "loss": 2.3151,
+      "step": 5959
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.651966626936829e-07,
+      "loss": 1.8944,
+      "step": 5960
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.651569328565752e-07,
+      "loss": 1.8065,
+      "step": 5961
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.651172030194676e-07,
+      "loss": 1.8865,
+      "step": 5962
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.6507747318236e-07,
+      "loss": 1.7957,
+      "step": 5963
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.650377433452523e-07,
+      "loss": 1.8855,
+      "step": 5964
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.649980135081447e-07,
+      "loss": 1.5774,
+      "step": 5965
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.649582836710369e-07,
+      "loss": 1.514,
+      "step": 5966
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.649185538339292e-07,
+      "loss": 2.0103,
+      "step": 5967
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.648788239968215e-07,
+      "loss": 1.7004,
+      "step": 5968
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.648390941597139e-07,
+      "loss": 1.5381,
+      "step": 5969
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.647993643226063e-07,
+      "loss": 2.3468,
+      "step": 5970
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.647596344854986e-07,
+      "loss": 2.0911,
+      "step": 5971
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.647199046483909e-07,
+      "loss": 1.9188,
+      "step": 5972
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.646801748112832e-07,
+      "loss": 2.0217,
+      "step": 5973
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.646404449741756e-07,
+      "loss": 1.7383,
+      "step": 5974
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.646007151370679e-07,
+      "loss": 1.6453,
+      "step": 5975
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.645609852999603e-07,
+      "loss": 1.9331,
+      "step": 5976
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.645212554628526e-07,
+      "loss": 1.661,
+      "step": 5977
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.644815256257449e-07,
+      "loss": 1.8203,
+      "step": 5978
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.644417957886373e-07,
+      "loss": 1.8162,
+      "step": 5979
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.644020659515295e-07,
+      "loss": 1.5658,
+      "step": 5980
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.643623361144219e-07,
+      "loss": 1.5125,
+      "step": 5981
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.643226062773142e-07,
+      "loss": 1.746,
+      "step": 5982
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.642828764402066e-07,
+      "loss": 1.6762,
+      "step": 5983
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.642431466030988e-07,
+      "loss": 1.6416,
+      "step": 5984
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.642034167659912e-07,
+      "loss": 1.8911,
+      "step": 5985
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.641636869288836e-07,
+      "loss": 1.6677,
+      "step": 5986
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.641239570917759e-07,
+      "loss": 2.0913,
+      "step": 5987
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.640842272546683e-07,
+      "loss": 2.0448,
+      "step": 5988
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.640444974175606e-07,
+      "loss": 1.7497,
+      "step": 5989
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.640047675804529e-07,
+      "loss": 1.8452,
+      "step": 5990
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.639650377433452e-07,
+      "loss": 1.632,
+      "step": 5991
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.639253079062376e-07,
+      "loss": 2.1628,
+      "step": 5992
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.6388557806913e-07,
+      "loss": 1.6554,
+      "step": 5993
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.638458482320222e-07,
+      "loss": 1.4668,
+      "step": 5994
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.638061183949146e-07,
+      "loss": 1.9231,
+      "step": 5995
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.637663885578068e-07,
+      "loss": 1.3966,
+      "step": 5996
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.637266587206992e-07,
+      "loss": 1.387,
+      "step": 5997
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.636869288835915e-07,
+      "loss": 1.5679,
+      "step": 5998
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.636471990464839e-07,
+      "loss": 1.7942,
+      "step": 5999
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.636074692093762e-07,
+      "loss": 2.06,
+      "step": 6000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.635677393722686e-07,
+      "loss": 1.6561,
+      "step": 6001
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.635280095351609e-07,
+      "loss": 1.7403,
+      "step": 6002
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.634882796980532e-07,
+      "loss": 1.934,
+      "step": 6003
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.634485498609456e-07,
+      "loss": 1.8085,
+      "step": 6004
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.634088200238379e-07,
+      "loss": 1.6007,
+      "step": 6005
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.633690901867303e-07,
+      "loss": 1.4552,
+      "step": 6006
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.633293603496225e-07,
+      "loss": 1.8873,
+      "step": 6007
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 7.632896305125148e-07,
+      "loss": 1.6818,
+      "step": 6008
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.632499006754071e-07,
+      "loss": 1.6348,
+      "step": 6009
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.632101708382995e-07,
+      "loss": 1.8691,
+      "step": 6010
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.631704410011919e-07,
+      "loss": 2.2543,
+      "step": 6011
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.631307111640842e-07,
+      "loss": 1.8016,
+      "step": 6012
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.630909813269766e-07,
+      "loss": 1.5567,
+      "step": 6013
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.630512514898688e-07,
+      "loss": 1.6252,
+      "step": 6014
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.630115216527612e-07,
+      "loss": 2.0453,
+      "step": 6015
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.629717918156535e-07,
+      "loss": 2.1077,
+      "step": 6016
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.629320619785459e-07,
+      "loss": 1.6498,
+      "step": 6017
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.628923321414383e-07,
+      "loss": 1.849,
+      "step": 6018
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.628526023043306e-07,
+      "loss": 1.7386,
+      "step": 6019
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.628128724672228e-07,
+      "loss": 1.7332,
+      "step": 6020
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.627731426301151e-07,
+      "loss": 1.7533,
+      "step": 6021
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.627334127930075e-07,
+      "loss": 1.9382,
+      "step": 6022
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.626936829558998e-07,
+      "loss": 1.5859,
+      "step": 6023
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.626539531187922e-07,
+      "loss": 1.6888,
+      "step": 6024
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.626142232816846e-07,
+      "loss": 1.4881,
+      "step": 6025
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.625744934445768e-07,
+      "loss": 1.8653,
+      "step": 6026
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.625347636074692e-07,
+      "loss": 1.8788,
+      "step": 6027
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.624950337703615e-07,
+      "loss": 1.7718,
+      "step": 6028
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.624553039332539e-07,
+      "loss": 1.8653,
+      "step": 6029
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.624155740961462e-07,
+      "loss": 1.7603,
+      "step": 6030
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.623758442590386e-07,
+      "loss": 1.903,
+      "step": 6031
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.623361144219308e-07,
+      "loss": 1.699,
+      "step": 6032
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.622963845848231e-07,
+      "loss": 1.9865,
+      "step": 6033
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.622566547477155e-07,
+      "loss": 1.6584,
+      "step": 6034
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.622169249106078e-07,
+      "loss": 1.674,
+      "step": 6035
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.621771950735002e-07,
+      "loss": 1.6018,
+      "step": 6036
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.621374652363925e-07,
+      "loss": 1.8231,
+      "step": 6037
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.620977353992848e-07,
+      "loss": 1.7686,
+      "step": 6038
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.620580055621771e-07,
+      "loss": 1.6147,
+      "step": 6039
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.620182757250695e-07,
+      "loss": 1.8383,
+      "step": 6040
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.619785458879619e-07,
+      "loss": 1.7395,
+      "step": 6041
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.619388160508542e-07,
+      "loss": 1.7576,
+      "step": 6042
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.618990862137465e-07,
+      "loss": 1.4559,
+      "step": 6043
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.618593563766388e-07,
+      "loss": 1.5967,
+      "step": 6044
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.618196265395312e-07,
+      "loss": 1.7066,
+      "step": 6045
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 7.617798967024234e-07,
+      "loss": 1.4849,
+      "step": 6046
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.617401668653158e-07,
+      "loss": 1.6097,
+      "step": 6047
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.617004370282082e-07,
+      "loss": 2.091,
+      "step": 6048
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.616607071911004e-07,
+      "loss": 1.7323,
+      "step": 6049
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.616209773539928e-07,
+      "loss": 1.7549,
+      "step": 6050
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.615812475168851e-07,
+      "loss": 1.7471,
+      "step": 6051
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.615415176797775e-07,
+      "loss": 1.392,
+      "step": 6052
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.615017878426698e-07,
+      "loss": 1.3033,
+      "step": 6053
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.614620580055622e-07,
+      "loss": 1.7521,
+      "step": 6054
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.614223281684544e-07,
+      "loss": 1.7967,
+      "step": 6055
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.613825983313468e-07,
+      "loss": 1.4971,
+      "step": 6056
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.613428684942392e-07,
+      "loss": 1.821,
+      "step": 6057
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.613031386571315e-07,
+      "loss": 1.99,
+      "step": 6058
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.612634088200238e-07,
+      "loss": 1.7791,
+      "step": 6059
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.612236789829161e-07,
+      "loss": 1.8686,
+      "step": 6060
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.611839491458084e-07,
+      "loss": 1.8583,
+      "step": 6061
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.611442193087007e-07,
+      "loss": 1.6042,
+      "step": 6062
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.611044894715931e-07,
+      "loss": 1.6286,
+      "step": 6063
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.610647596344855e-07,
+      "loss": 1.4628,
+      "step": 6064
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.610250297973778e-07,
+      "loss": 1.734,
+      "step": 6065
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.609852999602702e-07,
+      "loss": 1.6589,
+      "step": 6066
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.609455701231624e-07,
+      "loss": 1.4143,
+      "step": 6067
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.609058402860548e-07,
+      "loss": 1.4015,
+      "step": 6068
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.608661104489471e-07,
+      "loss": 1.7768,
+      "step": 6069
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.608263806118395e-07,
+      "loss": 2.0338,
+      "step": 6070
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.607866507747319e-07,
+      "loss": 1.8509,
+      "step": 6071
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.607469209376242e-07,
+      "loss": 1.8647,
+      "step": 6072
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.607071911005164e-07,
+      "loss": 1.6096,
+      "step": 6073
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.606674612634087e-07,
+      "loss": 1.571,
+      "step": 6074
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.606277314263011e-07,
+      "loss": 1.4095,
+      "step": 6075
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.605880015891934e-07,
+      "loss": 2.0765,
+      "step": 6076
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.605482717520858e-07,
+      "loss": 1.4702,
+      "step": 6077
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.605085419149781e-07,
+      "loss": 1.8719,
+      "step": 6078
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.604688120778704e-07,
+      "loss": 2.0288,
+      "step": 6079
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.604290822407628e-07,
+      "loss": 1.7382,
+      "step": 6080
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.603893524036551e-07,
+      "loss": 1.8674,
+      "step": 6081
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.603496225665475e-07,
+      "loss": 1.5057,
+      "step": 6082
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.603098927294398e-07,
+      "loss": 1.7774,
+      "step": 6083
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.602701628923322e-07,
+      "loss": 1.6893,
+      "step": 6084
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.602304330552243e-07,
+      "loss": 1.9376,
+      "step": 6085
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.601907032181167e-07,
+      "loss": 1.7613,
+      "step": 6086
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.60150973381009e-07,
+      "loss": 1.833,
+      "step": 6087
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.601112435439014e-07,
+      "loss": 1.3484,
+      "step": 6088
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.600715137067938e-07,
+      "loss": 1.5942,
+      "step": 6089
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.600317838696861e-07,
+      "loss": 1.8997,
+      "step": 6090
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.599920540325784e-07,
+      "loss": 1.6643,
+      "step": 6091
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.599523241954707e-07,
+      "loss": 1.6496,
+      "step": 6092
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.599125943583631e-07,
+      "loss": 1.6141,
+      "step": 6093
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.598728645212554e-07,
+      "loss": 1.5035,
+      "step": 6094
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.598331346841478e-07,
+      "loss": 1.5719,
+      "step": 6095
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.597934048470402e-07,
+      "loss": 1.7769,
+      "step": 6096
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.597536750099324e-07,
+      "loss": 1.5054,
+      "step": 6097
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.597139451728248e-07,
+      "loss": 1.8423,
+      "step": 6098
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.59674215335717e-07,
+      "loss": 1.8586,
+      "step": 6099
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.596344854986094e-07,
+      "loss": 1.7195,
+      "step": 6100
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.595947556615017e-07,
+      "loss": 1.5542,
+      "step": 6101
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.595550258243941e-07,
+      "loss": 1.7748,
+      "step": 6102
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.595152959872864e-07,
+      "loss": 1.6353,
+      "step": 6103
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.594755661501787e-07,
+      "loss": 1.5774,
+      "step": 6104
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.594358363130711e-07,
+      "loss": 2.0752,
+      "step": 6105
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.593961064759634e-07,
+      "loss": 1.4835,
+      "step": 6106
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.593563766388558e-07,
+      "loss": 1.675,
+      "step": 6107
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.593166468017481e-07,
+      "loss": 1.8736,
+      "step": 6108
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.592769169646404e-07,
+      "loss": 1.6543,
+      "step": 6109
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.592371871275327e-07,
+      "loss": 1.71,
+      "step": 6110
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.591974572904251e-07,
+      "loss": 1.4985,
+      "step": 6111
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.591577274533174e-07,
+      "loss": 1.762,
+      "step": 6112
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.591179976162097e-07,
+      "loss": 1.8286,
+      "step": 6113
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.590782677791021e-07,
+      "loss": 2.2739,
+      "step": 6114
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.590385379419943e-07,
+      "loss": 1.5962,
+      "step": 6115
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.589988081048867e-07,
+      "loss": 1.902,
+      "step": 6116
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.58959078267779e-07,
+      "loss": 1.8657,
+      "step": 6117
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.589193484306714e-07,
+      "loss": 1.888,
+      "step": 6118
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.588796185935638e-07,
+      "loss": 1.3209,
+      "step": 6119
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.588398887564561e-07,
+      "loss": 2.0683,
+      "step": 6120
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.588001589193484e-07,
+      "loss": 2.0286,
+      "step": 6121
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.587604290822407e-07,
+      "loss": 1.8192,
+      "step": 6122
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.587206992451331e-07,
+      "loss": 1.6449,
+      "step": 6123
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.586809694080254e-07,
+      "loss": 1.7564,
+      "step": 6124
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.586412395709177e-07,
+      "loss": 1.6303,
+      "step": 6125
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.5860150973381e-07,
+      "loss": 1.3177,
+      "step": 6126
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.585617798967023e-07,
+      "loss": 2.0998,
+      "step": 6127
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.585220500595947e-07,
+      "loss": 2.0622,
+      "step": 6128
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.58482320222487e-07,
+      "loss": 1.8665,
+      "step": 6129
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.584425903853794e-07,
+      "loss": 1.4449,
+      "step": 6130
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.584028605482717e-07,
+      "loss": 1.8214,
+      "step": 6131
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.583631307111641e-07,
+      "loss": 1.5681,
+      "step": 6132
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.583234008740563e-07,
+      "loss": 1.5971,
+      "step": 6133
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.582836710369487e-07,
+      "loss": 2.0222,
+      "step": 6134
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.582439411998411e-07,
+      "loss": 2.1121,
+      "step": 6135
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.582042113627334e-07,
+      "loss": 1.8967,
+      "step": 6136
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.581644815256258e-07,
+      "loss": 1.7432,
+      "step": 6137
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.581247516885181e-07,
+      "loss": 1.5808,
+      "step": 6138
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.580850218514103e-07,
+      "loss": 1.6649,
+      "step": 6139
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.580452920143026e-07,
+      "loss": 1.8622,
+      "step": 6140
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.58005562177195e-07,
+      "loss": 1.5978,
+      "step": 6141
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.579658323400874e-07,
+      "loss": 1.6048,
+      "step": 6142
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.579261025029797e-07,
+      "loss": 1.6609,
+      "step": 6143
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.578863726658721e-07,
+      "loss": 1.8315,
+      "step": 6144
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.578466428287643e-07,
+      "loss": 1.676,
+      "step": 6145
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.578069129916567e-07,
+      "loss": 1.8274,
+      "step": 6146
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.57767183154549e-07,
+      "loss": 1.998,
+      "step": 6147
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.577274533174414e-07,
+      "loss": 1.9426,
+      "step": 6148
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.576877234803338e-07,
+      "loss": 1.5109,
+      "step": 6149
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.576479936432261e-07,
+      "loss": 1.8583,
+      "step": 6150
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.576082638061183e-07,
+      "loss": 1.6966,
+      "step": 6151
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.575685339690106e-07,
+      "loss": 1.9745,
+      "step": 6152
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.57528804131903e-07,
+      "loss": 1.7409,
+      "step": 6153
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.574890742947953e-07,
+      "loss": 1.6258,
+      "step": 6154
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.574493444576877e-07,
+      "loss": 1.6208,
+      "step": 6155
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.5740961462058e-07,
+      "loss": 1.6261,
+      "step": 6156
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.573698847834723e-07,
+      "loss": 1.8577,
+      "step": 6157
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.573301549463647e-07,
+      "loss": 1.5856,
+      "step": 6158
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.57290425109257e-07,
+      "loss": 1.5682,
+      "step": 6159
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.572506952721494e-07,
+      "loss": 1.9296,
+      "step": 6160
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.572109654350417e-07,
+      "loss": 1.5188,
+      "step": 6161
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.571712355979341e-07,
+      "loss": 1.6481,
+      "step": 6162
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.571315057608263e-07,
+      "loss": 1.8181,
+      "step": 6163
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.570917759237187e-07,
+      "loss": 1.7858,
+      "step": 6164
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.57052046086611e-07,
+      "loss": 1.8351,
+      "step": 6165
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.570123162495033e-07,
+      "loss": 1.4591,
+      "step": 6166
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.569725864123957e-07,
+      "loss": 1.5888,
+      "step": 6167
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.56932856575288e-07,
+      "loss": 1.4516,
+      "step": 6168
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.568931267381803e-07,
+      "loss": 1.879,
+      "step": 6169
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.568533969010726e-07,
+      "loss": 1.6477,
+      "step": 6170
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.56813667063965e-07,
+      "loss": 1.7927,
+      "step": 6171
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.567739372268573e-07,
+      "loss": 1.4793,
+      "step": 6172
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.567342073897497e-07,
+      "loss": 1.7773,
+      "step": 6173
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.566944775526421e-07,
+      "loss": 1.6705,
+      "step": 6174
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.566547477155343e-07,
+      "loss": 1.662,
+      "step": 6175
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.566150178784267e-07,
+      "loss": 1.4182,
+      "step": 6176
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.56575288041319e-07,
+      "loss": 1.7122,
+      "step": 6177
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.565355582042113e-07,
+      "loss": 1.9518,
+      "step": 6178
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.564958283671036e-07,
+      "loss": 1.2855,
+      "step": 6179
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.56456098529996e-07,
+      "loss": 1.8002,
+      "step": 6180
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.564163686928883e-07,
+      "loss": 1.8533,
+      "step": 6181
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.563766388557806e-07,
+      "loss": 1.383,
+      "step": 6182
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.56336909018673e-07,
+      "loss": 1.6443,
+      "step": 6183
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.562971791815653e-07,
+      "loss": 1.7928,
+      "step": 6184
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.562574493444577e-07,
+      "loss": 1.6057,
+      "step": 6185
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.5621771950735e-07,
+      "loss": 1.5928,
+      "step": 6186
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.561779896702423e-07,
+      "loss": 1.8973,
+      "step": 6187
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.561382598331346e-07,
+      "loss": 1.8184,
+      "step": 6188
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.56098529996027e-07,
+      "loss": 1.5562,
+      "step": 6189
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.560588001589194e-07,
+      "loss": 1.8983,
+      "step": 6190
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.560190703218116e-07,
+      "loss": 1.6676,
+      "step": 6191
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.55979340484704e-07,
+      "loss": 1.9041,
+      "step": 6192
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.559396106475962e-07,
+      "loss": 1.8943,
+      "step": 6193
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.558998808104886e-07,
+      "loss": 1.729,
+      "step": 6194
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.558601509733809e-07,
+      "loss": 1.7096,
+      "step": 6195
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.558204211362733e-07,
+      "loss": 1.6016,
+      "step": 6196
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.557806912991657e-07,
+      "loss": 1.5082,
+      "step": 6197
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.55740961462058e-07,
+      "loss": 1.8253,
+      "step": 6198
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.557012316249503e-07,
+      "loss": 1.7288,
+      "step": 6199
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.556615017878426e-07,
+      "loss": 1.7955,
+      "step": 6200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.55621771950735e-07,
+      "loss": 1.5478,
+      "step": 6201
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.555820421136273e-07,
+      "loss": 1.8174,
+      "step": 6202
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.555423122765197e-07,
+      "loss": 1.6519,
+      "step": 6203
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.555025824394121e-07,
+      "loss": 1.4099,
+      "step": 6204
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.554628526023042e-07,
+      "loss": 1.7907,
+      "step": 6205
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.554231227651966e-07,
+      "loss": 2.1243,
+      "step": 6206
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.553833929280889e-07,
+      "loss": 1.5194,
+      "step": 6207
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.553436630909813e-07,
+      "loss": 1.6685,
+      "step": 6208
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.553039332538736e-07,
+      "loss": 1.8619,
+      "step": 6209
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.55264203416766e-07,
+      "loss": 1.55,
+      "step": 6210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.552244735796582e-07,
+      "loss": 1.8622,
+      "step": 6211
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.551847437425506e-07,
+      "loss": 1.7755,
+      "step": 6212
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.55145013905443e-07,
+      "loss": 1.6095,
+      "step": 6213
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.551052840683353e-07,
+      "loss": 1.9114,
+      "step": 6214
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.550655542312277e-07,
+      "loss": 1.9124,
+      "step": 6215
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.5502582439412e-07,
+      "loss": 1.2424,
+      "step": 6216
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.549860945570122e-07,
+      "loss": 1.7663,
+      "step": 6217
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.549463647199045e-07,
+      "loss": 1.7878,
+      "step": 6218
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.549066348827969e-07,
+      "loss": 1.4629,
+      "step": 6219
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.548669050456893e-07,
+      "loss": 1.2884,
+      "step": 6220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.548271752085816e-07,
+      "loss": 2.1948,
+      "step": 6221
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.54787445371474e-07,
+      "loss": 1.9594,
+      "step": 6222
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.547477155343662e-07,
+      "loss": 1.98,
+      "step": 6223
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.547079856972586e-07,
+      "loss": 1.9507,
+      "step": 6224
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.546682558601509e-07,
+      "loss": 1.729,
+      "step": 6225
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.546285260230433e-07,
+      "loss": 1.7187,
+      "step": 6226
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.545887961859357e-07,
+      "loss": 2.0669,
+      "step": 6227
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.54549066348828e-07,
+      "loss": 1.5736,
+      "step": 6228
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.545093365117203e-07,
+      "loss": 1.6785,
+      "step": 6229
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.544696066746126e-07,
+      "loss": 2.035,
+      "step": 6230
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.544298768375049e-07,
+      "loss": 1.6296,
+      "step": 6231
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.543901470003972e-07,
+      "loss": 1.7397,
+      "step": 6232
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.543504171632896e-07,
+      "loss": 1.3702,
+      "step": 6233
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.543106873261819e-07,
+      "loss": 1.8333,
+      "step": 6234
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.542709574890742e-07,
+      "loss": 1.6686,
+      "step": 6235
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.542312276519666e-07,
+      "loss": 1.7387,
+      "step": 6236
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.541914978148589e-07,
+      "loss": 1.6025,
+      "step": 6237
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.541517679777513e-07,
+      "loss": 1.9401,
+      "step": 6238
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.541120381406436e-07,
+      "loss": 1.8293,
+      "step": 6239
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.54072308303536e-07,
+      "loss": 1.8164,
+      "step": 6240
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.540325784664282e-07,
+      "loss": 1.4234,
+      "step": 6241
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.539928486293206e-07,
+      "loss": 1.9647,
+      "step": 6242
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.53953118792213e-07,
+      "loss": 1.4583,
+      "step": 6243
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.539133889551052e-07,
+      "loss": 1.668,
+      "step": 6244
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.538736591179976e-07,
+      "loss": 1.93,
+      "step": 6245
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.538339292808899e-07,
+      "loss": 1.5433,
+      "step": 6246
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.537941994437822e-07,
+      "loss": 2.0768,
+      "step": 6247
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.537544696066745e-07,
+      "loss": 1.3051,
+      "step": 6248
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.537147397695669e-07,
+      "loss": 1.6501,
+      "step": 6249
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.536750099324592e-07,
+      "loss": 1.5408,
+      "step": 6250
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.536352800953516e-07,
+      "loss": 1.8134,
+      "step": 6251
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.53595550258244e-07,
+      "loss": 1.7473,
+      "step": 6252
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.535558204211362e-07,
+      "loss": 1.7648,
+      "step": 6253
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.535160905840286e-07,
+      "loss": 1.5069,
+      "step": 6254
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.534763607469209e-07,
+      "loss": 1.4985,
+      "step": 6255
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.534366309098133e-07,
+      "loss": 2.0334,
+      "step": 6256
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.533969010727055e-07,
+      "loss": 1.5931,
+      "step": 6257
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.533571712355979e-07,
+      "loss": 1.5633,
+      "step": 6258
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.533174413984902e-07,
+      "loss": 1.9164,
+      "step": 6259
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.532777115613825e-07,
+      "loss": 1.6512,
+      "step": 6260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.532379817242749e-07,
+      "loss": 1.7918,
+      "step": 6261
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.531982518871672e-07,
+      "loss": 2.1046,
+      "step": 6262
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.531585220500596e-07,
+      "loss": 1.8231,
+      "step": 6263
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.531187922129519e-07,
+      "loss": 1.9302,
+      "step": 6264
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.530790623758442e-07,
+      "loss": 1.4469,
+      "step": 6265
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.530393325387365e-07,
+      "loss": 1.6287,
+      "step": 6266
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.529996027016289e-07,
+      "loss": 1.7117,
+      "step": 6267
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.529598728645213e-07,
+      "loss": 2.0606,
+      "step": 6268
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.529201430274136e-07,
+      "loss": 1.8873,
+      "step": 6269
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.52880413190306e-07,
+      "loss": 1.8721,
+      "step": 6270
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.528406833531981e-07,
+      "loss": 1.832,
+      "step": 6271
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.528009535160905e-07,
+      "loss": 1.9351,
+      "step": 6272
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.527612236789828e-07,
+      "loss": 1.5402,
+      "step": 6273
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.527214938418752e-07,
+      "loss": 1.7179,
+      "step": 6274
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.526817640047676e-07,
+      "loss": 1.5854,
+      "step": 6275
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.526420341676599e-07,
+      "loss": 1.7606,
+      "step": 6276
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.526023043305522e-07,
+      "loss": 1.6725,
+      "step": 6277
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.525625744934445e-07,
+      "loss": 1.6463,
+      "step": 6278
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.525228446563369e-07,
+      "loss": 2.202,
+      "step": 6279
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.524831148192292e-07,
+      "loss": 1.34,
+      "step": 6280
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.524433849821216e-07,
+      "loss": 1.5197,
+      "step": 6281
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.52403655145014e-07,
+      "loss": 1.5559,
+      "step": 6282
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.523639253079061e-07,
+      "loss": 2.1496,
+      "step": 6283
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.523241954707985e-07,
+      "loss": 1.5566,
+      "step": 6284
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.522844656336908e-07,
+      "loss": 1.717,
+      "step": 6285
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.522447357965832e-07,
+      "loss": 1.4488,
+      "step": 6286
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.522050059594755e-07,
+      "loss": 1.83,
+      "step": 6287
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.521652761223679e-07,
+      "loss": 1.8441,
+      "step": 6288
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.521255462852601e-07,
+      "loss": 1.5864,
+      "step": 6289
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.520858164481525e-07,
+      "loss": 1.9415,
+      "step": 6290
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.520460866110449e-07,
+      "loss": 1.6101,
+      "step": 6291
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.520063567739372e-07,
+      "loss": 1.8444,
+      "step": 6292
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.519666269368296e-07,
+      "loss": 1.6455,
+      "step": 6293
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.519268970997219e-07,
+      "loss": 1.7096,
+      "step": 6294
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.518871672626142e-07,
+      "loss": 1.8987,
+      "step": 6295
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.518474374255065e-07,
+      "loss": 2.0039,
+      "step": 6296
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.518077075883988e-07,
+      "loss": 1.647,
+      "step": 6297
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.517679777512912e-07,
+      "loss": 1.5663,
+      "step": 6298
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.517282479141835e-07,
+      "loss": 1.8577,
+      "step": 6299
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.516885180770759e-07,
+      "loss": 1.5697,
+      "step": 6300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.516487882399681e-07,
+      "loss": 1.6521,
+      "step": 6301
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.516090584028605e-07,
+      "loss": 1.4712,
+      "step": 6302
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.515693285657528e-07,
+      "loss": 1.6402,
+      "step": 6303
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.515295987286452e-07,
+      "loss": 1.643,
+      "step": 6304
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.514898688915376e-07,
+      "loss": 1.5965,
+      "step": 6305
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.35448057097541635,
+      "eval_loss": 1.761055588722229,
+      "eval_runtime": 6.4274,
+      "eval_samples_per_second": 392.38,
+      "eval_steps_per_second": 6.223,
+      "step": 6305
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.514501390544299e-07,
+      "loss": 1.7873,
+      "step": 6306
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.514104092173222e-07,
+      "loss": 1.9516,
+      "step": 6307
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.513706793802145e-07,
+      "loss": 1.7952,
+      "step": 6308
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.513309495431069e-07,
+      "loss": 1.9246,
+      "step": 6309
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.512912197059991e-07,
+      "loss": 1.5644,
+      "step": 6310
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.512514898688915e-07,
+      "loss": 1.7879,
+      "step": 6311
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.512117600317838e-07,
+      "loss": 1.8009,
+      "step": 6312
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.511720301946761e-07,
+      "loss": 1.9365,
+      "step": 6313
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.511323003575685e-07,
+      "loss": 1.6134,
+      "step": 6314
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.510925705204608e-07,
+      "loss": 1.4701,
+      "step": 6315
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.510528406833532e-07,
+      "loss": 1.8876,
+      "step": 6316
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.510131108462455e-07,
+      "loss": 1.6397,
+      "step": 6317
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.509733810091379e-07,
+      "loss": 1.5327,
+      "step": 6318
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.509336511720301e-07,
+      "loss": 1.7216,
+      "step": 6319
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.508939213349225e-07,
+      "loss": 1.4579,
+      "step": 6320
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.508541914978149e-07,
+      "loss": 1.6603,
+      "step": 6321
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.508144616607072e-07,
+      "loss": 1.9249,
+      "step": 6322
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.507747318235995e-07,
+      "loss": 2.0231,
+      "step": 6323
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.507350019864918e-07,
+      "loss": 1.8322,
+      "step": 6324
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.506952721493841e-07,
+      "loss": 1.9658,
+      "step": 6325
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.506555423122764e-07,
+      "loss": 1.6795,
+      "step": 6326
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.506158124751688e-07,
+      "loss": 1.6774,
+      "step": 6327
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.505760826380611e-07,
+      "loss": 1.6334,
+      "step": 6328
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.505363528009535e-07,
+      "loss": 1.6793,
+      "step": 6329
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.504966229638459e-07,
+      "loss": 1.5494,
+      "step": 6330
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.504568931267381e-07,
+      "loss": 1.7791,
+      "step": 6331
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.504171632896305e-07,
+      "loss": 1.7522,
+      "step": 6332
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.503774334525228e-07,
+      "loss": 1.8334,
+      "step": 6333
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.503377036154152e-07,
+      "loss": 1.6455,
+      "step": 6334
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.502979737783075e-07,
+      "loss": 1.5333,
+      "step": 6335
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.502582439411999e-07,
+      "loss": 1.5122,
+      "step": 6336
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.50218514104092e-07,
+      "loss": 1.984,
+      "step": 6337
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.501787842669844e-07,
+      "loss": 1.7828,
+      "step": 6338
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.501390544298768e-07,
+      "loss": 2.0225,
+      "step": 6339
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.500993245927691e-07,
+      "loss": 1.8866,
+      "step": 6340
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.500595947556615e-07,
+      "loss": 2.0818,
+      "step": 6341
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.500198649185538e-07,
+      "loss": 1.6478,
+      "step": 6342
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.499801350814461e-07,
+      "loss": 1.7732,
+      "step": 6343
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.499404052443384e-07,
+      "loss": 1.6959,
+      "step": 6344
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.499006754072308e-07,
+      "loss": 1.4564,
+      "step": 6345
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.498609455701232e-07,
+      "loss": 1.5907,
+      "step": 6346
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.498212157330155e-07,
+      "loss": 1.8079,
+      "step": 6347
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.497814858959079e-07,
+      "loss": 1.7509,
+      "step": 6348
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.497417560588e-07,
+      "loss": 1.8468,
+      "step": 6349
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.497020262216924e-07,
+      "loss": 1.7949,
+      "step": 6350
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.496622963845847e-07,
+      "loss": 1.7676,
+      "step": 6351
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.496225665474771e-07,
+      "loss": 2.0158,
+      "step": 6352
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.495828367103695e-07,
+      "loss": 1.6467,
+      "step": 6353
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.495431068732618e-07,
+      "loss": 1.7125,
+      "step": 6354
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.495033770361541e-07,
+      "loss": 2.1743,
+      "step": 6355
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.494636471990464e-07,
+      "loss": 1.3757,
+      "step": 6356
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.494239173619388e-07,
+      "loss": 1.6422,
+      "step": 6357
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.493841875248311e-07,
+      "loss": 1.616,
+      "step": 6358
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.493444576877235e-07,
+      "loss": 1.8464,
+      "step": 6359
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.493047278506159e-07,
+      "loss": 1.7305,
+      "step": 6360
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.492649980135081e-07,
+      "loss": 1.6614,
+      "step": 6361
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.492252681764005e-07,
+      "loss": 1.5151,
+      "step": 6362
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.491855383392927e-07,
+      "loss": 1.3817,
+      "step": 6363
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.491458085021851e-07,
+      "loss": 1.9884,
+      "step": 6364
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.491060786650774e-07,
+      "loss": 1.5047,
+      "step": 6365
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.490663488279698e-07,
+      "loss": 1.4859,
+      "step": 6366
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.49026618990862e-07,
+      "loss": 1.867,
+      "step": 6367
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.489868891537544e-07,
+      "loss": 1.6677,
+      "step": 6368
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.489471593166468e-07,
+      "loss": 1.6843,
+      "step": 6369
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.489074294795391e-07,
+      "loss": 1.3782,
+      "step": 6370
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.488676996424315e-07,
+      "loss": 1.6656,
+      "step": 6371
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.488279698053238e-07,
+      "loss": 1.7549,
+      "step": 6372
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.487882399682161e-07,
+      "loss": 1.2833,
+      "step": 6373
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.487485101311084e-07,
+      "loss": 1.4486,
+      "step": 6374
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.487087802940008e-07,
+      "loss": 1.5925,
+      "step": 6375
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.486690504568931e-07,
+      "loss": 1.6053,
+      "step": 6376
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.486293206197854e-07,
+      "loss": 1.7318,
+      "step": 6377
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.485895907826778e-07,
+      "loss": 1.6618,
+      "step": 6378
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.4854986094557e-07,
+      "loss": 1.6157,
+      "step": 6379
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.485101311084624e-07,
+      "loss": 1.854,
+      "step": 6380
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.484704012713547e-07,
+      "loss": 1.779,
+      "step": 6381
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.484306714342471e-07,
+      "loss": 1.7943,
+      "step": 6382
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.483909415971395e-07,
+      "loss": 1.3454,
+      "step": 6383
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.483512117600318e-07,
+      "loss": 1.7912,
+      "step": 6384
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.483114819229241e-07,
+      "loss": 1.7327,
+      "step": 6385
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.482717520858164e-07,
+      "loss": 1.3788,
+      "step": 6386
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.482320222487088e-07,
+      "loss": 1.3357,
+      "step": 6387
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.481922924116011e-07,
+      "loss": 1.8856,
+      "step": 6388
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.481525625744934e-07,
+      "loss": 1.7979,
+      "step": 6389
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.481128327373857e-07,
+      "loss": 1.7442,
+      "step": 6390
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.48073102900278e-07,
+      "loss": 1.7156,
+      "step": 6391
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.480333730631704e-07,
+      "loss": 1.7519,
+      "step": 6392
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.479936432260627e-07,
+      "loss": 1.7442,
+      "step": 6393
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.479539133889551e-07,
+      "loss": 1.5388,
+      "step": 6394
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.479141835518474e-07,
+      "loss": 1.6709,
+      "step": 6395
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.478744537147398e-07,
+      "loss": 1.8508,
+      "step": 6396
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.47834723877632e-07,
+      "loss": 1.9889,
+      "step": 6397
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.477949940405244e-07,
+      "loss": 1.6486,
+      "step": 6398
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.477552642034168e-07,
+      "loss": 1.8986,
+      "step": 6399
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.477155343663091e-07,
+      "loss": 1.7397,
+      "step": 6400
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.476758045292015e-07,
+      "loss": 1.6029,
+      "step": 6401
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.476360746920938e-07,
+      "loss": 1.692,
+      "step": 6402
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.47596344854986e-07,
+      "loss": 1.6075,
+      "step": 6403
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.475566150178783e-07,
+      "loss": 1.7772,
+      "step": 6404
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.475168851807707e-07,
+      "loss": 1.6059,
+      "step": 6405
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.47477155343663e-07,
+      "loss": 2.232,
+      "step": 6406
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.474374255065554e-07,
+      "loss": 1.563,
+      "step": 6407
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.473976956694478e-07,
+      "loss": 1.5275,
+      "step": 6408
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.4735796583234e-07,
+      "loss": 2.0529,
+      "step": 6409
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.473182359952324e-07,
+      "loss": 1.7551,
+      "step": 6410
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.472785061581247e-07,
+      "loss": 1.7142,
+      "step": 6411
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.472387763210171e-07,
+      "loss": 1.8418,
+      "step": 6412
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.471990464839094e-07,
+      "loss": 1.6263,
+      "step": 6413
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.471593166468018e-07,
+      "loss": 1.7865,
+      "step": 6414
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.47119586809694e-07,
+      "loss": 1.8236,
+      "step": 6415
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.470798569725863e-07,
+      "loss": 1.7556,
+      "step": 6416
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.470401271354787e-07,
+      "loss": 1.6153,
+      "step": 6417
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.47000397298371e-07,
+      "loss": 1.478,
+      "step": 6418
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.469606674612634e-07,
+      "loss": 1.9031,
+      "step": 6419
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.469209376241557e-07,
+      "loss": 1.8389,
+      "step": 6420
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.46881207787048e-07,
+      "loss": 1.672,
+      "step": 6421
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.468414779499404e-07,
+      "loss": 1.727,
+      "step": 6422
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.468017481128327e-07,
+      "loss": 1.3781,
+      "step": 6423
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.467620182757251e-07,
+      "loss": 1.5081,
+      "step": 6424
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.467222884386174e-07,
+      "loss": 1.6484,
+      "step": 6425
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.466825586015098e-07,
+      "loss": 1.7005,
+      "step": 6426
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.46642828764402e-07,
+      "loss": 1.8804,
+      "step": 6427
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.466030989272944e-07,
+      "loss": 1.667,
+      "step": 6428
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.465633690901866e-07,
+      "loss": 1.57,
+      "step": 6429
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.46523639253079e-07,
+      "loss": 1.4948,
+      "step": 6430
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.464839094159714e-07,
+      "loss": 1.6718,
+      "step": 6431
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.464441795788637e-07,
+      "loss": 2.0354,
+      "step": 6432
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.46404449741756e-07,
+      "loss": 1.3995,
+      "step": 6433
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.463647199046483e-07,
+      "loss": 1.9503,
+      "step": 6434
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.463249900675407e-07,
+      "loss": 1.7426,
+      "step": 6435
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.46285260230433e-07,
+      "loss": 1.7927,
+      "step": 6436
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.462455303933254e-07,
+      "loss": 1.6658,
+      "step": 6437
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.462058005562178e-07,
+      "loss": 1.9926,
+      "step": 6438
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.4616607071911e-07,
+      "loss": 1.5079,
+      "step": 6439
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.461263408820024e-07,
+      "loss": 1.5737,
+      "step": 6440
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.460866110448947e-07,
+      "loss": 1.8313,
+      "step": 6441
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.46046881207787e-07,
+      "loss": 1.641,
+      "step": 6442
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.460071513706793e-07,
+      "loss": 2.1611,
+      "step": 6443
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.459674215335717e-07,
+      "loss": 1.6802,
+      "step": 6444
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.459276916964639e-07,
+      "loss": 1.6487,
+      "step": 6445
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.458879618593563e-07,
+      "loss": 1.4485,
+      "step": 6446
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.458482320222487e-07,
+      "loss": 1.6562,
+      "step": 6447
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.45808502185141e-07,
+      "loss": 1.3547,
+      "step": 6448
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.457687723480334e-07,
+      "loss": 2.2874,
+      "step": 6449
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.457290425109257e-07,
+      "loss": 1.6096,
+      "step": 6450
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.45689312673818e-07,
+      "loss": 1.5907,
+      "step": 6451
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.456495828367103e-07,
+      "loss": 1.8441,
+      "step": 6452
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.456098529996027e-07,
+      "loss": 1.6815,
+      "step": 6453
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.455701231624951e-07,
+      "loss": 1.7818,
+      "step": 6454
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.455303933253873e-07,
+      "loss": 1.8754,
+      "step": 6455
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.454906634882797e-07,
+      "loss": 1.9693,
+      "step": 6456
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.454509336511719e-07,
+      "loss": 1.1711,
+      "step": 6457
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.454112038140643e-07,
+      "loss": 1.5946,
+      "step": 6458
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.453714739769566e-07,
+      "loss": 1.5803,
+      "step": 6459
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.45331744139849e-07,
+      "loss": 1.6998,
+      "step": 6460
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.452920143027414e-07,
+      "loss": 1.4175,
+      "step": 6461
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.452522844656337e-07,
+      "loss": 1.7626,
+      "step": 6462
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.45212554628526e-07,
+      "loss": 1.6315,
+      "step": 6463
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.451728247914183e-07,
+      "loss": 1.6659,
+      "step": 6464
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.451330949543107e-07,
+      "loss": 1.4105,
+      "step": 6465
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.45093365117203e-07,
+      "loss": 1.766,
+      "step": 6466
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.450536352800954e-07,
+      "loss": 1.5964,
+      "step": 6467
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.450139054429878e-07,
+      "loss": 1.7111,
+      "step": 6468
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.449741756058799e-07,
+      "loss": 1.1088,
+      "step": 6469
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.449344457687723e-07,
+      "loss": 1.2166,
+      "step": 6470
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.448947159316646e-07,
+      "loss": 1.573,
+      "step": 6471
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.44854986094557e-07,
+      "loss": 1.821,
+      "step": 6472
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.448152562574493e-07,
+      "loss": 1.7197,
+      "step": 6473
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.447755264203417e-07,
+      "loss": 1.7708,
+      "step": 6474
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.447357965832339e-07,
+      "loss": 1.6516,
+      "step": 6475
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.446960667461263e-07,
+      "loss": 1.5217,
+      "step": 6476
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.446563369090187e-07,
+      "loss": 1.9419,
+      "step": 6477
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.44616607071911e-07,
+      "loss": 1.8086,
+      "step": 6478
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.445768772348034e-07,
+      "loss": 1.6821,
+      "step": 6479
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.445371473976957e-07,
+      "loss": 1.6358,
+      "step": 6480
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.444974175605879e-07,
+      "loss": 1.8238,
+      "step": 6481
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.444576877234802e-07,
+      "loss": 1.455,
+      "step": 6482
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.444179578863726e-07,
+      "loss": 1.3868,
+      "step": 6483
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.44378228049265e-07,
+      "loss": 1.9903,
+      "step": 6484
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.443384982121573e-07,
+      "loss": 2.2735,
+      "step": 6485
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.442987683750497e-07,
+      "loss": 2.1177,
+      "step": 6486
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.442590385379419e-07,
+      "loss": 1.21,
+      "step": 6487
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.442193087008343e-07,
+      "loss": 1.897,
+      "step": 6488
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.441795788637266e-07,
+      "loss": 1.8077,
+      "step": 6489
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.44139849026619e-07,
+      "loss": 1.6345,
+      "step": 6490
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.441001191895113e-07,
+      "loss": 1.8931,
+      "step": 6491
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.440603893524037e-07,
+      "loss": 1.9616,
+      "step": 6492
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.44020659515296e-07,
+      "loss": 1.6014,
+      "step": 6493
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.439809296781883e-07,
+      "loss": 1.3808,
+      "step": 6494
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.439411998410806e-07,
+      "loss": 1.9039,
+      "step": 6495
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.439014700039729e-07,
+      "loss": 1.5928,
+      "step": 6496
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.438617401668653e-07,
+      "loss": 1.6002,
+      "step": 6497
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.438220103297576e-07,
+      "loss": 2.004,
+      "step": 6498
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.437822804926499e-07,
+      "loss": 1.9509,
+      "step": 6499
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.437425506555423e-07,
+      "loss": 1.8585,
+      "step": 6500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.437028208184346e-07,
+      "loss": 1.6964,
+      "step": 6501
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.43663090981327e-07,
+      "loss": 1.6782,
+      "step": 6502
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.436233611442193e-07,
+      "loss": 1.7988,
+      "step": 6503
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.435836313071117e-07,
+      "loss": 1.457,
+      "step": 6504
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.435439014700039e-07,
+      "loss": 1.6482,
+      "step": 6505
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.435041716328963e-07,
+      "loss": 2.0482,
+      "step": 6506
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.434644417957886e-07,
+      "loss": 1.672,
+      "step": 6507
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.434247119586809e-07,
+      "loss": 1.4152,
+      "step": 6508
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.433849821215733e-07,
+      "loss": 1.5146,
+      "step": 6509
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.433452522844656e-07,
+      "loss": 1.8756,
+      "step": 6510
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.433055224473579e-07,
+      "loss": 1.5231,
+      "step": 6511
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.432657926102502e-07,
+      "loss": 1.6107,
+      "step": 6512
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.432260627731426e-07,
+      "loss": 1.5426,
+      "step": 6513
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.431863329360349e-07,
+      "loss": 1.5265,
+      "step": 6514
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.431466030989273e-07,
+      "loss": 1.6049,
+      "step": 6515
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.431068732618197e-07,
+      "loss": 1.7619,
+      "step": 6516
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.430671434247119e-07,
+      "loss": 1.8121,
+      "step": 6517
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.430274135876043e-07,
+      "loss": 1.2376,
+      "step": 6518
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.429876837504966e-07,
+      "loss": 1.6813,
+      "step": 6519
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.42947953913389e-07,
+      "loss": 1.5814,
+      "step": 6520
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.429082240762812e-07,
+      "loss": 1.9458,
+      "step": 6521
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.428684942391736e-07,
+      "loss": 1.8853,
+      "step": 6522
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.428287644020658e-07,
+      "loss": 1.71,
+      "step": 6523
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.427890345649582e-07,
+      "loss": 1.7557,
+      "step": 6524
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.427493047278506e-07,
+      "loss": 1.7182,
+      "step": 6525
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.427095748907429e-07,
+      "loss": 1.8897,
+      "step": 6526
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.426698450536353e-07,
+      "loss": 1.8381,
+      "step": 6527
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.426301152165276e-07,
+      "loss": 1.3772,
+      "step": 6528
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.425903853794199e-07,
+      "loss": 1.4113,
+      "step": 6529
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.425506555423122e-07,
+      "loss": 1.7147,
+      "step": 6530
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.425109257052046e-07,
+      "loss": 1.4674,
+      "step": 6531
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.42471195868097e-07,
+      "loss": 1.7223,
+      "step": 6532
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.424314660309893e-07,
+      "loss": 1.4467,
+      "step": 6533
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.423917361938817e-07,
+      "loss": 1.7247,
+      "step": 6534
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.423520063567738e-07,
+      "loss": 1.2871,
+      "step": 6535
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.423122765196662e-07,
+      "loss": 1.5132,
+      "step": 6536
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.422725466825585e-07,
+      "loss": 1.8906,
+      "step": 6537
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.422328168454509e-07,
+      "loss": 1.814,
+      "step": 6538
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.421930870083433e-07,
+      "loss": 1.5269,
+      "step": 6539
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.421533571712356e-07,
+      "loss": 1.272,
+      "step": 6540
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.421136273341279e-07,
+      "loss": 1.7862,
+      "step": 6541
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.420738974970202e-07,
+      "loss": 1.8393,
+      "step": 6542
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.420341676599126e-07,
+      "loss": 1.7638,
+      "step": 6543
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.419944378228049e-07,
+      "loss": 2.0079,
+      "step": 6544
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.419547079856973e-07,
+      "loss": 2.063,
+      "step": 6545
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.419149781485897e-07,
+      "loss": 1.4353,
+      "step": 6546
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.418752483114818e-07,
+      "loss": 1.5055,
+      "step": 6547
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.418355184743742e-07,
+      "loss": 1.7484,
+      "step": 6548
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.417957886372665e-07,
+      "loss": 1.8253,
+      "step": 6549
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.417560588001589e-07,
+      "loss": 1.932,
+      "step": 6550
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.417163289630512e-07,
+      "loss": 1.7311,
+      "step": 6551
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.416765991259436e-07,
+      "loss": 1.7813,
+      "step": 6552
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.416368692888358e-07,
+      "loss": 1.7181,
+      "step": 6553
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.415971394517282e-07,
+      "loss": 1.7413,
+      "step": 6554
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.415574096146206e-07,
+      "loss": 1.8408,
+      "step": 6555
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.415176797775129e-07,
+      "loss": 1.4761,
+      "step": 6556
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.414779499404053e-07,
+      "loss": 1.6651,
+      "step": 6557
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.414382201032976e-07,
+      "loss": 1.6152,
+      "step": 6558
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.413984902661899e-07,
+      "loss": 1.6745,
+      "step": 6559
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.413587604290822e-07,
+      "loss": 1.5503,
+      "step": 6560
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.413190305919745e-07,
+      "loss": 1.1234,
+      "step": 6561
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.412793007548668e-07,
+      "loss": 1.483,
+      "step": 6562
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.412395709177592e-07,
+      "loss": 1.781,
+      "step": 6563
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.411998410806516e-07,
+      "loss": 1.383,
+      "step": 6564
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.411601112435438e-07,
+      "loss": 1.7242,
+      "step": 6565
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.411203814064362e-07,
+      "loss": 1.2339,
+      "step": 6566
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.410806515693285e-07,
+      "loss": 2.2294,
+      "step": 6567
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.410409217322209e-07,
+      "loss": 2.1073,
+      "step": 6568
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.410011918951132e-07,
+      "loss": 1.9735,
+      "step": 6569
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.409614620580056e-07,
+      "loss": 1.9392,
+      "step": 6570
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.409217322208979e-07,
+      "loss": 1.7946,
+      "step": 6571
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.408820023837902e-07,
+      "loss": 1.645,
+      "step": 6572
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.408422725466826e-07,
+      "loss": 1.5678,
+      "step": 6573
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.408025427095748e-07,
+      "loss": 1.9602,
+      "step": 6574
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.407628128724672e-07,
+      "loss": 1.6702,
+      "step": 6575
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.407230830353595e-07,
+      "loss": 1.779,
+      "step": 6576
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.406833531982518e-07,
+      "loss": 1.4028,
+      "step": 6577
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.406436233611442e-07,
+      "loss": 1.6701,
+      "step": 6578
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.406038935240365e-07,
+      "loss": 1.9425,
+      "step": 6579
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.405641636869289e-07,
+      "loss": 1.7336,
+      "step": 6580
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.405244338498212e-07,
+      "loss": 1.694,
+      "step": 6581
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.404847040127136e-07,
+      "loss": 1.8498,
+      "step": 6582
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.404449741756058e-07,
+      "loss": 1.5889,
+      "step": 6583
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.404052443384982e-07,
+      "loss": 1.4894,
+      "step": 6584
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.403655145013905e-07,
+      "loss": 1.9547,
+      "step": 6585
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.403257846642829e-07,
+      "loss": 1.8547,
+      "step": 6586
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.402860548271752e-07,
+      "loss": 1.8162,
+      "step": 6587
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.402463249900675e-07,
+      "loss": 1.261,
+      "step": 6588
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.402065951529598e-07,
+      "loss": 1.6489,
+      "step": 6589
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.401668653158521e-07,
+      "loss": 1.3461,
+      "step": 6590
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.401271354787445e-07,
+      "loss": 1.7132,
+      "step": 6591
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.400874056416368e-07,
+      "loss": 1.4122,
+      "step": 6592
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.400476758045292e-07,
+      "loss": 1.7759,
+      "step": 6593
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.400079459674216e-07,
+      "loss": 1.6838,
+      "step": 6594
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.399682161303138e-07,
+      "loss": 1.6999,
+      "step": 6595
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.399284862932062e-07,
+      "loss": 1.4912,
+      "step": 6596
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.398887564560985e-07,
+      "loss": 1.6049,
+      "step": 6597
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.398490266189909e-07,
+      "loss": 1.844,
+      "step": 6598
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.398092967818832e-07,
+      "loss": 1.4646,
+      "step": 6599
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.397695669447756e-07,
+      "loss": 1.6198,
+      "step": 6600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.397298371076677e-07,
+      "loss": 1.5831,
+      "step": 6601
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.396901072705601e-07,
+      "loss": 1.847,
+      "step": 6602
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.396503774334525e-07,
+      "loss": 1.9767,
+      "step": 6603
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.396106475963448e-07,
+      "loss": 1.7508,
+      "step": 6604
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.395709177592372e-07,
+      "loss": 1.803,
+      "step": 6605
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.395311879221295e-07,
+      "loss": 1.6625,
+      "step": 6606
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.394914580850218e-07,
+      "loss": 1.652,
+      "step": 6607
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.394517282479141e-07,
+      "loss": 1.7605,
+      "step": 6608
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.394119984108065e-07,
+      "loss": 1.6729,
+      "step": 6609
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.393722685736989e-07,
+      "loss": 1.6293,
+      "step": 6610
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.393325387365912e-07,
+      "loss": 1.4872,
+      "step": 6611
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.392928088994836e-07,
+      "loss": 1.8407,
+      "step": 6612
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.392530790623757e-07,
+      "loss": 1.4927,
+      "step": 6613
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.392133492252681e-07,
+      "loss": 1.5887,
+      "step": 6614
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.391736193881604e-07,
+      "loss": 1.3848,
+      "step": 6615
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.391338895510528e-07,
+      "loss": 1.4614,
+      "step": 6616
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.390941597139452e-07,
+      "loss": 1.534,
+      "step": 6617
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.390544298768375e-07,
+      "loss": 1.3703,
+      "step": 6618
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.390147000397298e-07,
+      "loss": 1.8225,
+      "step": 6619
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.389749702026221e-07,
+      "loss": 1.8854,
+      "step": 6620
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.389352403655145e-07,
+      "loss": 1.8611,
+      "step": 6621
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.388955105284068e-07,
+      "loss": 1.5596,
+      "step": 6622
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.388557806912992e-07,
+      "loss": 1.9571,
+      "step": 6623
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.388160508541916e-07,
+      "loss": 1.7143,
+      "step": 6624
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.387763210170838e-07,
+      "loss": 1.5464,
+      "step": 6625
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.387365911799762e-07,
+      "loss": 2.0405,
+      "step": 6626
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.386968613428684e-07,
+      "loss": 1.5013,
+      "step": 6627
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.386571315057608e-07,
+      "loss": 1.3928,
+      "step": 6628
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.386174016686531e-07,
+      "loss": 1.8585,
+      "step": 6629
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.385776718315455e-07,
+      "loss": 1.7864,
+      "step": 6630
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.385379419944377e-07,
+      "loss": 2.0194,
+      "step": 6631
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.384982121573301e-07,
+      "loss": 1.2207,
+      "step": 6632
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.384584823202225e-07,
+      "loss": 1.6799,
+      "step": 6633
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.384187524831148e-07,
+      "loss": 2.1494,
+      "step": 6634
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.383790226460072e-07,
+      "loss": 1.5056,
+      "step": 6635
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.383392928088995e-07,
+      "loss": 1.3366,
+      "step": 6636
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.382995629717918e-07,
+      "loss": 1.8105,
+      "step": 6637
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.382598331346841e-07,
+      "loss": 1.8333,
+      "step": 6638
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.382201032975765e-07,
+      "loss": 1.9123,
+      "step": 6639
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.381803734604687e-07,
+      "loss": 1.377,
+      "step": 6640
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.381406436233611e-07,
+      "loss": 1.3953,
+      "step": 6641
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.381009137862535e-07,
+      "loss": 1.3927,
+      "step": 6642
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.380611839491457e-07,
+      "loss": 1.9061,
+      "step": 6643
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.380214541120381e-07,
+      "loss": 1.7438,
+      "step": 6644
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.379817242749304e-07,
+      "loss": 2.066,
+      "step": 6645
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.379419944378228e-07,
+      "loss": 1.5949,
+      "step": 6646
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.379022646007151e-07,
+      "loss": 1.55,
+      "step": 6647
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.378625347636074e-07,
+      "loss": 1.5984,
+      "step": 6648
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.378228049264998e-07,
+      "loss": 1.8107,
+      "step": 6649
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.377830750893921e-07,
+      "loss": 1.9501,
+      "step": 6650
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.377433452522845e-07,
+      "loss": 1.7149,
+      "step": 6651
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.377036154151768e-07,
+      "loss": 1.7291,
+      "step": 6652
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.376638855780691e-07,
+      "loss": 1.4952,
+      "step": 6653
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.376241557409613e-07,
+      "loss": 1.5507,
+      "step": 6654
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.375844259038537e-07,
+      "loss": 1.5413,
+      "step": 6655
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.37544696066746e-07,
+      "loss": 1.6405,
+      "step": 6656
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.375049662296384e-07,
+      "loss": 1.4994,
+      "step": 6657
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.374652363925308e-07,
+      "loss": 2.2155,
+      "step": 6658
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.374255065554231e-07,
+      "loss": 1.4089,
+      "step": 6659
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.373857767183154e-07,
+      "loss": 1.7358,
+      "step": 6660
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.373460468812077e-07,
+      "loss": 1.7393,
+      "step": 6661
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.373063170441001e-07,
+      "loss": 2.0437,
+      "step": 6662
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.372665872069924e-07,
+      "loss": 1.4388,
+      "step": 6663
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.372268573698848e-07,
+      "loss": 1.4431,
+      "step": 6664
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.371871275327772e-07,
+      "loss": 1.5829,
+      "step": 6665
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.371473976956693e-07,
+      "loss": 2.095,
+      "step": 6666
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.371076678585617e-07,
+      "loss": 1.654,
+      "step": 6667
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.37067938021454e-07,
+      "loss": 1.7217,
+      "step": 6668
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.370282081843464e-07,
+      "loss": 1.5962,
+      "step": 6669
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.369884783472387e-07,
+      "loss": 2.0482,
+      "step": 6670
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.369487485101311e-07,
+      "loss": 1.6681,
+      "step": 6671
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.369090186730234e-07,
+      "loss": 1.5967,
+      "step": 6672
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.368692888359157e-07,
+      "loss": 1.3493,
+      "step": 6673
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.368295589988081e-07,
+      "loss": 1.6984,
+      "step": 6674
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.367898291617004e-07,
+      "loss": 1.9583,
+      "step": 6675
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.367500993245928e-07,
+      "loss": 1.7505,
+      "step": 6676
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.367103694874851e-07,
+      "loss": 1.7314,
+      "step": 6677
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.366706396503774e-07,
+      "loss": 1.634,
+      "step": 6678
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.366309098132696e-07,
+      "loss": 1.2388,
+      "step": 6679
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.36591179976162e-07,
+      "loss": 1.7935,
+      "step": 6680
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.365514501390544e-07,
+      "loss": 1.8288,
+      "step": 6681
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.365117203019467e-07,
+      "loss": 1.9922,
+      "step": 6682
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.364719904648391e-07,
+      "loss": 2.2203,
+      "step": 6683
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.364322606277313e-07,
+      "loss": 1.6635,
+      "step": 6684
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.363925307906237e-07,
+      "loss": 1.8181,
+      "step": 6685
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.36352800953516e-07,
+      "loss": 1.683,
+      "step": 6686
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.363130711164084e-07,
+      "loss": 1.3316,
+      "step": 6687
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.362733412793008e-07,
+      "loss": 1.6403,
+      "step": 6688
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.362336114421931e-07,
+      "loss": 1.916,
+      "step": 6689
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.361938816050854e-07,
+      "loss": 1.7922,
+      "step": 6690
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.361541517679777e-07,
+      "loss": 1.514,
+      "step": 6691
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.3611442193087e-07,
+      "loss": 1.9449,
+      "step": 6692
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.360746920937623e-07,
+      "loss": 1.7196,
+      "step": 6693
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.360349622566547e-07,
+      "loss": 1.5814,
+      "step": 6694
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.359952324195471e-07,
+      "loss": 1.8566,
+      "step": 6695
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.359555025824393e-07,
+      "loss": 1.4973,
+      "step": 6696
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.359157727453317e-07,
+      "loss": 2.0876,
+      "step": 6697
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.35876042908224e-07,
+      "loss": 1.5944,
+      "step": 6698
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.358363130711164e-07,
+      "loss": 2.0132,
+      "step": 6699
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.357965832340087e-07,
+      "loss": 2.0605,
+      "step": 6700
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.357568533969011e-07,
+      "loss": 1.7762,
+      "step": 6701
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.357171235597933e-07,
+      "loss": 1.5476,
+      "step": 6702
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.356773937226857e-07,
+      "loss": 1.7648,
+      "step": 6703
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.356376638855781e-07,
+      "loss": 1.8923,
+      "step": 6704
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.355979340484704e-07,
+      "loss": 1.7181,
+      "step": 6705
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.355582042113627e-07,
+      "loss": 1.4948,
+      "step": 6706
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.35518474374255e-07,
+      "loss": 1.5432,
+      "step": 6707
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.354787445371473e-07,
+      "loss": 1.7079,
+      "step": 6708
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.354390147000396e-07,
+      "loss": 1.8521,
+      "step": 6709
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.35399284862932e-07,
+      "loss": 2.0414,
+      "step": 6710
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.353595550258244e-07,
+      "loss": 1.5736,
+      "step": 6711
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.353198251887167e-07,
+      "loss": 1.9546,
+      "step": 6712
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.352800953516091e-07,
+      "loss": 2.4005,
+      "step": 6713
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.352403655145013e-07,
+      "loss": 1.6615,
+      "step": 6714
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.352006356773937e-07,
+      "loss": 1.7706,
+      "step": 6715
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.35160905840286e-07,
+      "loss": 1.4951,
+      "step": 6716
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.351211760031784e-07,
+      "loss": 1.991,
+      "step": 6717
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.350814461660708e-07,
+      "loss": 1.5365,
+      "step": 6718
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.35041716328963e-07,
+      "loss": 1.6782,
+      "step": 6719
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.350019864918553e-07,
+      "loss": 1.6383,
+      "step": 6720
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.349622566547476e-07,
+      "loss": 1.3899,
+      "step": 6721
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.3492252681764e-07,
+      "loss": 1.7223,
+      "step": 6722
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.348827969805323e-07,
+      "loss": 1.4464,
+      "step": 6723
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.348430671434247e-07,
+      "loss": 1.6252,
+      "step": 6724
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.34803337306317e-07,
+      "loss": 1.698,
+      "step": 6725
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.347636074692093e-07,
+      "loss": 1.8293,
+      "step": 6726
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.347238776321017e-07,
+      "loss": 1.7959,
+      "step": 6727
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.34684147794994e-07,
+      "loss": 1.67,
+      "step": 6728
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.346444179578864e-07,
+      "loss": 1.1285,
+      "step": 6729
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.346046881207787e-07,
+      "loss": 1.6938,
+      "step": 6730
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.345649582836711e-07,
+      "loss": 1.8983,
+      "step": 6731
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.345252284465632e-07,
+      "loss": 1.3588,
+      "step": 6732
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.344854986094556e-07,
+      "loss": 1.4972,
+      "step": 6733
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.34445768772348e-07,
+      "loss": 1.7132,
+      "step": 6734
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.344060389352403e-07,
+      "loss": 1.6445,
+      "step": 6735
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.343663090981327e-07,
+      "loss": 1.7476,
+      "step": 6736
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.34326579261025e-07,
+      "loss": 1.9756,
+      "step": 6737
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.342868494239173e-07,
+      "loss": 2.0142,
+      "step": 6738
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.342471195868096e-07,
+      "loss": 1.7629,
+      "step": 6739
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.34207389749702e-07,
+      "loss": 1.8715,
+      "step": 6740
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.341676599125944e-07,
+      "loss": 1.7134,
+      "step": 6741
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.341279300754867e-07,
+      "loss": 1.8088,
+      "step": 6742
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.340882002383791e-07,
+      "loss": 1.4457,
+      "step": 6743
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.340484704012713e-07,
+      "loss": 1.8728,
+      "step": 6744
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.340087405641636e-07,
+      "loss": 1.4037,
+      "step": 6745
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.339690107270559e-07,
+      "loss": 1.6638,
+      "step": 6746
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.339292808899483e-07,
+      "loss": 1.3863,
+      "step": 6747
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.338895510528406e-07,
+      "loss": 2.0049,
+      "step": 6748
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.33849821215733e-07,
+      "loss": 1.6102,
+      "step": 6749
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.338100913786253e-07,
+      "loss": 1.8096,
+      "step": 6750
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.337703615415176e-07,
+      "loss": 1.4428,
+      "step": 6751
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.3373063170441e-07,
+      "loss": 1.7089,
+      "step": 6752
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.336909018673023e-07,
+      "loss": 1.7938,
+      "step": 6753
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.336511720301947e-07,
+      "loss": 1.7778,
+      "step": 6754
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.33611442193087e-07,
+      "loss": 1.8806,
+      "step": 6755
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.335717123559793e-07,
+      "loss": 1.6171,
+      "step": 6756
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.335319825188717e-07,
+      "loss": 1.6008,
+      "step": 6757
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.334922526817639e-07,
+      "loss": 1.5947,
+      "step": 6758
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.334525228446563e-07,
+      "loss": 1.8079,
+      "step": 6759
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.334127930075486e-07,
+      "loss": 1.6373,
+      "step": 6760
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.33373063170441e-07,
+      "loss": 1.7719,
+      "step": 6761
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.333333333333332e-07,
+      "loss": 1.7145,
+      "step": 6762
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.332936034962256e-07,
+      "loss": 1.4676,
+      "step": 6763
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.332538736591179e-07,
+      "loss": 1.785,
+      "step": 6764
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.332141438220103e-07,
+      "loss": 1.5562,
+      "step": 6765
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.331744139849027e-07,
+      "loss": 1.6864,
+      "step": 6766
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.33134684147795e-07,
+      "loss": 1.8348,
+      "step": 6767
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.330949543106873e-07,
+      "loss": 1.6393,
+      "step": 6768
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.330552244735796e-07,
+      "loss": 2.0502,
+      "step": 6769
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.33015494636472e-07,
+      "loss": 1.7554,
+      "step": 6770
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.329757647993643e-07,
+      "loss": 1.92,
+      "step": 6771
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.329360349622566e-07,
+      "loss": 1.3299,
+      "step": 6772
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.32896305125149e-07,
+      "loss": 1.6516,
+      "step": 6773
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.328565752880412e-07,
+      "loss": 1.3887,
+      "step": 6774
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.328168454509336e-07,
+      "loss": 1.6621,
+      "step": 6775
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.327771156138259e-07,
+      "loss": 1.8623,
+      "step": 6776
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.327373857767183e-07,
+      "loss": 1.7166,
+      "step": 6777
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.326976559396106e-07,
+      "loss": 1.9474,
+      "step": 6778
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.32657926102503e-07,
+      "loss": 1.601,
+      "step": 6779
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.326181962653952e-07,
+      "loss": 1.5358,
+      "step": 6780
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.325784664282876e-07,
+      "loss": 1.8647,
+      "step": 6781
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.3253873659118e-07,
+      "loss": 1.6811,
+      "step": 6782
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.324990067540723e-07,
+      "loss": 1.5715,
+      "step": 6783
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.324592769169647e-07,
+      "loss": 1.7392,
+      "step": 6784
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.324195470798569e-07,
+      "loss": 1.3605,
+      "step": 6785
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.323798172427492e-07,
+      "loss": 1.7446,
+      "step": 6786
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.323400874056415e-07,
+      "loss": 1.6795,
+      "step": 6787
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.323003575685339e-07,
+      "loss": 1.4563,
+      "step": 6788
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.322606277314263e-07,
+      "loss": 1.9879,
+      "step": 6789
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.322208978943186e-07,
+      "loss": 1.4714,
+      "step": 6790
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.32181168057211e-07,
+      "loss": 1.7247,
+      "step": 6791
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.321414382201032e-07,
+      "loss": 1.7194,
+      "step": 6792
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.321017083829956e-07,
+      "loss": 1.8534,
+      "step": 6793
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.320619785458879e-07,
+      "loss": 1.9139,
+      "step": 6794
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.320222487087803e-07,
+      "loss": 1.9944,
+      "step": 6795
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.319825188716727e-07,
+      "loss": 1.8184,
+      "step": 6796
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.31942789034565e-07,
+      "loss": 1.5813,
+      "step": 6797
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.319030591974572e-07,
+      "loss": 2.1223,
+      "step": 6798
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.318633293603495e-07,
+      "loss": 1.6745,
+      "step": 6799
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.318235995232419e-07,
+      "loss": 1.5949,
+      "step": 6800
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.317838696861342e-07,
+      "loss": 1.5597,
+      "step": 6801
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.317441398490266e-07,
+      "loss": 1.7595,
+      "step": 6802
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.31704410011919e-07,
+      "loss": 1.4553,
+      "step": 6803
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.316646801748112e-07,
+      "loss": 1.7064,
+      "step": 6804
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.316249503377036e-07,
+      "loss": 1.8284,
+      "step": 6805
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.315852205005959e-07,
+      "loss": 1.5906,
+      "step": 6806
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.315454906634883e-07,
+      "loss": 1.4863,
+      "step": 6807
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.315057608263806e-07,
+      "loss": 1.6243,
+      "step": 6808
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.31466030989273e-07,
+      "loss": 1.9654,
+      "step": 6809
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.314263011521652e-07,
+      "loss": 1.7634,
+      "step": 6810
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.313865713150575e-07,
+      "loss": 1.6739,
+      "step": 6811
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.313468414779499e-07,
+      "loss": 2.01,
+      "step": 6812
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.313071116408422e-07,
+      "loss": 1.6257,
+      "step": 6813
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.312673818037346e-07,
+      "loss": 1.8463,
+      "step": 6814
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.312276519666269e-07,
+      "loss": 1.675,
+      "step": 6815
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.311879221295192e-07,
+      "loss": 1.825,
+      "step": 6816
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.311481922924115e-07,
+      "loss": 1.994,
+      "step": 6817
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.311084624553039e-07,
+      "loss": 1.8318,
+      "step": 6818
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.310687326181963e-07,
+      "loss": 1.5279,
+      "step": 6819
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.310290027810886e-07,
+      "loss": 1.593,
+      "step": 6820
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.30989272943981e-07,
+      "loss": 1.7569,
+      "step": 6821
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.309495431068732e-07,
+      "loss": 1.7594,
+      "step": 6822
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.309098132697656e-07,
+      "loss": 2.4153,
+      "step": 6823
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.308700834326578e-07,
+      "loss": 1.6494,
+      "step": 6824
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.308303535955502e-07,
+      "loss": 1.9057,
+      "step": 6825
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.307906237584425e-07,
+      "loss": 1.6758,
+      "step": 6826
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.307508939213349e-07,
+      "loss": 1.6388,
+      "step": 6827
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.307111640842272e-07,
+      "loss": 1.9139,
+      "step": 6828
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.306714342471195e-07,
+      "loss": 1.4148,
+      "step": 6829
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.306317044100119e-07,
+      "loss": 2.0174,
+      "step": 6830
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.305919745729042e-07,
+      "loss": 1.8203,
+      "step": 6831
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.305522447357966e-07,
+      "loss": 1.6062,
+      "step": 6832
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.305125148986889e-07,
+      "loss": 1.6871,
+      "step": 6833
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.304727850615812e-07,
+      "loss": 2.1226,
+      "step": 6834
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.304330552244736e-07,
+      "loss": 1.4861,
+      "step": 6835
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.303933253873659e-07,
+      "loss": 1.4219,
+      "step": 6836
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.303535955502583e-07,
+      "loss": 1.7251,
+      "step": 6837
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.303138657131505e-07,
+      "loss": 1.665,
+      "step": 6838
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.302741358760429e-07,
+      "loss": 1.6612,
+      "step": 6839
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.302344060389351e-07,
+      "loss": 1.7865,
+      "step": 6840
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.301946762018275e-07,
+      "loss": 1.77,
+      "step": 6841
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.301549463647198e-07,
+      "loss": 1.7368,
+      "step": 6842
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.301152165276122e-07,
+      "loss": 1.5076,
+      "step": 6843
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.300754866905046e-07,
+      "loss": 1.6583,
+      "step": 6844
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.300357568533969e-07,
+      "loss": 1.7345,
+      "step": 6845
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.299960270162892e-07,
+      "loss": 1.498,
+      "step": 6846
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.299562971791815e-07,
+      "loss": 1.6887,
+      "step": 6847
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.299165673420739e-07,
+      "loss": 1.305,
+      "step": 6848
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.298768375049662e-07,
+      "loss": 1.5413,
+      "step": 6849
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.298371076678586e-07,
+      "loss": 1.2501,
+      "step": 6850
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.297973778307509e-07,
+      "loss": 1.6558,
+      "step": 6851
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.297576479936431e-07,
+      "loss": 1.7258,
+      "step": 6852
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.297179181565355e-07,
+      "loss": 1.9172,
+      "step": 6853
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.296781883194278e-07,
+      "loss": 1.7786,
+      "step": 6854
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.296384584823202e-07,
+      "loss": 1.8313,
+      "step": 6855
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.295987286452125e-07,
+      "loss": 1.8879,
+      "step": 6856
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.295589988081049e-07,
+      "loss": 1.5218,
+      "step": 6857
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.295192689709971e-07,
+      "loss": 1.4826,
+      "step": 6858
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.294795391338895e-07,
+      "loss": 1.6829,
+      "step": 6859
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.294398092967819e-07,
+      "loss": 1.7993,
+      "step": 6860
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.294000794596742e-07,
+      "loss": 1.6008,
+      "step": 6861
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.293603496225666e-07,
+      "loss": 1.271,
+      "step": 6862
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.293206197854589e-07,
+      "loss": 1.7045,
+      "step": 6863
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.292808899483511e-07,
+      "loss": 1.8956,
+      "step": 6864
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.292411601112434e-07,
+      "loss": 1.6228,
+      "step": 6865
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.292014302741358e-07,
+      "loss": 1.4432,
+      "step": 6866
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.291617004370282e-07,
+      "loss": 1.2833,
+      "step": 6867
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.291219705999205e-07,
+      "loss": 1.7299,
+      "step": 6868
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.290822407628129e-07,
+      "loss": 1.3113,
+      "step": 6869
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.290425109257051e-07,
+      "loss": 1.7822,
+      "step": 6870
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.290027810885975e-07,
+      "loss": 1.5956,
+      "step": 6871
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.289630512514898e-07,
+      "loss": 1.6743,
+      "step": 6872
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.289233214143822e-07,
+      "loss": 1.8126,
+      "step": 6873
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.288835915772746e-07,
+      "loss": 1.8175,
+      "step": 6874
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.288438617401669e-07,
+      "loss": 1.7392,
+      "step": 6875
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.288041319030592e-07,
+      "loss": 1.7133,
+      "step": 6876
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.287644020659514e-07,
+      "loss": 1.7382,
+      "step": 6877
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.287246722288438e-07,
+      "loss": 1.5239,
+      "step": 6878
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.286849423917361e-07,
+      "loss": 1.777,
+      "step": 6879
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.286452125546285e-07,
+      "loss": 1.5045,
+      "step": 6880
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.286054827175208e-07,
+      "loss": 1.7298,
+      "step": 6881
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.285657528804131e-07,
+      "loss": 1.8524,
+      "step": 6882
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.285260230433055e-07,
+      "loss": 1.765,
+      "step": 6883
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.284862932061978e-07,
+      "loss": 1.8077,
+      "step": 6884
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.284465633690902e-07,
+      "loss": 1.574,
+      "step": 6885
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.284068335319825e-07,
+      "loss": 1.4676,
+      "step": 6886
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.283671036948749e-07,
+      "loss": 1.8261,
+      "step": 6887
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.283273738577671e-07,
+      "loss": 1.7662,
+      "step": 6888
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.282876440206595e-07,
+      "loss": 1.8066,
+      "step": 6889
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.282479141835518e-07,
+      "loss": 1.8301,
+      "step": 6890
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.282081843464441e-07,
+      "loss": 2.0211,
+      "step": 6891
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.281684545093365e-07,
+      "loss": 2.0807,
+      "step": 6892
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.281287246722288e-07,
+      "loss": 1.2666,
+      "step": 6893
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.280889948351211e-07,
+      "loss": 1.7708,
+      "step": 6894
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.280492649980134e-07,
+      "loss": 1.6217,
+      "step": 6895
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.280095351609058e-07,
+      "loss": 1.4731,
+      "step": 6896
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.279698053237982e-07,
+      "loss": 1.3638,
+      "step": 6897
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.279300754866905e-07,
+      "loss": 1.8949,
+      "step": 6898
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.278903456495829e-07,
+      "loss": 1.8589,
+      "step": 6899
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.278506158124751e-07,
+      "loss": 1.6347,
+      "step": 6900
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.278108859753675e-07,
+      "loss": 1.9227,
+      "step": 6901
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.277711561382598e-07,
+      "loss": 1.7848,
+      "step": 6902
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.277314263011522e-07,
+      "loss": 1.4845,
+      "step": 6903
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.276916964640444e-07,
+      "loss": 1.6787,
+      "step": 6904
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.276519666269368e-07,
+      "loss": 1.8039,
+      "step": 6905
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.276122367898291e-07,
+      "loss": 1.6225,
+      "step": 6906
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.275725069527214e-07,
+      "loss": 1.9012,
+      "step": 6907
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.275327771156138e-07,
+      "loss": 1.4868,
+      "step": 6908
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.274930472785061e-07,
+      "loss": 1.8099,
+      "step": 6909
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.274533174413985e-07,
+      "loss": 1.5106,
+      "step": 6910
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.274135876042908e-07,
+      "loss": 1.8053,
+      "step": 6911
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.273738577671831e-07,
+      "loss": 1.6693,
+      "step": 6912
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.273341279300755e-07,
+      "loss": 1.5118,
+      "step": 6913
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.272943980929678e-07,
+      "loss": 1.5996,
+      "step": 6914
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.272546682558602e-07,
+      "loss": 1.851,
+      "step": 6915
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.272149384187525e-07,
+      "loss": 1.8235,
+      "step": 6916
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.271752085816448e-07,
+      "loss": 1.3734,
+      "step": 6917
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.27135478744537e-07,
+      "loss": 1.6042,
+      "step": 6918
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.270957489074294e-07,
+      "loss": 1.7034,
+      "step": 6919
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.270560190703217e-07,
+      "loss": 1.6014,
+      "step": 6920
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.270162892332141e-07,
+      "loss": 1.9275,
+      "step": 6921
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.269765593961065e-07,
+      "loss": 1.7144,
+      "step": 6922
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.269368295589988e-07,
+      "loss": 1.596,
+      "step": 6923
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.268970997218911e-07,
+      "loss": 1.1808,
+      "step": 6924
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.268573698847834e-07,
+      "loss": 1.4541,
+      "step": 6925
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.268176400476758e-07,
+      "loss": 1.631,
+      "step": 6926
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.267779102105681e-07,
+      "loss": 1.7721,
+      "step": 6927
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.267381803734605e-07,
+      "loss": 1.6832,
+      "step": 6928
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.266984505363529e-07,
+      "loss": 1.6735,
+      "step": 6929
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.26658720699245e-07,
+      "loss": 1.4551,
+      "step": 6930
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.266189908621374e-07,
+      "loss": 1.9999,
+      "step": 6931
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.265792610250297e-07,
+      "loss": 1.7708,
+      "step": 6932
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.265395311879221e-07,
+      "loss": 1.918,
+      "step": 6933
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.264998013508144e-07,
+      "loss": 2.0719,
+      "step": 6934
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.264600715137068e-07,
+      "loss": 1.611,
+      "step": 6935
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.26420341676599e-07,
+      "loss": 1.6651,
+      "step": 6936
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.263806118394914e-07,
+      "loss": 1.8908,
+      "step": 6937
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.263408820023838e-07,
+      "loss": 1.5798,
+      "step": 6938
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.263011521652761e-07,
+      "loss": 1.4752,
+      "step": 6939
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.262614223281685e-07,
+      "loss": 1.6691,
+      "step": 6940
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.262216924910608e-07,
+      "loss": 1.9431,
+      "step": 6941
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.261819626539531e-07,
+      "loss": 2.0016,
+      "step": 6942
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.261422328168453e-07,
+      "loss": 1.6229,
+      "step": 6943
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.261025029797377e-07,
+      "loss": 1.6751,
+      "step": 6944
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.260627731426301e-07,
+      "loss": 1.6636,
+      "step": 6945
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.260230433055224e-07,
+      "loss": 1.6752,
+      "step": 6946
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.259833134684148e-07,
+      "loss": 1.6371,
+      "step": 6947
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.25943583631307e-07,
+      "loss": 1.6198,
+      "step": 6948
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.259038537941994e-07,
+      "loss": 1.8383,
+      "step": 6949
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.258641239570917e-07,
+      "loss": 1.3358,
+      "step": 6950
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.258243941199841e-07,
+      "loss": 1.783,
+      "step": 6951
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.257846642828765e-07,
+      "loss": 1.3053,
+      "step": 6952
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.257449344457688e-07,
+      "loss": 1.5798,
+      "step": 6953
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.257052046086611e-07,
+      "loss": 1.8546,
+      "step": 6954
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.256654747715534e-07,
+      "loss": 1.3735,
+      "step": 6955
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.256257449344457e-07,
+      "loss": 1.4759,
+      "step": 6956
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.25586015097338e-07,
+      "loss": 1.4025,
+      "step": 6957
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.255462852602304e-07,
+      "loss": 1.592,
+      "step": 6958
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.255065554231227e-07,
+      "loss": 2.0458,
+      "step": 6959
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.25466825586015e-07,
+      "loss": 1.6618,
+      "step": 6960
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.254270957489074e-07,
+      "loss": 1.8429,
+      "step": 6961
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.253873659117997e-07,
+      "loss": 1.8229,
+      "step": 6962
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.253476360746921e-07,
+      "loss": 1.7155,
+      "step": 6963
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.253079062375844e-07,
+      "loss": 1.6678,
+      "step": 6964
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.252681764004768e-07,
+      "loss": 1.5424,
+      "step": 6965
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.25228446563369e-07,
+      "loss": 1.6518,
+      "step": 6966
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.251887167262614e-07,
+      "loss": 1.5112,
+      "step": 6967
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.251489868891538e-07,
+      "loss": 1.902,
+      "step": 6968
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.251092570520461e-07,
+      "loss": 1.6861,
+      "step": 6969
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.250695272149384e-07,
+      "loss": 2.0725,
+      "step": 6970
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.250297973778307e-07,
+      "loss": 1.1956,
+      "step": 6971
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.24990067540723e-07,
+      "loss": 1.5207,
+      "step": 6972
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.249503377036153e-07,
+      "loss": 1.5429,
+      "step": 6973
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.249106078665077e-07,
+      "loss": 2.0555,
+      "step": 6974
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.248708780294e-07,
+      "loss": 1.9252,
+      "step": 6975
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.248311481922924e-07,
+      "loss": 1.9842,
+      "step": 6976
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.247914183551848e-07,
+      "loss": 1.6584,
+      "step": 6977
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.24751688518077e-07,
+      "loss": 1.6876,
+      "step": 6978
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.247119586809694e-07,
+      "loss": 1.5451,
+      "step": 6979
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.246722288438617e-07,
+      "loss": 1.3601,
+      "step": 6980
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.246324990067541e-07,
+      "loss": 1.5894,
+      "step": 6981
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.245927691696464e-07,
+      "loss": 1.3293,
+      "step": 6982
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.245530393325387e-07,
+      "loss": 1.3647,
+      "step": 6983
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.24513309495431e-07,
+      "loss": 2.204,
+      "step": 6984
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.244735796583233e-07,
+      "loss": 1.7436,
+      "step": 6985
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.244338498212157e-07,
+      "loss": 1.8925,
+      "step": 6986
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.24394119984108e-07,
+      "loss": 1.4916,
+      "step": 6987
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.243543901470004e-07,
+      "loss": 1.9686,
+      "step": 6988
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.243146603098927e-07,
+      "loss": 1.7585,
+      "step": 6989
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.24274930472785e-07,
+      "loss": 1.9406,
+      "step": 6990
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.242352006356774e-07,
+      "loss": 1.9326,
+      "step": 6991
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.241954707985697e-07,
+      "loss": 1.5223,
+      "step": 6992
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.241557409614621e-07,
+      "loss": 1.7581,
+      "step": 6993
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.241160111243544e-07,
+      "loss": 1.4895,
+      "step": 6994
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.240762812872468e-07,
+      "loss": 1.5742,
+      "step": 6995
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.240365514501389e-07,
+      "loss": 2.0519,
+      "step": 6996
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.239968216130313e-07,
+      "loss": 1.8348,
+      "step": 6997
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.239570917759236e-07,
+      "loss": 1.881,
+      "step": 6998
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.23917361938816e-07,
+      "loss": 1.6644,
+      "step": 6999
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.238776321017084e-07,
+      "loss": 1.7048,
+      "step": 7000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.238379022646007e-07,
+      "loss": 1.856,
+      "step": 7001
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.23798172427493e-07,
+      "loss": 2.0793,
+      "step": 7002
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.237584425903853e-07,
+      "loss": 1.8614,
+      "step": 7003
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.237187127532777e-07,
+      "loss": 2.0726,
+      "step": 7004
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.2367898291617e-07,
+      "loss": 1.7717,
+      "step": 7005
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.236392530790624e-07,
+      "loss": 1.421,
+      "step": 7006
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.235995232419548e-07,
+      "loss": 1.8936,
+      "step": 7007
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.23559793404847e-07,
+      "loss": 1.9655,
+      "step": 7008
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.235200635677393e-07,
+      "loss": 1.6026,
+      "step": 7009
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.234803337306316e-07,
+      "loss": 1.8846,
+      "step": 7010
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.23440603893524e-07,
+      "loss": 1.9701,
+      "step": 7011
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.234008740564163e-07,
+      "loss": 1.6696,
+      "step": 7012
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.233611442193087e-07,
+      "loss": 1.5195,
+      "step": 7013
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.23321414382201e-07,
+      "loss": 1.672,
+      "step": 7014
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.232816845450933e-07,
+      "loss": 2.0803,
+      "step": 7015
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.232419547079857e-07,
+      "loss": 1.7597,
+      "step": 7016
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.23202224870878e-07,
+      "loss": 1.578,
+      "step": 7017
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.231624950337704e-07,
+      "loss": 1.7942,
+      "step": 7018
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.231227651966627e-07,
+      "loss": 1.4888,
+      "step": 7019
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.23083035359555e-07,
+      "loss": 1.6325,
+      "step": 7020
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.230433055224473e-07,
+      "loss": 1.6649,
+      "step": 7021
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.230035756853396e-07,
+      "loss": 1.6321,
+      "step": 7022
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.22963845848232e-07,
+      "loss": 1.4659,
+      "step": 7023
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.229241160111243e-07,
+      "loss": 1.4485,
+      "step": 7024
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.228843861740167e-07,
+      "loss": 1.567,
+      "step": 7025
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.228446563369089e-07,
+      "loss": 1.9436,
+      "step": 7026
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.228049264998013e-07,
+      "loss": 1.6612,
+      "step": 7027
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.227651966626936e-07,
+      "loss": 1.6315,
+      "step": 7028
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.22725466825586e-07,
+      "loss": 2.3276,
+      "step": 7029
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.226857369884784e-07,
+      "loss": 1.5289,
+      "step": 7030
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.226460071513707e-07,
+      "loss": 1.6936,
+      "step": 7031
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.22606277314263e-07,
+      "loss": 2.2761,
+      "step": 7032
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.225665474771553e-07,
+      "loss": 1.6253,
+      "step": 7033
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.225268176400477e-07,
+      "loss": 1.8703,
+      "step": 7034
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.2248708780294e-07,
+      "loss": 1.6707,
+      "step": 7035
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.224473579658323e-07,
+      "loss": 1.5038,
+      "step": 7036
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.224076281287246e-07,
+      "loss": 1.9243,
+      "step": 7037
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.223678982916169e-07,
+      "loss": 1.903,
+      "step": 7038
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.223281684545093e-07,
+      "loss": 1.5515,
+      "step": 7039
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.222884386174016e-07,
+      "loss": 1.768,
+      "step": 7040
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.22248708780294e-07,
+      "loss": 1.6371,
+      "step": 7041
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.222089789431863e-07,
+      "loss": 1.7605,
+      "step": 7042
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.221692491060787e-07,
+      "loss": 1.904,
+      "step": 7043
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.221295192689709e-07,
+      "loss": 1.7548,
+      "step": 7044
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.220897894318633e-07,
+      "loss": 1.5619,
+      "step": 7045
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.220500595947557e-07,
+      "loss": 1.9251,
+      "step": 7046
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.22010329757648e-07,
+      "loss": 1.7329,
+      "step": 7047
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.219705999205404e-07,
+      "loss": 1.5092,
+      "step": 7048
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.219308700834326e-07,
+      "loss": 1.8151,
+      "step": 7049
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.218911402463249e-07,
+      "loss": 1.5862,
+      "step": 7050
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.218514104092172e-07,
+      "loss": 1.7337,
+      "step": 7051
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.218116805721096e-07,
+      "loss": 1.806,
+      "step": 7052
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.21771950735002e-07,
+      "loss": 1.8622,
+      "step": 7053
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.217322208978943e-07,
+      "loss": 1.5551,
+      "step": 7054
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.216924910607867e-07,
+      "loss": 1.2538,
+      "step": 7055
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.216527612236789e-07,
+      "loss": 1.6336,
+      "step": 7056
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.216130313865713e-07,
+      "loss": 2.092,
+      "step": 7057
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.215733015494636e-07,
+      "loss": 1.5401,
+      "step": 7058
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.21533571712356e-07,
+      "loss": 1.8926,
+      "step": 7059
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.214938418752484e-07,
+      "loss": 1.7405,
+      "step": 7060
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.214541120381407e-07,
+      "loss": 1.7861,
+      "step": 7061
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.214143822010329e-07,
+      "loss": 1.8497,
+      "step": 7062
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.213746523639252e-07,
+      "loss": 1.7719,
+      "step": 7063
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.213349225268176e-07,
+      "loss": 1.9165,
+      "step": 7064
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.212951926897099e-07,
+      "loss": 2.3456,
+      "step": 7065
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.212554628526023e-07,
+      "loss": 1.561,
+      "step": 7066
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.212157330154946e-07,
+      "loss": 1.8296,
+      "step": 7067
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.211760031783869e-07,
+      "loss": 1.6131,
+      "step": 7068
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.211362733412793e-07,
+      "loss": 1.6819,
+      "step": 7069
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.210965435041716e-07,
+      "loss": 1.7815,
+      "step": 7070
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.21056813667064e-07,
+      "loss": 1.7625,
+      "step": 7071
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.210170838299563e-07,
+      "loss": 1.8053,
+      "step": 7072
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.209773539928487e-07,
+      "loss": 1.8777,
+      "step": 7073
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.209376241557409e-07,
+      "loss": 1.7536,
+      "step": 7074
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.208978943186332e-07,
+      "loss": 2.0712,
+      "step": 7075
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.208581644815255e-07,
+      "loss": 1.8518,
+      "step": 7076
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.208184346444179e-07,
+      "loss": 1.5877,
+      "step": 7077
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.207787048073103e-07,
+      "loss": 1.7283,
+      "step": 7078
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.207389749702026e-07,
+      "loss": 1.6725,
+      "step": 7079
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.206992451330949e-07,
+      "loss": 1.5867,
+      "step": 7080
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.206595152959872e-07,
+      "loss": 1.7059,
+      "step": 7081
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.206197854588796e-07,
+      "loss": 1.9532,
+      "step": 7082
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.205800556217719e-07,
+      "loss": 1.3574,
+      "step": 7083
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.205403257846643e-07,
+      "loss": 1.4153,
+      "step": 7084
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.205005959475567e-07,
+      "loss": 1.5614,
+      "step": 7085
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.204608661104489e-07,
+      "loss": 1.461,
+      "step": 7086
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.204211362733413e-07,
+      "loss": 1.6811,
+      "step": 7087
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.203814064362335e-07,
+      "loss": 2.1575,
+      "step": 7088
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.203416765991259e-07,
+      "loss": 1.6827,
+      "step": 7089
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.203019467620182e-07,
+      "loss": 1.4492,
+      "step": 7090
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.202622169249106e-07,
+      "loss": 1.7817,
+      "step": 7091
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.202224870878028e-07,
+      "loss": 1.5811,
+      "step": 7092
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.201827572506952e-07,
+      "loss": 1.8767,
+      "step": 7093
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.201430274135876e-07,
+      "loss": 2.0213,
+      "step": 7094
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.201032975764799e-07,
+      "loss": 2.0755,
+      "step": 7095
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.200635677393723e-07,
+      "loss": 1.7404,
+      "step": 7096
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.200238379022646e-07,
+      "loss": 1.5346,
+      "step": 7097
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.199841080651569e-07,
+      "loss": 1.8735,
+      "step": 7098
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.199443782280492e-07,
+      "loss": 1.9536,
+      "step": 7099
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.199046483909416e-07,
+      "loss": 1.7734,
+      "step": 7100
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.19864918553834e-07,
+      "loss": 1.7161,
+      "step": 7101
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.198251887167262e-07,
+      "loss": 1.7112,
+      "step": 7102
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.197854588796186e-07,
+      "loss": 1.723,
+      "step": 7103
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.197457290425108e-07,
+      "loss": 1.4982,
+      "step": 7104
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.197059992054032e-07,
+      "loss": 1.6403,
+      "step": 7105
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.196662693682955e-07,
+      "loss": 1.5317,
+      "step": 7106
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.196265395311879e-07,
+      "loss": 1.7647,
+      "step": 7107
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.195868096940803e-07,
+      "loss": 1.6637,
+      "step": 7108
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.195470798569726e-07,
+      "loss": 1.6382,
+      "step": 7109
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.195073500198649e-07,
+      "loss": 2.0179,
+      "step": 7110
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.194676201827572e-07,
+      "loss": 1.5472,
+      "step": 7111
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.194278903456496e-07,
+      "loss": 1.4667,
+      "step": 7112
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.193881605085419e-07,
+      "loss": 1.8103,
+      "step": 7113
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.193484306714343e-07,
+      "loss": 1.2437,
+      "step": 7114
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.193087008343265e-07,
+      "loss": 1.8144,
+      "step": 7115
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.192689709972188e-07,
+      "loss": 1.6243,
+      "step": 7116
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.192292411601112e-07,
+      "loss": 1.6898,
+      "step": 7117
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.191895113230035e-07,
+      "loss": 1.6424,
+      "step": 7118
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.191497814858959e-07,
+      "loss": 1.9839,
+      "step": 7119
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.191100516487882e-07,
+      "loss": 1.7198,
+      "step": 7120
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.190703218116806e-07,
+      "loss": 1.6242,
+      "step": 7121
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.190305919745728e-07,
+      "loss": 1.8094,
+      "step": 7122
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.189908621374652e-07,
+      "loss": 1.9243,
+      "step": 7123
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.189511323003576e-07,
+      "loss": 1.7418,
+      "step": 7124
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.189114024632499e-07,
+      "loss": 2.0842,
+      "step": 7125
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.188716726261423e-07,
+      "loss": 1.5692,
+      "step": 7126
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.188319427890346e-07,
+      "loss": 1.839,
+      "step": 7127
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.187922129519268e-07,
+      "loss": 1.5194,
+      "step": 7128
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.187524831148191e-07,
+      "loss": 1.821,
+      "step": 7129
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.187127532777115e-07,
+      "loss": 1.4983,
+      "step": 7130
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.186730234406039e-07,
+      "loss": 1.81,
+      "step": 7131
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.186332936034962e-07,
+      "loss": 1.6731,
+      "step": 7132
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.185935637663886e-07,
+      "loss": 1.9365,
+      "step": 7133
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.185538339292808e-07,
+      "loss": 1.4424,
+      "step": 7134
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.185141040921732e-07,
+      "loss": 2.0188,
+      "step": 7135
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.184743742550655e-07,
+      "loss": 1.6667,
+      "step": 7136
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.184346444179579e-07,
+      "loss": 2.0613,
+      "step": 7137
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.183949145808503e-07,
+      "loss": 1.4084,
+      "step": 7138
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.183551847437426e-07,
+      "loss": 1.8236,
+      "step": 7139
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.183154549066349e-07,
+      "loss": 1.6898,
+      "step": 7140
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.182757250695271e-07,
+      "loss": 1.8021,
+      "step": 7141
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.182359952324195e-07,
+      "loss": 1.9867,
+      "step": 7142
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.181962653953118e-07,
+      "loss": 1.3793,
+      "step": 7143
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.181565355582042e-07,
+      "loss": 1.8456,
+      "step": 7144
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.181168057210965e-07,
+      "loss": 1.4854,
+      "step": 7145
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.180770758839888e-07,
+      "loss": 2.0336,
+      "step": 7146
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.180373460468812e-07,
+      "loss": 1.732,
+      "step": 7147
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.179976162097735e-07,
+      "loss": 1.3967,
+      "step": 7148
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.179578863726659e-07,
+      "loss": 1.698,
+      "step": 7149
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.179181565355582e-07,
+      "loss": 1.5292,
+      "step": 7150
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.178784266984506e-07,
+      "loss": 1.6592,
+      "step": 7151
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.178386968613428e-07,
+      "loss": 1.6545,
+      "step": 7152
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.177989670242352e-07,
+      "loss": 1.436,
+      "step": 7153
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.177592371871274e-07,
+      "loss": 1.7068,
+      "step": 7154
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.177195073500198e-07,
+      "loss": 1.4155,
+      "step": 7155
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.176797775129122e-07,
+      "loss": 1.7974,
+      "step": 7156
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.176400476758045e-07,
+      "loss": 1.4044,
+      "step": 7157
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.176003178386968e-07,
+      "loss": 1.8321,
+      "step": 7158
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.175605880015891e-07,
+      "loss": 1.7007,
+      "step": 7159
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.175208581644815e-07,
+      "loss": 1.5121,
+      "step": 7160
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.174811283273738e-07,
+      "loss": 1.4102,
+      "step": 7161
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.174413984902662e-07,
+      "loss": 1.8565,
+      "step": 7162
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.174016686531586e-07,
+      "loss": 1.4457,
+      "step": 7163
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.173619388160508e-07,
+      "loss": 1.7679,
+      "step": 7164
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.173222089789432e-07,
+      "loss": 1.2695,
+      "step": 7165
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.172824791418355e-07,
+      "loss": 1.8029,
+      "step": 7166
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.172427493047279e-07,
+      "loss": 1.6394,
+      "step": 7167
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.172030194676201e-07,
+      "loss": 1.7948,
+      "step": 7168
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.171632896305125e-07,
+      "loss": 1.8916,
+      "step": 7169
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.171235597934047e-07,
+      "loss": 1.4814,
+      "step": 7170
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.170838299562971e-07,
+      "loss": 1.6528,
+      "step": 7171
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.170441001191895e-07,
+      "loss": 1.5369,
+      "step": 7172
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.170043702820818e-07,
+      "loss": 1.6633,
+      "step": 7173
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.169646404449742e-07,
+      "loss": 2.0949,
+      "step": 7174
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.169249106078665e-07,
+      "loss": 1.7447,
+      "step": 7175
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.168851807707588e-07,
+      "loss": 1.8533,
+      "step": 7176
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.168454509336511e-07,
+      "loss": 1.7717,
+      "step": 7177
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.168057210965435e-07,
+      "loss": 1.8087,
+      "step": 7178
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.167659912594359e-07,
+      "loss": 1.4661,
+      "step": 7179
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.167262614223282e-07,
+      "loss": 1.9781,
+      "step": 7180
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.166865315852205e-07,
+      "loss": 1.544,
+      "step": 7181
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.166468017481127e-07,
+      "loss": 1.6176,
+      "step": 7182
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.166070719110051e-07,
+      "loss": 1.5536,
+      "step": 7183
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.165673420738974e-07,
+      "loss": 1.6807,
+      "step": 7184
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.165276122367898e-07,
+      "loss": 2.0504,
+      "step": 7185
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.164878823996822e-07,
+      "loss": 2.231,
+      "step": 7186
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.164481525625745e-07,
+      "loss": 2.1026,
+      "step": 7187
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.164084227254668e-07,
+      "loss": 1.8365,
+      "step": 7188
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.163686928883591e-07,
+      "loss": 1.8834,
+      "step": 7189
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.163289630512515e-07,
+      "loss": 1.8676,
+      "step": 7190
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.162892332141438e-07,
+      "loss": 1.8643,
+      "step": 7191
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.162495033770362e-07,
+      "loss": 1.6062,
+      "step": 7192
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.162097735399286e-07,
+      "loss": 1.8633,
+      "step": 7193
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.161700437028207e-07,
+      "loss": 1.5189,
+      "step": 7194
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.161303138657131e-07,
+      "loss": 1.5615,
+      "step": 7195
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.160905840286054e-07,
+      "loss": 1.829,
+      "step": 7196
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.160508541914978e-07,
+      "loss": 1.8953,
+      "step": 7197
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.160111243543901e-07,
+      "loss": 1.6922,
+      "step": 7198
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.159713945172825e-07,
+      "loss": 2.0818,
+      "step": 7199
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.159316646801747e-07,
+      "loss": 1.5105,
+      "step": 7200
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.158919348430671e-07,
+      "loss": 1.3103,
+      "step": 7201
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.158522050059595e-07,
+      "loss": 1.5577,
+      "step": 7202
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.158124751688518e-07,
+      "loss": 1.6555,
+      "step": 7203
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.157727453317442e-07,
+      "loss": 1.8335,
+      "step": 7204
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.157330154946365e-07,
+      "loss": 1.6268,
+      "step": 7205
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.156932856575288e-07,
+      "loss": 1.5986,
+      "step": 7206
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.15653555820421e-07,
+      "loss": 1.5006,
+      "step": 7207
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.156138259833134e-07,
+      "loss": 1.4699,
+      "step": 7208
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.155740961462058e-07,
+      "loss": 1.9568,
+      "step": 7209
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.155343663090981e-07,
+      "loss": 1.3432,
+      "step": 7210
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.154946364719905e-07,
+      "loss": 1.3784,
+      "step": 7211
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.154549066348827e-07,
+      "loss": 1.735,
+      "step": 7212
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.154151767977751e-07,
+      "loss": 1.507,
+      "step": 7213
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.153754469606674e-07,
+      "loss": 1.779,
+      "step": 7214
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.153357171235598e-07,
+      "loss": 1.7391,
+      "step": 7215
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.152959872864522e-07,
+      "loss": 1.67,
+      "step": 7216
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.152562574493445e-07,
+      "loss": 1.4932,
+      "step": 7217
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.152165276122368e-07,
+      "loss": 1.9809,
+      "step": 7218
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.151767977751291e-07,
+      "loss": 1.8188,
+      "step": 7219
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.151370679380214e-07,
+      "loss": 1.9876,
+      "step": 7220
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.150973381009137e-07,
+      "loss": 1.8274,
+      "step": 7221
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.150576082638061e-07,
+      "loss": 1.2794,
+      "step": 7222
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.150178784266984e-07,
+      "loss": 1.5694,
+      "step": 7223
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.149781485895907e-07,
+      "loss": 1.7149,
+      "step": 7224
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.149384187524831e-07,
+      "loss": 1.796,
+      "step": 7225
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.148986889153754e-07,
+      "loss": 1.635,
+      "step": 7226
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.148589590782678e-07,
+      "loss": 1.2117,
+      "step": 7227
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.148192292411601e-07,
+      "loss": 1.5733,
+      "step": 7228
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.147794994040525e-07,
+      "loss": 1.4103,
+      "step": 7229
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.147397695669447e-07,
+      "loss": 1.7531,
+      "step": 7230
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.147000397298371e-07,
+      "loss": 1.54,
+      "step": 7231
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.146603098927295e-07,
+      "loss": 1.9234,
+      "step": 7232
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.146205800556218e-07,
+      "loss": 1.6948,
+      "step": 7233
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.145808502185141e-07,
+      "loss": 1.8924,
+      "step": 7234
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.145411203814064e-07,
+      "loss": 1.7832,
+      "step": 7235
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.145013905442987e-07,
+      "loss": 1.1763,
+      "step": 7236
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.14461660707191e-07,
+      "loss": 1.8585,
+      "step": 7237
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.144219308700834e-07,
+      "loss": 1.9222,
+      "step": 7238
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.143822010329757e-07,
+      "loss": 1.6535,
+      "step": 7239
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.143424711958681e-07,
+      "loss": 1.4792,
+      "step": 7240
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.143027413587605e-07,
+      "loss": 1.9359,
+      "step": 7241
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.142630115216527e-07,
+      "loss": 1.6511,
+      "step": 7242
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.142232816845451e-07,
+      "loss": 1.902,
+      "step": 7243
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.141835518474374e-07,
+      "loss": 1.6869,
+      "step": 7244
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.141438220103298e-07,
+      "loss": 1.6318,
+      "step": 7245
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.141040921732221e-07,
+      "loss": 1.5933,
+      "step": 7246
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.140643623361143e-07,
+      "loss": 1.7372,
+      "step": 7247
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.140246324990066e-07,
+      "loss": 1.65,
+      "step": 7248
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.13984902661899e-07,
+      "loss": 2.0182,
+      "step": 7249
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.139451728247914e-07,
+      "loss": 2.0187,
+      "step": 7250
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.139054429876837e-07,
+      "loss": 1.8541,
+      "step": 7251
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.138657131505761e-07,
+      "loss": 1.4949,
+      "step": 7252
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.138259833134683e-07,
+      "loss": 2.0938,
+      "step": 7253
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.137862534763607e-07,
+      "loss": 1.6997,
+      "step": 7254
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.13746523639253e-07,
+      "loss": 1.3249,
+      "step": 7255
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.137067938021454e-07,
+      "loss": 1.6523,
+      "step": 7256
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.136670639650378e-07,
+      "loss": 1.6125,
+      "step": 7257
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.136273341279301e-07,
+      "loss": 1.2289,
+      "step": 7258
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.135876042908224e-07,
+      "loss": 1.7456,
+      "step": 7259
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.135478744537146e-07,
+      "loss": 1.6567,
+      "step": 7260
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.13508144616607e-07,
+      "loss": 2.0708,
+      "step": 7261
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.134684147794993e-07,
+      "loss": 1.8008,
+      "step": 7262
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.134286849423917e-07,
+      "loss": 1.6141,
+      "step": 7263
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.133889551052841e-07,
+      "loss": 1.7323,
+      "step": 7264
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.133492252681763e-07,
+      "loss": 1.5926,
+      "step": 7265
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.133094954310687e-07,
+      "loss": 1.9041,
+      "step": 7266
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.13269765593961e-07,
+      "loss": 1.5419,
+      "step": 7267
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.132300357568534e-07,
+      "loss": 1.8957,
+      "step": 7268
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.131903059197457e-07,
+      "loss": 1.8762,
+      "step": 7269
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.131505760826381e-07,
+      "loss": 1.8934,
+      "step": 7270
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.131108462455303e-07,
+      "loss": 1.8381,
+      "step": 7271
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.130711164084227e-07,
+      "loss": 1.9356,
+      "step": 7272
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.13031386571315e-07,
+      "loss": 1.8048,
+      "step": 7273
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.129916567342073e-07,
+      "loss": 1.6179,
+      "step": 7274
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.129519268970997e-07,
+      "loss": 1.3021,
+      "step": 7275
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.12912197059992e-07,
+      "loss": 1.7138,
+      "step": 7276
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.128724672228843e-07,
+      "loss": 1.7692,
+      "step": 7277
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.128327373857766e-07,
+      "loss": 1.5354,
+      "step": 7278
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.12793007548669e-07,
+      "loss": 1.8983,
+      "step": 7279
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.127532777115614e-07,
+      "loss": 1.6159,
+      "step": 7280
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.127135478744537e-07,
+      "loss": 2.0957,
+      "step": 7281
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.126738180373461e-07,
+      "loss": 1.8873,
+      "step": 7282
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.126340882002383e-07,
+      "loss": 1.8341,
+      "step": 7283
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.125943583631307e-07,
+      "loss": 1.7851,
+      "step": 7284
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.12554628526023e-07,
+      "loss": 2.0867,
+      "step": 7285
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.125148986889153e-07,
+      "loss": 1.5697,
+      "step": 7286
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.124751688518077e-07,
+      "loss": 1.7304,
+      "step": 7287
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.124354390147e-07,
+      "loss": 1.2268,
+      "step": 7288
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.123957091775923e-07,
+      "loss": 1.4572,
+      "step": 7289
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.123559793404846e-07,
+      "loss": 1.5402,
+      "step": 7290
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.12316249503377e-07,
+      "loss": 1.4465,
+      "step": 7291
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.122765196662693e-07,
+      "loss": 1.5367,
+      "step": 7292
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.122367898291617e-07,
+      "loss": 1.6615,
+      "step": 7293
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.12197059992054e-07,
+      "loss": 1.3262,
+      "step": 7294
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.121573301549463e-07,
+      "loss": 1.5676,
+      "step": 7295
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.121176003178387e-07,
+      "loss": 1.8944,
+      "step": 7296
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.12077870480731e-07,
+      "loss": 1.6123,
+      "step": 7297
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.120381406436234e-07,
+      "loss": 1.5685,
+      "step": 7298
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.119984108065157e-07,
+      "loss": 1.4926,
+      "step": 7299
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.11958680969408e-07,
+      "loss": 1.8753,
+      "step": 7300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.119189511323002e-07,
+      "loss": 1.5687,
+      "step": 7301
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.118792212951926e-07,
+      "loss": 2.0407,
+      "step": 7302
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.11839491458085e-07,
+      "loss": 1.6847,
+      "step": 7303
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.117997616209773e-07,
+      "loss": 1.7849,
+      "step": 7304
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.117600317838697e-07,
+      "loss": 1.9152,
+      "step": 7305
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.11720301946762e-07,
+      "loss": 1.5587,
+      "step": 7306
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.116805721096543e-07,
+      "loss": 1.9159,
+      "step": 7307
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.116408422725466e-07,
+      "loss": 1.645,
+      "step": 7308
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.11601112435439e-07,
+      "loss": 2.0816,
+      "step": 7309
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.115613825983314e-07,
+      "loss": 1.5591,
+      "step": 7310
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.115216527612237e-07,
+      "loss": 1.4118,
+      "step": 7311
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.114819229241161e-07,
+      "loss": 1.8224,
+      "step": 7312
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.114421930870082e-07,
+      "loss": 1.8129,
+      "step": 7313
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.114024632499006e-07,
+      "loss": 1.7494,
+      "step": 7314
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.113627334127929e-07,
+      "loss": 1.8626,
+      "step": 7315
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.113230035756853e-07,
+      "loss": 1.4726,
+      "step": 7316
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.112832737385776e-07,
+      "loss": 1.8095,
+      "step": 7317
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.1124354390147e-07,
+      "loss": 1.5392,
+      "step": 7318
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.112038140643623e-07,
+      "loss": 1.8041,
+      "step": 7319
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.111640842272546e-07,
+      "loss": 1.2717,
+      "step": 7320
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.11124354390147e-07,
+      "loss": 1.9493,
+      "step": 7321
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.110846245530393e-07,
+      "loss": 1.8761,
+      "step": 7322
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.110448947159317e-07,
+      "loss": 1.9086,
+      "step": 7323
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.11005164878824e-07,
+      "loss": 1.7928,
+      "step": 7324
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.109654350417162e-07,
+      "loss": 1.556,
+      "step": 7325
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.109257052046085e-07,
+      "loss": 1.9346,
+      "step": 7326
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.108859753675009e-07,
+      "loss": 1.3604,
+      "step": 7327
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.108462455303933e-07,
+      "loss": 1.5888,
+      "step": 7328
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.108065156932856e-07,
+      "loss": 1.8643,
+      "step": 7329
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.10766785856178e-07,
+      "loss": 1.8667,
+      "step": 7330
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.107270560190702e-07,
+      "loss": 1.5032,
+      "step": 7331
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.106873261819626e-07,
+      "loss": 1.7195,
+      "step": 7332
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.10647596344855e-07,
+      "loss": 1.4149,
+      "step": 7333
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.106078665077473e-07,
+      "loss": 2.138,
+      "step": 7334
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.105681366706397e-07,
+      "loss": 1.284,
+      "step": 7335
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.10528406833532e-07,
+      "loss": 1.4648,
+      "step": 7336
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.104886769964243e-07,
+      "loss": 1.0376,
+      "step": 7337
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.104489471593166e-07,
+      "loss": 1.3669,
+      "step": 7338
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.104092173222089e-07,
+      "loss": 2.0191,
+      "step": 7339
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.103694874851012e-07,
+      "loss": 1.8476,
+      "step": 7340
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.103297576479936e-07,
+      "loss": 1.8439,
+      "step": 7341
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.10290027810886e-07,
+      "loss": 1.3647,
+      "step": 7342
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.102502979737782e-07,
+      "loss": 1.5914,
+      "step": 7343
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.102105681366706e-07,
+      "loss": 1.6016,
+      "step": 7344
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.101708382995629e-07,
+      "loss": 1.8189,
+      "step": 7345
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.101311084624553e-07,
+      "loss": 1.9483,
+      "step": 7346
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.100913786253476e-07,
+      "loss": 1.6607,
+      "step": 7347
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.1005164878824e-07,
+      "loss": 1.5207,
+      "step": 7348
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.100119189511322e-07,
+      "loss": 1.5414,
+      "step": 7349
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.099721891140246e-07,
+      "loss": 2.0535,
+      "step": 7350
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.09932459276917e-07,
+      "loss": 1.6961,
+      "step": 7351
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.098927294398092e-07,
+      "loss": 2.0283,
+      "step": 7352
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.098529996027016e-07,
+      "loss": 1.9512,
+      "step": 7353
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.098132697655939e-07,
+      "loss": 1.7767,
+      "step": 7354
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.097735399284862e-07,
+      "loss": 1.6263,
+      "step": 7355
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.097338100913785e-07,
+      "loss": 1.4319,
+      "step": 7356
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.096940802542709e-07,
+      "loss": 1.7229,
+      "step": 7357
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.096543504171633e-07,
+      "loss": 1.4256,
+      "step": 7358
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.096146205800556e-07,
+      "loss": 1.6585,
+      "step": 7359
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.09574890742948e-07,
+      "loss": 1.7211,
+      "step": 7360
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.095351609058402e-07,
+      "loss": 1.7464,
+      "step": 7361
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.094954310687326e-07,
+      "loss": 1.7363,
+      "step": 7362
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.094557012316249e-07,
+      "loss": 1.4719,
+      "step": 7363
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.094159713945173e-07,
+      "loss": 1.8028,
+      "step": 7364
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.093762415574096e-07,
+      "loss": 1.7608,
+      "step": 7365
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.093365117203019e-07,
+      "loss": 1.7839,
+      "step": 7366
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.092967818831942e-07,
+      "loss": 1.8953,
+      "step": 7367
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.092570520460865e-07,
+      "loss": 1.3518,
+      "step": 7368
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.092173222089789e-07,
+      "loss": 1.8443,
+      "step": 7369
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.091775923718712e-07,
+      "loss": 1.8199,
+      "step": 7370
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.091378625347636e-07,
+      "loss": 1.4866,
+      "step": 7371
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.09098132697656e-07,
+      "loss": 1.9145,
+      "step": 7372
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.090584028605482e-07,
+      "loss": 1.5404,
+      "step": 7373
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.090186730234406e-07,
+      "loss": 1.7205,
+      "step": 7374
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.089789431863329e-07,
+      "loss": 1.7637,
+      "step": 7375
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.089392133492253e-07,
+      "loss": 1.2997,
+      "step": 7376
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.088994835121176e-07,
+      "loss": 1.6861,
+      "step": 7377
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.0885975367501e-07,
+      "loss": 1.7756,
+      "step": 7378
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.088200238379021e-07,
+      "loss": 1.5716,
+      "step": 7379
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.087802940007945e-07,
+      "loss": 1.4139,
+      "step": 7380
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.087405641636869e-07,
+      "loss": 2.0741,
+      "step": 7381
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.087008343265792e-07,
+      "loss": 1.7266,
+      "step": 7382
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.086611044894716e-07,
+      "loss": 1.5792,
+      "step": 7383
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.086213746523639e-07,
+      "loss": 1.5252,
+      "step": 7384
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.085816448152562e-07,
+      "loss": 1.5855,
+      "step": 7385
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.085419149781485e-07,
+      "loss": 2.1579,
+      "step": 7386
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.085021851410409e-07,
+      "loss": 1.5292,
+      "step": 7387
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.084624553039333e-07,
+      "loss": 1.5879,
+      "step": 7388
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.084227254668256e-07,
+      "loss": 1.5241,
+      "step": 7389
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.08382995629718e-07,
+      "loss": 1.3324,
+      "step": 7390
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.083432657926101e-07,
+      "loss": 1.7363,
+      "step": 7391
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.083035359555025e-07,
+      "loss": 1.5572,
+      "step": 7392
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.082638061183948e-07,
+      "loss": 1.7857,
+      "step": 7393
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.082240762812872e-07,
+      "loss": 1.6396,
+      "step": 7394
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.081843464441795e-07,
+      "loss": 1.879,
+      "step": 7395
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.081446166070719e-07,
+      "loss": 1.6775,
+      "step": 7396
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.081048867699642e-07,
+      "loss": 1.8139,
+      "step": 7397
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.080651569328565e-07,
+      "loss": 1.8381,
+      "step": 7398
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.080254270957489e-07,
+      "loss": 1.7386,
+      "step": 7399
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.079856972586412e-07,
+      "loss": 1.2656,
+      "step": 7400
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.079459674215336e-07,
+      "loss": 1.7701,
+      "step": 7401
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.079062375844259e-07,
+      "loss": 1.43,
+      "step": 7402
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.078665077473182e-07,
+      "loss": 1.9047,
+      "step": 7403
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.078267779102106e-07,
+      "loss": 1.7359,
+      "step": 7404
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.077870480731028e-07,
+      "loss": 1.7912,
+      "step": 7405
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.077473182359952e-07,
+      "loss": 1.8668,
+      "step": 7406
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.077075883988875e-07,
+      "loss": 1.4563,
+      "step": 7407
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.076678585617799e-07,
+      "loss": 1.8371,
+      "step": 7408
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.076281287246721e-07,
+      "loss": 1.5752,
+      "step": 7409
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.075883988875645e-07,
+      "loss": 1.9148,
+      "step": 7410
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.075486690504568e-07,
+      "loss": 1.4828,
+      "step": 7411
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.075089392133492e-07,
+      "loss": 1.5821,
+      "step": 7412
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.074692093762416e-07,
+      "loss": 1.7672,
+      "step": 7413
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.074294795391339e-07,
+      "loss": 1.5402,
+      "step": 7414
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.073897497020262e-07,
+      "loss": 1.83,
+      "step": 7415
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.073500198649185e-07,
+      "loss": 1.8791,
+      "step": 7416
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.073102900278109e-07,
+      "loss": 1.3878,
+      "step": 7417
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.072705601907031e-07,
+      "loss": 2.1024,
+      "step": 7418
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.072308303535955e-07,
+      "loss": 1.4488,
+      "step": 7419
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.071911005164879e-07,
+      "loss": 1.6848,
+      "step": 7420
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.071513706793801e-07,
+      "loss": 1.59,
+      "step": 7421
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.071116408422725e-07,
+      "loss": 1.5786,
+      "step": 7422
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.070719110051648e-07,
+      "loss": 1.5876,
+      "step": 7423
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.070321811680572e-07,
+      "loss": 1.5563,
+      "step": 7424
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.069924513309495e-07,
+      "loss": 1.8949,
+      "step": 7425
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.069527214938419e-07,
+      "loss": 1.8605,
+      "step": 7426
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.069129916567342e-07,
+      "loss": 2.2168,
+      "step": 7427
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.068732618196265e-07,
+      "loss": 1.4834,
+      "step": 7428
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.068335319825189e-07,
+      "loss": 1.5257,
+      "step": 7429
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.067938021454112e-07,
+      "loss": 1.8033,
+      "step": 7430
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.067540723083035e-07,
+      "loss": 2.1088,
+      "step": 7431
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.067143424711958e-07,
+      "loss": 1.524,
+      "step": 7432
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.066746126340881e-07,
+      "loss": 1.8119,
+      "step": 7433
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.066348827969804e-07,
+      "loss": 1.633,
+      "step": 7434
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.065951529598728e-07,
+      "loss": 1.4934,
+      "step": 7435
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.065554231227652e-07,
+      "loss": 1.6155,
+      "step": 7436
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.065156932856575e-07,
+      "loss": 1.7291,
+      "step": 7437
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.064759634485499e-07,
+      "loss": 1.9996,
+      "step": 7438
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.064362336114421e-07,
+      "loss": 1.6706,
+      "step": 7439
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.063965037743345e-07,
+      "loss": 1.5744,
+      "step": 7440
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.063567739372268e-07,
+      "loss": 1.6856,
+      "step": 7441
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.063170441001192e-07,
+      "loss": 1.5686,
+      "step": 7442
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.062773142630116e-07,
+      "loss": 1.8524,
+      "step": 7443
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.062375844259039e-07,
+      "loss": 1.2835,
+      "step": 7444
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.061978545887961e-07,
+      "loss": 1.5495,
+      "step": 7445
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.061581247516884e-07,
+      "loss": 1.5688,
+      "step": 7446
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.061183949145808e-07,
+      "loss": 1.7237,
+      "step": 7447
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.060786650774731e-07,
+      "loss": 1.5593,
+      "step": 7448
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.060389352403655e-07,
+      "loss": 1.6605,
+      "step": 7449
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.059992054032579e-07,
+      "loss": 1.9275,
+      "step": 7450
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.059594755661501e-07,
+      "loss": 1.7898,
+      "step": 7451
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.059197457290425e-07,
+      "loss": 1.3157,
+      "step": 7452
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.058800158919348e-07,
+      "loss": 1.7592,
+      "step": 7453
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.058402860548272e-07,
+      "loss": 2.0135,
+      "step": 7454
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.058005562177195e-07,
+      "loss": 1.7318,
+      "step": 7455
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.057608263806119e-07,
+      "loss": 1.4259,
+      "step": 7456
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.05721096543504e-07,
+      "loss": 1.8764,
+      "step": 7457
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.056813667063964e-07,
+      "loss": 2.0189,
+      "step": 7458
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.056416368692888e-07,
+      "loss": 1.4605,
+      "step": 7459
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.056019070321811e-07,
+      "loss": 2.0141,
+      "step": 7460
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.055621771950735e-07,
+      "loss": 1.5134,
+      "step": 7461
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.055224473579658e-07,
+      "loss": 1.641,
+      "step": 7462
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.054827175208581e-07,
+      "loss": 1.6151,
+      "step": 7463
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.054429876837504e-07,
+      "loss": 1.1713,
+      "step": 7464
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.054032578466428e-07,
+      "loss": 2.0835,
+      "step": 7465
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.053635280095352e-07,
+      "loss": 1.8038,
+      "step": 7466
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.053237981724275e-07,
+      "loss": 1.3157,
+      "step": 7467
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.052840683353199e-07,
+      "loss": 1.787,
+      "step": 7468
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.052443384982121e-07,
+      "loss": 1.6711,
+      "step": 7469
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.052046086611045e-07,
+      "loss": 1.8547,
+      "step": 7470
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.051648788239967e-07,
+      "loss": 1.5621,
+      "step": 7471
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.051251489868891e-07,
+      "loss": 2.0841,
+      "step": 7472
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.050854191497814e-07,
+      "loss": 1.5793,
+      "step": 7473
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.050456893126738e-07,
+      "loss": 1.7249,
+      "step": 7474
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.050059594755661e-07,
+      "loss": 1.4006,
+      "step": 7475
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.049662296384584e-07,
+      "loss": 1.7108,
+      "step": 7476
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.049264998013508e-07,
+      "loss": 1.4756,
+      "step": 7477
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.048867699642431e-07,
+      "loss": 1.3939,
+      "step": 7478
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.048470401271355e-07,
+      "loss": 1.5476,
+      "step": 7479
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.048073102900278e-07,
+      "loss": 1.6226,
+      "step": 7480
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.047675804529201e-07,
+      "loss": 1.3885,
+      "step": 7481
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.047278506158125e-07,
+      "loss": 1.9135,
+      "step": 7482
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.046881207787048e-07,
+      "loss": 1.7449,
+      "step": 7483
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.046483909415971e-07,
+      "loss": 1.7048,
+      "step": 7484
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.046086611044894e-07,
+      "loss": 1.8827,
+      "step": 7485
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.045689312673818e-07,
+      "loss": 1.3849,
+      "step": 7486
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.04529201430274e-07,
+      "loss": 1.8564,
+      "step": 7487
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.044894715931664e-07,
+      "loss": 2.1266,
+      "step": 7488
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.044497417560587e-07,
+      "loss": 1.5886,
+      "step": 7489
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.044100119189511e-07,
+      "loss": 1.8869,
+      "step": 7490
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.043702820818435e-07,
+      "loss": 1.8485,
+      "step": 7491
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.043305522447358e-07,
+      "loss": 1.8504,
+      "step": 7492
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.042908224076281e-07,
+      "loss": 1.6305,
+      "step": 7493
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.042510925705204e-07,
+      "loss": 1.6886,
+      "step": 7494
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.042113627334128e-07,
+      "loss": 1.771,
+      "step": 7495
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.041716328963051e-07,
+      "loss": 1.6198,
+      "step": 7496
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.041319030591974e-07,
+      "loss": 1.5971,
+      "step": 7497
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.040921732220898e-07,
+      "loss": 1.3829,
+      "step": 7498
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.04052443384982e-07,
+      "loss": 1.8857,
+      "step": 7499
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.040127135478744e-07,
+      "loss": 1.5099,
+      "step": 7500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.039729837107667e-07,
+      "loss": 1.6306,
+      "step": 7501
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.039332538736591e-07,
+      "loss": 1.6668,
+      "step": 7502
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.038935240365514e-07,
+      "loss": 2.1559,
+      "step": 7503
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.038537941994438e-07,
+      "loss": 1.6935,
+      "step": 7504
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.03814064362336e-07,
+      "loss": 1.6927,
+      "step": 7505
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.037743345252284e-07,
+      "loss": 1.7407,
+      "step": 7506
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.037346046881208e-07,
+      "loss": 1.8941,
+      "step": 7507
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.036948748510131e-07,
+      "loss": 1.6523,
+      "step": 7508
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.036551450139055e-07,
+      "loss": 1.8146,
+      "step": 7509
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.036154151767978e-07,
+      "loss": 1.5546,
+      "step": 7510
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.0357568533969e-07,
+      "loss": 1.8491,
+      "step": 7511
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.035359555025823e-07,
+      "loss": 1.6328,
+      "step": 7512
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.034962256654747e-07,
+      "loss": 1.6595,
+      "step": 7513
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.034564958283671e-07,
+      "loss": 1.932,
+      "step": 7514
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.034167659912594e-07,
+      "loss": 1.6889,
+      "step": 7515
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.033770361541518e-07,
+      "loss": 1.6898,
+      "step": 7516
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.03337306317044e-07,
+      "loss": 2.0128,
+      "step": 7517
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.032975764799364e-07,
+      "loss": 1.573,
+      "step": 7518
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.032578466428287e-07,
+      "loss": 1.7799,
+      "step": 7519
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.032181168057211e-07,
+      "loss": 1.5427,
+      "step": 7520
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.031783869686135e-07,
+      "loss": 1.7175,
+      "step": 7521
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.031386571315058e-07,
+      "loss": 1.6694,
+      "step": 7522
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.03098927294398e-07,
+      "loss": 1.781,
+      "step": 7523
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.030591974572903e-07,
+      "loss": 1.4639,
+      "step": 7524
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.030194676201827e-07,
+      "loss": 1.4823,
+      "step": 7525
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.02979737783075e-07,
+      "loss": 1.7415,
+      "step": 7526
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.029400079459674e-07,
+      "loss": 1.5316,
+      "step": 7527
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.029002781088598e-07,
+      "loss": 1.552,
+      "step": 7528
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.02860548271752e-07,
+      "loss": 1.4673,
+      "step": 7529
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.028208184346444e-07,
+      "loss": 1.7824,
+      "step": 7530
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.027810885975367e-07,
+      "loss": 2.0128,
+      "step": 7531
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.027413587604291e-07,
+      "loss": 1.885,
+      "step": 7532
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.027016289233214e-07,
+      "loss": 1.6618,
+      "step": 7533
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.026618990862138e-07,
+      "loss": 1.4427,
+      "step": 7534
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.02622169249106e-07,
+      "loss": 1.6177,
+      "step": 7535
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.025824394119984e-07,
+      "loss": 2.0404,
+      "step": 7536
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.025427095748907e-07,
+      "loss": 1.6772,
+      "step": 7537
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.02502979737783e-07,
+      "loss": 1.7627,
+      "step": 7538
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.024632499006754e-07,
+      "loss": 1.7385,
+      "step": 7539
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.024235200635677e-07,
+      "loss": 1.6741,
+      "step": 7540
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.0238379022646e-07,
+      "loss": 1.511,
+      "step": 7541
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.023440603893523e-07,
+      "loss": 1.7282,
+      "step": 7542
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.023043305522447e-07,
+      "loss": 1.5067,
+      "step": 7543
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.022646007151371e-07,
+      "loss": 1.5706,
+      "step": 7544
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.022248708780294e-07,
+      "loss": 2.1949,
+      "step": 7545
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.021851410409218e-07,
+      "loss": 1.4927,
+      "step": 7546
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.02145411203814e-07,
+      "loss": 1.5441,
+      "step": 7547
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.021056813667064e-07,
+      "loss": 1.7537,
+      "step": 7548
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.020659515295987e-07,
+      "loss": 2.1477,
+      "step": 7549
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.02026221692491e-07,
+      "loss": 1.5307,
+      "step": 7550
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.019864918553833e-07,
+      "loss": 1.905,
+      "step": 7551
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.019467620182757e-07,
+      "loss": 1.584,
+      "step": 7552
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.01907032181168e-07,
+      "loss": 1.4458,
+      "step": 7553
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.018673023440603e-07,
+      "loss": 2.1911,
+      "step": 7554
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.018275725069527e-07,
+      "loss": 1.3465,
+      "step": 7555
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.01787842669845e-07,
+      "loss": 2.0939,
+      "step": 7556
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.017481128327374e-07,
+      "loss": 1.689,
+      "step": 7557
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.017083829956297e-07,
+      "loss": 1.8063,
+      "step": 7558
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.01668653158522e-07,
+      "loss": 1.8802,
+      "step": 7559
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.016289233214144e-07,
+      "loss": 2.0673,
+      "step": 7560
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.015891934843067e-07,
+      "loss": 1.6337,
+      "step": 7561
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.015494636471991e-07,
+      "loss": 1.5003,
+      "step": 7562
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.015097338100913e-07,
+      "loss": 1.6719,
+      "step": 7563
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.014700039729837e-07,
+      "loss": 1.676,
+      "step": 7564
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.014302741358759e-07,
+      "loss": 1.9884,
+      "step": 7565
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.013905442987683e-07,
+      "loss": 1.6707,
+      "step": 7566
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.35963521015067407,
+      "eval_loss": 1.746188998222351,
+      "eval_runtime": 6.4737,
+      "eval_samples_per_second": 389.577,
+      "eval_steps_per_second": 6.179,
+      "step": 7566
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.013508144616606e-07,
+      "loss": 1.4994,
+      "step": 7567
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.01311084624553e-07,
+      "loss": 1.6001,
+      "step": 7568
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.012713547874454e-07,
+      "loss": 1.5145,
+      "step": 7569
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.012316249503377e-07,
+      "loss": 1.585,
+      "step": 7570
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.0119189511323e-07,
+      "loss": 1.8704,
+      "step": 7571
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.011521652761223e-07,
+      "loss": 1.9234,
+      "step": 7572
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.011124354390147e-07,
+      "loss": 1.431,
+      "step": 7573
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.01072705601907e-07,
+      "loss": 1.4208,
+      "step": 7574
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.010329757647994e-07,
+      "loss": 1.5641,
+      "step": 7575
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.009932459276918e-07,
+      "loss": 1.5309,
+      "step": 7576
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.009535160905839e-07,
+      "loss": 1.4077,
+      "step": 7577
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.009137862534763e-07,
+      "loss": 1.4246,
+      "step": 7578
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.008740564163686e-07,
+      "loss": 1.3954,
+      "step": 7579
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.00834326579261e-07,
+      "loss": 1.6836,
+      "step": 7580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.007945967421533e-07,
+      "loss": 1.5851,
+      "step": 7581
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.007548669050457e-07,
+      "loss": 1.8886,
+      "step": 7582
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.00715137067938e-07,
+      "loss": 1.7488,
+      "step": 7583
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.006754072308303e-07,
+      "loss": 1.217,
+      "step": 7584
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.006356773937227e-07,
+      "loss": 1.8373,
+      "step": 7585
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.00595947556615e-07,
+      "loss": 1.4612,
+      "step": 7586
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.005562177195074e-07,
+      "loss": 1.643,
+      "step": 7587
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.005164878823997e-07,
+      "loss": 1.5442,
+      "step": 7588
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.004767580452919e-07,
+      "loss": 2.3239,
+      "step": 7589
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.004370282081842e-07,
+      "loss": 1.6473,
+      "step": 7590
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.003972983710766e-07,
+      "loss": 1.4989,
+      "step": 7591
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.00357568533969e-07,
+      "loss": 1.7087,
+      "step": 7592
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.003178386968613e-07,
+      "loss": 1.7373,
+      "step": 7593
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.002781088597537e-07,
+      "loss": 1.6631,
+      "step": 7594
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.002383790226459e-07,
+      "loss": 1.6842,
+      "step": 7595
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.001986491855383e-07,
+      "loss": 1.7377,
+      "step": 7596
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.001589193484306e-07,
+      "loss": 1.477,
+      "step": 7597
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.00119189511323e-07,
+      "loss": 1.7234,
+      "step": 7598
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.000794596742154e-07,
+      "loss": 1.6073,
+      "step": 7599
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.000397298371077e-07,
+      "loss": 1.7217,
+      "step": 7600
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7e-07,
+      "loss": 1.8383,
+      "step": 7601
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.999602701628923e-07,
+      "loss": 1.3308,
+      "step": 7602
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.999205403257846e-07,
+      "loss": 1.527,
+      "step": 7603
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.998808104886769e-07,
+      "loss": 1.5942,
+      "step": 7604
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.998410806515693e-07,
+      "loss": 2.2479,
+      "step": 7605
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.998013508144617e-07,
+      "loss": 1.7822,
+      "step": 7606
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.997616209773539e-07,
+      "loss": 1.8488,
+      "step": 7607
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.997218911402463e-07,
+      "loss": 1.3482,
+      "step": 7608
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.996821613031386e-07,
+      "loss": 1.7669,
+      "step": 7609
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.99642431466031e-07,
+      "loss": 1.457,
+      "step": 7610
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.996027016289233e-07,
+      "loss": 1.8106,
+      "step": 7611
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.995629717918157e-07,
+      "loss": 1.9292,
+      "step": 7612
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.995232419547079e-07,
+      "loss": 1.6104,
+      "step": 7613
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.994835121176003e-07,
+      "loss": 1.4482,
+      "step": 7614
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.994437822804927e-07,
+      "loss": 1.673,
+      "step": 7615
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.994040524433849e-07,
+      "loss": 1.5689,
+      "step": 7616
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.993643226062773e-07,
+      "loss": 1.4068,
+      "step": 7617
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.993245927691696e-07,
+      "loss": 1.9532,
+      "step": 7618
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.992848629320619e-07,
+      "loss": 1.6443,
+      "step": 7619
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.992451330949542e-07,
+      "loss": 1.4392,
+      "step": 7620
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.992054032578466e-07,
+      "loss": 1.4014,
+      "step": 7621
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.99165673420739e-07,
+      "loss": 1.7547,
+      "step": 7622
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.991259435836313e-07,
+      "loss": 1.6391,
+      "step": 7623
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.990862137465237e-07,
+      "loss": 1.7263,
+      "step": 7624
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.990464839094159e-07,
+      "loss": 1.6192,
+      "step": 7625
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.990067540723083e-07,
+      "loss": 1.4632,
+      "step": 7626
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.989670242352006e-07,
+      "loss": 1.6542,
+      "step": 7627
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.98927294398093e-07,
+      "loss": 1.8146,
+      "step": 7628
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.988875645609852e-07,
+      "loss": 1.6722,
+      "step": 7629
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.988478347238776e-07,
+      "loss": 1.3838,
+      "step": 7630
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.988081048867699e-07,
+      "loss": 2.0007,
+      "step": 7631
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.987683750496622e-07,
+      "loss": 2.3374,
+      "step": 7632
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.987286452125546e-07,
+      "loss": 1.7502,
+      "step": 7633
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.986889153754469e-07,
+      "loss": 1.6935,
+      "step": 7634
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.986491855383393e-07,
+      "loss": 1.3141,
+      "step": 7635
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.986094557012316e-07,
+      "loss": 1.614,
+      "step": 7636
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.985697258641239e-07,
+      "loss": 1.6131,
+      "step": 7637
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.985299960270163e-07,
+      "loss": 1.7203,
+      "step": 7638
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.984902661899086e-07,
+      "loss": 1.1788,
+      "step": 7639
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.98450536352801e-07,
+      "loss": 1.759,
+      "step": 7640
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.984108065156933e-07,
+      "loss": 1.4829,
+      "step": 7641
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.983710766785857e-07,
+      "loss": 2.165,
+      "step": 7642
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.983313468414778e-07,
+      "loss": 1.5325,
+      "step": 7643
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.982916170043702e-07,
+      "loss": 1.9165,
+      "step": 7644
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.982518871672625e-07,
+      "loss": 1.8993,
+      "step": 7645
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.982121573301549e-07,
+      "loss": 2.3416,
+      "step": 7646
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.981724274930473e-07,
+      "loss": 1.5787,
+      "step": 7647
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.981326976559396e-07,
+      "loss": 1.2935,
+      "step": 7648
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.980929678188319e-07,
+      "loss": 1.9063,
+      "step": 7649
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.980532379817242e-07,
+      "loss": 1.6099,
+      "step": 7650
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.980135081446166e-07,
+      "loss": 1.6787,
+      "step": 7651
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.97973778307509e-07,
+      "loss": 1.707,
+      "step": 7652
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.979340484704013e-07,
+      "loss": 1.4995,
+      "step": 7653
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.978943186332937e-07,
+      "loss": 1.7541,
+      "step": 7654
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.978545887961858e-07,
+      "loss": 1.6109,
+      "step": 7655
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.978148589590782e-07,
+      "loss": 1.6656,
+      "step": 7656
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.977751291219705e-07,
+      "loss": 1.6333,
+      "step": 7657
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.977353992848629e-07,
+      "loss": 1.9354,
+      "step": 7658
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.976956694477552e-07,
+      "loss": 1.786,
+      "step": 7659
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.976559396106476e-07,
+      "loss": 2.1293,
+      "step": 7660
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.976162097735399e-07,
+      "loss": 1.8838,
+      "step": 7661
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.975764799364322e-07,
+      "loss": 1.4727,
+      "step": 7662
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.975367500993246e-07,
+      "loss": 1.3802,
+      "step": 7663
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.974970202622169e-07,
+      "loss": 1.734,
+      "step": 7664
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.974572904251093e-07,
+      "loss": 1.9799,
+      "step": 7665
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.974175605880016e-07,
+      "loss": 1.7788,
+      "step": 7666
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.973778307508939e-07,
+      "loss": 1.355,
+      "step": 7667
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.973381009137862e-07,
+      "loss": 1.5835,
+      "step": 7668
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.972983710766785e-07,
+      "loss": 1.5417,
+      "step": 7669
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.972586412395709e-07,
+      "loss": 1.3101,
+      "step": 7670
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.972189114024632e-07,
+      "loss": 1.6966,
+      "step": 7671
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.971791815653556e-07,
+      "loss": 1.4809,
+      "step": 7672
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.971394517282478e-07,
+      "loss": 1.6915,
+      "step": 7673
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.970997218911402e-07,
+      "loss": 1.5547,
+      "step": 7674
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.970599920540325e-07,
+      "loss": 1.6594,
+      "step": 7675
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.970202622169249e-07,
+      "loss": 1.5341,
+      "step": 7676
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.969805323798173e-07,
+      "loss": 1.4256,
+      "step": 7677
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.969408025427096e-07,
+      "loss": 1.723,
+      "step": 7678
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.969010727056019e-07,
+      "loss": 1.799,
+      "step": 7679
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.968613428684942e-07,
+      "loss": 1.5367,
+      "step": 7680
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.968216130313866e-07,
+      "loss": 1.7149,
+      "step": 7681
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.967818831942788e-07,
+      "loss": 2.1062,
+      "step": 7682
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.967421533571712e-07,
+      "loss": 1.8583,
+      "step": 7683
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.967024235200636e-07,
+      "loss": 1.9166,
+      "step": 7684
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.966626936829558e-07,
+      "loss": 1.7656,
+      "step": 7685
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.966229638458482e-07,
+      "loss": 1.6965,
+      "step": 7686
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.965832340087405e-07,
+      "loss": 1.7421,
+      "step": 7687
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.965435041716329e-07,
+      "loss": 1.65,
+      "step": 7688
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.965037743345252e-07,
+      "loss": 1.2596,
+      "step": 7689
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.964640444974176e-07,
+      "loss": 1.8131,
+      "step": 7690
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.964243146603098e-07,
+      "loss": 1.7319,
+      "step": 7691
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.963845848232022e-07,
+      "loss": 1.9713,
+      "step": 7692
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.963448549860946e-07,
+      "loss": 1.8014,
+      "step": 7693
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.963051251489869e-07,
+      "loss": 1.773,
+      "step": 7694
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.962653953118792e-07,
+      "loss": 1.7683,
+      "step": 7695
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.962256654747715e-07,
+      "loss": 1.6763,
+      "step": 7696
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.961859356376638e-07,
+      "loss": 2.0577,
+      "step": 7697
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.961462058005561e-07,
+      "loss": 1.7331,
+      "step": 7698
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.961064759634485e-07,
+      "loss": 1.6176,
+      "step": 7699
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.960667461263409e-07,
+      "loss": 1.7567,
+      "step": 7700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.960270162892332e-07,
+      "loss": 1.751,
+      "step": 7701
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.959872864521256e-07,
+      "loss": 1.8849,
+      "step": 7702
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.959475566150178e-07,
+      "loss": 1.7484,
+      "step": 7703
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.959078267779102e-07,
+      "loss": 1.7255,
+      "step": 7704
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.958680969408025e-07,
+      "loss": 1.787,
+      "step": 7705
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.958283671036949e-07,
+      "loss": 1.7154,
+      "step": 7706
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.957886372665873e-07,
+      "loss": 1.3647,
+      "step": 7707
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.957489074294796e-07,
+      "loss": 1.7847,
+      "step": 7708
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.957091775923718e-07,
+      "loss": 1.6504,
+      "step": 7709
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.956694477552641e-07,
+      "loss": 1.4791,
+      "step": 7710
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.956297179181565e-07,
+      "loss": 1.6428,
+      "step": 7711
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.955899880810488e-07,
+      "loss": 2.4155,
+      "step": 7712
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.955502582439412e-07,
+      "loss": 1.744,
+      "step": 7713
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.955105284068335e-07,
+      "loss": 2.0587,
+      "step": 7714
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.954707985697258e-07,
+      "loss": 2.0095,
+      "step": 7715
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.954310687326182e-07,
+      "loss": 1.525,
+      "step": 7716
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.953913388955105e-07,
+      "loss": 1.3864,
+      "step": 7717
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.953516090584029e-07,
+      "loss": 1.5447,
+      "step": 7718
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.953118792212952e-07,
+      "loss": 1.9802,
+      "step": 7719
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.952721493841876e-07,
+      "loss": 1.794,
+      "step": 7720
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.952324195470797e-07,
+      "loss": 1.3424,
+      "step": 7721
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.951926897099721e-07,
+      "loss": 1.7073,
+      "step": 7722
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.951529598728644e-07,
+      "loss": 1.7318,
+      "step": 7723
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.951132300357568e-07,
+      "loss": 1.7782,
+      "step": 7724
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.950735001986492e-07,
+      "loss": 1.8315,
+      "step": 7725
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.950337703615415e-07,
+      "loss": 1.6299,
+      "step": 7726
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.949940405244338e-07,
+      "loss": 1.6098,
+      "step": 7727
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.949543106873261e-07,
+      "loss": 1.6629,
+      "step": 7728
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.949145808502185e-07,
+      "loss": 2.4283,
+      "step": 7729
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.948748510131108e-07,
+      "loss": 1.4085,
+      "step": 7730
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.948351211760032e-07,
+      "loss": 1.587,
+      "step": 7731
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.947953913388956e-07,
+      "loss": 1.5386,
+      "step": 7732
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.947556615017878e-07,
+      "loss": 1.7415,
+      "step": 7733
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.947159316646802e-07,
+      "loss": 2.0217,
+      "step": 7734
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.946762018275724e-07,
+      "loss": 1.9007,
+      "step": 7735
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.946364719904648e-07,
+      "loss": 1.8655,
+      "step": 7736
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.945967421533571e-07,
+      "loss": 1.9393,
+      "step": 7737
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.945570123162495e-07,
+      "loss": 1.6026,
+      "step": 7738
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.945172824791418e-07,
+      "loss": 2.0738,
+      "step": 7739
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.944775526420341e-07,
+      "loss": 1.5,
+      "step": 7740
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.944378228049265e-07,
+      "loss": 1.6325,
+      "step": 7741
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.943980929678188e-07,
+      "loss": 1.3235,
+      "step": 7742
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.943583631307112e-07,
+      "loss": 1.5573,
+      "step": 7743
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.943186332936035e-07,
+      "loss": 1.6119,
+      "step": 7744
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.942789034564958e-07,
+      "loss": 1.8676,
+      "step": 7745
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.942391736193882e-07,
+      "loss": 1.8162,
+      "step": 7746
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.941994437822805e-07,
+      "loss": 1.6468,
+      "step": 7747
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.941597139451728e-07,
+      "loss": 1.5151,
+      "step": 7748
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.941199841080651e-07,
+      "loss": 1.8789,
+      "step": 7749
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.940802542709575e-07,
+      "loss": 1.8129,
+      "step": 7750
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.940405244338497e-07,
+      "loss": 1.6436,
+      "step": 7751
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.940007945967421e-07,
+      "loss": 1.2864,
+      "step": 7752
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.939610647596344e-07,
+      "loss": 1.4842,
+      "step": 7753
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.939213349225268e-07,
+      "loss": 1.492,
+      "step": 7754
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.938816050854192e-07,
+      "loss": 1.5369,
+      "step": 7755
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.938418752483115e-07,
+      "loss": 1.952,
+      "step": 7756
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.938021454112038e-07,
+      "loss": 1.6277,
+      "step": 7757
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.937624155740961e-07,
+      "loss": 1.6821,
+      "step": 7758
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.937226857369885e-07,
+      "loss": 1.3431,
+      "step": 7759
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.936829558998808e-07,
+      "loss": 1.7434,
+      "step": 7760
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.936432260627731e-07,
+      "loss": 1.7333,
+      "step": 7761
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.936034962256655e-07,
+      "loss": 1.4682,
+      "step": 7762
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.935637663885577e-07,
+      "loss": 1.5573,
+      "step": 7763
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.935240365514501e-07,
+      "loss": 1.7322,
+      "step": 7764
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.934843067143424e-07,
+      "loss": 2.1305,
+      "step": 7765
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.934445768772348e-07,
+      "loss": 2.0259,
+      "step": 7766
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.934048470401271e-07,
+      "loss": 1.8299,
+      "step": 7767
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.933651172030195e-07,
+      "loss": 1.664,
+      "step": 7768
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.933253873659117e-07,
+      "loss": 2.039,
+      "step": 7769
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.932856575288041e-07,
+      "loss": 2.0467,
+      "step": 7770
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.932459276916965e-07,
+      "loss": 1.5798,
+      "step": 7771
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.932061978545888e-07,
+      "loss": 1.6012,
+      "step": 7772
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.931664680174812e-07,
+      "loss": 1.4652,
+      "step": 7773
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.931267381803735e-07,
+      "loss": 1.6286,
+      "step": 7774
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.930870083432657e-07,
+      "loss": 1.8316,
+      "step": 7775
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.93047278506158e-07,
+      "loss": 1.8698,
+      "step": 7776
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.930075486690504e-07,
+      "loss": 2.4699,
+      "step": 7777
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.929678188319428e-07,
+      "loss": 1.7921,
+      "step": 7778
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.929280889948351e-07,
+      "loss": 1.6247,
+      "step": 7779
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.928883591577275e-07,
+      "loss": 1.3978,
+      "step": 7780
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.928486293206197e-07,
+      "loss": 1.5345,
+      "step": 7781
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.928088994835121e-07,
+      "loss": 1.6817,
+      "step": 7782
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.927691696464044e-07,
+      "loss": 1.8304,
+      "step": 7783
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.927294398092968e-07,
+      "loss": 1.7309,
+      "step": 7784
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.926897099721892e-07,
+      "loss": 1.8441,
+      "step": 7785
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.926499801350815e-07,
+      "loss": 1.759,
+      "step": 7786
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.926102502979737e-07,
+      "loss": 1.7852,
+      "step": 7787
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.92570520460866e-07,
+      "loss": 1.41,
+      "step": 7788
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.925307906237584e-07,
+      "loss": 1.8635,
+      "step": 7789
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.924910607866507e-07,
+      "loss": 1.7273,
+      "step": 7790
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.924513309495431e-07,
+      "loss": 1.5586,
+      "step": 7791
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.924116011124354e-07,
+      "loss": 1.7679,
+      "step": 7792
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.923718712753277e-07,
+      "loss": 1.7422,
+      "step": 7793
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.923321414382201e-07,
+      "loss": 1.8753,
+      "step": 7794
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.922924116011124e-07,
+      "loss": 1.8448,
+      "step": 7795
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.922526817640048e-07,
+      "loss": 1.9003,
+      "step": 7796
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.922129519268971e-07,
+      "loss": 1.7641,
+      "step": 7797
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.921732220897895e-07,
+      "loss": 1.5071,
+      "step": 7798
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.921334922526817e-07,
+      "loss": 1.6805,
+      "step": 7799
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.920937624155741e-07,
+      "loss": 1.704,
+      "step": 7800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.920540325784663e-07,
+      "loss": 1.7923,
+      "step": 7801
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.920143027413587e-07,
+      "loss": 1.541,
+      "step": 7802
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.919745729042511e-07,
+      "loss": 1.7642,
+      "step": 7803
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.919348430671434e-07,
+      "loss": 1.625,
+      "step": 7804
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.918951132300357e-07,
+      "loss": 1.7586,
+      "step": 7805
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.91855383392928e-07,
+      "loss": 1.6091,
+      "step": 7806
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.918156535558204e-07,
+      "loss": 1.5784,
+      "step": 7807
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.917759237187127e-07,
+      "loss": 1.6447,
+      "step": 7808
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.917361938816051e-07,
+      "loss": 1.0166,
+      "step": 7809
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.916964640444975e-07,
+      "loss": 1.712,
+      "step": 7810
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.916567342073897e-07,
+      "loss": 1.708,
+      "step": 7811
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.916170043702821e-07,
+      "loss": 2.0055,
+      "step": 7812
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.915772745331744e-07,
+      "loss": 1.9189,
+      "step": 7813
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.915375446960667e-07,
+      "loss": 1.9399,
+      "step": 7814
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.91497814858959e-07,
+      "loss": 1.6943,
+      "step": 7815
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.914580850218514e-07,
+      "loss": 1.7506,
+      "step": 7816
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.914183551847437e-07,
+      "loss": 1.7056,
+      "step": 7817
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.91378625347636e-07,
+      "loss": 1.6329,
+      "step": 7818
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.913388955105284e-07,
+      "loss": 1.6916,
+      "step": 7819
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.912991656734207e-07,
+      "loss": 1.9857,
+      "step": 7820
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.912594358363131e-07,
+      "loss": 2.0443,
+      "step": 7821
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.912197059992054e-07,
+      "loss": 1.3995,
+      "step": 7822
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.911799761620977e-07,
+      "loss": 1.327,
+      "step": 7823
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.9114024632499e-07,
+      "loss": 1.8451,
+      "step": 7824
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.911005164878824e-07,
+      "loss": 1.807,
+      "step": 7825
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.910607866507748e-07,
+      "loss": 1.4081,
+      "step": 7826
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.91021056813667e-07,
+      "loss": 1.8734,
+      "step": 7827
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.909813269765594e-07,
+      "loss": 1.391,
+      "step": 7828
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.909415971394516e-07,
+      "loss": 1.6453,
+      "step": 7829
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.90901867302344e-07,
+      "loss": 1.8092,
+      "step": 7830
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.908621374652363e-07,
+      "loss": 1.2753,
+      "step": 7831
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.908224076281287e-07,
+      "loss": 1.6285,
+      "step": 7832
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.907826777910211e-07,
+      "loss": 1.6639,
+      "step": 7833
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.907429479539134e-07,
+      "loss": 1.7805,
+      "step": 7834
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.907032181168057e-07,
+      "loss": 1.8032,
+      "step": 7835
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.90663488279698e-07,
+      "loss": 1.176,
+      "step": 7836
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.906237584425904e-07,
+      "loss": 1.7649,
+      "step": 7837
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.905840286054827e-07,
+      "loss": 1.182,
+      "step": 7838
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.905442987683751e-07,
+      "loss": 1.8068,
+      "step": 7839
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.905045689312675e-07,
+      "loss": 1.7932,
+      "step": 7840
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.904648390941596e-07,
+      "loss": 1.9162,
+      "step": 7841
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.90425109257052e-07,
+      "loss": 1.3965,
+      "step": 7842
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.903853794199443e-07,
+      "loss": 1.7579,
+      "step": 7843
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.903456495828367e-07,
+      "loss": 1.5563,
+      "step": 7844
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.90305919745729e-07,
+      "loss": 1.5031,
+      "step": 7845
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.902661899086213e-07,
+      "loss": 1.8442,
+      "step": 7846
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.902264600715136e-07,
+      "loss": 1.7241,
+      "step": 7847
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.90186730234406e-07,
+      "loss": 1.6944,
+      "step": 7848
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.901470003972984e-07,
+      "loss": 1.7528,
+      "step": 7849
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.901072705601907e-07,
+      "loss": 1.883,
+      "step": 7850
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.900675407230831e-07,
+      "loss": 1.4583,
+      "step": 7851
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.900278108859753e-07,
+      "loss": 1.6789,
+      "step": 7852
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.899880810488676e-07,
+      "loss": 1.6248,
+      "step": 7853
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.899483512117599e-07,
+      "loss": 1.7461,
+      "step": 7854
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.899086213746523e-07,
+      "loss": 1.3687,
+      "step": 7855
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.898688915375447e-07,
+      "loss": 1.4338,
+      "step": 7856
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.89829161700437e-07,
+      "loss": 1.7082,
+      "step": 7857
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.897894318633293e-07,
+      "loss": 1.7682,
+      "step": 7858
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.897497020262216e-07,
+      "loss": 1.9335,
+      "step": 7859
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.89709972189114e-07,
+      "loss": 1.8239,
+      "step": 7860
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.896702423520063e-07,
+      "loss": 2.1638,
+      "step": 7861
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.896305125148987e-07,
+      "loss": 1.3317,
+      "step": 7862
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.895907826777911e-07,
+      "loss": 2.0178,
+      "step": 7863
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.895510528406833e-07,
+      "loss": 1.4152,
+      "step": 7864
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.895113230035757e-07,
+      "loss": 1.6993,
+      "step": 7865
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.89471593166468e-07,
+      "loss": 1.9244,
+      "step": 7866
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.894318633293603e-07,
+      "loss": 1.3619,
+      "step": 7867
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.893921334922526e-07,
+      "loss": 1.4659,
+      "step": 7868
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.89352403655145e-07,
+      "loss": 2.2673,
+      "step": 7869
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.893126738180372e-07,
+      "loss": 1.8126,
+      "step": 7870
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.892729439809296e-07,
+      "loss": 1.4575,
+      "step": 7871
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.89233214143822e-07,
+      "loss": 1.0748,
+      "step": 7872
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.891934843067143e-07,
+      "loss": 1.897,
+      "step": 7873
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.891537544696067e-07,
+      "loss": 1.6445,
+      "step": 7874
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.89114024632499e-07,
+      "loss": 1.2005,
+      "step": 7875
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.890742947953913e-07,
+      "loss": 1.8893,
+      "step": 7876
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.890345649582836e-07,
+      "loss": 1.9941,
+      "step": 7877
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.88994835121176e-07,
+      "loss": 1.6595,
+      "step": 7878
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.889551052840684e-07,
+      "loss": 2.037,
+      "step": 7879
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.889153754469606e-07,
+      "loss": 1.5827,
+      "step": 7880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.88875645609853e-07,
+      "loss": 1.6732,
+      "step": 7881
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.888359157727452e-07,
+      "loss": 1.4736,
+      "step": 7882
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.887961859356376e-07,
+      "loss": 1.7616,
+      "step": 7883
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.887564560985299e-07,
+      "loss": 1.6268,
+      "step": 7884
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.887167262614223e-07,
+      "loss": 1.7376,
+      "step": 7885
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.886769964243146e-07,
+      "loss": 1.6209,
+      "step": 7886
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.88637266587207e-07,
+      "loss": 1.6831,
+      "step": 7887
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.885975367500993e-07,
+      "loss": 1.6705,
+      "step": 7888
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.885578069129916e-07,
+      "loss": 1.8349,
+      "step": 7889
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.88518077075884e-07,
+      "loss": 1.6691,
+      "step": 7890
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.884783472387763e-07,
+      "loss": 1.754,
+      "step": 7891
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.884386174016687e-07,
+      "loss": 1.4655,
+      "step": 7892
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.883988875645609e-07,
+      "loss": 1.7706,
+      "step": 7893
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.883591577274532e-07,
+      "loss": 1.9301,
+      "step": 7894
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.883194278903456e-07,
+      "loss": 1.3177,
+      "step": 7895
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.882796980532379e-07,
+      "loss": 1.7196,
+      "step": 7896
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.882399682161303e-07,
+      "loss": 1.83,
+      "step": 7897
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.882002383790226e-07,
+      "loss": 1.5762,
+      "step": 7898
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.88160508541915e-07,
+      "loss": 1.6587,
+      "step": 7899
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.881207787048072e-07,
+      "loss": 1.3424,
+      "step": 7900
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.880810488676996e-07,
+      "loss": 1.833,
+      "step": 7901
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.88041319030592e-07,
+      "loss": 1.0026,
+      "step": 7902
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.880015891934843e-07,
+      "loss": 1.6826,
+      "step": 7903
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.879618593563767e-07,
+      "loss": 1.7758,
+      "step": 7904
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.87922129519269e-07,
+      "loss": 1.7491,
+      "step": 7905
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.878823996821612e-07,
+      "loss": 1.6524,
+      "step": 7906
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.878426698450535e-07,
+      "loss": 1.8549,
+      "step": 7907
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.878029400079459e-07,
+      "loss": 1.776,
+      "step": 7908
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.877632101708382e-07,
+      "loss": 1.6729,
+      "step": 7909
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.877234803337306e-07,
+      "loss": 1.6463,
+      "step": 7910
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.87683750496623e-07,
+      "loss": 1.6676,
+      "step": 7911
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.876440206595152e-07,
+      "loss": 1.4484,
+      "step": 7912
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.876042908224076e-07,
+      "loss": 1.3766,
+      "step": 7913
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.875645609852999e-07,
+      "loss": 1.5595,
+      "step": 7914
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.875248311481923e-07,
+      "loss": 1.9063,
+      "step": 7915
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.874851013110846e-07,
+      "loss": 1.3315,
+      "step": 7916
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.87445371473977e-07,
+      "loss": 2.3717,
+      "step": 7917
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.874056416368693e-07,
+      "loss": 1.9381,
+      "step": 7918
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.873659117997615e-07,
+      "loss": 2.0617,
+      "step": 7919
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.873261819626539e-07,
+      "loss": 1.5808,
+      "step": 7920
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.872864521255462e-07,
+      "loss": 1.8675,
+      "step": 7921
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.872467222884386e-07,
+      "loss": 1.6671,
+      "step": 7922
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.872069924513309e-07,
+      "loss": 2.3145,
+      "step": 7923
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.871672626142232e-07,
+      "loss": 1.4818,
+      "step": 7924
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.871275327771155e-07,
+      "loss": 1.8294,
+      "step": 7925
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.870878029400079e-07,
+      "loss": 1.6921,
+      "step": 7926
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.870480731029003e-07,
+      "loss": 1.5283,
+      "step": 7927
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.870083432657926e-07,
+      "loss": 1.6373,
+      "step": 7928
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.86968613428685e-07,
+      "loss": 1.9038,
+      "step": 7929
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.869288835915772e-07,
+      "loss": 1.6588,
+      "step": 7930
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.868891537544696e-07,
+      "loss": 1.7738,
+      "step": 7931
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.868494239173619e-07,
+      "loss": 1.8014,
+      "step": 7932
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.868096940802542e-07,
+      "loss": 1.525,
+      "step": 7933
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.867699642431466e-07,
+      "loss": 1.7608,
+      "step": 7934
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.867302344060389e-07,
+      "loss": 1.7643,
+      "step": 7935
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.866905045689312e-07,
+      "loss": 1.5957,
+      "step": 7936
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.866507747318235e-07,
+      "loss": 1.3983,
+      "step": 7937
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.866110448947159e-07,
+      "loss": 1.6874,
+      "step": 7938
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.865713150576082e-07,
+      "loss": 1.4576,
+      "step": 7939
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.865315852205006e-07,
+      "loss": 1.363,
+      "step": 7940
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.86491855383393e-07,
+      "loss": 1.6274,
+      "step": 7941
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.864521255462852e-07,
+      "loss": 1.5783,
+      "step": 7942
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.864123957091776e-07,
+      "loss": 1.6772,
+      "step": 7943
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.863726658720699e-07,
+      "loss": 1.5948,
+      "step": 7944
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.863329360349623e-07,
+      "loss": 1.9576,
+      "step": 7945
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.862932061978545e-07,
+      "loss": 1.5562,
+      "step": 7946
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.862534763607469e-07,
+      "loss": 1.277,
+      "step": 7947
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.862137465236391e-07,
+      "loss": 1.668,
+      "step": 7948
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.861740166865315e-07,
+      "loss": 1.9013,
+      "step": 7949
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.861342868494239e-07,
+      "loss": 1.2545,
+      "step": 7950
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.860945570123162e-07,
+      "loss": 1.7834,
+      "step": 7951
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.860548271752086e-07,
+      "loss": 1.2978,
+      "step": 7952
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.860150973381009e-07,
+      "loss": 1.6705,
+      "step": 7953
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.859753675009932e-07,
+      "loss": 1.7461,
+      "step": 7954
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.859356376638855e-07,
+      "loss": 1.7283,
+      "step": 7955
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.858959078267779e-07,
+      "loss": 1.7466,
+      "step": 7956
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.858561779896703e-07,
+      "loss": 1.8067,
+      "step": 7957
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.858164481525626e-07,
+      "loss": 1.6551,
+      "step": 7958
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.857767183154549e-07,
+      "loss": 1.3622,
+      "step": 7959
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.857369884783471e-07,
+      "loss": 1.6438,
+      "step": 7960
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.856972586412395e-07,
+      "loss": 2.0337,
+      "step": 7961
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.856575288041318e-07,
+      "loss": 1.5161,
+      "step": 7962
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.856177989670242e-07,
+      "loss": 1.6517,
+      "step": 7963
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.855780691299165e-07,
+      "loss": 1.8651,
+      "step": 7964
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.855383392928089e-07,
+      "loss": 1.476,
+      "step": 7965
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.854986094557012e-07,
+      "loss": 1.6136,
+      "step": 7966
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.854588796185935e-07,
+      "loss": 1.6336,
+      "step": 7967
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.854191497814859e-07,
+      "loss": 1.2105,
+      "step": 7968
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.853794199443782e-07,
+      "loss": 1.7721,
+      "step": 7969
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.853396901072706e-07,
+      "loss": 1.1485,
+      "step": 7970
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.85299960270163e-07,
+      "loss": 1.8521,
+      "step": 7971
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.852602304330551e-07,
+      "loss": 1.4148,
+      "step": 7972
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.852205005959475e-07,
+      "loss": 1.6258,
+      "step": 7973
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.851807707588398e-07,
+      "loss": 1.9469,
+      "step": 7974
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.851410409217322e-07,
+      "loss": 1.8383,
+      "step": 7975
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.851013110846245e-07,
+      "loss": 1.6579,
+      "step": 7976
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.850615812475169e-07,
+      "loss": 1.6959,
+      "step": 7977
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.850218514104091e-07,
+      "loss": 1.7169,
+      "step": 7978
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.849821215733015e-07,
+      "loss": 1.8636,
+      "step": 7979
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.849423917361939e-07,
+      "loss": 1.606,
+      "step": 7980
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.849026618990862e-07,
+      "loss": 1.6234,
+      "step": 7981
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.848629320619786e-07,
+      "loss": 1.8855,
+      "step": 7982
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.848232022248709e-07,
+      "loss": 1.6447,
+      "step": 7983
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.847834723877632e-07,
+      "loss": 1.8406,
+      "step": 7984
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.847437425506554e-07,
+      "loss": 1.4467,
+      "step": 7985
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.847040127135478e-07,
+      "loss": 1.8298,
+      "step": 7986
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.846642828764401e-07,
+      "loss": 2.3653,
+      "step": 7987
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.846245530393325e-07,
+      "loss": 1.36,
+      "step": 7988
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.845848232022249e-07,
+      "loss": 1.8005,
+      "step": 7989
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.845450933651171e-07,
+      "loss": 1.5059,
+      "step": 7990
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.845053635280095e-07,
+      "loss": 1.5083,
+      "step": 7991
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.844656336909018e-07,
+      "loss": 1.8074,
+      "step": 7992
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.844259038537942e-07,
+      "loss": 1.7861,
+      "step": 7993
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.843861740166865e-07,
+      "loss": 1.7881,
+      "step": 7994
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.843464441795789e-07,
+      "loss": 1.3881,
+      "step": 7995
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.843067143424712e-07,
+      "loss": 1.8647,
+      "step": 7996
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.842669845053635e-07,
+      "loss": 1.563,
+      "step": 7997
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.842272546682558e-07,
+      "loss": 1.6738,
+      "step": 7998
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.841875248311481e-07,
+      "loss": 1.2838,
+      "step": 7999
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.841477949940405e-07,
+      "loss": 1.5594,
+      "step": 8000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.841080651569328e-07,
+      "loss": 1.4626,
+      "step": 8001
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.840683353198251e-07,
+      "loss": 1.6141,
+      "step": 8002
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.840286054827174e-07,
+      "loss": 1.7079,
+      "step": 8003
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.839888756456098e-07,
+      "loss": 1.6833,
+      "step": 8004
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.839491458085022e-07,
+      "loss": 1.3477,
+      "step": 8005
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.839094159713945e-07,
+      "loss": 1.8074,
+      "step": 8006
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.838696861342869e-07,
+      "loss": 1.8717,
+      "step": 8007
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.838299562971791e-07,
+      "loss": 1.7792,
+      "step": 8008
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.837902264600715e-07,
+      "loss": 1.3634,
+      "step": 8009
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.837504966229638e-07,
+      "loss": 1.8006,
+      "step": 8010
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.837107667858562e-07,
+      "loss": 1.95,
+      "step": 8011
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.836710369487485e-07,
+      "loss": 2.0379,
+      "step": 8012
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.836313071116408e-07,
+      "loss": 1.7162,
+      "step": 8013
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.835915772745331e-07,
+      "loss": 1.5592,
+      "step": 8014
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.835518474374254e-07,
+      "loss": 2.23,
+      "step": 8015
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.835121176003178e-07,
+      "loss": 1.6683,
+      "step": 8016
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.834723877632101e-07,
+      "loss": 1.6953,
+      "step": 8017
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.834326579261025e-07,
+      "loss": 1.6954,
+      "step": 8018
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.833929280889949e-07,
+      "loss": 1.6801,
+      "step": 8019
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.833531982518871e-07,
+      "loss": 1.7063,
+      "step": 8020
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.833134684147795e-07,
+      "loss": 1.7362,
+      "step": 8021
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.832737385776718e-07,
+      "loss": 1.7848,
+      "step": 8022
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.832340087405642e-07,
+      "loss": 1.5795,
+      "step": 8023
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.831942789034565e-07,
+      "loss": 1.349,
+      "step": 8024
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.831545490663488e-07,
+      "loss": 1.2613,
+      "step": 8025
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.83114819229241e-07,
+      "loss": 1.5408,
+      "step": 8026
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.830750893921334e-07,
+      "loss": 1.7481,
+      "step": 8027
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.830353595550258e-07,
+      "loss": 1.7968,
+      "step": 8028
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.829956297179181e-07,
+      "loss": 1.9181,
+      "step": 8029
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.829558998808105e-07,
+      "loss": 2.3822,
+      "step": 8030
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.829161700437028e-07,
+      "loss": 1.8067,
+      "step": 8031
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.828764402065951e-07,
+      "loss": 1.9696,
+      "step": 8032
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.828367103694874e-07,
+      "loss": 1.4606,
+      "step": 8033
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.827969805323798e-07,
+      "loss": 1.675,
+      "step": 8034
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.827572506952722e-07,
+      "loss": 1.4443,
+      "step": 8035
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.827175208581645e-07,
+      "loss": 1.3961,
+      "step": 8036
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.826777910210569e-07,
+      "loss": 1.3817,
+      "step": 8037
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.82638061183949e-07,
+      "loss": 1.7071,
+      "step": 8038
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.825983313468414e-07,
+      "loss": 1.4959,
+      "step": 8039
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.825586015097337e-07,
+      "loss": 1.7316,
+      "step": 8040
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.825188716726261e-07,
+      "loss": 1.6407,
+      "step": 8041
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.824791418355184e-07,
+      "loss": 1.5035,
+      "step": 8042
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.824394119984108e-07,
+      "loss": 1.9517,
+      "step": 8043
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.823996821613031e-07,
+      "loss": 1.2676,
+      "step": 8044
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.823599523241954e-07,
+      "loss": 1.7908,
+      "step": 8045
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.823202224870878e-07,
+      "loss": 2.0169,
+      "step": 8046
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.822804926499801e-07,
+      "loss": 1.324,
+      "step": 8047
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.822407628128725e-07,
+      "loss": 1.8642,
+      "step": 8048
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.822010329757648e-07,
+      "loss": 1.3125,
+      "step": 8049
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.821613031386571e-07,
+      "loss": 2.1458,
+      "step": 8050
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.821215733015494e-07,
+      "loss": 2.0309,
+      "step": 8051
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.820818434644417e-07,
+      "loss": 1.668,
+      "step": 8052
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.820421136273341e-07,
+      "loss": 1.9347,
+      "step": 8053
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.820023837902264e-07,
+      "loss": 1.6892,
+      "step": 8054
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.819626539531188e-07,
+      "loss": 1.8041,
+      "step": 8055
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.81922924116011e-07,
+      "loss": 1.7398,
+      "step": 8056
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.818831942789034e-07,
+      "loss": 1.3676,
+      "step": 8057
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.818434644417958e-07,
+      "loss": 1.5416,
+      "step": 8058
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.818037346046881e-07,
+      "loss": 1.9198,
+      "step": 8059
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.817640047675805e-07,
+      "loss": 1.572,
+      "step": 8060
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.817242749304728e-07,
+      "loss": 1.6343,
+      "step": 8061
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.816845450933651e-07,
+      "loss": 1.7914,
+      "step": 8062
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.816448152562574e-07,
+      "loss": 1.8167,
+      "step": 8063
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.816050854191497e-07,
+      "loss": 1.8783,
+      "step": 8064
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.81565355582042e-07,
+      "loss": 1.3207,
+      "step": 8065
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.815256257449344e-07,
+      "loss": 1.5967,
+      "step": 8066
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.814858959078268e-07,
+      "loss": 1.4922,
+      "step": 8067
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.81446166070719e-07,
+      "loss": 1.6234,
+      "step": 8068
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.814064362336114e-07,
+      "loss": 1.6694,
+      "step": 8069
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.813667063965037e-07,
+      "loss": 1.5995,
+      "step": 8070
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.813269765593961e-07,
+      "loss": 1.7704,
+      "step": 8071
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.812872467222884e-07,
+      "loss": 1.618,
+      "step": 8072
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.812475168851808e-07,
+      "loss": 1.3711,
+      "step": 8073
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.812077870480731e-07,
+      "loss": 1.6254,
+      "step": 8074
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.811680572109654e-07,
+      "loss": 2.1209,
+      "step": 8075
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.811283273738578e-07,
+      "loss": 1.7629,
+      "step": 8076
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.810885975367501e-07,
+      "loss": 1.9139,
+      "step": 8077
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.810488676996424e-07,
+      "loss": 1.4053,
+      "step": 8078
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.810091378625347e-07,
+      "loss": 1.6336,
+      "step": 8079
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.80969408025427e-07,
+      "loss": 1.3374,
+      "step": 8080
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.809296781883193e-07,
+      "loss": 1.211,
+      "step": 8081
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.808899483512117e-07,
+      "loss": 1.6512,
+      "step": 8082
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.808502185141041e-07,
+      "loss": 1.8664,
+      "step": 8083
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.808104886769964e-07,
+      "loss": 1.5437,
+      "step": 8084
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.807707588398888e-07,
+      "loss": 1.8043,
+      "step": 8085
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.80731029002781e-07,
+      "loss": 1.657,
+      "step": 8086
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.806912991656734e-07,
+      "loss": 1.4953,
+      "step": 8087
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.806515693285657e-07,
+      "loss": 1.4134,
+      "step": 8088
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.806118394914581e-07,
+      "loss": 1.5482,
+      "step": 8089
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.805721096543505e-07,
+      "loss": 1.7434,
+      "step": 8090
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.805323798172427e-07,
+      "loss": 1.8536,
+      "step": 8091
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.80492649980135e-07,
+      "loss": 1.6905,
+      "step": 8092
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.804529201430273e-07,
+      "loss": 1.8451,
+      "step": 8093
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.804131903059197e-07,
+      "loss": 1.8009,
+      "step": 8094
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.80373460468812e-07,
+      "loss": 1.8843,
+      "step": 8095
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.803337306317044e-07,
+      "loss": 1.861,
+      "step": 8096
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.802940007945968e-07,
+      "loss": 1.614,
+      "step": 8097
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.80254270957489e-07,
+      "loss": 1.8037,
+      "step": 8098
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.802145411203814e-07,
+      "loss": 1.7493,
+      "step": 8099
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.801748112832737e-07,
+      "loss": 1.65,
+      "step": 8100
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.801350814461661e-07,
+      "loss": 1.6346,
+      "step": 8101
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.800953516090584e-07,
+      "loss": 1.9779,
+      "step": 8102
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.800556217719508e-07,
+      "loss": 1.8582,
+      "step": 8103
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.800158919348429e-07,
+      "loss": 1.3707,
+      "step": 8104
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.799761620977353e-07,
+      "loss": 1.5167,
+      "step": 8105
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.799364322606277e-07,
+      "loss": 1.6429,
+      "step": 8106
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.7989670242352e-07,
+      "loss": 1.9286,
+      "step": 8107
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.798569725864124e-07,
+      "loss": 1.6189,
+      "step": 8108
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.798172427493047e-07,
+      "loss": 1.531,
+      "step": 8109
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.79777512912197e-07,
+      "loss": 1.3959,
+      "step": 8110
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.797377830750893e-07,
+      "loss": 1.5554,
+      "step": 8111
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.796980532379817e-07,
+      "loss": 1.8898,
+      "step": 8112
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.796583234008741e-07,
+      "loss": 2.0223,
+      "step": 8113
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.796185935637664e-07,
+      "loss": 1.9642,
+      "step": 8114
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.795788637266588e-07,
+      "loss": 1.5296,
+      "step": 8115
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.79539133889551e-07,
+      "loss": 1.3663,
+      "step": 8116
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.794994040524433e-07,
+      "loss": 1.7946,
+      "step": 8117
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.794596742153356e-07,
+      "loss": 1.19,
+      "step": 8118
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.79419944378228e-07,
+      "loss": 1.4835,
+      "step": 8119
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.793802145411203e-07,
+      "loss": 1.7653,
+      "step": 8120
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.793404847040127e-07,
+      "loss": 1.8339,
+      "step": 8121
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.79300754866905e-07,
+      "loss": 1.7215,
+      "step": 8122
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.792610250297973e-07,
+      "loss": 1.6223,
+      "step": 8123
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.792212951926897e-07,
+      "loss": 1.6405,
+      "step": 8124
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.79181565355582e-07,
+      "loss": 1.7306,
+      "step": 8125
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.791418355184744e-07,
+      "loss": 1.5587,
+      "step": 8126
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.791021056813667e-07,
+      "loss": 1.567,
+      "step": 8127
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.79062375844259e-07,
+      "loss": 1.9451,
+      "step": 8128
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.790226460071514e-07,
+      "loss": 1.6801,
+      "step": 8129
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.789829161700436e-07,
+      "loss": 1.477,
+      "step": 8130
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.78943186332936e-07,
+      "loss": 1.8855,
+      "step": 8131
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.789034564958283e-07,
+      "loss": 1.3801,
+      "step": 8132
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.788637266587207e-07,
+      "loss": 1.9162,
+      "step": 8133
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.788239968216129e-07,
+      "loss": 1.844,
+      "step": 8134
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.787842669845053e-07,
+      "loss": 1.4279,
+      "step": 8135
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.787445371473977e-07,
+      "loss": 1.4168,
+      "step": 8136
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.7870480731029e-07,
+      "loss": 1.4437,
+      "step": 8137
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.786650774731824e-07,
+      "loss": 1.7992,
+      "step": 8138
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.786253476360747e-07,
+      "loss": 1.6536,
+      "step": 8139
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.78585617798967e-07,
+      "loss": 1.6805,
+      "step": 8140
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.785458879618593e-07,
+      "loss": 1.6315,
+      "step": 8141
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.785061581247517e-07,
+      "loss": 1.7333,
+      "step": 8142
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.78466428287644e-07,
+      "loss": 1.6375,
+      "step": 8143
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.784266984505363e-07,
+      "loss": 1.8357,
+      "step": 8144
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.783869686134287e-07,
+      "loss": 1.4164,
+      "step": 8145
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.783472387763209e-07,
+      "loss": 1.5439,
+      "step": 8146
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.783075089392133e-07,
+      "loss": 1.4205,
+      "step": 8147
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.782677791021056e-07,
+      "loss": 1.7397,
+      "step": 8148
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.78228049264998e-07,
+      "loss": 1.7461,
+      "step": 8149
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.781883194278903e-07,
+      "loss": 1.6185,
+      "step": 8150
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.781485895907827e-07,
+      "loss": 1.2423,
+      "step": 8151
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 6.78108859753675e-07,
+      "loss": 1.3951,
+      "step": 8152
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.780691299165673e-07,
+      "loss": 1.5615,
+      "step": 8153
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.780294000794597e-07,
+      "loss": 1.5479,
+      "step": 8154
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.77989670242352e-07,
+      "loss": 1.3794,
+      "step": 8155
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.779499404052444e-07,
+      "loss": 1.6461,
+      "step": 8156
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.779102105681366e-07,
+      "loss": 1.6154,
+      "step": 8157
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.778704807310289e-07,
+      "loss": 1.9842,
+      "step": 8158
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.778307508939212e-07,
+      "loss": 1.5483,
+      "step": 8159
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.777910210568136e-07,
+      "loss": 1.3067,
+      "step": 8160
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.77751291219706e-07,
+      "loss": 1.7,
+      "step": 8161
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.777115613825983e-07,
+      "loss": 1.5495,
+      "step": 8162
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.776718315454907e-07,
+      "loss": 1.6259,
+      "step": 8163
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.776321017083829e-07,
+      "loss": 1.6954,
+      "step": 8164
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.775923718712753e-07,
+      "loss": 1.7125,
+      "step": 8165
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.775526420341676e-07,
+      "loss": 1.4869,
+      "step": 8166
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.7751291219706e-07,
+      "loss": 1.5052,
+      "step": 8167
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.774731823599524e-07,
+      "loss": 1.5569,
+      "step": 8168
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.774334525228447e-07,
+      "loss": 1.7002,
+      "step": 8169
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.773937226857369e-07,
+      "loss": 2.2212,
+      "step": 8170
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.773539928486292e-07,
+      "loss": 1.3588,
+      "step": 8171
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.773142630115216e-07,
+      "loss": 1.9216,
+      "step": 8172
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.772745331744139e-07,
+      "loss": 1.9071,
+      "step": 8173
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.772348033373063e-07,
+      "loss": 1.4449,
+      "step": 8174
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.771950735001987e-07,
+      "loss": 1.8218,
+      "step": 8175
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.771553436630909e-07,
+      "loss": 1.5785,
+      "step": 8176
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.771156138259833e-07,
+      "loss": 1.8445,
+      "step": 8177
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.770758839888756e-07,
+      "loss": 1.7534,
+      "step": 8178
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.77036154151768e-07,
+      "loss": 1.8524,
+      "step": 8179
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.769964243146603e-07,
+      "loss": 1.5817,
+      "step": 8180
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.769566944775527e-07,
+      "loss": 1.3322,
+      "step": 8181
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.76916964640445e-07,
+      "loss": 1.5836,
+      "step": 8182
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.768772348033372e-07,
+      "loss": 1.5167,
+      "step": 8183
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.768375049662296e-07,
+      "loss": 1.9215,
+      "step": 8184
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.767977751291219e-07,
+      "loss": 1.615,
+      "step": 8185
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.767580452920143e-07,
+      "loss": 1.5035,
+      "step": 8186
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.767183154549066e-07,
+      "loss": 1.6104,
+      "step": 8187
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.766785856177989e-07,
+      "loss": 1.8421,
+      "step": 8188
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.766388557806912e-07,
+      "loss": 1.846,
+      "step": 8189
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 6.765991259435836e-07,
+      "loss": 1.274,
+      "step": 8190
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.76559396106476e-07,
+      "loss": 1.7206,
+      "step": 8191
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.765196662693683e-07,
+      "loss": 2.0244,
+      "step": 8192
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.764799364322607e-07,
+      "loss": 1.5846,
+      "step": 8193
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.764402065951529e-07,
+      "loss": 1.6328,
+      "step": 8194
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.764004767580453e-07,
+      "loss": 1.747,
+      "step": 8195
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.763607469209375e-07,
+      "loss": 2.099,
+      "step": 8196
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.763210170838299e-07,
+      "loss": 1.7457,
+      "step": 8197
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.762812872467222e-07,
+      "loss": 1.0537,
+      "step": 8198
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.762415574096146e-07,
+      "loss": 1.7532,
+      "step": 8199
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.762018275725069e-07,
+      "loss": 1.8172,
+      "step": 8200
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.761620977353992e-07,
+      "loss": 1.2818,
+      "step": 8201
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.761223678982916e-07,
+      "loss": 1.2477,
+      "step": 8202
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.760826380611839e-07,
+      "loss": 1.9047,
+      "step": 8203
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.760429082240763e-07,
+      "loss": 1.6028,
+      "step": 8204
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.760031783869686e-07,
+      "loss": 1.6881,
+      "step": 8205
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.759634485498609e-07,
+      "loss": 1.6334,
+      "step": 8206
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.759237187127533e-07,
+      "loss": 1.6532,
+      "step": 8207
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.758839888756456e-07,
+      "loss": 1.5259,
+      "step": 8208
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.75844259038538e-07,
+      "loss": 1.3854,
+      "step": 8209
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.758045292014302e-07,
+      "loss": 1.7206,
+      "step": 8210
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.757647993643226e-07,
+      "loss": 1.2722,
+      "step": 8211
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.757250695272148e-07,
+      "loss": 2.0863,
+      "step": 8212
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.756853396901072e-07,
+      "loss": 1.4212,
+      "step": 8213
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.756456098529996e-07,
+      "loss": 1.5545,
+      "step": 8214
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 6.756058800158919e-07,
+      "loss": 1.3433,
+      "step": 8215
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.755661501787843e-07,
+      "loss": 1.521,
+      "step": 8216
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.755264203416766e-07,
+      "loss": 1.7462,
+      "step": 8217
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.754866905045689e-07,
+      "loss": 1.7879,
+      "step": 8218
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.754469606674612e-07,
+      "loss": 1.4088,
+      "step": 8219
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.754072308303536e-07,
+      "loss": 1.4487,
+      "step": 8220
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.75367500993246e-07,
+      "loss": 1.9861,
+      "step": 8221
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.753277711561383e-07,
+      "loss": 1.8742,
+      "step": 8222
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.752880413190306e-07,
+      "loss": 1.5134,
+      "step": 8223
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.752483114819228e-07,
+      "loss": 1.4621,
+      "step": 8224
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.752085816448152e-07,
+      "loss": 2.1887,
+      "step": 8225
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.751688518077075e-07,
+      "loss": 1.6423,
+      "step": 8226
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.751291219705999e-07,
+      "loss": 1.7609,
+      "step": 8227
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.750893921334922e-07,
+      "loss": 1.7737,
+      "step": 8228
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.750496622963846e-07,
+      "loss": 2.2904,
+      "step": 8229
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.750099324592769e-07,
+      "loss": 1.3661,
+      "step": 8230
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.749702026221692e-07,
+      "loss": 1.864,
+      "step": 8231
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.749304727850616e-07,
+      "loss": 1.498,
+      "step": 8232
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.748907429479539e-07,
+      "loss": 1.5081,
+      "step": 8233
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.748510131108463e-07,
+      "loss": 1.4205,
+      "step": 8234
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.748112832737386e-07,
+      "loss": 1.6613,
+      "step": 8235
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.747715534366308e-07,
+      "loss": 2.0516,
+      "step": 8236
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.747318235995231e-07,
+      "loss": 1.611,
+      "step": 8237
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.746920937624155e-07,
+      "loss": 1.4869,
+      "step": 8238
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.746523639253079e-07,
+      "loss": 1.5305,
+      "step": 8239
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.746126340882002e-07,
+      "loss": 1.66,
+      "step": 8240
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.745729042510926e-07,
+      "loss": 1.7292,
+      "step": 8241
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.745331744139848e-07,
+      "loss": 1.2136,
+      "step": 8242
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.744934445768772e-07,
+      "loss": 1.7218,
+      "step": 8243
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.744537147397695e-07,
+      "loss": 1.3069,
+      "step": 8244
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.744139849026619e-07,
+      "loss": 1.5625,
+      "step": 8245
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.743742550655543e-07,
+      "loss": 1.7305,
+      "step": 8246
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.743345252284466e-07,
+      "loss": 1.3213,
+      "step": 8247
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.742947953913389e-07,
+      "loss": 1.3735,
+      "step": 8248
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.742550655542311e-07,
+      "loss": 1.2436,
+      "step": 8249
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.742153357171235e-07,
+      "loss": 1.6595,
+      "step": 8250
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.741756058800158e-07,
+      "loss": 1.878,
+      "step": 8251
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.741358760429082e-07,
+      "loss": 1.7346,
+      "step": 8252
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 6.740961462058006e-07,
+      "loss": 1.4155,
+      "step": 8253
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.740564163686928e-07,
+      "loss": 1.8358,
+      "step": 8254
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.740166865315852e-07,
+      "loss": 1.6847,
+      "step": 8255
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.739769566944775e-07,
+      "loss": 1.6559,
+      "step": 8256
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.739372268573699e-07,
+      "loss": 1.6672,
+      "step": 8257
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.738974970202622e-07,
+      "loss": 1.8367,
+      "step": 8258
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.738577671831546e-07,
+      "loss": 1.0201,
+      "step": 8259
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.738180373460468e-07,
+      "loss": 1.4549,
+      "step": 8260
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.737783075089392e-07,
+      "loss": 1.6002,
+      "step": 8261
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.737385776718315e-07,
+      "loss": 1.3889,
+      "step": 8262
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.736988478347238e-07,
+      "loss": 1.761,
+      "step": 8263
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.736591179976162e-07,
+      "loss": 1.5994,
+      "step": 8264
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.736193881605085e-07,
+      "loss": 1.8575,
+      "step": 8265
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.735796583234008e-07,
+      "loss": 1.9847,
+      "step": 8266
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.735399284862931e-07,
+      "loss": 1.544,
+      "step": 8267
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.735001986491855e-07,
+      "loss": 1.7533,
+      "step": 8268
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.734604688120779e-07,
+      "loss": 1.8264,
+      "step": 8269
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.734207389749702e-07,
+      "loss": 1.6049,
+      "step": 8270
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.733810091378626e-07,
+      "loss": 1.6448,
+      "step": 8271
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.733412793007548e-07,
+      "loss": 1.503,
+      "step": 8272
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.733015494636472e-07,
+      "loss": 1.7378,
+      "step": 8273
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.732618196265395e-07,
+      "loss": 1.5941,
+      "step": 8274
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.732220897894319e-07,
+      "loss": 1.7242,
+      "step": 8275
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.731823599523241e-07,
+      "loss": 1.5922,
+      "step": 8276
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.731426301152165e-07,
+      "loss": 1.6322,
+      "step": 8277
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.731029002781088e-07,
+      "loss": 2.0325,
+      "step": 8278
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.730631704410011e-07,
+      "loss": 1.796,
+      "step": 8279
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.730234406038935e-07,
+      "loss": 1.5592,
+      "step": 8280
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.729837107667858e-07,
+      "loss": 1.5187,
+      "step": 8281
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.729439809296782e-07,
+      "loss": 1.288,
+      "step": 8282
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.729042510925705e-07,
+      "loss": 1.8147,
+      "step": 8283
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.728645212554628e-07,
+      "loss": 1.6307,
+      "step": 8284
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.728247914183552e-07,
+      "loss": 1.8002,
+      "step": 8285
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.727850615812475e-07,
+      "loss": 1.3851,
+      "step": 8286
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.727453317441399e-07,
+      "loss": 1.5011,
+      "step": 8287
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.727056019070322e-07,
+      "loss": 1.2487,
+      "step": 8288
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.726658720699245e-07,
+      "loss": 1.144,
+      "step": 8289
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.726261422328167e-07,
+      "loss": 1.718,
+      "step": 8290
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 6.725864123957091e-07,
+      "loss": 1.9656,
+      "step": 8291
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.725466825586015e-07,
+      "loss": 1.7416,
+      "step": 8292
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.725069527214938e-07,
+      "loss": 1.625,
+      "step": 8293
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.724672228843862e-07,
+      "loss": 1.8964,
+      "step": 8294
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.724274930472785e-07,
+      "loss": 1.5074,
+      "step": 8295
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.723877632101708e-07,
+      "loss": 1.9219,
+      "step": 8296
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.723480333730631e-07,
+      "loss": 1.878,
+      "step": 8297
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.723083035359555e-07,
+      "loss": 1.9441,
+      "step": 8298
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.722685736988479e-07,
+      "loss": 1.4786,
+      "step": 8299
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.722288438617402e-07,
+      "loss": 1.6576,
+      "step": 8300
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.721891140246326e-07,
+      "loss": 1.6797,
+      "step": 8301
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.721493841875247e-07,
+      "loss": 1.4868,
+      "step": 8302
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.721096543504171e-07,
+      "loss": 1.7842,
+      "step": 8303
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.720699245133094e-07,
+      "loss": 1.9901,
+      "step": 8304
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.720301946762018e-07,
+      "loss": 1.9882,
+      "step": 8305
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.719904648390941e-07,
+      "loss": 1.8101,
+      "step": 8306
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.719507350019865e-07,
+      "loss": 1.5711,
+      "step": 8307
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.719110051648788e-07,
+      "loss": 1.5295,
+      "step": 8308
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.718712753277711e-07,
+      "loss": 2.0191,
+      "step": 8309
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.718315454906635e-07,
+      "loss": 1.967,
+      "step": 8310
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.717918156535558e-07,
+      "loss": 1.7845,
+      "step": 8311
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.717520858164482e-07,
+      "loss": 1.2426,
+      "step": 8312
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.717123559793405e-07,
+      "loss": 1.4779,
+      "step": 8313
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.716726261422328e-07,
+      "loss": 1.6519,
+      "step": 8314
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.71632896305125e-07,
+      "loss": 1.3695,
+      "step": 8315
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.715931664680174e-07,
+      "loss": 1.3498,
+      "step": 8316
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.715534366309098e-07,
+      "loss": 1.5224,
+      "step": 8317
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.715137067938021e-07,
+      "loss": 1.3728,
+      "step": 8318
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.714739769566945e-07,
+      "loss": 2.2903,
+      "step": 8319
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.714342471195867e-07,
+      "loss": 1.6822,
+      "step": 8320
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.713945172824791e-07,
+      "loss": 1.3991,
+      "step": 8321
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.713547874453714e-07,
+      "loss": 1.2938,
+      "step": 8322
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.713150576082638e-07,
+      "loss": 1.7095,
+      "step": 8323
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.712753277711562e-07,
+      "loss": 2.422,
+      "step": 8324
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.712355979340485e-07,
+      "loss": 2.1054,
+      "step": 8325
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.711958680969408e-07,
+      "loss": 1.6269,
+      "step": 8326
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.711561382598331e-07,
+      "loss": 1.8456,
+      "step": 8327
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 6.711164084227254e-07,
+      "loss": 1.4302,
+      "step": 8328
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.710766785856177e-07,
+      "loss": 1.6657,
+      "step": 8329
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.710369487485101e-07,
+      "loss": 1.3825,
+      "step": 8330
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.709972189114025e-07,
+      "loss": 1.5135,
+      "step": 8331
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.709574890742947e-07,
+      "loss": 1.4067,
+      "step": 8332
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.709177592371871e-07,
+      "loss": 1.188,
+      "step": 8333
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.708780294000794e-07,
+      "loss": 1.6617,
+      "step": 8334
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.708382995629718e-07,
+      "loss": 1.6247,
+      "step": 8335
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.707985697258641e-07,
+      "loss": 1.4134,
+      "step": 8336
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.707588398887565e-07,
+      "loss": 1.8054,
+      "step": 8337
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.707191100516487e-07,
+      "loss": 1.9956,
+      "step": 8338
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.706793802145411e-07,
+      "loss": 1.8546,
+      "step": 8339
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.706396503774335e-07,
+      "loss": 2.3251,
+      "step": 8340
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.705999205403258e-07,
+      "loss": 1.7167,
+      "step": 8341
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.705601907032181e-07,
+      "loss": 1.7318,
+      "step": 8342
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.705204608661104e-07,
+      "loss": 1.9107,
+      "step": 8343
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.704807310290027e-07,
+      "loss": 1.6297,
+      "step": 8344
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.70441001191895e-07,
+      "loss": 1.8508,
+      "step": 8345
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.704012713547874e-07,
+      "loss": 1.7588,
+      "step": 8346
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.703615415176798e-07,
+      "loss": 1.6582,
+      "step": 8347
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.703218116805721e-07,
+      "loss": 1.7012,
+      "step": 8348
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.702820818434645e-07,
+      "loss": 1.837,
+      "step": 8349
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.702423520063567e-07,
+      "loss": 1.5406,
+      "step": 8350
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.702026221692491e-07,
+      "loss": 2.0646,
+      "step": 8351
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.701628923321414e-07,
+      "loss": 1.4744,
+      "step": 8352
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.701231624950338e-07,
+      "loss": 1.7523,
+      "step": 8353
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.700834326579262e-07,
+      "loss": 1.9654,
+      "step": 8354
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.700437028208184e-07,
+      "loss": 1.3656,
+      "step": 8355
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.700039729837107e-07,
+      "loss": 1.6077,
+      "step": 8356
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.69964243146603e-07,
+      "loss": 1.9636,
+      "step": 8357
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.699245133094954e-07,
+      "loss": 1.8874,
+      "step": 8358
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.698847834723877e-07,
+      "loss": 1.6058,
+      "step": 8359
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.698450536352801e-07,
+      "loss": 1.7977,
+      "step": 8360
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.698053237981724e-07,
+      "loss": 1.6759,
+      "step": 8361
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.697655939610647e-07,
+      "loss": 1.4203,
+      "step": 8362
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.697258641239571e-07,
+      "loss": 1.6789,
+      "step": 8363
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.696861342868494e-07,
+      "loss": 1.9475,
+      "step": 8364
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.696464044497418e-07,
+      "loss": 1.8954,
+      "step": 8365
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 6.696066746126341e-07,
+      "loss": 2.3933,
+      "step": 8366
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.695669447755265e-07,
+      "loss": 2.1308,
+      "step": 8367
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.695272149384186e-07,
+      "loss": 1.8175,
+      "step": 8368
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.69487485101311e-07,
+      "loss": 1.2139,
+      "step": 8369
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.694477552642034e-07,
+      "loss": 1.8383,
+      "step": 8370
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.694080254270957e-07,
+      "loss": 1.7596,
+      "step": 8371
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.693682955899881e-07,
+      "loss": 1.684,
+      "step": 8372
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.693285657528804e-07,
+      "loss": 1.7507,
+      "step": 8373
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.692888359157727e-07,
+      "loss": 2.1007,
+      "step": 8374
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.69249106078665e-07,
+      "loss": 1.458,
+      "step": 8375
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.692093762415574e-07,
+      "loss": 1.5967,
+      "step": 8376
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.691696464044498e-07,
+      "loss": 1.7089,
+      "step": 8377
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.691299165673421e-07,
+      "loss": 1.9022,
+      "step": 8378
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.690901867302345e-07,
+      "loss": 1.6587,
+      "step": 8379
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.690504568931267e-07,
+      "loss": 1.755,
+      "step": 8380
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.69010727056019e-07,
+      "loss": 1.8386,
+      "step": 8381
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.689709972189113e-07,
+      "loss": 1.3743,
+      "step": 8382
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.689312673818037e-07,
+      "loss": 1.5133,
+      "step": 8383
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.68891537544696e-07,
+      "loss": 1.8376,
+      "step": 8384
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.688518077075884e-07,
+      "loss": 1.6075,
+      "step": 8385
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.688120778704807e-07,
+      "loss": 2.1259,
+      "step": 8386
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.68772348033373e-07,
+      "loss": 1.8595,
+      "step": 8387
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.687326181962654e-07,
+      "loss": 1.8802,
+      "step": 8388
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.686928883591577e-07,
+      "loss": 1.5364,
+      "step": 8389
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.686531585220501e-07,
+      "loss": 1.3961,
+      "step": 8390
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 6.686134286849424e-07,
+      "loss": 1.8405,
+      "step": 8391
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.685736988478347e-07,
+      "loss": 1.671,
+      "step": 8392
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.685339690107271e-07,
+      "loss": 1.5921,
+      "step": 8393
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.684942391736193e-07,
+      "loss": 1.8413,
+      "step": 8394
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.684545093365117e-07,
+      "loss": 2.0433,
+      "step": 8395
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.68414779499404e-07,
+      "loss": 2.044,
+      "step": 8396
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.683750496622964e-07,
+      "loss": 1.3914,
+      "step": 8397
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.683353198251886e-07,
+      "loss": 1.4298,
+      "step": 8398
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.68295589988081e-07,
+      "loss": 1.6729,
+      "step": 8399
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.682558601509733e-07,
+      "loss": 1.5526,
+      "step": 8400
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.682161303138657e-07,
+      "loss": 1.3502,
+      "step": 8401
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.681764004767581e-07,
+      "loss": 1.7121,
+      "step": 8402
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.681366706396504e-07,
+      "loss": 1.413,
+      "step": 8403
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.680969408025427e-07,
+      "loss": 1.6342,
+      "step": 8404
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.68057210965435e-07,
+      "loss": 1.7445,
+      "step": 8405
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.680174811283274e-07,
+      "loss": 1.5247,
+      "step": 8406
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.679777512912197e-07,
+      "loss": 1.5767,
+      "step": 8407
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.67938021454112e-07,
+      "loss": 1.7637,
+      "step": 8408
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.678982916170044e-07,
+      "loss": 1.8385,
+      "step": 8409
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.678585617798966e-07,
+      "loss": 1.7727,
+      "step": 8410
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.67818831942789e-07,
+      "loss": 1.4916,
+      "step": 8411
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.677791021056813e-07,
+      "loss": 1.7337,
+      "step": 8412
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.677393722685737e-07,
+      "loss": 1.5757,
+      "step": 8413
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.67699642431466e-07,
+      "loss": 1.7033,
+      "step": 8414
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.676599125943584e-07,
+      "loss": 1.6403,
+      "step": 8415
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.676201827572506e-07,
+      "loss": 1.5988,
+      "step": 8416
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.67580452920143e-07,
+      "loss": 1.4118,
+      "step": 8417
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.675407230830354e-07,
+      "loss": 2.2779,
+      "step": 8418
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.675009932459277e-07,
+      "loss": 1.2347,
+      "step": 8419
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.674612634088201e-07,
+      "loss": 1.4366,
+      "step": 8420
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.674215335717123e-07,
+      "loss": 1.6665,
+      "step": 8421
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.673818037346046e-07,
+      "loss": 1.6747,
+      "step": 8422
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.673420738974969e-07,
+      "loss": 2.0574,
+      "step": 8423
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.673023440603893e-07,
+      "loss": 1.3089,
+      "step": 8424
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.672626142232817e-07,
+      "loss": 1.5803,
+      "step": 8425
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.67222884386174e-07,
+      "loss": 1.5916,
+      "step": 8426
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.671831545490664e-07,
+      "loss": 1.3942,
+      "step": 8427
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.671434247119586e-07,
+      "loss": 1.586,
+      "step": 8428
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 6.67103694874851e-07,
+      "loss": 1.8432,
+      "step": 8429
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.670639650377433e-07,
+      "loss": 1.6645,
+      "step": 8430
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.670242352006357e-07,
+      "loss": 1.2603,
+      "step": 8431
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.669845053635281e-07,
+      "loss": 1.6586,
+      "step": 8432
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.669447755264204e-07,
+      "loss": 1.6781,
+      "step": 8433
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.669050456893126e-07,
+      "loss": 1.8671,
+      "step": 8434
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.668653158522049e-07,
+      "loss": 1.7088,
+      "step": 8435
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.668255860150973e-07,
+      "loss": 2.0615,
+      "step": 8436
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.667858561779896e-07,
+      "loss": 1.7417,
+      "step": 8437
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.66746126340882e-07,
+      "loss": 1.526,
+      "step": 8438
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.667063965037743e-07,
+      "loss": 1.326,
+      "step": 8439
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.666666666666666e-07,
+      "loss": 2.0148,
+      "step": 8440
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.66626936829559e-07,
+      "loss": 1.6958,
+      "step": 8441
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.665872069924513e-07,
+      "loss": 2.2068,
+      "step": 8442
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.665474771553437e-07,
+      "loss": 1.7841,
+      "step": 8443
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.66507747318236e-07,
+      "loss": 1.9349,
+      "step": 8444
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.664680174811283e-07,
+      "loss": 1.6249,
+      "step": 8445
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.664282876440206e-07,
+      "loss": 1.5294,
+      "step": 8446
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.663885578069129e-07,
+      "loss": 1.5772,
+      "step": 8447
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.663488279698053e-07,
+      "loss": 1.8661,
+      "step": 8448
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.663090981326976e-07,
+      "loss": 1.7277,
+      "step": 8449
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.6626936829559e-07,
+      "loss": 1.5674,
+      "step": 8450
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.662296384584822e-07,
+      "loss": 1.5178,
+      "step": 8451
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.661899086213746e-07,
+      "loss": 1.5779,
+      "step": 8452
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.661501787842669e-07,
+      "loss": 1.5027,
+      "step": 8453
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.661104489471593e-07,
+      "loss": 2.2665,
+      "step": 8454
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.660707191100517e-07,
+      "loss": 1.8187,
+      "step": 8455
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.66030989272944e-07,
+      "loss": 1.751,
+      "step": 8456
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.659912594358363e-07,
+      "loss": 1.4274,
+      "step": 8457
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.659515295987286e-07,
+      "loss": 1.9503,
+      "step": 8458
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.65911799761621e-07,
+      "loss": 2.1057,
+      "step": 8459
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.658720699245132e-07,
+      "loss": 1.441,
+      "step": 8460
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.658323400874056e-07,
+      "loss": 1.4667,
+      "step": 8461
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.657926102502979e-07,
+      "loss": 1.7254,
+      "step": 8462
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.657528804131902e-07,
+      "loss": 1.8976,
+      "step": 8463
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.657131505760826e-07,
+      "loss": 1.6399,
+      "step": 8464
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.656734207389749e-07,
+      "loss": 1.8619,
+      "step": 8465
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.656336909018673e-07,
+      "loss": 1.7194,
+      "step": 8466
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 6.655939610647596e-07,
+      "loss": 1.6237,
+      "step": 8467
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.65554231227652e-07,
+      "loss": 1.4535,
+      "step": 8468
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.655145013905442e-07,
+      "loss": 1.6329,
+      "step": 8469
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.654747715534366e-07,
+      "loss": 1.7253,
+      "step": 8470
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.65435041716329e-07,
+      "loss": 1.3627,
+      "step": 8471
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.653953118792213e-07,
+      "loss": 2.2508,
+      "step": 8472
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.653555820421137e-07,
+      "loss": 1.584,
+      "step": 8473
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.653158522050059e-07,
+      "loss": 1.8932,
+      "step": 8474
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.652761223678982e-07,
+      "loss": 2.2199,
+      "step": 8475
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.652363925307905e-07,
+      "loss": 1.4823,
+      "step": 8476
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.651966626936829e-07,
+      "loss": 1.8327,
+      "step": 8477
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.651569328565752e-07,
+      "loss": 1.6722,
+      "step": 8478
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.651172030194676e-07,
+      "loss": 1.5452,
+      "step": 8479
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.6507747318236e-07,
+      "loss": 1.4246,
+      "step": 8480
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.650377433452522e-07,
+      "loss": 1.8389,
+      "step": 8481
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.649980135081446e-07,
+      "loss": 1.87,
+      "step": 8482
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.649582836710369e-07,
+      "loss": 1.4787,
+      "step": 8483
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.649185538339293e-07,
+      "loss": 1.3965,
+      "step": 8484
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.648788239968216e-07,
+      "loss": 1.5962,
+      "step": 8485
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.64839094159714e-07,
+      "loss": 1.5994,
+      "step": 8486
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.647993643226061e-07,
+      "loss": 1.3591,
+      "step": 8487
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.647596344854985e-07,
+      "loss": 1.7197,
+      "step": 8488
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.647199046483909e-07,
+      "loss": 1.8933,
+      "step": 8489
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.646801748112832e-07,
+      "loss": 1.8201,
+      "step": 8490
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.646404449741756e-07,
+      "loss": 1.9514,
+      "step": 8491
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.646007151370679e-07,
+      "loss": 1.8375,
+      "step": 8492
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.645609852999602e-07,
+      "loss": 1.8328,
+      "step": 8493
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.645212554628525e-07,
+      "loss": 1.7915,
+      "step": 8494
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.644815256257449e-07,
+      "loss": 1.555,
+      "step": 8495
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.644417957886373e-07,
+      "loss": 1.765,
+      "step": 8496
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.644020659515296e-07,
+      "loss": 1.3254,
+      "step": 8497
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.64362336114422e-07,
+      "loss": 1.6294,
+      "step": 8498
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.643226062773142e-07,
+      "loss": 1.2765,
+      "step": 8499
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.642828764402065e-07,
+      "loss": 1.5162,
+      "step": 8500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.642431466030988e-07,
+      "loss": 1.8477,
+      "step": 8501
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.642034167659912e-07,
+      "loss": 1.5285,
+      "step": 8502
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.641636869288836e-07,
+      "loss": 1.6725,
+      "step": 8503
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.641239570917759e-07,
+      "loss": 1.8241,
+      "step": 8504
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 6.640842272546682e-07,
+      "loss": 1.8029,
+      "step": 8505
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.640444974175605e-07,
+      "loss": 1.8884,
+      "step": 8506
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.640047675804529e-07,
+      "loss": 1.4063,
+      "step": 8507
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.639650377433452e-07,
+      "loss": 1.7395,
+      "step": 8508
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.639253079062376e-07,
+      "loss": 1.0306,
+      "step": 8509
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.6388557806913e-07,
+      "loss": 1.9164,
+      "step": 8510
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.638458482320222e-07,
+      "loss": 1.5309,
+      "step": 8511
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.638061183949146e-07,
+      "loss": 1.3977,
+      "step": 8512
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.637663885578068e-07,
+      "loss": 1.5699,
+      "step": 8513
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.637266587206992e-07,
+      "loss": 1.906,
+      "step": 8514
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.636869288835915e-07,
+      "loss": 1.3458,
+      "step": 8515
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.636471990464839e-07,
+      "loss": 1.765,
+      "step": 8516
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.636074692093761e-07,
+      "loss": 1.8688,
+      "step": 8517
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.635677393722685e-07,
+      "loss": 1.5066,
+      "step": 8518
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.635280095351609e-07,
+      "loss": 1.6023,
+      "step": 8519
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.634882796980532e-07,
+      "loss": 1.4779,
+      "step": 8520
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.634485498609456e-07,
+      "loss": 2.0346,
+      "step": 8521
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.634088200238379e-07,
+      "loss": 1.6638,
+      "step": 8522
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.633690901867302e-07,
+      "loss": 1.5766,
+      "step": 8523
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.633293603496225e-07,
+      "loss": 1.8566,
+      "step": 8524
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.632896305125149e-07,
+      "loss": 1.7786,
+      "step": 8525
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.632499006754072e-07,
+      "loss": 1.6399,
+      "step": 8526
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.632101708382995e-07,
+      "loss": 1.4622,
+      "step": 8527
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.631704410011919e-07,
+      "loss": 1.7754,
+      "step": 8528
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.631307111640841e-07,
+      "loss": 1.8181,
+      "step": 8529
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 6.630909813269765e-07,
+      "loss": 2.0455,
+      "step": 8530
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.630512514898688e-07,
+      "loss": 1.8104,
+      "step": 8531
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.630115216527612e-07,
+      "loss": 1.6554,
+      "step": 8532
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.629717918156536e-07,
+      "loss": 1.6447,
+      "step": 8533
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.629320619785459e-07,
+      "loss": 1.7611,
+      "step": 8534
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.628923321414382e-07,
+      "loss": 1.8717,
+      "step": 8535
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.628526023043305e-07,
+      "loss": 1.7119,
+      "step": 8536
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.628128724672229e-07,
+      "loss": 1.3863,
+      "step": 8537
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.627731426301152e-07,
+      "loss": 1.5856,
+      "step": 8538
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.627334127930076e-07,
+      "loss": 1.5796,
+      "step": 8539
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.626936829558998e-07,
+      "loss": 1.1875,
+      "step": 8540
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.626539531187921e-07,
+      "loss": 1.6478,
+      "step": 8541
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.626142232816845e-07,
+      "loss": 1.8169,
+      "step": 8542
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.625744934445768e-07,
+      "loss": 1.9055,
+      "step": 8543
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.625347636074692e-07,
+      "loss": 1.6226,
+      "step": 8544
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.624950337703615e-07,
+      "loss": 1.598,
+      "step": 8545
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.624553039332539e-07,
+      "loss": 1.9643,
+      "step": 8546
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.624155740961461e-07,
+      "loss": 1.7842,
+      "step": 8547
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.623758442590385e-07,
+      "loss": 1.7512,
+      "step": 8548
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.623361144219309e-07,
+      "loss": 1.6887,
+      "step": 8549
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.622963845848232e-07,
+      "loss": 1.4614,
+      "step": 8550
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.622566547477156e-07,
+      "loss": 1.8985,
+      "step": 8551
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.622169249106079e-07,
+      "loss": 1.954,
+      "step": 8552
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.621771950735001e-07,
+      "loss": 1.6716,
+      "step": 8553
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.621374652363924e-07,
+      "loss": 1.7773,
+      "step": 8554
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.620977353992848e-07,
+      "loss": 1.6501,
+      "step": 8555
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.620580055621771e-07,
+      "loss": 1.6859,
+      "step": 8556
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.620182757250695e-07,
+      "loss": 1.4568,
+      "step": 8557
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.619785458879619e-07,
+      "loss": 1.5935,
+      "step": 8558
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.619388160508541e-07,
+      "loss": 1.5413,
+      "step": 8559
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.618990862137465e-07,
+      "loss": 1.8371,
+      "step": 8560
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.618593563766388e-07,
+      "loss": 1.5251,
+      "step": 8561
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.618196265395312e-07,
+      "loss": 1.6627,
+      "step": 8562
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.617798967024235e-07,
+      "loss": 1.5682,
+      "step": 8563
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.617401668653159e-07,
+      "loss": 2.1725,
+      "step": 8564
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.617004370282082e-07,
+      "loss": 1.4718,
+      "step": 8565
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.616607071911004e-07,
+      "loss": 1.8107,
+      "step": 8566
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.616209773539928e-07,
+      "loss": 1.6044,
+      "step": 8567
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.615812475168851e-07,
+      "loss": 1.1712,
+      "step": 8568
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.615415176797775e-07,
+      "loss": 1.8064,
+      "step": 8569
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.615017878426698e-07,
+      "loss": 1.7237,
+      "step": 8570
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.614620580055621e-07,
+      "loss": 2.0487,
+      "step": 8571
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.614223281684544e-07,
+      "loss": 1.266,
+      "step": 8572
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.613825983313468e-07,
+      "loss": 1.7879,
+      "step": 8573
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.613428684942392e-07,
+      "loss": 1.7493,
+      "step": 8574
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.613031386571315e-07,
+      "loss": 1.8613,
+      "step": 8575
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.612634088200239e-07,
+      "loss": 2.0037,
+      "step": 8576
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.612236789829161e-07,
+      "loss": 1.6722,
+      "step": 8577
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.611839491458085e-07,
+      "loss": 1.5023,
+      "step": 8578
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.611442193087007e-07,
+      "loss": 1.2254,
+      "step": 8579
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.611044894715931e-07,
+      "loss": 1.3855,
+      "step": 8580
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.610647596344855e-07,
+      "loss": 1.9455,
+      "step": 8581
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.610250297973778e-07,
+      "loss": 1.3592,
+      "step": 8582
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.609852999602701e-07,
+      "loss": 1.4686,
+      "step": 8583
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.609455701231624e-07,
+      "loss": 1.5213,
+      "step": 8584
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.609058402860548e-07,
+      "loss": 1.8662,
+      "step": 8585
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.608661104489471e-07,
+      "loss": 1.8576,
+      "step": 8586
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.608263806118395e-07,
+      "loss": 1.5499,
+      "step": 8587
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.607866507747319e-07,
+      "loss": 1.9012,
+      "step": 8588
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.607469209376241e-07,
+      "loss": 1.989,
+      "step": 8589
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.607071911005165e-07,
+      "loss": 1.5245,
+      "step": 8590
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.606674612634088e-07,
+      "loss": 1.9275,
+      "step": 8591
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.606277314263011e-07,
+      "loss": 1.6797,
+      "step": 8592
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.605880015891934e-07,
+      "loss": 1.2879,
+      "step": 8593
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.605482717520858e-07,
+      "loss": 1.6783,
+      "step": 8594
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.60508541914978e-07,
+      "loss": 1.6224,
+      "step": 8595
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.604688120778704e-07,
+      "loss": 2.0843,
+      "step": 8596
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.604290822407628e-07,
+      "loss": 1.8207,
+      "step": 8597
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.603893524036551e-07,
+      "loss": 1.8619,
+      "step": 8598
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.603496225665475e-07,
+      "loss": 1.7129,
+      "step": 8599
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.603098927294398e-07,
+      "loss": 1.8968,
+      "step": 8600
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.602701628923321e-07,
+      "loss": 1.2363,
+      "step": 8601
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.602304330552244e-07,
+      "loss": 1.4038,
+      "step": 8602
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.601907032181168e-07,
+      "loss": 1.3614,
+      "step": 8603
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.601509733810092e-07,
+      "loss": 1.7459,
+      "step": 8604
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.601112435439015e-07,
+      "loss": 1.4851,
+      "step": 8605
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.600715137067938e-07,
+      "loss": 1.7789,
+      "step": 8606
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.60031783869686e-07,
+      "loss": 1.8169,
+      "step": 8607
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.599920540325784e-07,
+      "loss": 1.6932,
+      "step": 8608
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.599523241954707e-07,
+      "loss": 1.8311,
+      "step": 8609
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.599125943583631e-07,
+      "loss": 1.3894,
+      "step": 8610
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.598728645212555e-07,
+      "loss": 1.7788,
+      "step": 8611
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.598331346841478e-07,
+      "loss": 1.7498,
+      "step": 8612
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.597934048470401e-07,
+      "loss": 1.4975,
+      "step": 8613
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.597536750099324e-07,
+      "loss": 1.6058,
+      "step": 8614
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.597139451728248e-07,
+      "loss": 1.7937,
+      "step": 8615
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.596742153357171e-07,
+      "loss": 1.3255,
+      "step": 8616
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.596344854986095e-07,
+      "loss": 1.7672,
+      "step": 8617
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.595947556615019e-07,
+      "loss": 1.3355,
+      "step": 8618
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.59555025824394e-07,
+      "loss": 1.6063,
+      "step": 8619
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.595152959872864e-07,
+      "loss": 1.4579,
+      "step": 8620
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.594755661501787e-07,
+      "loss": 1.823,
+      "step": 8621
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.594358363130711e-07,
+      "loss": 1.6735,
+      "step": 8622
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.593961064759634e-07,
+      "loss": 1.6303,
+      "step": 8623
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.593563766388558e-07,
+      "loss": 1.9933,
+      "step": 8624
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.59316646801748e-07,
+      "loss": 1.8114,
+      "step": 8625
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.592769169646404e-07,
+      "loss": 1.438,
+      "step": 8626
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.592371871275328e-07,
+      "loss": 1.6919,
+      "step": 8627
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.591974572904251e-07,
+      "loss": 1.8844,
+      "step": 8628
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.591577274533175e-07,
+      "loss": 1.5377,
+      "step": 8629
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.591179976162098e-07,
+      "loss": 1.684,
+      "step": 8630
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.59078267779102e-07,
+      "loss": 1.4161,
+      "step": 8631
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.590385379419943e-07,
+      "loss": 1.3318,
+      "step": 8632
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.589988081048867e-07,
+      "loss": 2.1253,
+      "step": 8633
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.58959078267779e-07,
+      "loss": 1.5689,
+      "step": 8634
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.589193484306714e-07,
+      "loss": 1.7939,
+      "step": 8635
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.588796185935638e-07,
+      "loss": 1.6496,
+      "step": 8636
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.58839888756456e-07,
+      "loss": 1.6489,
+      "step": 8637
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.588001589193484e-07,
+      "loss": 1.7253,
+      "step": 8638
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.587604290822407e-07,
+      "loss": 1.6059,
+      "step": 8639
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.587206992451331e-07,
+      "loss": 2.3164,
+      "step": 8640
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.586809694080254e-07,
+      "loss": 1.5488,
+      "step": 8641
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.586412395709178e-07,
+      "loss": 1.9004,
+      "step": 8642
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.586015097338101e-07,
+      "loss": 1.685,
+      "step": 8643
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.585617798967024e-07,
+      "loss": 1.9341,
+      "step": 8644
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.585220500595947e-07,
+      "loss": 1.609,
+      "step": 8645
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.58482320222487e-07,
+      "loss": 1.5748,
+      "step": 8646
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.584425903853794e-07,
+      "loss": 1.4279,
+      "step": 8647
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.584028605482717e-07,
+      "loss": 1.848,
+      "step": 8648
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.58363130711164e-07,
+      "loss": 1.7043,
+      "step": 8649
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.583234008740563e-07,
+      "loss": 1.4913,
+      "step": 8650
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.582836710369487e-07,
+      "loss": 1.9038,
+      "step": 8651
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.582439411998411e-07,
+      "loss": 1.6146,
+      "step": 8652
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.582042113627334e-07,
+      "loss": 1.9763,
+      "step": 8653
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.581644815256258e-07,
+      "loss": 1.7256,
+      "step": 8654
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.58124751688518e-07,
+      "loss": 1.513,
+      "step": 8655
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.580850218514104e-07,
+      "loss": 1.753,
+      "step": 8656
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.580452920143027e-07,
+      "loss": 1.6355,
+      "step": 8657
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.58005562177195e-07,
+      "loss": 1.5105,
+      "step": 8658
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.579658323400874e-07,
+      "loss": 1.7501,
+      "step": 8659
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.579261025029797e-07,
+      "loss": 1.5629,
+      "step": 8660
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.57886372665872e-07,
+      "loss": 1.5883,
+      "step": 8661
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.578466428287643e-07,
+      "loss": 1.5479,
+      "step": 8662
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.578069129916567e-07,
+      "loss": 1.7617,
+      "step": 8663
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.57767183154549e-07,
+      "loss": 1.639,
+      "step": 8664
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.577274533174414e-07,
+      "loss": 1.738,
+      "step": 8665
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.576877234803338e-07,
+      "loss": 1.6875,
+      "step": 8666
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.57647993643226e-07,
+      "loss": 1.4396,
+      "step": 8667
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.576082638061184e-07,
+      "loss": 1.7065,
+      "step": 8668
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.575685339690107e-07,
+      "loss": 1.9204,
+      "step": 8669
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.575288041319031e-07,
+      "loss": 1.6159,
+      "step": 8670
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.574890742947954e-07,
+      "loss": 1.2305,
+      "step": 8671
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.574493444576877e-07,
+      "loss": 1.8099,
+      "step": 8672
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.574096146205799e-07,
+      "loss": 1.4421,
+      "step": 8673
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.573698847834723e-07,
+      "loss": 1.5921,
+      "step": 8674
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.573301549463647e-07,
+      "loss": 1.6762,
+      "step": 8675
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.57290425109257e-07,
+      "loss": 1.5399,
+      "step": 8676
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.572506952721494e-07,
+      "loss": 1.3558,
+      "step": 8677
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.572109654350417e-07,
+      "loss": 1.8625,
+      "step": 8678
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.57171235597934e-07,
+      "loss": 1.5981,
+      "step": 8679
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.571315057608263e-07,
+      "loss": 1.7203,
+      "step": 8680
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.570917759237187e-07,
+      "loss": 0.9504,
+      "step": 8681
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.570520460866111e-07,
+      "loss": 1.4093,
+      "step": 8682
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.570123162495034e-07,
+      "loss": 1.7202,
+      "step": 8683
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.569725864123958e-07,
+      "loss": 1.885,
+      "step": 8684
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.569328565752879e-07,
+      "loss": 1.4577,
+      "step": 8685
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.568931267381803e-07,
+      "loss": 1.6087,
+      "step": 8686
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.568533969010726e-07,
+      "loss": 1.6877,
+      "step": 8687
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.56813667063965e-07,
+      "loss": 1.2808,
+      "step": 8688
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.567739372268574e-07,
+      "loss": 1.327,
+      "step": 8689
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.567342073897497e-07,
+      "loss": 1.9931,
+      "step": 8690
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.56694477552642e-07,
+      "loss": 1.4847,
+      "step": 8691
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.566547477155343e-07,
+      "loss": 2.186,
+      "step": 8692
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.566150178784267e-07,
+      "loss": 1.8453,
+      "step": 8693
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.56575288041319e-07,
+      "loss": 1.9464,
+      "step": 8694
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.565355582042114e-07,
+      "loss": 1.7594,
+      "step": 8695
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.564958283671038e-07,
+      "loss": 1.725,
+      "step": 8696
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.564560985299959e-07,
+      "loss": 2.3517,
+      "step": 8697
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.564163686928883e-07,
+      "loss": 1.6545,
+      "step": 8698
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.563766388557806e-07,
+      "loss": 1.4139,
+      "step": 8699
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.56336909018673e-07,
+      "loss": 1.6801,
+      "step": 8700
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.562971791815653e-07,
+      "loss": 1.6483,
+      "step": 8701
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.562574493444577e-07,
+      "loss": 1.5212,
+      "step": 8702
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.562177195073499e-07,
+      "loss": 1.5911,
+      "step": 8703
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.561779896702423e-07,
+      "loss": 1.7382,
+      "step": 8704
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.561382598331347e-07,
+      "loss": 1.4658,
+      "step": 8705
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.56098529996027e-07,
+      "loss": 1.5974,
+      "step": 8706
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.560588001589194e-07,
+      "loss": 1.3911,
+      "step": 8707
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.560190703218117e-07,
+      "loss": 1.467,
+      "step": 8708
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.55979340484704e-07,
+      "loss": 2.0368,
+      "step": 8709
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.559396106475963e-07,
+      "loss": 1.6662,
+      "step": 8710
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.558998808104886e-07,
+      "loss": 1.8305,
+      "step": 8711
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.55860150973381e-07,
+      "loss": 1.4755,
+      "step": 8712
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.558204211362733e-07,
+      "loss": 1.7944,
+      "step": 8713
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.557806912991657e-07,
+      "loss": 1.8077,
+      "step": 8714
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.557409614620579e-07,
+      "loss": 1.5918,
+      "step": 8715
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.557012316249503e-07,
+      "loss": 1.5769,
+      "step": 8716
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.556615017878426e-07,
+      "loss": 1.6032,
+      "step": 8717
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.55621771950735e-07,
+      "loss": 1.376,
+      "step": 8718
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.555820421136273e-07,
+      "loss": 1.4181,
+      "step": 8719
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.555423122765197e-07,
+      "loss": 1.6408,
+      "step": 8720
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.55502582439412e-07,
+      "loss": 1.7414,
+      "step": 8721
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.554628526023043e-07,
+      "loss": 2.2269,
+      "step": 8722
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.554231227651967e-07,
+      "loss": 1.9008,
+      "step": 8723
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.553833929280889e-07,
+      "loss": 2.1527,
+      "step": 8724
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.553436630909813e-07,
+      "loss": 1.8188,
+      "step": 8725
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.553039332538736e-07,
+      "loss": 1.7018,
+      "step": 8726
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.552642034167659e-07,
+      "loss": 1.6411,
+      "step": 8727
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.552244735796582e-07,
+      "loss": 1.5895,
+      "step": 8728
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.551847437425506e-07,
+      "loss": 1.9253,
+      "step": 8729
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.55145013905443e-07,
+      "loss": 1.5191,
+      "step": 8730
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.551052840683353e-07,
+      "loss": 1.3964,
+      "step": 8731
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.550655542312277e-07,
+      "loss": 1.9527,
+      "step": 8732
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.550258243941199e-07,
+      "loss": 1.5632,
+      "step": 8733
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.549860945570123e-07,
+      "loss": 1.9043,
+      "step": 8734
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.549463647199046e-07,
+      "loss": 1.4726,
+      "step": 8735
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.54906634882797e-07,
+      "loss": 1.8825,
+      "step": 8736
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.548669050456893e-07,
+      "loss": 1.6275,
+      "step": 8737
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.548271752085816e-07,
+      "loss": 1.593,
+      "step": 8738
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.547874453714739e-07,
+      "loss": 1.5405,
+      "step": 8739
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.547477155343662e-07,
+      "loss": 1.6416,
+      "step": 8740
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.547079856972586e-07,
+      "loss": 1.5967,
+      "step": 8741
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.546682558601509e-07,
+      "loss": 1.5238,
+      "step": 8742
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.546285260230433e-07,
+      "loss": 1.4365,
+      "step": 8743
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.545887961859357e-07,
+      "loss": 1.6114,
+      "step": 8744
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.545490663488279e-07,
+      "loss": 1.3473,
+      "step": 8745
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.545093365117203e-07,
+      "loss": 1.4405,
+      "step": 8746
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.544696066746126e-07,
+      "loss": 1.7937,
+      "step": 8747
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.54429876837505e-07,
+      "loss": 1.481,
+      "step": 8748
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.543901470003973e-07,
+      "loss": 2.0195,
+      "step": 8749
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.543504171632897e-07,
+      "loss": 1.6905,
+      "step": 8750
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.543106873261818e-07,
+      "loss": 1.86,
+      "step": 8751
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.542709574890742e-07,
+      "loss": 2.0046,
+      "step": 8752
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.542312276519666e-07,
+      "loss": 1.6364,
+      "step": 8753
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.541914978148589e-07,
+      "loss": 1.8563,
+      "step": 8754
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.541517679777513e-07,
+      "loss": 1.8416,
+      "step": 8755
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.541120381406436e-07,
+      "loss": 1.5555,
+      "step": 8756
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.540723083035359e-07,
+      "loss": 1.6669,
+      "step": 8757
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.540325784664282e-07,
+      "loss": 1.9294,
+      "step": 8758
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.539928486293206e-07,
+      "loss": 1.5201,
+      "step": 8759
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.53953118792213e-07,
+      "loss": 1.7632,
+      "step": 8760
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.539133889551053e-07,
+      "loss": 2.0465,
+      "step": 8761
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.538736591179977e-07,
+      "loss": 1.7757,
+      "step": 8762
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.538339292808898e-07,
+      "loss": 1.6215,
+      "step": 8763
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.537941994437822e-07,
+      "loss": 1.5488,
+      "step": 8764
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.537544696066745e-07,
+      "loss": 1.8718,
+      "step": 8765
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.537147397695669e-07,
+      "loss": 1.7236,
+      "step": 8766
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.536750099324593e-07,
+      "loss": 1.5054,
+      "step": 8767
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.536352800953516e-07,
+      "loss": 1.0149,
+      "step": 8768
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.535955502582439e-07,
+      "loss": 1.7453,
+      "step": 8769
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.535558204211362e-07,
+      "loss": 1.6064,
+      "step": 8770
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.535160905840286e-07,
+      "loss": 1.7822,
+      "step": 8771
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.534763607469209e-07,
+      "loss": 1.8022,
+      "step": 8772
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.534366309098133e-07,
+      "loss": 1.6317,
+      "step": 8773
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.533969010727057e-07,
+      "loss": 1.8003,
+      "step": 8774
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.533571712355979e-07,
+      "loss": 1.8484,
+      "step": 8775
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.533174413984903e-07,
+      "loss": 1.292,
+      "step": 8776
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.532777115613825e-07,
+      "loss": 1.5466,
+      "step": 8777
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.532379817242749e-07,
+      "loss": 1.7676,
+      "step": 8778
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.531982518871672e-07,
+      "loss": 1.8259,
+      "step": 8779
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.531585220500596e-07,
+      "loss": 1.7023,
+      "step": 8780
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.531187922129518e-07,
+      "loss": 1.7515,
+      "step": 8781
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.530790623758442e-07,
+      "loss": 1.3936,
+      "step": 8782
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.530393325387366e-07,
+      "loss": 1.6186,
+      "step": 8783
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.529996027016289e-07,
+      "loss": 1.6031,
+      "step": 8784
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.529598728645213e-07,
+      "loss": 1.8818,
+      "step": 8785
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.529201430274136e-07,
+      "loss": 1.8393,
+      "step": 8786
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.528804131903059e-07,
+      "loss": 2.0941,
+      "step": 8787
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.528406833531982e-07,
+      "loss": 1.9681,
+      "step": 8788
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.528009535160906e-07,
+      "loss": 1.3978,
+      "step": 8789
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.527612236789828e-07,
+      "loss": 1.7858,
+      "step": 8790
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.527214938418752e-07,
+      "loss": 1.7024,
+      "step": 8791
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.526817640047676e-07,
+      "loss": 1.583,
+      "step": 8792
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.526420341676598e-07,
+      "loss": 1.5415,
+      "step": 8793
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.526023043305522e-07,
+      "loss": 1.8259,
+      "step": 8794
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.525625744934445e-07,
+      "loss": 1.4533,
+      "step": 8795
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.525228446563369e-07,
+      "loss": 1.6213,
+      "step": 8796
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.524831148192292e-07,
+      "loss": 1.8733,
+      "step": 8797
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.524433849821216e-07,
+      "loss": 1.5174,
+      "step": 8798
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.524036551450139e-07,
+      "loss": 1.5084,
+      "step": 8799
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.523639253079062e-07,
+      "loss": 1.5648,
+      "step": 8800
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.523241954707986e-07,
+      "loss": 1.8375,
+      "step": 8801
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.522844656336909e-07,
+      "loss": 2.4075,
+      "step": 8802
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.522447357965832e-07,
+      "loss": 1.534,
+      "step": 8803
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.522050059594755e-07,
+      "loss": 1.9919,
+      "step": 8804
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.521652761223678e-07,
+      "loss": 2.0086,
+      "step": 8805
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.521255462852601e-07,
+      "loss": 1.778,
+      "step": 8806
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.520858164481525e-07,
+      "loss": 1.716,
+      "step": 8807
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.520460866110449e-07,
+      "loss": 1.3249,
+      "step": 8808
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.520063567739372e-07,
+      "loss": 1.8971,
+      "step": 8809
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.519666269368296e-07,
+      "loss": 1.8268,
+      "step": 8810
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.519268970997218e-07,
+      "loss": 1.8744,
+      "step": 8811
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.518871672626142e-07,
+      "loss": 1.6206,
+      "step": 8812
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.518474374255065e-07,
+      "loss": 1.7549,
+      "step": 8813
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.518077075883989e-07,
+      "loss": 1.5516,
+      "step": 8814
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.517679777512913e-07,
+      "loss": 1.886,
+      "step": 8815
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.517282479141836e-07,
+      "loss": 1.8523,
+      "step": 8816
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.516885180770758e-07,
+      "loss": 1.8185,
+      "step": 8817
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.516487882399681e-07,
+      "loss": 1.593,
+      "step": 8818
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.516090584028605e-07,
+      "loss": 1.4245,
+      "step": 8819
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.515693285657528e-07,
+      "loss": 1.7416,
+      "step": 8820
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.515295987286452e-07,
+      "loss": 1.2379,
+      "step": 8821
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.514898688915376e-07,
+      "loss": 1.5936,
+      "step": 8822
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.514501390544298e-07,
+      "loss": 1.5708,
+      "step": 8823
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.514104092173222e-07,
+      "loss": 1.2621,
+      "step": 8824
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.513706793802145e-07,
+      "loss": 1.9868,
+      "step": 8825
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.513309495431069e-07,
+      "loss": 1.7887,
+      "step": 8826
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.512912197059992e-07,
+      "loss": 1.6626,
+      "step": 8827
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.35725614591593974,
+      "eval_loss": 1.7358545064926147,
+      "eval_runtime": 6.425,
+      "eval_samples_per_second": 392.532,
+      "eval_steps_per_second": 6.226,
+      "step": 8827
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.512514898688916e-07,
+      "loss": 1.6293,
+      "step": 8828
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.512117600317837e-07,
+      "loss": 1.4875,
+      "step": 8829
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.511720301946761e-07,
+      "loss": 1.6666,
+      "step": 8830
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.511323003575685e-07,
+      "loss": 1.6434,
+      "step": 8831
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.510925705204608e-07,
+      "loss": 1.4626,
+      "step": 8832
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.510528406833532e-07,
+      "loss": 1.0759,
+      "step": 8833
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.510131108462455e-07,
+      "loss": 1.8394,
+      "step": 8834
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.509733810091378e-07,
+      "loss": 1.8055,
+      "step": 8835
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.509336511720301e-07,
+      "loss": 1.5717,
+      "step": 8836
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.508939213349225e-07,
+      "loss": 1.6671,
+      "step": 8837
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.508541914978149e-07,
+      "loss": 1.6951,
+      "step": 8838
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.508144616607072e-07,
+      "loss": 1.9312,
+      "step": 8839
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.507747318235996e-07,
+      "loss": 1.7085,
+      "step": 8840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.507350019864918e-07,
+      "loss": 1.7838,
+      "step": 8841
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.506952721493842e-07,
+      "loss": 1.5479,
+      "step": 8842
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.506555423122764e-07,
+      "loss": 1.6037,
+      "step": 8843
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.506158124751688e-07,
+      "loss": 1.5245,
+      "step": 8844
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.505760826380612e-07,
+      "loss": 1.3578,
+      "step": 8845
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.505363528009535e-07,
+      "loss": 1.3036,
+      "step": 8846
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.504966229638458e-07,
+      "loss": 1.8694,
+      "step": 8847
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.504568931267381e-07,
+      "loss": 1.698,
+      "step": 8848
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.504171632896305e-07,
+      "loss": 1.5344,
+      "step": 8849
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.503774334525228e-07,
+      "loss": 1.662,
+      "step": 8850
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.503377036154152e-07,
+      "loss": 1.901,
+      "step": 8851
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.502979737783076e-07,
+      "loss": 1.6578,
+      "step": 8852
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.502582439411998e-07,
+      "loss": 1.8246,
+      "step": 8853
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.502185141040922e-07,
+      "loss": 1.6886,
+      "step": 8854
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.501787842669845e-07,
+      "loss": 1.9993,
+      "step": 8855
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.501390544298768e-07,
+      "loss": 1.9724,
+      "step": 8856
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.500993245927691e-07,
+      "loss": 1.5104,
+      "step": 8857
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.500595947556615e-07,
+      "loss": 1.8718,
+      "step": 8858
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.500198649185537e-07,
+      "loss": 1.4448,
+      "step": 8859
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.499801350814461e-07,
+      "loss": 1.7318,
+      "step": 8860
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.499404052443385e-07,
+      "loss": 1.6049,
+      "step": 8861
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.499006754072308e-07,
+      "loss": 1.8361,
+      "step": 8862
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.498609455701232e-07,
+      "loss": 1.6245,
+      "step": 8863
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.498212157330155e-07,
+      "loss": 1.6187,
+      "step": 8864
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.497814858959078e-07,
+      "loss": 1.7388,
+      "step": 8865
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.497417560588001e-07,
+      "loss": 1.5264,
+      "step": 8866
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.497020262216925e-07,
+      "loss": 2.0828,
+      "step": 8867
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.496622963845849e-07,
+      "loss": 1.9096,
+      "step": 8868
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.496225665474771e-07,
+      "loss": 1.4773,
+      "step": 8869
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.495828367103695e-07,
+      "loss": 1.8756,
+      "step": 8870
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.495431068732617e-07,
+      "loss": 1.0924,
+      "step": 8871
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.495033770361541e-07,
+      "loss": 1.9317,
+      "step": 8872
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.494636471990464e-07,
+      "loss": 1.8321,
+      "step": 8873
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.494239173619388e-07,
+      "loss": 1.6835,
+      "step": 8874
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.493841875248311e-07,
+      "loss": 1.5354,
+      "step": 8875
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.493444576877235e-07,
+      "loss": 1.8897,
+      "step": 8876
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.493047278506158e-07,
+      "loss": 1.8559,
+      "step": 8877
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.492649980135081e-07,
+      "loss": 1.0954,
+      "step": 8878
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.492252681764005e-07,
+      "loss": 1.647,
+      "step": 8879
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.491855383392928e-07,
+      "loss": 1.4931,
+      "step": 8880
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.491458085021852e-07,
+      "loss": 1.8869,
+      "step": 8881
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.491060786650775e-07,
+      "loss": 1.8156,
+      "step": 8882
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.490663488279697e-07,
+      "loss": 1.6563,
+      "step": 8883
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.49026618990862e-07,
+      "loss": 1.9455,
+      "step": 8884
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.489868891537544e-07,
+      "loss": 1.8503,
+      "step": 8885
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.489471593166468e-07,
+      "loss": 1.4728,
+      "step": 8886
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.489074294795391e-07,
+      "loss": 1.564,
+      "step": 8887
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.488676996424315e-07,
+      "loss": 1.5033,
+      "step": 8888
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.488279698053237e-07,
+      "loss": 1.4902,
+      "step": 8889
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.487882399682161e-07,
+      "loss": 1.7051,
+      "step": 8890
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.487485101311084e-07,
+      "loss": 1.2398,
+      "step": 8891
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.487087802940008e-07,
+      "loss": 1.5597,
+      "step": 8892
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.486690504568932e-07,
+      "loss": 1.3592,
+      "step": 8893
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.486293206197855e-07,
+      "loss": 2.0555,
+      "step": 8894
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.485895907826777e-07,
+      "loss": 1.9217,
+      "step": 8895
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.4854986094557e-07,
+      "loss": 2.0234,
+      "step": 8896
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.485101311084624e-07,
+      "loss": 1.4735,
+      "step": 8897
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.484704012713547e-07,
+      "loss": 1.9175,
+      "step": 8898
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.484306714342471e-07,
+      "loss": 1.8776,
+      "step": 8899
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.483909415971395e-07,
+      "loss": 1.9524,
+      "step": 8900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.483512117600317e-07,
+      "loss": 1.374,
+      "step": 8901
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.483114819229241e-07,
+      "loss": 2.2055,
+      "step": 8902
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.482717520858164e-07,
+      "loss": 1.8869,
+      "step": 8903
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.482320222487088e-07,
+      "loss": 1.5678,
+      "step": 8904
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.481922924116011e-07,
+      "loss": 1.6623,
+      "step": 8905
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.481525625744935e-07,
+      "loss": 1.6572,
+      "step": 8906
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.481128327373858e-07,
+      "loss": 1.6074,
+      "step": 8907
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.480731029002781e-07,
+      "loss": 1.5079,
+      "step": 8908
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.480333730631704e-07,
+      "loss": 2.1347,
+      "step": 8909
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.479936432260627e-07,
+      "loss": 2.0779,
+      "step": 8910
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.479539133889551e-07,
+      "loss": 1.878,
+      "step": 8911
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.479141835518474e-07,
+      "loss": 1.6099,
+      "step": 8912
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.478744537147397e-07,
+      "loss": 1.7287,
+      "step": 8913
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.47834723877632e-07,
+      "loss": 1.4498,
+      "step": 8914
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.477949940405244e-07,
+      "loss": 1.5018,
+      "step": 8915
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.477552642034168e-07,
+      "loss": 1.4928,
+      "step": 8916
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.477155343663091e-07,
+      "loss": 1.7573,
+      "step": 8917
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.476758045292015e-07,
+      "loss": 1.8739,
+      "step": 8918
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.476360746920937e-07,
+      "loss": 1.6016,
+      "step": 8919
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.475963448549861e-07,
+      "loss": 1.8541,
+      "step": 8920
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.475566150178784e-07,
+      "loss": 1.6204,
+      "step": 8921
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.475168851807707e-07,
+      "loss": 1.067,
+      "step": 8922
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.47477155343663e-07,
+      "loss": 1.5573,
+      "step": 8923
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.474374255065554e-07,
+      "loss": 1.7394,
+      "step": 8924
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.473976956694477e-07,
+      "loss": 1.66,
+      "step": 8925
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.4735796583234e-07,
+      "loss": 1.3013,
+      "step": 8926
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.473182359952324e-07,
+      "loss": 1.6664,
+      "step": 8927
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.472785061581247e-07,
+      "loss": 1.5847,
+      "step": 8928
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.472387763210171e-07,
+      "loss": 1.5449,
+      "step": 8929
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.471990464839095e-07,
+      "loss": 1.3121,
+      "step": 8930
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.471593166468017e-07,
+      "loss": 1.8671,
+      "step": 8931
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.471195868096941e-07,
+      "loss": 1.7652,
+      "step": 8932
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.470798569725864e-07,
+      "loss": 1.7218,
+      "step": 8933
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.470401271354788e-07,
+      "loss": 1.4859,
+      "step": 8934
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.47000397298371e-07,
+      "loss": 1.566,
+      "step": 8935
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.469606674612634e-07,
+      "loss": 1.5641,
+      "step": 8936
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.469209376241556e-07,
+      "loss": 1.7383,
+      "step": 8937
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.46881207787048e-07,
+      "loss": 1.5099,
+      "step": 8938
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.468414779499404e-07,
+      "loss": 1.6386,
+      "step": 8939
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.468017481128327e-07,
+      "loss": 1.598,
+      "step": 8940
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.467620182757251e-07,
+      "loss": 2.1935,
+      "step": 8941
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.467222884386174e-07,
+      "loss": 2.0401,
+      "step": 8942
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.466825586015097e-07,
+      "loss": 2.0355,
+      "step": 8943
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.46642828764402e-07,
+      "loss": 1.6637,
+      "step": 8944
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.466030989272944e-07,
+      "loss": 1.5055,
+      "step": 8945
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.465633690901868e-07,
+      "loss": 1.7149,
+      "step": 8946
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.465236392530791e-07,
+      "loss": 1.7386,
+      "step": 8947
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.464839094159715e-07,
+      "loss": 2.0065,
+      "step": 8948
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.464441795788636e-07,
+      "loss": 1.7104,
+      "step": 8949
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.46404449741756e-07,
+      "loss": 1.5127,
+      "step": 8950
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.463647199046483e-07,
+      "loss": 1.6838,
+      "step": 8951
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.463249900675407e-07,
+      "loss": 1.8737,
+      "step": 8952
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.46285260230433e-07,
+      "loss": 1.5639,
+      "step": 8953
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.462455303933254e-07,
+      "loss": 1.6484,
+      "step": 8954
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.462058005562177e-07,
+      "loss": 1.6127,
+      "step": 8955
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.4616607071911e-07,
+      "loss": 1.998,
+      "step": 8956
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.461263408820024e-07,
+      "loss": 1.8232,
+      "step": 8957
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.460866110448947e-07,
+      "loss": 1.8491,
+      "step": 8958
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.460468812077871e-07,
+      "loss": 1.3421,
+      "step": 8959
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.460071513706794e-07,
+      "loss": 1.2919,
+      "step": 8960
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.459674215335716e-07,
+      "loss": 2.0159,
+      "step": 8961
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.45927691696464e-07,
+      "loss": 1.226,
+      "step": 8962
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.458879618593563e-07,
+      "loss": 1.5694,
+      "step": 8963
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.458482320222487e-07,
+      "loss": 1.1191,
+      "step": 8964
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.45808502185141e-07,
+      "loss": 1.5449,
+      "step": 8965
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.457687723480334e-07,
+      "loss": 1.5747,
+      "step": 8966
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.457290425109256e-07,
+      "loss": 1.4319,
+      "step": 8967
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.45689312673818e-07,
+      "loss": 1.5926,
+      "step": 8968
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.456495828367103e-07,
+      "loss": 1.7309,
+      "step": 8969
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.456098529996027e-07,
+      "loss": 1.9369,
+      "step": 8970
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.455701231624951e-07,
+      "loss": 1.3761,
+      "step": 8971
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.455303933253874e-07,
+      "loss": 1.1392,
+      "step": 8972
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.454906634882797e-07,
+      "loss": 1.4492,
+      "step": 8973
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.45450933651172e-07,
+      "loss": 1.6792,
+      "step": 8974
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.454112038140643e-07,
+      "loss": 1.9156,
+      "step": 8975
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.453714739769566e-07,
+      "loss": 1.8316,
+      "step": 8976
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.45331744139849e-07,
+      "loss": 1.8111,
+      "step": 8977
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.452920143027414e-07,
+      "loss": 1.3997,
+      "step": 8978
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.452522844656336e-07,
+      "loss": 2.1125,
+      "step": 8979
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.45212554628526e-07,
+      "loss": 1.2727,
+      "step": 8980
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.451728247914183e-07,
+      "loss": 1.6974,
+      "step": 8981
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.451330949543107e-07,
+      "loss": 1.6924,
+      "step": 8982
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.45093365117203e-07,
+      "loss": 1.9399,
+      "step": 8983
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.450536352800954e-07,
+      "loss": 1.6247,
+      "step": 8984
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.450139054429877e-07,
+      "loss": 1.3544,
+      "step": 8985
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.4497417560588e-07,
+      "loss": 1.7226,
+      "step": 8986
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.449344457687724e-07,
+      "loss": 1.1266,
+      "step": 8987
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.448947159316646e-07,
+      "loss": 1.9143,
+      "step": 8988
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.44854986094557e-07,
+      "loss": 1.6028,
+      "step": 8989
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.448152562574493e-07,
+      "loss": 1.7179,
+      "step": 8990
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.447755264203416e-07,
+      "loss": 1.8112,
+      "step": 8991
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.447357965832339e-07,
+      "loss": 1.5234,
+      "step": 8992
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.446960667461263e-07,
+      "loss": 1.5976,
+      "step": 8993
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.446563369090187e-07,
+      "loss": 1.8997,
+      "step": 8994
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.44616607071911e-07,
+      "loss": 1.6862,
+      "step": 8995
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.445768772348034e-07,
+      "loss": 1.927,
+      "step": 8996
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.445371473976956e-07,
+      "loss": 1.3926,
+      "step": 8997
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.44497417560588e-07,
+      "loss": 1.4476,
+      "step": 8998
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.444576877234803e-07,
+      "loss": 1.8513,
+      "step": 8999
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.444179578863727e-07,
+      "loss": 1.8766,
+      "step": 9000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.44378228049265e-07,
+      "loss": 1.5455,
+      "step": 9001
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.443384982121573e-07,
+      "loss": 1.9583,
+      "step": 9002
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.442987683750496e-07,
+      "loss": 1.6449,
+      "step": 9003
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.442590385379419e-07,
+      "loss": 2.1376,
+      "step": 9004
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.442193087008343e-07,
+      "loss": 1.3074,
+      "step": 9005
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.441795788637266e-07,
+      "loss": 1.6332,
+      "step": 9006
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.44139849026619e-07,
+      "loss": 1.6752,
+      "step": 9007
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.441001191895114e-07,
+      "loss": 1.2338,
+      "step": 9008
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.440603893524036e-07,
+      "loss": 2.0349,
+      "step": 9009
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.44020659515296e-07,
+      "loss": 1.8279,
+      "step": 9010
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.439809296781883e-07,
+      "loss": 1.7359,
+      "step": 9011
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.439411998410807e-07,
+      "loss": 1.6197,
+      "step": 9012
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.43901470003973e-07,
+      "loss": 1.4344,
+      "step": 9013
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.438617401668654e-07,
+      "loss": 1.4767,
+      "step": 9014
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.438220103297575e-07,
+      "loss": 1.6026,
+      "step": 9015
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.437822804926499e-07,
+      "loss": 1.5437,
+      "step": 9016
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.437425506555423e-07,
+      "loss": 2.0828,
+      "step": 9017
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.437028208184346e-07,
+      "loss": 1.6777,
+      "step": 9018
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.43663090981327e-07,
+      "loss": 1.4806,
+      "step": 9019
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.436233611442193e-07,
+      "loss": 1.289,
+      "step": 9020
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.435836313071116e-07,
+      "loss": 1.8428,
+      "step": 9021
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.435439014700039e-07,
+      "loss": 1.8682,
+      "step": 9022
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.435041716328963e-07,
+      "loss": 1.833,
+      "step": 9023
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.434644417957887e-07,
+      "loss": 1.1368,
+      "step": 9024
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.43424711958681e-07,
+      "loss": 2.1405,
+      "step": 9025
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.433849821215734e-07,
+      "loss": 1.1568,
+      "step": 9026
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.433452522844655e-07,
+      "loss": 1.5777,
+      "step": 9027
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.433055224473579e-07,
+      "loss": 1.8015,
+      "step": 9028
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.432657926102502e-07,
+      "loss": 1.5059,
+      "step": 9029
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.432260627731426e-07,
+      "loss": 1.6477,
+      "step": 9030
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.43186332936035e-07,
+      "loss": 1.8588,
+      "step": 9031
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.431466030989273e-07,
+      "loss": 1.6005,
+      "step": 9032
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.431068732618196e-07,
+      "loss": 1.3303,
+      "step": 9033
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.430671434247119e-07,
+      "loss": 1.9292,
+      "step": 9034
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.430274135876043e-07,
+      "loss": 1.7149,
+      "step": 9035
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.429876837504966e-07,
+      "loss": 2.1587,
+      "step": 9036
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.42947953913389e-07,
+      "loss": 1.4551,
+      "step": 9037
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.429082240762813e-07,
+      "loss": 1.6574,
+      "step": 9038
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.428684942391736e-07,
+      "loss": 1.5034,
+      "step": 9039
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.42828764402066e-07,
+      "loss": 1.4048,
+      "step": 9040
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.427890345649582e-07,
+      "loss": 1.7604,
+      "step": 9041
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.427493047278506e-07,
+      "loss": 1.6944,
+      "step": 9042
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.427095748907429e-07,
+      "loss": 1.7972,
+      "step": 9043
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.426698450536352e-07,
+      "loss": 1.7958,
+      "step": 9044
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.426301152165275e-07,
+      "loss": 1.7647,
+      "step": 9045
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.425903853794199e-07,
+      "loss": 1.6324,
+      "step": 9046
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.425506555423122e-07,
+      "loss": 1.4022,
+      "step": 9047
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.425109257052046e-07,
+      "loss": 1.6918,
+      "step": 9048
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.42471195868097e-07,
+      "loss": 1.9542,
+      "step": 9049
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.424314660309892e-07,
+      "loss": 1.466,
+      "step": 9050
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.423917361938816e-07,
+      "loss": 1.279,
+      "step": 9051
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.423520063567739e-07,
+      "loss": 1.6326,
+      "step": 9052
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.423122765196663e-07,
+      "loss": 1.4096,
+      "step": 9053
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.422725466825585e-07,
+      "loss": 1.7144,
+      "step": 9054
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.422328168454509e-07,
+      "loss": 1.6905,
+      "step": 9055
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.421930870083432e-07,
+      "loss": 1.7664,
+      "step": 9056
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.421533571712355e-07,
+      "loss": 1.7943,
+      "step": 9057
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.421136273341279e-07,
+      "loss": 1.6756,
+      "step": 9058
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.420738974970202e-07,
+      "loss": 1.6551,
+      "step": 9059
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.420341676599126e-07,
+      "loss": 1.8032,
+      "step": 9060
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.419944378228049e-07,
+      "loss": 1.5613,
+      "step": 9061
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.419547079856972e-07,
+      "loss": 1.6862,
+      "step": 9062
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.419149781485896e-07,
+      "loss": 1.8845,
+      "step": 9063
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.418752483114819e-07,
+      "loss": 1.857,
+      "step": 9064
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.418355184743743e-07,
+      "loss": 1.8789,
+      "step": 9065
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.417957886372666e-07,
+      "loss": 1.7056,
+      "step": 9066
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.417560588001589e-07,
+      "loss": 1.5256,
+      "step": 9067
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.417163289630511e-07,
+      "loss": 1.2816,
+      "step": 9068
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.416765991259435e-07,
+      "loss": 1.6102,
+      "step": 9069
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.416368692888358e-07,
+      "loss": 1.1819,
+      "step": 9070
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.415971394517282e-07,
+      "loss": 1.9011,
+      "step": 9071
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.415574096146206e-07,
+      "loss": 1.5207,
+      "step": 9072
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.415176797775129e-07,
+      "loss": 1.3987,
+      "step": 9073
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.414779499404052e-07,
+      "loss": 1.4069,
+      "step": 9074
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.414382201032975e-07,
+      "loss": 1.763,
+      "step": 9075
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.413984902661899e-07,
+      "loss": 1.6717,
+      "step": 9076
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.413587604290822e-07,
+      "loss": 1.9913,
+      "step": 9077
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.413190305919746e-07,
+      "loss": 1.3572,
+      "step": 9078
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.41279300754867e-07,
+      "loss": 1.4889,
+      "step": 9079
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.412395709177591e-07,
+      "loss": 1.3832,
+      "step": 9080
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.411998410806515e-07,
+      "loss": 1.9767,
+      "step": 9081
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.411601112435438e-07,
+      "loss": 1.9407,
+      "step": 9082
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.411203814064362e-07,
+      "loss": 1.6243,
+      "step": 9083
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.410806515693285e-07,
+      "loss": 1.5996,
+      "step": 9084
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.410409217322209e-07,
+      "loss": 1.4139,
+      "step": 9085
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.410011918951131e-07,
+      "loss": 2.1848,
+      "step": 9086
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.409614620580055e-07,
+      "loss": 1.559,
+      "step": 9087
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.409217322208979e-07,
+      "loss": 1.4959,
+      "step": 9088
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.408820023837902e-07,
+      "loss": 1.5154,
+      "step": 9089
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.408422725466826e-07,
+      "loss": 2.3406,
+      "step": 9090
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.408025427095749e-07,
+      "loss": 1.7538,
+      "step": 9091
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.407628128724672e-07,
+      "loss": 1.9353,
+      "step": 9092
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.407230830353594e-07,
+      "loss": 1.5477,
+      "step": 9093
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.406833531982518e-07,
+      "loss": 1.6086,
+      "step": 9094
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.406436233611442e-07,
+      "loss": 1.6299,
+      "step": 9095
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.406038935240365e-07,
+      "loss": 1.4009,
+      "step": 9096
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.405641636869289e-07,
+      "loss": 1.6576,
+      "step": 9097
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.405244338498211e-07,
+      "loss": 1.8506,
+      "step": 9098
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.404847040127135e-07,
+      "loss": 1.708,
+      "step": 9099
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.404449741756058e-07,
+      "loss": 2.0235,
+      "step": 9100
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.404052443384982e-07,
+      "loss": 1.2497,
+      "step": 9101
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.403655145013906e-07,
+      "loss": 2.1552,
+      "step": 9102
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.403257846642829e-07,
+      "loss": 1.6403,
+      "step": 9103
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.402860548271752e-07,
+      "loss": 1.335,
+      "step": 9104
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.402463249900675e-07,
+      "loss": 1.1338,
+      "step": 9105
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.402065951529599e-07,
+      "loss": 1.5649,
+      "step": 9106
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.401668653158521e-07,
+      "loss": 1.3758,
+      "step": 9107
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.401271354787445e-07,
+      "loss": 1.7215,
+      "step": 9108
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.400874056416368e-07,
+      "loss": 1.6238,
+      "step": 9109
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.400476758045291e-07,
+      "loss": 1.4176,
+      "step": 9110
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.400079459674215e-07,
+      "loss": 2.0482,
+      "step": 9111
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.399682161303138e-07,
+      "loss": 1.4192,
+      "step": 9112
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.399284862932062e-07,
+      "loss": 2.1897,
+      "step": 9113
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.398887564560985e-07,
+      "loss": 1.8699,
+      "step": 9114
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.398490266189909e-07,
+      "loss": 1.9682,
+      "step": 9115
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.398092967818831e-07,
+      "loss": 1.6791,
+      "step": 9116
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.397695669447755e-07,
+      "loss": 1.5668,
+      "step": 9117
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.397298371076679e-07,
+      "loss": 1.3573,
+      "step": 9118
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.396901072705602e-07,
+      "loss": 1.318,
+      "step": 9119
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.396503774334525e-07,
+      "loss": 1.8001,
+      "step": 9120
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.396106475963448e-07,
+      "loss": 1.5325,
+      "step": 9121
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.395709177592371e-07,
+      "loss": 2.5767,
+      "step": 9122
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.395311879221294e-07,
+      "loss": 1.7333,
+      "step": 9123
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.394914580850218e-07,
+      "loss": 1.6174,
+      "step": 9124
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.394517282479141e-07,
+      "loss": 1.6263,
+      "step": 9125
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.394119984108065e-07,
+      "loss": 1.5461,
+      "step": 9126
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.393722685736989e-07,
+      "loss": 1.6753,
+      "step": 9127
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.393325387365911e-07,
+      "loss": 1.9894,
+      "step": 9128
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.392928088994835e-07,
+      "loss": 1.5371,
+      "step": 9129
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.392530790623758e-07,
+      "loss": 1.4573,
+      "step": 9130
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.392133492252682e-07,
+      "loss": 1.3055,
+      "step": 9131
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.391736193881605e-07,
+      "loss": 1.5516,
+      "step": 9132
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.391338895510528e-07,
+      "loss": 1.433,
+      "step": 9133
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.39094159713945e-07,
+      "loss": 2.0482,
+      "step": 9134
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.390544298768374e-07,
+      "loss": 1.6802,
+      "step": 9135
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.390147000397298e-07,
+      "loss": 1.7307,
+      "step": 9136
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.389749702026221e-07,
+      "loss": 1.6229,
+      "step": 9137
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.389352403655145e-07,
+      "loss": 1.6813,
+      "step": 9138
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.388955105284068e-07,
+      "loss": 1.6076,
+      "step": 9139
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.388557806912991e-07,
+      "loss": 1.4419,
+      "step": 9140
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.388160508541915e-07,
+      "loss": 1.513,
+      "step": 9141
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.387763210170838e-07,
+      "loss": 1.5135,
+      "step": 9142
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.387365911799762e-07,
+      "loss": 1.6563,
+      "step": 9143
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.386968613428685e-07,
+      "loss": 1.8764,
+      "step": 9144
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.386571315057609e-07,
+      "loss": 1.7144,
+      "step": 9145
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.38617401668653e-07,
+      "loss": 1.3823,
+      "step": 9146
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.385776718315454e-07,
+      "loss": 1.7527,
+      "step": 9147
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.385379419944377e-07,
+      "loss": 1.4167,
+      "step": 9148
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.384982121573301e-07,
+      "loss": 1.907,
+      "step": 9149
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.384584823202225e-07,
+      "loss": 1.9717,
+      "step": 9150
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.384187524831148e-07,
+      "loss": 1.5097,
+      "step": 9151
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.383790226460071e-07,
+      "loss": 1.4396,
+      "step": 9152
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.383392928088994e-07,
+      "loss": 1.5313,
+      "step": 9153
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.382995629717918e-07,
+      "loss": 1.5671,
+      "step": 9154
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.382598331346841e-07,
+      "loss": 1.4654,
+      "step": 9155
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.382201032975765e-07,
+      "loss": 1.6498,
+      "step": 9156
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.381803734604689e-07,
+      "loss": 1.5046,
+      "step": 9157
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.381406436233611e-07,
+      "loss": 1.6517,
+      "step": 9158
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.381009137862534e-07,
+      "loss": 1.2812,
+      "step": 9159
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.380611839491457e-07,
+      "loss": 1.696,
+      "step": 9160
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.380214541120381e-07,
+      "loss": 1.2844,
+      "step": 9161
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.379817242749304e-07,
+      "loss": 1.6495,
+      "step": 9162
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.379419944378228e-07,
+      "loss": 1.4879,
+      "step": 9163
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.37902264600715e-07,
+      "loss": 1.5396,
+      "step": 9164
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.378625347636074e-07,
+      "loss": 1.6926,
+      "step": 9165
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.378228049264998e-07,
+      "loss": 1.61,
+      "step": 9166
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.377830750893921e-07,
+      "loss": 1.9051,
+      "step": 9167
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.377433452522845e-07,
+      "loss": 1.9654,
+      "step": 9168
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.377036154151768e-07,
+      "loss": 1.8558,
+      "step": 9169
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.376638855780691e-07,
+      "loss": 1.5676,
+      "step": 9170
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.376241557409614e-07,
+      "loss": 1.8966,
+      "step": 9171
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.375844259038538e-07,
+      "loss": 2.2347,
+      "step": 9172
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.375446960667461e-07,
+      "loss": 1.4451,
+      "step": 9173
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.375049662296384e-07,
+      "loss": 1.7343,
+      "step": 9174
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.374652363925308e-07,
+      "loss": 1.8745,
+      "step": 9175
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.37425506555423e-07,
+      "loss": 1.7994,
+      "step": 9176
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.373857767183154e-07,
+      "loss": 1.7181,
+      "step": 9177
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.373460468812077e-07,
+      "loss": 1.5477,
+      "step": 9178
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.373063170441001e-07,
+      "loss": 1.6974,
+      "step": 9179
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.372665872069925e-07,
+      "loss": 1.4098,
+      "step": 9180
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.372268573698848e-07,
+      "loss": 1.6296,
+      "step": 9181
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.371871275327771e-07,
+      "loss": 1.6404,
+      "step": 9182
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.371473976956694e-07,
+      "loss": 1.7874,
+      "step": 9183
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.371076678585618e-07,
+      "loss": 1.4909,
+      "step": 9184
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.370679380214541e-07,
+      "loss": 1.6918,
+      "step": 9185
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.370282081843464e-07,
+      "loss": 1.3612,
+      "step": 9186
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.369884783472387e-07,
+      "loss": 1.4986,
+      "step": 9187
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.36948748510131e-07,
+      "loss": 1.7631,
+      "step": 9188
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.369090186730234e-07,
+      "loss": 1.824,
+      "step": 9189
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.368692888359157e-07,
+      "loss": 1.5131,
+      "step": 9190
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.368295589988081e-07,
+      "loss": 1.6658,
+      "step": 9191
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.367898291617004e-07,
+      "loss": 1.6122,
+      "step": 9192
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.367500993245928e-07,
+      "loss": 1.3813,
+      "step": 9193
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.36710369487485e-07,
+      "loss": 1.6487,
+      "step": 9194
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.366706396503774e-07,
+      "loss": 1.7762,
+      "step": 9195
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.366309098132698e-07,
+      "loss": 1.6826,
+      "step": 9196
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.365911799761621e-07,
+      "loss": 1.3865,
+      "step": 9197
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.365514501390545e-07,
+      "loss": 1.7684,
+      "step": 9198
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.365117203019467e-07,
+      "loss": 1.4571,
+      "step": 9199
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.36471990464839e-07,
+      "loss": 1.4754,
+      "step": 9200
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.364322606277313e-07,
+      "loss": 1.657,
+      "step": 9201
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.363925307906237e-07,
+      "loss": 1.8941,
+      "step": 9202
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.36352800953516e-07,
+      "loss": 2.1243,
+      "step": 9203
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.363130711164084e-07,
+      "loss": 1.361,
+      "step": 9204
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.362733412793008e-07,
+      "loss": 1.9311,
+      "step": 9205
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.36233611442193e-07,
+      "loss": 1.6993,
+      "step": 9206
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.361938816050854e-07,
+      "loss": 1.6837,
+      "step": 9207
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.361541517679777e-07,
+      "loss": 1.2781,
+      "step": 9208
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.361144219308701e-07,
+      "loss": 1.9162,
+      "step": 9209
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.360746920937624e-07,
+      "loss": 1.9251,
+      "step": 9210
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.360349622566548e-07,
+      "loss": 1.4317,
+      "step": 9211
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.35995232419547e-07,
+      "loss": 1.9066,
+      "step": 9212
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.359555025824393e-07,
+      "loss": 1.5631,
+      "step": 9213
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.359157727453317e-07,
+      "loss": 1.7796,
+      "step": 9214
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.35876042908224e-07,
+      "loss": 1.5547,
+      "step": 9215
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.358363130711164e-07,
+      "loss": 2.3414,
+      "step": 9216
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.357965832340087e-07,
+      "loss": 1.4221,
+      "step": 9217
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.35756853396901e-07,
+      "loss": 1.5266,
+      "step": 9218
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.357171235597934e-07,
+      "loss": 1.2308,
+      "step": 9219
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.356773937226857e-07,
+      "loss": 1.6711,
+      "step": 9220
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.356376638855781e-07,
+      "loss": 1.4899,
+      "step": 9221
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.355979340484704e-07,
+      "loss": 1.7227,
+      "step": 9222
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.355582042113628e-07,
+      "loss": 1.3293,
+      "step": 9223
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.35518474374255e-07,
+      "loss": 1.666,
+      "step": 9224
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.354787445371473e-07,
+      "loss": 1.5835,
+      "step": 9225
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.354390147000396e-07,
+      "loss": 1.8169,
+      "step": 9226
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.35399284862932e-07,
+      "loss": 1.7626,
+      "step": 9227
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.353595550258244e-07,
+      "loss": 1.8439,
+      "step": 9228
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.353198251887167e-07,
+      "loss": 1.1805,
+      "step": 9229
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.35280095351609e-07,
+      "loss": 1.7067,
+      "step": 9230
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.352403655145013e-07,
+      "loss": 1.264,
+      "step": 9231
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.352006356773937e-07,
+      "loss": 1.05,
+      "step": 9232
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.35160905840286e-07,
+      "loss": 1.9592,
+      "step": 9233
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.351211760031784e-07,
+      "loss": 1.4593,
+      "step": 9234
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.350814461660708e-07,
+      "loss": 1.5074,
+      "step": 9235
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.35041716328963e-07,
+      "loss": 1.7801,
+      "step": 9236
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.350019864918554e-07,
+      "loss": 1.6949,
+      "step": 9237
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.349622566547477e-07,
+      "loss": 1.3521,
+      "step": 9238
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.3492252681764e-07,
+      "loss": 1.9584,
+      "step": 9239
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.348827969805323e-07,
+      "loss": 1.6217,
+      "step": 9240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.348430671434247e-07,
+      "loss": 1.4591,
+      "step": 9241
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.348033373063169e-07,
+      "loss": 1.5601,
+      "step": 9242
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.347636074692093e-07,
+      "loss": 1.7614,
+      "step": 9243
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.347238776321017e-07,
+      "loss": 1.5456,
+      "step": 9244
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.34684147794994e-07,
+      "loss": 1.2934,
+      "step": 9245
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.346444179578864e-07,
+      "loss": 1.7467,
+      "step": 9246
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.346046881207787e-07,
+      "loss": 2.1337,
+      "step": 9247
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.34564958283671e-07,
+      "loss": 1.6818,
+      "step": 9248
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.345252284465633e-07,
+      "loss": 1.4195,
+      "step": 9249
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.344854986094557e-07,
+      "loss": 1.4436,
+      "step": 9250
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.344457687723481e-07,
+      "loss": 1.5853,
+      "step": 9251
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.344060389352403e-07,
+      "loss": 1.5337,
+      "step": 9252
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.343663090981327e-07,
+      "loss": 1.679,
+      "step": 9253
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.343265792610249e-07,
+      "loss": 2.0515,
+      "step": 9254
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.342868494239173e-07,
+      "loss": 1.9709,
+      "step": 9255
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.342471195868096e-07,
+      "loss": 1.5904,
+      "step": 9256
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.34207389749702e-07,
+      "loss": 1.2263,
+      "step": 9257
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.341676599125944e-07,
+      "loss": 1.7685,
+      "step": 9258
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.341279300754867e-07,
+      "loss": 1.5171,
+      "step": 9259
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.34088200238379e-07,
+      "loss": 1.5469,
+      "step": 9260
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.340484704012713e-07,
+      "loss": 2.1657,
+      "step": 9261
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.340087405641637e-07,
+      "loss": 1.5525,
+      "step": 9262
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.33969010727056e-07,
+      "loss": 1.7349,
+      "step": 9263
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.339292808899484e-07,
+      "loss": 1.1977,
+      "step": 9264
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.338895510528406e-07,
+      "loss": 1.9121,
+      "step": 9265
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.338498212157329e-07,
+      "loss": 1.9324,
+      "step": 9266
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.338100913786253e-07,
+      "loss": 1.3388,
+      "step": 9267
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.337703615415176e-07,
+      "loss": 1.5987,
+      "step": 9268
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.3373063170441e-07,
+      "loss": 1.5689,
+      "step": 9269
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.336909018673023e-07,
+      "loss": 2.0349,
+      "step": 9270
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.336511720301947e-07,
+      "loss": 1.8869,
+      "step": 9271
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.336114421930869e-07,
+      "loss": 1.9647,
+      "step": 9272
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.335717123559793e-07,
+      "loss": 1.7156,
+      "step": 9273
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.335319825188717e-07,
+      "loss": 1.5954,
+      "step": 9274
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.33492252681764e-07,
+      "loss": 1.542,
+      "step": 9275
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.334525228446564e-07,
+      "loss": 2.0352,
+      "step": 9276
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.334127930075487e-07,
+      "loss": 1.7469,
+      "step": 9277
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.333730631704409e-07,
+      "loss": 2.0027,
+      "step": 9278
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.333333333333332e-07,
+      "loss": 1.6389,
+      "step": 9279
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.332936034962256e-07,
+      "loss": 1.7333,
+      "step": 9280
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.33253873659118e-07,
+      "loss": 1.5871,
+      "step": 9281
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.332141438220103e-07,
+      "loss": 1.5655,
+      "step": 9282
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.331744139849027e-07,
+      "loss": 1.4856,
+      "step": 9283
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.331346841477949e-07,
+      "loss": 1.8744,
+      "step": 9284
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.330949543106873e-07,
+      "loss": 1.5392,
+      "step": 9285
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.330552244735796e-07,
+      "loss": 1.9002,
+      "step": 9286
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.33015494636472e-07,
+      "loss": 1.5513,
+      "step": 9287
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.329757647993643e-07,
+      "loss": 1.6065,
+      "step": 9288
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.329360349622567e-07,
+      "loss": 2.0505,
+      "step": 9289
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.32896305125149e-07,
+      "loss": 1.5354,
+      "step": 9290
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.328565752880412e-07,
+      "loss": 1.5435,
+      "step": 9291
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.328168454509336e-07,
+      "loss": 1.6806,
+      "step": 9292
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.327771156138259e-07,
+      "loss": 1.641,
+      "step": 9293
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.327373857767183e-07,
+      "loss": 1.556,
+      "step": 9294
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.326976559396106e-07,
+      "loss": 1.9178,
+      "step": 9295
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.326579261025029e-07,
+      "loss": 2.0465,
+      "step": 9296
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.326181962653953e-07,
+      "loss": 1.4359,
+      "step": 9297
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.325784664282876e-07,
+      "loss": 1.4036,
+      "step": 9298
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.3253873659118e-07,
+      "loss": 1.5623,
+      "step": 9299
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.324990067540723e-07,
+      "loss": 1.7053,
+      "step": 9300
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.324592769169647e-07,
+      "loss": 1.3726,
+      "step": 9301
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.324195470798569e-07,
+      "loss": 1.2377,
+      "step": 9302
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.323798172427493e-07,
+      "loss": 1.9893,
+      "step": 9303
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.323400874056417e-07,
+      "loss": 1.6546,
+      "step": 9304
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.323003575685339e-07,
+      "loss": 1.6236,
+      "step": 9305
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.322606277314263e-07,
+      "loss": 1.722,
+      "step": 9306
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.322208978943186e-07,
+      "loss": 1.6517,
+      "step": 9307
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.321811680572109e-07,
+      "loss": 1.5509,
+      "step": 9308
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.321414382201032e-07,
+      "loss": 1.5234,
+      "step": 9309
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.321017083829956e-07,
+      "loss": 1.2799,
+      "step": 9310
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.320619785458879e-07,
+      "loss": 1.9348,
+      "step": 9311
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.320222487087803e-07,
+      "loss": 1.8678,
+      "step": 9312
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.319825188716727e-07,
+      "loss": 1.7304,
+      "step": 9313
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.319427890345649e-07,
+      "loss": 1.1367,
+      "step": 9314
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.319030591974573e-07,
+      "loss": 1.6478,
+      "step": 9315
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.318633293603496e-07,
+      "loss": 1.4901,
+      "step": 9316
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.31823599523242e-07,
+      "loss": 1.7984,
+      "step": 9317
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.317838696861342e-07,
+      "loss": 1.7944,
+      "step": 9318
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.317441398490266e-07,
+      "loss": 1.8658,
+      "step": 9319
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.317044100119188e-07,
+      "loss": 1.5447,
+      "step": 9320
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.316646801748112e-07,
+      "loss": 1.7004,
+      "step": 9321
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.316249503377036e-07,
+      "loss": 1.8363,
+      "step": 9322
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.315852205005959e-07,
+      "loss": 1.5536,
+      "step": 9323
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.315454906634883e-07,
+      "loss": 1.8482,
+      "step": 9324
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.315057608263806e-07,
+      "loss": 1.0586,
+      "step": 9325
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.314660309892729e-07,
+      "loss": 1.3925,
+      "step": 9326
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.314263011521652e-07,
+      "loss": 1.8044,
+      "step": 9327
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.313865713150576e-07,
+      "loss": 1.6782,
+      "step": 9328
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.3134684147795e-07,
+      "loss": 1.7226,
+      "step": 9329
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.313071116408423e-07,
+      "loss": 1.7431,
+      "step": 9330
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.312673818037346e-07,
+      "loss": 1.5919,
+      "step": 9331
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.312276519666268e-07,
+      "loss": 1.3788,
+      "step": 9332
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.311879221295192e-07,
+      "loss": 1.4871,
+      "step": 9333
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.311481922924115e-07,
+      "loss": 1.4679,
+      "step": 9334
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.311084624553039e-07,
+      "loss": 1.7406,
+      "step": 9335
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.310687326181963e-07,
+      "loss": 1.3342,
+      "step": 9336
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.310290027810886e-07,
+      "loss": 1.8944,
+      "step": 9337
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.309892729439809e-07,
+      "loss": 1.3099,
+      "step": 9338
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.309495431068732e-07,
+      "loss": 1.3816,
+      "step": 9339
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.309098132697656e-07,
+      "loss": 1.8436,
+      "step": 9340
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.308700834326579e-07,
+      "loss": 2.4182,
+      "step": 9341
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.308303535955503e-07,
+      "loss": 1.502,
+      "step": 9342
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.307906237584427e-07,
+      "loss": 1.3324,
+      "step": 9343
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.307508939213348e-07,
+      "loss": 1.2106,
+      "step": 9344
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.307111640842272e-07,
+      "loss": 1.79,
+      "step": 9345
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.306714342471195e-07,
+      "loss": 1.0089,
+      "step": 9346
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.306317044100119e-07,
+      "loss": 1.7605,
+      "step": 9347
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.305919745729042e-07,
+      "loss": 2.1603,
+      "step": 9348
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.305522447357966e-07,
+      "loss": 1.8084,
+      "step": 9349
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.305125148986888e-07,
+      "loss": 1.8265,
+      "step": 9350
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.304727850615812e-07,
+      "loss": 2.0197,
+      "step": 9351
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.304330552244736e-07,
+      "loss": 1.4618,
+      "step": 9352
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.303933253873659e-07,
+      "loss": 1.4804,
+      "step": 9353
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.303535955502583e-07,
+      "loss": 1.8456,
+      "step": 9354
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.303138657131506e-07,
+      "loss": 1.6333,
+      "step": 9355
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.302741358760429e-07,
+      "loss": 1.2134,
+      "step": 9356
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.302344060389351e-07,
+      "loss": 1.5592,
+      "step": 9357
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.301946762018275e-07,
+      "loss": 1.6445,
+      "step": 9358
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.301549463647199e-07,
+      "loss": 1.9391,
+      "step": 9359
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.301152165276122e-07,
+      "loss": 1.582,
+      "step": 9360
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.300754866905046e-07,
+      "loss": 1.7053,
+      "step": 9361
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.300357568533968e-07,
+      "loss": 1.7039,
+      "step": 9362
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.299960270162892e-07,
+      "loss": 1.9658,
+      "step": 9363
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.299562971791815e-07,
+      "loss": 1.5515,
+      "step": 9364
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.299165673420739e-07,
+      "loss": 1.9065,
+      "step": 9365
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.298768375049662e-07,
+      "loss": 1.6332,
+      "step": 9366
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.298371076678586e-07,
+      "loss": 1.6818,
+      "step": 9367
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.297973778307509e-07,
+      "loss": 1.6342,
+      "step": 9368
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.297576479936432e-07,
+      "loss": 1.5867,
+      "step": 9369
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.297179181565355e-07,
+      "loss": 1.9461,
+      "step": 9370
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.296781883194278e-07,
+      "loss": 1.876,
+      "step": 9371
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.296384584823202e-07,
+      "loss": 1.2964,
+      "step": 9372
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.295987286452125e-07,
+      "loss": 1.9354,
+      "step": 9373
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.295589988081048e-07,
+      "loss": 1.7856,
+      "step": 9374
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.295192689709972e-07,
+      "loss": 1.3586,
+      "step": 9375
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.294795391338895e-07,
+      "loss": 1.528,
+      "step": 9376
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.294398092967819e-07,
+      "loss": 1.4566,
+      "step": 9377
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.294000794596742e-07,
+      "loss": 1.3736,
+      "step": 9378
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.293603496225666e-07,
+      "loss": 1.6414,
+      "step": 9379
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.293206197854588e-07,
+      "loss": 1.7675,
+      "step": 9380
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.292808899483512e-07,
+      "loss": 1.4691,
+      "step": 9381
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.292411601112436e-07,
+      "loss": 1.9389,
+      "step": 9382
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.292014302741359e-07,
+      "loss": 1.7269,
+      "step": 9383
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.291617004370282e-07,
+      "loss": 1.644,
+      "step": 9384
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.291219705999205e-07,
+      "loss": 1.6299,
+      "step": 9385
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.290822407628128e-07,
+      "loss": 1.2826,
+      "step": 9386
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.290425109257051e-07,
+      "loss": 1.9744,
+      "step": 9387
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.290027810885975e-07,
+      "loss": 1.7153,
+      "step": 9388
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.289630512514898e-07,
+      "loss": 1.9362,
+      "step": 9389
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.289233214143822e-07,
+      "loss": 1.641,
+      "step": 9390
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.288835915772746e-07,
+      "loss": 1.6113,
+      "step": 9391
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.288438617401668e-07,
+      "loss": 1.8749,
+      "step": 9392
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.288041319030592e-07,
+      "loss": 1.8053,
+      "step": 9393
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.287644020659515e-07,
+      "loss": 1.6031,
+      "step": 9394
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.287246722288439e-07,
+      "loss": 1.8079,
+      "step": 9395
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.286849423917362e-07,
+      "loss": 1.4384,
+      "step": 9396
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.286452125546285e-07,
+      "loss": 1.9117,
+      "step": 9397
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.286054827175207e-07,
+      "loss": 1.8221,
+      "step": 9398
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.285657528804131e-07,
+      "loss": 1.7643,
+      "step": 9399
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.285260230433055e-07,
+      "loss": 1.4042,
+      "step": 9400
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.284862932061978e-07,
+      "loss": 1.2705,
+      "step": 9401
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.284465633690902e-07,
+      "loss": 1.5359,
+      "step": 9402
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.284068335319825e-07,
+      "loss": 1.4187,
+      "step": 9403
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.283671036948748e-07,
+      "loss": 1.7968,
+      "step": 9404
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.283273738577671e-07,
+      "loss": 2.1164,
+      "step": 9405
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.282876440206595e-07,
+      "loss": 1.8425,
+      "step": 9406
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.282479141835519e-07,
+      "loss": 1.9,
+      "step": 9407
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.282081843464442e-07,
+      "loss": 1.9142,
+      "step": 9408
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.281684545093366e-07,
+      "loss": 1.3684,
+      "step": 9409
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.281287246722287e-07,
+      "loss": 1.3353,
+      "step": 9410
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.280889948351211e-07,
+      "loss": 1.394,
+      "step": 9411
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.280492649980134e-07,
+      "loss": 1.6216,
+      "step": 9412
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.280095351609058e-07,
+      "loss": 1.6389,
+      "step": 9413
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.279698053237982e-07,
+      "loss": 1.3494,
+      "step": 9414
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.279300754866905e-07,
+      "loss": 1.8987,
+      "step": 9415
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.278903456495828e-07,
+      "loss": 2.2066,
+      "step": 9416
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.278506158124751e-07,
+      "loss": 1.8035,
+      "step": 9417
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.278108859753675e-07,
+      "loss": 1.8342,
+      "step": 9418
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.277711561382598e-07,
+      "loss": 1.8632,
+      "step": 9419
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.277314263011522e-07,
+      "loss": 1.8075,
+      "step": 9420
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.276916964640446e-07,
+      "loss": 1.3273,
+      "step": 9421
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.276519666269368e-07,
+      "loss": 1.5534,
+      "step": 9422
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.276122367898291e-07,
+      "loss": 1.6557,
+      "step": 9423
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.275725069527214e-07,
+      "loss": 2.0133,
+      "step": 9424
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.275327771156138e-07,
+      "loss": 1.7546,
+      "step": 9425
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.274930472785061e-07,
+      "loss": 1.6259,
+      "step": 9426
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.274533174413985e-07,
+      "loss": 1.2732,
+      "step": 9427
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.274135876042907e-07,
+      "loss": 1.5661,
+      "step": 9428
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.273738577671831e-07,
+      "loss": 1.7541,
+      "step": 9429
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.273341279300755e-07,
+      "loss": 1.382,
+      "step": 9430
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.272943980929678e-07,
+      "loss": 1.6232,
+      "step": 9431
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.272546682558602e-07,
+      "loss": 1.5953,
+      "step": 9432
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.272149384187525e-07,
+      "loss": 1.8574,
+      "step": 9433
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.271752085816448e-07,
+      "loss": 1.5256,
+      "step": 9434
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.271354787445371e-07,
+      "loss": 1.7719,
+      "step": 9435
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.270957489074294e-07,
+      "loss": 1.8186,
+      "step": 9436
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.270560190703218e-07,
+      "loss": 2.1092,
+      "step": 9437
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.270162892332141e-07,
+      "loss": 1.6765,
+      "step": 9438
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.269765593961065e-07,
+      "loss": 2.1378,
+      "step": 9439
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.269368295589987e-07,
+      "loss": 1.5239,
+      "step": 9440
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.268970997218911e-07,
+      "loss": 1.7891,
+      "step": 9441
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.268573698847834e-07,
+      "loss": 1.9366,
+      "step": 9442
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.268176400476758e-07,
+      "loss": 1.4315,
+      "step": 9443
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.267779102105681e-07,
+      "loss": 1.9395,
+      "step": 9444
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.267381803734605e-07,
+      "loss": 1.3879,
+      "step": 9445
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.266984505363528e-07,
+      "loss": 1.9627,
+      "step": 9446
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.266587206992451e-07,
+      "loss": 1.6126,
+      "step": 9447
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.266189908621375e-07,
+      "loss": 1.5823,
+      "step": 9448
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.265792610250298e-07,
+      "loss": 1.689,
+      "step": 9449
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.265395311879221e-07,
+      "loss": 1.5902,
+      "step": 9450
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.264998013508144e-07,
+      "loss": 1.473,
+      "step": 9451
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.264600715137067e-07,
+      "loss": 1.9002,
+      "step": 9452
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.26420341676599e-07,
+      "loss": 1.5595,
+      "step": 9453
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.263806118394914e-07,
+      "loss": 1.5127,
+      "step": 9454
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.263408820023838e-07,
+      "loss": 1.3957,
+      "step": 9455
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.263011521652761e-07,
+      "loss": 1.8035,
+      "step": 9456
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.262614223281685e-07,
+      "loss": 1.1354,
+      "step": 9457
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.262216924910607e-07,
+      "loss": 1.9684,
+      "step": 9458
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.261819626539531e-07,
+      "loss": 1.9239,
+      "step": 9459
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.261422328168455e-07,
+      "loss": 2.0115,
+      "step": 9460
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.261025029797378e-07,
+      "loss": 1.4239,
+      "step": 9461
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.260627731426302e-07,
+      "loss": 1.8082,
+      "step": 9462
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.260230433055224e-07,
+      "loss": 1.7475,
+      "step": 9463
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.259833134684147e-07,
+      "loss": 1.7518,
+      "step": 9464
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.25943583631307e-07,
+      "loss": 1.3868,
+      "step": 9465
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.259038537941994e-07,
+      "loss": 2.058,
+      "step": 9466
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.258641239570917e-07,
+      "loss": 1.8724,
+      "step": 9467
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.258243941199841e-07,
+      "loss": 1.7692,
+      "step": 9468
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.257846642828765e-07,
+      "loss": 1.6176,
+      "step": 9469
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.257449344457687e-07,
+      "loss": 1.6538,
+      "step": 9470
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.257052046086611e-07,
+      "loss": 1.5932,
+      "step": 9471
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.256654747715534e-07,
+      "loss": 1.6255,
+      "step": 9472
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.256257449344458e-07,
+      "loss": 1.6406,
+      "step": 9473
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.255860150973381e-07,
+      "loss": 1.6968,
+      "step": 9474
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.255462852602305e-07,
+      "loss": 1.6583,
+      "step": 9475
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.255065554231226e-07,
+      "loss": 1.7661,
+      "step": 9476
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.25466825586015e-07,
+      "loss": 1.852,
+      "step": 9477
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.254270957489074e-07,
+      "loss": 1.4213,
+      "step": 9478
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.253873659117997e-07,
+      "loss": 1.4735,
+      "step": 9479
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.253476360746921e-07,
+      "loss": 1.3068,
+      "step": 9480
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.253079062375844e-07,
+      "loss": 1.8438,
+      "step": 9481
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.252681764004767e-07,
+      "loss": 1.6311,
+      "step": 9482
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.25228446563369e-07,
+      "loss": 1.3508,
+      "step": 9483
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.251887167262614e-07,
+      "loss": 1.4878,
+      "step": 9484
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.251489868891538e-07,
+      "loss": 1.6923,
+      "step": 9485
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.251092570520461e-07,
+      "loss": 1.8931,
+      "step": 9486
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.250695272149385e-07,
+      "loss": 1.869,
+      "step": 9487
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.250297973778307e-07,
+      "loss": 1.8491,
+      "step": 9488
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.24990067540723e-07,
+      "loss": 2.1066,
+      "step": 9489
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.249503377036153e-07,
+      "loss": 1.7682,
+      "step": 9490
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.249106078665077e-07,
+      "loss": 1.9903,
+      "step": 9491
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.248708780294001e-07,
+      "loss": 1.444,
+      "step": 9492
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.248311481922924e-07,
+      "loss": 1.3108,
+      "step": 9493
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.247914183551847e-07,
+      "loss": 1.5354,
+      "step": 9494
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.24751688518077e-07,
+      "loss": 1.7162,
+      "step": 9495
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.247119586809694e-07,
+      "loss": 1.4976,
+      "step": 9496
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.246722288438617e-07,
+      "loss": 1.3602,
+      "step": 9497
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.246324990067541e-07,
+      "loss": 1.8442,
+      "step": 9498
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.245927691696465e-07,
+      "loss": 1.8554,
+      "step": 9499
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.245530393325387e-07,
+      "loss": 1.8576,
+      "step": 9500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.245133094954311e-07,
+      "loss": 1.7601,
+      "step": 9501
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.244735796583233e-07,
+      "loss": 1.5524,
+      "step": 9502
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.244338498212157e-07,
+      "loss": 1.7753,
+      "step": 9503
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.24394119984108e-07,
+      "loss": 1.4948,
+      "step": 9504
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.243543901470004e-07,
+      "loss": 1.6923,
+      "step": 9505
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.243146603098926e-07,
+      "loss": 1.687,
+      "step": 9506
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.24274930472785e-07,
+      "loss": 1.5162,
+      "step": 9507
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.242352006356774e-07,
+      "loss": 1.4997,
+      "step": 9508
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.241954707985697e-07,
+      "loss": 1.6587,
+      "step": 9509
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.241557409614621e-07,
+      "loss": 1.6812,
+      "step": 9510
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.241160111243544e-07,
+      "loss": 2.0201,
+      "step": 9511
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.240762812872467e-07,
+      "loss": 1.6997,
+      "step": 9512
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.24036551450139e-07,
+      "loss": 1.6947,
+      "step": 9513
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.239968216130314e-07,
+      "loss": 1.439,
+      "step": 9514
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.239570917759238e-07,
+      "loss": 1.3579,
+      "step": 9515
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.23917361938816e-07,
+      "loss": 1.4945,
+      "step": 9516
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.238776321017084e-07,
+      "loss": 2.0651,
+      "step": 9517
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.238379022646006e-07,
+      "loss": 1.4381,
+      "step": 9518
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.23798172427493e-07,
+      "loss": 1.5931,
+      "step": 9519
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.237584425903853e-07,
+      "loss": 1.5778,
+      "step": 9520
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.237187127532777e-07,
+      "loss": 1.796,
+      "step": 9521
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.2367898291617e-07,
+      "loss": 1.4606,
+      "step": 9522
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.236392530790624e-07,
+      "loss": 1.7541,
+      "step": 9523
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.235995232419547e-07,
+      "loss": 1.3651,
+      "step": 9524
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.23559793404847e-07,
+      "loss": 1.7029,
+      "step": 9525
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.235200635677394e-07,
+      "loss": 1.5146,
+      "step": 9526
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.234803337306317e-07,
+      "loss": 1.3811,
+      "step": 9527
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.234406038935241e-07,
+      "loss": 2.0688,
+      "step": 9528
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.234008740564163e-07,
+      "loss": 1.282,
+      "step": 9529
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.233611442193086e-07,
+      "loss": 1.2816,
+      "step": 9530
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.23321414382201e-07,
+      "loss": 1.8985,
+      "step": 9531
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.232816845450933e-07,
+      "loss": 1.6904,
+      "step": 9532
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.232419547079857e-07,
+      "loss": 1.6513,
+      "step": 9533
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.23202224870878e-07,
+      "loss": 1.8369,
+      "step": 9534
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.231624950337704e-07,
+      "loss": 1.7371,
+      "step": 9535
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.231227651966626e-07,
+      "loss": 1.8925,
+      "step": 9536
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.23083035359555e-07,
+      "loss": 1.9056,
+      "step": 9537
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.230433055224474e-07,
+      "loss": 1.411,
+      "step": 9538
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.230035756853397e-07,
+      "loss": 1.7541,
+      "step": 9539
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.229638458482321e-07,
+      "loss": 1.5947,
+      "step": 9540
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.229241160111244e-07,
+      "loss": 1.6416,
+      "step": 9541
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.228843861740166e-07,
+      "loss": 1.7504,
+      "step": 9542
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.228446563369089e-07,
+      "loss": 1.2947,
+      "step": 9543
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.228049264998013e-07,
+      "loss": 1.9538,
+      "step": 9544
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.227651966626936e-07,
+      "loss": 1.1374,
+      "step": 9545
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.22725466825586e-07,
+      "loss": 1.8209,
+      "step": 9546
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.226857369884784e-07,
+      "loss": 1.2342,
+      "step": 9547
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.226460071513706e-07,
+      "loss": 1.7381,
+      "step": 9548
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.22606277314263e-07,
+      "loss": 1.5579,
+      "step": 9549
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.225665474771553e-07,
+      "loss": 1.6804,
+      "step": 9550
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.225268176400477e-07,
+      "loss": 1.6196,
+      "step": 9551
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.2248708780294e-07,
+      "loss": 1.8638,
+      "step": 9552
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.224473579658324e-07,
+      "loss": 2.0824,
+      "step": 9553
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.224076281287247e-07,
+      "loss": 1.7146,
+      "step": 9554
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.223678982916169e-07,
+      "loss": 1.5037,
+      "step": 9555
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.223281684545093e-07,
+      "loss": 1.6544,
+      "step": 9556
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.222884386174016e-07,
+      "loss": 1.6072,
+      "step": 9557
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.22248708780294e-07,
+      "loss": 1.9973,
+      "step": 9558
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.222089789431863e-07,
+      "loss": 1.5228,
+      "step": 9559
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.221692491060786e-07,
+      "loss": 1.9297,
+      "step": 9560
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.221295192689709e-07,
+      "loss": 2.2041,
+      "step": 9561
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.220897894318633e-07,
+      "loss": 1.7468,
+      "step": 9562
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.220500595947557e-07,
+      "loss": 1.6887,
+      "step": 9563
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.22010329757648e-07,
+      "loss": 1.8199,
+      "step": 9564
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.219705999205404e-07,
+      "loss": 1.5531,
+      "step": 9565
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.219308700834326e-07,
+      "loss": 1.3629,
+      "step": 9566
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.21891140246325e-07,
+      "loss": 2.005,
+      "step": 9567
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.218514104092172e-07,
+      "loss": 2.2163,
+      "step": 9568
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.218116805721096e-07,
+      "loss": 1.8638,
+      "step": 9569
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.21771950735002e-07,
+      "loss": 1.1889,
+      "step": 9570
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.217322208978943e-07,
+      "loss": 1.8875,
+      "step": 9571
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.216924910607866e-07,
+      "loss": 1.6716,
+      "step": 9572
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.216527612236789e-07,
+      "loss": 1.457,
+      "step": 9573
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.216130313865713e-07,
+      "loss": 1.5945,
+      "step": 9574
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.215733015494636e-07,
+      "loss": 1.7206,
+      "step": 9575
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.21533571712356e-07,
+      "loss": 1.31,
+      "step": 9576
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.214938418752484e-07,
+      "loss": 1.6019,
+      "step": 9577
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.214541120381406e-07,
+      "loss": 1.7148,
+      "step": 9578
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.21414382201033e-07,
+      "loss": 1.5492,
+      "step": 9579
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.213746523639253e-07,
+      "loss": 1.7364,
+      "step": 9580
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.213349225268177e-07,
+      "loss": 1.5097,
+      "step": 9581
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.212951926897099e-07,
+      "loss": 1.777,
+      "step": 9582
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.212554628526023e-07,
+      "loss": 1.512,
+      "step": 9583
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.212157330154945e-07,
+      "loss": 1.6207,
+      "step": 9584
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.211760031783869e-07,
+      "loss": 1.344,
+      "step": 9585
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.211362733412793e-07,
+      "loss": 2.2569,
+      "step": 9586
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.210965435041716e-07,
+      "loss": 1.5193,
+      "step": 9587
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.21056813667064e-07,
+      "loss": 1.777,
+      "step": 9588
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.210170838299563e-07,
+      "loss": 1.4808,
+      "step": 9589
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.209773539928486e-07,
+      "loss": 1.6552,
+      "step": 9590
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.209376241557409e-07,
+      "loss": 1.3428,
+      "step": 9591
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.208978943186333e-07,
+      "loss": 1.53,
+      "step": 9592
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.208581644815257e-07,
+      "loss": 1.5915,
+      "step": 9593
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.20818434644418e-07,
+      "loss": 1.3629,
+      "step": 9594
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.207787048073103e-07,
+      "loss": 1.5733,
+      "step": 9595
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.207389749702025e-07,
+      "loss": 1.4596,
+      "step": 9596
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.206992451330949e-07,
+      "loss": 1.6503,
+      "step": 9597
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.206595152959872e-07,
+      "loss": 1.5987,
+      "step": 9598
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.206197854588796e-07,
+      "loss": 1.8872,
+      "step": 9599
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.20580055621772e-07,
+      "loss": 1.1115,
+      "step": 9600
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.205403257846643e-07,
+      "loss": 1.9376,
+      "step": 9601
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.205005959475566e-07,
+      "loss": 1.4658,
+      "step": 9602
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.204608661104489e-07,
+      "loss": 1.4385,
+      "step": 9603
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.204211362733413e-07,
+      "loss": 1.9911,
+      "step": 9604
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.203814064362336e-07,
+      "loss": 1.6079,
+      "step": 9605
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.20341676599126e-07,
+      "loss": 1.5615,
+      "step": 9606
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.203019467620183e-07,
+      "loss": 1.3094,
+      "step": 9607
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.202622169249105e-07,
+      "loss": 1.5004,
+      "step": 9608
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.202224870878029e-07,
+      "loss": 1.8705,
+      "step": 9609
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.201827572506952e-07,
+      "loss": 1.5322,
+      "step": 9610
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.201430274135876e-07,
+      "loss": 1.5302,
+      "step": 9611
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.201032975764799e-07,
+      "loss": 1.5607,
+      "step": 9612
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.200635677393723e-07,
+      "loss": 2.2157,
+      "step": 9613
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.200238379022645e-07,
+      "loss": 1.8328,
+      "step": 9614
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.199841080651569e-07,
+      "loss": 1.8052,
+      "step": 9615
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.199443782280493e-07,
+      "loss": 1.3979,
+      "step": 9616
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.199046483909416e-07,
+      "loss": 1.9339,
+      "step": 9617
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.19864918553834e-07,
+      "loss": 1.5512,
+      "step": 9618
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.198251887167263e-07,
+      "loss": 1.8145,
+      "step": 9619
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.197854588796186e-07,
+      "loss": 1.6819,
+      "step": 9620
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.197457290425108e-07,
+      "loss": 1.62,
+      "step": 9621
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.197059992054032e-07,
+      "loss": 1.2962,
+      "step": 9622
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.196662693682955e-07,
+      "loss": 1.6935,
+      "step": 9623
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.196265395311879e-07,
+      "loss": 1.6672,
+      "step": 9624
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.195868096940803e-07,
+      "loss": 1.9101,
+      "step": 9625
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.195470798569725e-07,
+      "loss": 1.4792,
+      "step": 9626
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.195073500198649e-07,
+      "loss": 1.689,
+      "step": 9627
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.194676201827572e-07,
+      "loss": 2.2985,
+      "step": 9628
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.194278903456496e-07,
+      "loss": 1.6157,
+      "step": 9629
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.193881605085419e-07,
+      "loss": 1.4837,
+      "step": 9630
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.193484306714343e-07,
+      "loss": 1.9973,
+      "step": 9631
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.193087008343266e-07,
+      "loss": 1.7882,
+      "step": 9632
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.192689709972189e-07,
+      "loss": 1.3202,
+      "step": 9633
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.192292411601112e-07,
+      "loss": 1.7915,
+      "step": 9634
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.191895113230035e-07,
+      "loss": 1.7823,
+      "step": 9635
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.191497814858959e-07,
+      "loss": 1.5669,
+      "step": 9636
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.191100516487882e-07,
+      "loss": 1.7769,
+      "step": 9637
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.190703218116805e-07,
+      "loss": 1.5005,
+      "step": 9638
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.190305919745728e-07,
+      "loss": 1.784,
+      "step": 9639
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.189908621374652e-07,
+      "loss": 1.8772,
+      "step": 9640
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.189511323003576e-07,
+      "loss": 1.7351,
+      "step": 9641
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.189114024632499e-07,
+      "loss": 1.7212,
+      "step": 9642
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.188716726261422e-07,
+      "loss": 1.3332,
+      "step": 9643
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.188319427890345e-07,
+      "loss": 1.4007,
+      "step": 9644
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.187922129519269e-07,
+      "loss": 1.8078,
+      "step": 9645
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.187524831148192e-07,
+      "loss": 1.6167,
+      "step": 9646
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.187127532777116e-07,
+      "loss": 1.5494,
+      "step": 9647
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.186730234406039e-07,
+      "loss": 1.3759,
+      "step": 9648
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.186332936034961e-07,
+      "loss": 1.6124,
+      "step": 9649
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.185935637663885e-07,
+      "loss": 1.1806,
+      "step": 9650
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.185538339292808e-07,
+      "loss": 1.6309,
+      "step": 9651
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.185141040921732e-07,
+      "loss": 1.5744,
+      "step": 9652
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.184743742550655e-07,
+      "loss": 1.2502,
+      "step": 9653
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.184346444179579e-07,
+      "loss": 1.6953,
+      "step": 9654
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.183949145808501e-07,
+      "loss": 1.7643,
+      "step": 9655
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.183551847437425e-07,
+      "loss": 1.7269,
+      "step": 9656
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.183154549066349e-07,
+      "loss": 1.6787,
+      "step": 9657
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.182757250695272e-07,
+      "loss": 1.7454,
+      "step": 9658
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.182359952324196e-07,
+      "loss": 1.7443,
+      "step": 9659
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.181962653953119e-07,
+      "loss": 1.3616,
+      "step": 9660
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.181565355582041e-07,
+      "loss": 1.5317,
+      "step": 9661
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.181168057210964e-07,
+      "loss": 1.8517,
+      "step": 9662
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.180770758839888e-07,
+      "loss": 1.3589,
+      "step": 9663
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.180373460468812e-07,
+      "loss": 1.5543,
+      "step": 9664
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.179976162097735e-07,
+      "loss": 1.615,
+      "step": 9665
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.179578863726659e-07,
+      "loss": 1.8348,
+      "step": 9666
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.179181565355581e-07,
+      "loss": 1.4139,
+      "step": 9667
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.178784266984505e-07,
+      "loss": 2.2369,
+      "step": 9668
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.178386968613428e-07,
+      "loss": 1.3336,
+      "step": 9669
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.177989670242352e-07,
+      "loss": 1.5598,
+      "step": 9670
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.177592371871276e-07,
+      "loss": 1.8414,
+      "step": 9671
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.177195073500199e-07,
+      "loss": 1.531,
+      "step": 9672
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.176797775129122e-07,
+      "loss": 1.7747,
+      "step": 9673
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.176400476758044e-07,
+      "loss": 1.8756,
+      "step": 9674
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.176003178386968e-07,
+      "loss": 1.6329,
+      "step": 9675
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.175605880015891e-07,
+      "loss": 1.534,
+      "step": 9676
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.175208581644815e-07,
+      "loss": 1.4572,
+      "step": 9677
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.174811283273738e-07,
+      "loss": 1.8751,
+      "step": 9678
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.174413984902661e-07,
+      "loss": 1.0986,
+      "step": 9679
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.174016686531585e-07,
+      "loss": 1.5174,
+      "step": 9680
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.173619388160508e-07,
+      "loss": 1.7054,
+      "step": 9681
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.173222089789432e-07,
+      "loss": 1.3016,
+      "step": 9682
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.172824791418355e-07,
+      "loss": 1.87,
+      "step": 9683
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.172427493047279e-07,
+      "loss": 1.7135,
+      "step": 9684
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.172030194676201e-07,
+      "loss": 1.402,
+      "step": 9685
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.171632896305125e-07,
+      "loss": 1.6631,
+      "step": 9686
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.171235597934048e-07,
+      "loss": 1.7761,
+      "step": 9687
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.170838299562971e-07,
+      "loss": 1.4771,
+      "step": 9688
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.170441001191895e-07,
+      "loss": 1.8771,
+      "step": 9689
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.170043702820818e-07,
+      "loss": 1.6228,
+      "step": 9690
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.169646404449741e-07,
+      "loss": 1.7759,
+      "step": 9691
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.169249106078664e-07,
+      "loss": 1.1423,
+      "step": 9692
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.168851807707588e-07,
+      "loss": 1.5122,
+      "step": 9693
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.168454509336512e-07,
+      "loss": 1.7674,
+      "step": 9694
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.168057210965435e-07,
+      "loss": 1.4449,
+      "step": 9695
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.167659912594359e-07,
+      "loss": 1.9214,
+      "step": 9696
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.167262614223281e-07,
+      "loss": 1.7478,
+      "step": 9697
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.166865315852205e-07,
+      "loss": 1.7812,
+      "step": 9698
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.166468017481128e-07,
+      "loss": 1.7392,
+      "step": 9699
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.166070719110051e-07,
+      "loss": 1.6566,
+      "step": 9700
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.165673420738974e-07,
+      "loss": 1.5565,
+      "step": 9701
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.165276122367898e-07,
+      "loss": 1.6346,
+      "step": 9702
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.164878823996821e-07,
+      "loss": 1.9723,
+      "step": 9703
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.164481525625744e-07,
+      "loss": 1.7198,
+      "step": 9704
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.164084227254668e-07,
+      "loss": 1.8038,
+      "step": 9705
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.163686928883591e-07,
+      "loss": 1.4225,
+      "step": 9706
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.163289630512515e-07,
+      "loss": 1.5208,
+      "step": 9707
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.162892332141438e-07,
+      "loss": 1.6844,
+      "step": 9708
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.162495033770361e-07,
+      "loss": 1.1627,
+      "step": 9709
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.162097735399285e-07,
+      "loss": 1.779,
+      "step": 9710
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.161700437028208e-07,
+      "loss": 1.6712,
+      "step": 9711
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.161303138657132e-07,
+      "loss": 1.2276,
+      "step": 9712
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.160905840286055e-07,
+      "loss": 1.6582,
+      "step": 9713
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.160508541914978e-07,
+      "loss": 1.6238,
+      "step": 9714
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.1601112435439e-07,
+      "loss": 1.8111,
+      "step": 9715
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.159713945172824e-07,
+      "loss": 1.779,
+      "step": 9716
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.159316646801747e-07,
+      "loss": 1.4963,
+      "step": 9717
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.158919348430671e-07,
+      "loss": 1.8586,
+      "step": 9718
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.158522050059595e-07,
+      "loss": 2.342,
+      "step": 9719
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.158124751688518e-07,
+      "loss": 1.7905,
+      "step": 9720
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.157727453317441e-07,
+      "loss": 1.5691,
+      "step": 9721
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.157330154946364e-07,
+      "loss": 1.2312,
+      "step": 9722
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.156932856575288e-07,
+      "loss": 1.6712,
+      "step": 9723
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.156535558204211e-07,
+      "loss": 1.3439,
+      "step": 9724
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.156138259833135e-07,
+      "loss": 1.6651,
+      "step": 9725
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.155740961462059e-07,
+      "loss": 1.813,
+      "step": 9726
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.15534366309098e-07,
+      "loss": 1.541,
+      "step": 9727
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.154946364719904e-07,
+      "loss": 1.6607,
+      "step": 9728
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.154549066348827e-07,
+      "loss": 1.2258,
+      "step": 9729
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.154151767977751e-07,
+      "loss": 1.8881,
+      "step": 9730
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.153754469606674e-07,
+      "loss": 1.5397,
+      "step": 9731
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.153357171235598e-07,
+      "loss": 1.6639,
+      "step": 9732
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.15295987286452e-07,
+      "loss": 1.8193,
+      "step": 9733
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.152562574493444e-07,
+      "loss": 1.6214,
+      "step": 9734
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.152165276122368e-07,
+      "loss": 1.3366,
+      "step": 9735
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.151767977751291e-07,
+      "loss": 1.4151,
+      "step": 9736
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.151370679380215e-07,
+      "loss": 1.8109,
+      "step": 9737
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.150973381009138e-07,
+      "loss": 1.6942,
+      "step": 9738
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.150576082638061e-07,
+      "loss": 1.7953,
+      "step": 9739
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.150178784266983e-07,
+      "loss": 1.4907,
+      "step": 9740
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.149781485895907e-07,
+      "loss": 1.2578,
+      "step": 9741
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.149384187524831e-07,
+      "loss": 1.7368,
+      "step": 9742
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.148986889153754e-07,
+      "loss": 1.5411,
+      "step": 9743
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.148589590782678e-07,
+      "loss": 1.6867,
+      "step": 9744
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.1481922924116e-07,
+      "loss": 1.7057,
+      "step": 9745
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.147794994040524e-07,
+      "loss": 1.6594,
+      "step": 9746
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.147397695669447e-07,
+      "loss": 1.5762,
+      "step": 9747
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.147000397298371e-07,
+      "loss": 1.6648,
+      "step": 9748
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.146603098927295e-07,
+      "loss": 1.7353,
+      "step": 9749
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.146205800556218e-07,
+      "loss": 1.5564,
+      "step": 9750
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.145808502185141e-07,
+      "loss": 1.7252,
+      "step": 9751
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.145411203814064e-07,
+      "loss": 1.5606,
+      "step": 9752
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.145013905442987e-07,
+      "loss": 1.6274,
+      "step": 9753
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.14461660707191e-07,
+      "loss": 1.5511,
+      "step": 9754
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.144219308700834e-07,
+      "loss": 1.7744,
+      "step": 9755
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.143822010329758e-07,
+      "loss": 1.3214,
+      "step": 9756
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.14342471195868e-07,
+      "loss": 1.936,
+      "step": 9757
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.143027413587604e-07,
+      "loss": 1.9175,
+      "step": 9758
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.142630115216527e-07,
+      "loss": 1.458,
+      "step": 9759
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.142232816845451e-07,
+      "loss": 1.4418,
+      "step": 9760
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.141835518474374e-07,
+      "loss": 1.5148,
+      "step": 9761
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.141438220103298e-07,
+      "loss": 1.756,
+      "step": 9762
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.14104092173222e-07,
+      "loss": 1.4792,
+      "step": 9763
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.140643623361144e-07,
+      "loss": 1.6548,
+      "step": 9764
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.140246324990068e-07,
+      "loss": 1.9448,
+      "step": 9765
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.13984902661899e-07,
+      "loss": 1.4994,
+      "step": 9766
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.139451728247914e-07,
+      "loss": 1.5298,
+      "step": 9767
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.139054429876837e-07,
+      "loss": 1.312,
+      "step": 9768
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.13865713150576e-07,
+      "loss": 1.7625,
+      "step": 9769
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.138259833134683e-07,
+      "loss": 1.565,
+      "step": 9770
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.137862534763607e-07,
+      "loss": 1.8858,
+      "step": 9771
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.13746523639253e-07,
+      "loss": 1.5693,
+      "step": 9772
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.137067938021454e-07,
+      "loss": 1.4088,
+      "step": 9773
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.136670639650378e-07,
+      "loss": 1.908,
+      "step": 9774
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.1362733412793e-07,
+      "loss": 1.9913,
+      "step": 9775
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.135876042908224e-07,
+      "loss": 1.4481,
+      "step": 9776
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.135478744537147e-07,
+      "loss": 2.0594,
+      "step": 9777
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.135081446166071e-07,
+      "loss": 2.1229,
+      "step": 9778
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.134684147794995e-07,
+      "loss": 1.4908,
+      "step": 9779
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.134286849423917e-07,
+      "loss": 1.7104,
+      "step": 9780
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.13388955105284e-07,
+      "loss": 2.0734,
+      "step": 9781
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.133492252681763e-07,
+      "loss": 1.9796,
+      "step": 9782
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.133094954310687e-07,
+      "loss": 1.9258,
+      "step": 9783
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.13269765593961e-07,
+      "loss": 1.6614,
+      "step": 9784
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.132300357568534e-07,
+      "loss": 1.9835,
+      "step": 9785
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.131903059197457e-07,
+      "loss": 1.5898,
+      "step": 9786
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.13150576082638e-07,
+      "loss": 1.582,
+      "step": 9787
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.131108462455304e-07,
+      "loss": 1.6198,
+      "step": 9788
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.130711164084227e-07,
+      "loss": 1.6358,
+      "step": 9789
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.130313865713151e-07,
+      "loss": 1.0775,
+      "step": 9790
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.129916567342074e-07,
+      "loss": 1.1794,
+      "step": 9791
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.129519268970998e-07,
+      "loss": 1.9891,
+      "step": 9792
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.129121970599919e-07,
+      "loss": 1.7606,
+      "step": 9793
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.128724672228843e-07,
+      "loss": 1.5407,
+      "step": 9794
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.128327373857766e-07,
+      "loss": 1.7248,
+      "step": 9795
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.12793007548669e-07,
+      "loss": 1.8906,
+      "step": 9796
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.127532777115614e-07,
+      "loss": 1.9156,
+      "step": 9797
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.127135478744537e-07,
+      "loss": 1.6824,
+      "step": 9798
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.12673818037346e-07,
+      "loss": 2.0562,
+      "step": 9799
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.126340882002383e-07,
+      "loss": 1.7148,
+      "step": 9800
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.125943583631307e-07,
+      "loss": 1.8443,
+      "step": 9801
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.12554628526023e-07,
+      "loss": 1.7636,
+      "step": 9802
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.125148986889154e-07,
+      "loss": 1.2786,
+      "step": 9803
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.124751688518078e-07,
+      "loss": 1.603,
+      "step": 9804
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.124354390147e-07,
+      "loss": 1.2607,
+      "step": 9805
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.123957091775923e-07,
+      "loss": 1.6241,
+      "step": 9806
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.123559793404846e-07,
+      "loss": 1.6585,
+      "step": 9807
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.12316249503377e-07,
+      "loss": 1.7467,
+      "step": 9808
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.122765196662693e-07,
+      "loss": 1.6462,
+      "step": 9809
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.122367898291617e-07,
+      "loss": 1.8582,
+      "step": 9810
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.12197059992054e-07,
+      "loss": 1.5713,
+      "step": 9811
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.121573301549463e-07,
+      "loss": 1.7271,
+      "step": 9812
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.121176003178387e-07,
+      "loss": 1.3255,
+      "step": 9813
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.12077870480731e-07,
+      "loss": 1.5175,
+      "step": 9814
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.120381406436234e-07,
+      "loss": 1.5651,
+      "step": 9815
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.119984108065157e-07,
+      "loss": 1.4526,
+      "step": 9816
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.11958680969408e-07,
+      "loss": 1.8862,
+      "step": 9817
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.119189511323003e-07,
+      "loss": 1.4269,
+      "step": 9818
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.118792212951926e-07,
+      "loss": 1.5906,
+      "step": 9819
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.11839491458085e-07,
+      "loss": 1.7737,
+      "step": 9820
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.117997616209773e-07,
+      "loss": 1.5806,
+      "step": 9821
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.117600317838697e-07,
+      "loss": 1.557,
+      "step": 9822
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.117203019467619e-07,
+      "loss": 1.9368,
+      "step": 9823
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.116805721096543e-07,
+      "loss": 1.2971,
+      "step": 9824
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.116408422725466e-07,
+      "loss": 1.4706,
+      "step": 9825
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.11601112435439e-07,
+      "loss": 1.7799,
+      "step": 9826
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.115613825983314e-07,
+      "loss": 1.5753,
+      "step": 9827
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.115216527612237e-07,
+      "loss": 2.0257,
+      "step": 9828
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.11481922924116e-07,
+      "loss": 2.1535,
+      "step": 9829
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.114421930870083e-07,
+      "loss": 1.7731,
+      "step": 9830
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.114024632499007e-07,
+      "loss": 1.269,
+      "step": 9831
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.113627334127929e-07,
+      "loss": 2.1158,
+      "step": 9832
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.113230035756853e-07,
+      "loss": 1.6182,
+      "step": 9833
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.112832737385777e-07,
+      "loss": 1.4573,
+      "step": 9834
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.112435439014699e-07,
+      "loss": 1.5788,
+      "step": 9835
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.112038140643623e-07,
+      "loss": 1.6748,
+      "step": 9836
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.111640842272546e-07,
+      "loss": 1.8846,
+      "step": 9837
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.11124354390147e-07,
+      "loss": 1.3963,
+      "step": 9838
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.110846245530393e-07,
+      "loss": 1.9712,
+      "step": 9839
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.110448947159317e-07,
+      "loss": 1.4296,
+      "step": 9840
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.110051648788239e-07,
+      "loss": 1.6278,
+      "step": 9841
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.109654350417163e-07,
+      "loss": 2.0804,
+      "step": 9842
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.109257052046087e-07,
+      "loss": 1.6746,
+      "step": 9843
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.10885975367501e-07,
+      "loss": 1.7333,
+      "step": 9844
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.108462455303934e-07,
+      "loss": 1.7797,
+      "step": 9845
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.108065156932856e-07,
+      "loss": 1.2843,
+      "step": 9846
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.107667858561779e-07,
+      "loss": 1.5373,
+      "step": 9847
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.107270560190702e-07,
+      "loss": 1.5989,
+      "step": 9848
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.106873261819626e-07,
+      "loss": 1.5468,
+      "step": 9849
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.10647596344855e-07,
+      "loss": 1.7131,
+      "step": 9850
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.106078665077473e-07,
+      "loss": 1.8003,
+      "step": 9851
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.105681366706397e-07,
+      "loss": 2.1093,
+      "step": 9852
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.105284068335319e-07,
+      "loss": 1.4672,
+      "step": 9853
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.104886769964243e-07,
+      "loss": 1.6545,
+      "step": 9854
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.104489471593166e-07,
+      "loss": 1.4828,
+      "step": 9855
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.10409217322209e-07,
+      "loss": 1.5912,
+      "step": 9856
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.103694874851014e-07,
+      "loss": 1.4845,
+      "step": 9857
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.103297576479937e-07,
+      "loss": 1.9873,
+      "step": 9858
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.102900278108859e-07,
+      "loss": 1.9094,
+      "step": 9859
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.102502979737782e-07,
+      "loss": 1.9732,
+      "step": 9860
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.102105681366706e-07,
+      "loss": 1.697,
+      "step": 9861
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.101708382995629e-07,
+      "loss": 1.6598,
+      "step": 9862
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.101311084624553e-07,
+      "loss": 1.3406,
+      "step": 9863
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.100913786253476e-07,
+      "loss": 1.6108,
+      "step": 9864
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.100516487882399e-07,
+      "loss": 1.2766,
+      "step": 9865
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.100119189511323e-07,
+      "loss": 2.1078,
+      "step": 9866
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.099721891140246e-07,
+      "loss": 1.4592,
+      "step": 9867
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.09932459276917e-07,
+      "loss": 2.1752,
+      "step": 9868
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.098927294398093e-07,
+      "loss": 1.6113,
+      "step": 9869
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.098529996027017e-07,
+      "loss": 1.5235,
+      "step": 9870
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.098132697655939e-07,
+      "loss": 1.5539,
+      "step": 9871
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.097735399284862e-07,
+      "loss": 1.9366,
+      "step": 9872
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.097338100913785e-07,
+      "loss": 1.6829,
+      "step": 9873
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.096940802542709e-07,
+      "loss": 1.6205,
+      "step": 9874
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.096543504171633e-07,
+      "loss": 1.4044,
+      "step": 9875
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.096146205800556e-07,
+      "loss": 1.2438,
+      "step": 9876
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.095748907429479e-07,
+      "loss": 1.6457,
+      "step": 9877
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.095351609058402e-07,
+      "loss": 1.6857,
+      "step": 9878
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.094954310687326e-07,
+      "loss": 1.2995,
+      "step": 9879
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.094557012316249e-07,
+      "loss": 1.597,
+      "step": 9880
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.094159713945173e-07,
+      "loss": 1.4586,
+      "step": 9881
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.093762415574097e-07,
+      "loss": 1.4311,
+      "step": 9882
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.093365117203019e-07,
+      "loss": 1.836,
+      "step": 9883
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.092967818831943e-07,
+      "loss": 1.9749,
+      "step": 9884
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.092570520460865e-07,
+      "loss": 1.9235,
+      "step": 9885
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.092173222089789e-07,
+      "loss": 1.4888,
+      "step": 9886
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.091775923718712e-07,
+      "loss": 1.6462,
+      "step": 9887
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.091378625347636e-07,
+      "loss": 1.577,
+      "step": 9888
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.090981326976558e-07,
+      "loss": 1.9372,
+      "step": 9889
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.090584028605482e-07,
+      "loss": 1.435,
+      "step": 9890
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.090186730234406e-07,
+      "loss": 1.263,
+      "step": 9891
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.089789431863329e-07,
+      "loss": 1.7845,
+      "step": 9892
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.089392133492253e-07,
+      "loss": 1.5888,
+      "step": 9893
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.088994835121176e-07,
+      "loss": 1.5675,
+      "step": 9894
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.088597536750099e-07,
+      "loss": 1.0205,
+      "step": 9895
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.088200238379022e-07,
+      "loss": 1.6839,
+      "step": 9896
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.087802940007946e-07,
+      "loss": 1.5299,
+      "step": 9897
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.087405641636869e-07,
+      "loss": 1.2451,
+      "step": 9898
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.087008343265792e-07,
+      "loss": 1.4432,
+      "step": 9899
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.086611044894716e-07,
+      "loss": 1.6542,
+      "step": 9900
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.086213746523638e-07,
+      "loss": 1.8922,
+      "step": 9901
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.085816448152562e-07,
+      "loss": 1.4609,
+      "step": 9902
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.085419149781485e-07,
+      "loss": 1.3252,
+      "step": 9903
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.085021851410409e-07,
+      "loss": 1.6499,
+      "step": 9904
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.084624553039333e-07,
+      "loss": 1.623,
+      "step": 9905
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.084227254668256e-07,
+      "loss": 1.6037,
+      "step": 9906
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.083829956297179e-07,
+      "loss": 1.7533,
+      "step": 9907
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.083432657926102e-07,
+      "loss": 1.6011,
+      "step": 9908
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.083035359555026e-07,
+      "loss": 1.6844,
+      "step": 9909
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.082638061183949e-07,
+      "loss": 1.3579,
+      "step": 9910
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.082240762812873e-07,
+      "loss": 1.6949,
+      "step": 9911
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.081843464441796e-07,
+      "loss": 1.5492,
+      "step": 9912
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.081446166070718e-07,
+      "loss": 1.65,
+      "step": 9913
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.081048867699642e-07,
+      "loss": 1.7667,
+      "step": 9914
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.080651569328565e-07,
+      "loss": 1.5878,
+      "step": 9915
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.080254270957489e-07,
+      "loss": 1.756,
+      "step": 9916
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.079856972586412e-07,
+      "loss": 1.5562,
+      "step": 9917
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.079459674215336e-07,
+      "loss": 1.5476,
+      "step": 9918
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.079062375844258e-07,
+      "loss": 1.2309,
+      "step": 9919
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.078665077473182e-07,
+      "loss": 1.4645,
+      "step": 9920
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.078267779102106e-07,
+      "loss": 1.5265,
+      "step": 9921
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.077870480731029e-07,
+      "loss": 1.8681,
+      "step": 9922
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.077473182359953e-07,
+      "loss": 1.6579,
+      "step": 9923
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.077075883988876e-07,
+      "loss": 1.6813,
+      "step": 9924
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.076678585617798e-07,
+      "loss": 1.4057,
+      "step": 9925
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.076281287246721e-07,
+      "loss": 1.4021,
+      "step": 9926
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.075883988875645e-07,
+      "loss": 1.7898,
+      "step": 9927
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.075486690504569e-07,
+      "loss": 1.433,
+      "step": 9928
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.075089392133492e-07,
+      "loss": 1.901,
+      "step": 9929
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.074692093762416e-07,
+      "loss": 1.7858,
+      "step": 9930
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.074294795391338e-07,
+      "loss": 1.7059,
+      "step": 9931
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.073897497020262e-07,
+      "loss": 1.9241,
+      "step": 9932
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.073500198649185e-07,
+      "loss": 1.3002,
+      "step": 9933
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.073102900278109e-07,
+      "loss": 1.5278,
+      "step": 9934
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.072705601907033e-07,
+      "loss": 1.5508,
+      "step": 9935
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.072308303535956e-07,
+      "loss": 1.9673,
+      "step": 9936
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.071911005164879e-07,
+      "loss": 1.2945,
+      "step": 9937
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.071513706793801e-07,
+      "loss": 1.6454,
+      "step": 9938
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.071116408422725e-07,
+      "loss": 1.6426,
+      "step": 9939
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.070719110051648e-07,
+      "loss": 2.378,
+      "step": 9940
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.070321811680572e-07,
+      "loss": 1.2369,
+      "step": 9941
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.069924513309495e-07,
+      "loss": 1.9551,
+      "step": 9942
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.069527214938418e-07,
+      "loss": 1.9243,
+      "step": 9943
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.069129916567342e-07,
+      "loss": 1.6133,
+      "step": 9944
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.068732618196265e-07,
+      "loss": 1.7052,
+      "step": 9945
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.068335319825189e-07,
+      "loss": 1.2231,
+      "step": 9946
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.067938021454112e-07,
+      "loss": 2.1476,
+      "step": 9947
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.067540723083036e-07,
+      "loss": 1.5841,
+      "step": 9948
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.067143424711958e-07,
+      "loss": 1.5478,
+      "step": 9949
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.066746126340882e-07,
+      "loss": 1.4661,
+      "step": 9950
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.066348827969804e-07,
+      "loss": 1.2547,
+      "step": 9951
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.065951529598728e-07,
+      "loss": 1.2484,
+      "step": 9952
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.065554231227652e-07,
+      "loss": 2.0029,
+      "step": 9953
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.065156932856575e-07,
+      "loss": 1.8047,
+      "step": 9954
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.064759634485498e-07,
+      "loss": 1.7576,
+      "step": 9955
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.064362336114421e-07,
+      "loss": 1.7617,
+      "step": 9956
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.063965037743345e-07,
+      "loss": 1.8979,
+      "step": 9957
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.063567739372268e-07,
+      "loss": 1.5265,
+      "step": 9958
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.063170441001192e-07,
+      "loss": 1.9987,
+      "step": 9959
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.062773142630116e-07,
+      "loss": 1.4675,
+      "step": 9960
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.062375844259038e-07,
+      "loss": 1.8459,
+      "step": 9961
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.061978545887962e-07,
+      "loss": 1.6434,
+      "step": 9962
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.061581247516885e-07,
+      "loss": 1.403,
+      "step": 9963
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.061183949145808e-07,
+      "loss": 1.9095,
+      "step": 9964
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.060786650774731e-07,
+      "loss": 1.6933,
+      "step": 9965
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.060389352403655e-07,
+      "loss": 1.7742,
+      "step": 9966
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.059992054032577e-07,
+      "loss": 1.7549,
+      "step": 9967
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.059594755661501e-07,
+      "loss": 1.7681,
+      "step": 9968
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.059197457290425e-07,
+      "loss": 1.743,
+      "step": 9969
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.058800158919348e-07,
+      "loss": 2.0151,
+      "step": 9970
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.058402860548272e-07,
+      "loss": 1.715,
+      "step": 9971
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.058005562177195e-07,
+      "loss": 1.6121,
+      "step": 9972
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.057608263806118e-07,
+      "loss": 1.5661,
+      "step": 9973
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.057210965435041e-07,
+      "loss": 1.2107,
+      "step": 9974
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.056813667063965e-07,
+      "loss": 1.3078,
+      "step": 9975
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.056416368692889e-07,
+      "loss": 1.8183,
+      "step": 9976
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.056019070321812e-07,
+      "loss": 1.9492,
+      "step": 9977
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.055621771950735e-07,
+      "loss": 1.5214,
+      "step": 9978
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.055224473579657e-07,
+      "loss": 1.4153,
+      "step": 9979
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.054827175208581e-07,
+      "loss": 2.1845,
+      "step": 9980
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.054429876837504e-07,
+      "loss": 1.6347,
+      "step": 9981
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.054032578466428e-07,
+      "loss": 1.5097,
+      "step": 9982
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.053635280095352e-07,
+      "loss": 2.0255,
+      "step": 9983
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.053237981724275e-07,
+      "loss": 1.8053,
+      "step": 9984
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.052840683353198e-07,
+      "loss": 1.5373,
+      "step": 9985
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.052443384982121e-07,
+      "loss": 1.3594,
+      "step": 9986
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.052046086611045e-07,
+      "loss": 1.3279,
+      "step": 9987
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.051648788239968e-07,
+      "loss": 1.3652,
+      "step": 9988
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.051251489868892e-07,
+      "loss": 1.6421,
+      "step": 9989
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.050854191497816e-07,
+      "loss": 1.5928,
+      "step": 9990
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.050456893126737e-07,
+      "loss": 1.6364,
+      "step": 9991
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.050059594755661e-07,
+      "loss": 1.5295,
+      "step": 9992
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.049662296384584e-07,
+      "loss": 1.574,
+      "step": 9993
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.049264998013508e-07,
+      "loss": 1.8897,
+      "step": 9994
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.048867699642431e-07,
+      "loss": 1.1075,
+      "step": 9995
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.048470401271355e-07,
+      "loss": 2.0381,
+      "step": 9996
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.048073102900277e-07,
+      "loss": 1.669,
+      "step": 9997
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.047675804529201e-07,
+      "loss": 1.4125,
+      "step": 9998
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.047278506158125e-07,
+      "loss": 2.3131,
+      "step": 9999
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.046881207787048e-07,
+      "loss": 1.483,
+      "step": 10000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.046483909415972e-07,
+      "loss": 1.811,
+      "step": 10001
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.046086611044895e-07,
+      "loss": 1.5105,
+      "step": 10002
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.045689312673817e-07,
+      "loss": 1.6794,
+      "step": 10003
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.04529201430274e-07,
+      "loss": 1.5904,
+      "step": 10004
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.044894715931664e-07,
+      "loss": 1.6688,
+      "step": 10005
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.044497417560588e-07,
+      "loss": 1.3777,
+      "step": 10006
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.044100119189511e-07,
+      "loss": 1.3849,
+      "step": 10007
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.043702820818435e-07,
+      "loss": 1.6719,
+      "step": 10008
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.043305522447357e-07,
+      "loss": 1.7285,
+      "step": 10009
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.042908224076281e-07,
+      "loss": 1.4371,
+      "step": 10010
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.042510925705204e-07,
+      "loss": 1.6873,
+      "step": 10011
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.042113627334128e-07,
+      "loss": 1.8137,
+      "step": 10012
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.041716328963052e-07,
+      "loss": 1.6126,
+      "step": 10013
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.041319030591975e-07,
+      "loss": 1.2497,
+      "step": 10014
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.040921732220898e-07,
+      "loss": 1.464,
+      "step": 10015
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.040524433849821e-07,
+      "loss": 1.6901,
+      "step": 10016
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.040127135478744e-07,
+      "loss": 1.8804,
+      "step": 10017
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.039729837107667e-07,
+      "loss": 1.7746,
+      "step": 10018
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.039332538736591e-07,
+      "loss": 1.2639,
+      "step": 10019
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.038935240365514e-07,
+      "loss": 2.1209,
+      "step": 10020
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.038537941994437e-07,
+      "loss": 1.725,
+      "step": 10021
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.038140643623361e-07,
+      "loss": 1.641,
+      "step": 10022
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.037743345252284e-07,
+      "loss": 1.694,
+      "step": 10023
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.037346046881208e-07,
+      "loss": 1.5414,
+      "step": 10024
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.036948748510131e-07,
+      "loss": 1.7925,
+      "step": 10025
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.036551450139055e-07,
+      "loss": 1.4334,
+      "step": 10026
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.036154151767977e-07,
+      "loss": 1.5839,
+      "step": 10027
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.035756853396901e-07,
+      "loss": 1.1163,
+      "step": 10028
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.035359555025825e-07,
+      "loss": 1.5774,
+      "step": 10029
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.034962256654747e-07,
+      "loss": 1.7208,
+      "step": 10030
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.034564958283671e-07,
+      "loss": 1.2603,
+      "step": 10031
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.034167659912594e-07,
+      "loss": 1.5783,
+      "step": 10032
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.033770361541517e-07,
+      "loss": 2.0437,
+      "step": 10033
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.03337306317044e-07,
+      "loss": 1.6715,
+      "step": 10034
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.032975764799364e-07,
+      "loss": 1.2655,
+      "step": 10035
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.032578466428287e-07,
+      "loss": 1.7078,
+      "step": 10036
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.032181168057211e-07,
+      "loss": 1.2646,
+      "step": 10037
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.031783869686135e-07,
+      "loss": 1.4589,
+      "step": 10038
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.031386571315057e-07,
+      "loss": 1.936,
+      "step": 10039
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.030989272943981e-07,
+      "loss": 1.9688,
+      "step": 10040
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.030591974572904e-07,
+      "loss": 1.5287,
+      "step": 10041
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.030194676201828e-07,
+      "loss": 1.8411,
+      "step": 10042
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.02979737783075e-07,
+      "loss": 1.9662,
+      "step": 10043
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.029400079459674e-07,
+      "loss": 1.644,
+      "step": 10044
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.029002781088597e-07,
+      "loss": 1.404,
+      "step": 10045
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.02860548271752e-07,
+      "loss": 1.4468,
+      "step": 10046
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.028208184346444e-07,
+      "loss": 1.3606,
+      "step": 10047
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.027810885975367e-07,
+      "loss": 1.0837,
+      "step": 10048
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.027413587604291e-07,
+      "loss": 1.9242,
+      "step": 10049
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.027016289233214e-07,
+      "loss": 1.3803,
+      "step": 10050
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.026618990862137e-07,
+      "loss": 1.571,
+      "step": 10051
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.02622169249106e-07,
+      "loss": 1.9948,
+      "step": 10052
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.025824394119984e-07,
+      "loss": 1.4911,
+      "step": 10053
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.025427095748908e-07,
+      "loss": 1.7837,
+      "step": 10054
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.025029797377831e-07,
+      "loss": 1.1038,
+      "step": 10055
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.024632499006755e-07,
+      "loss": 1.8504,
+      "step": 10056
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.024235200635676e-07,
+      "loss": 1.9935,
+      "step": 10057
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.0238379022646e-07,
+      "loss": 1.5731,
+      "step": 10058
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.023440603893523e-07,
+      "loss": 1.4241,
+      "step": 10059
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.023043305522447e-07,
+      "loss": 1.8551,
+      "step": 10060
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.022646007151371e-07,
+      "loss": 1.5002,
+      "step": 10061
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.022248708780294e-07,
+      "loss": 1.3324,
+      "step": 10062
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.021851410409217e-07,
+      "loss": 1.5156,
+      "step": 10063
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.02145411203814e-07,
+      "loss": 1.9003,
+      "step": 10064
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.021056813667064e-07,
+      "loss": 1.6568,
+      "step": 10065
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.020659515295987e-07,
+      "loss": 1.5811,
+      "step": 10066
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.020262216924911e-07,
+      "loss": 1.6128,
+      "step": 10067
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.019864918553835e-07,
+      "loss": 1.5739,
+      "step": 10068
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.019467620182756e-07,
+      "loss": 1.5159,
+      "step": 10069
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.01907032181168e-07,
+      "loss": 1.6403,
+      "step": 10070
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.018673023440603e-07,
+      "loss": 2.1146,
+      "step": 10071
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.018275725069527e-07,
+      "loss": 1.6276,
+      "step": 10072
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.01787842669845e-07,
+      "loss": 1.5836,
+      "step": 10073
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.017481128327374e-07,
+      "loss": 1.6925,
+      "step": 10074
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.017083829956296e-07,
+      "loss": 1.6802,
+      "step": 10075
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.01668653158522e-07,
+      "loss": 1.6644,
+      "step": 10076
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.016289233214144e-07,
+      "loss": 1.5172,
+      "step": 10077
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.015891934843067e-07,
+      "loss": 1.6617,
+      "step": 10078
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.015494636471991e-07,
+      "loss": 1.6916,
+      "step": 10079
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.015097338100914e-07,
+      "loss": 1.759,
+      "step": 10080
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.014700039729837e-07,
+      "loss": 1.5417,
+      "step": 10081
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.01430274135876e-07,
+      "loss": 1.7449,
+      "step": 10082
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.013905442987683e-07,
+      "loss": 1.7912,
+      "step": 10083
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.013508144616607e-07,
+      "loss": 1.7191,
+      "step": 10084
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.01311084624553e-07,
+      "loss": 1.4198,
+      "step": 10085
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.012713547874454e-07,
+      "loss": 1.576,
+      "step": 10086
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.012316249503376e-07,
+      "loss": 1.6801,
+      "step": 10087
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.0119189511323e-07,
+      "loss": 1.4559,
+      "step": 10088
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.3687549563838224,
+      "eval_loss": 1.73171865940094,
+      "eval_runtime": 6.4159,
+      "eval_samples_per_second": 393.085,
+      "eval_steps_per_second": 6.234,
+      "step": 10088
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.011521652761223e-07,
+      "loss": 1.5231,
+      "step": 10089
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.011124354390147e-07,
+      "loss": 1.957,
+      "step": 10090
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.01072705601907e-07,
+      "loss": 1.683,
+      "step": 10091
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.010329757647994e-07,
+      "loss": 1.5967,
+      "step": 10092
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.009932459276917e-07,
+      "loss": 1.2364,
+      "step": 10093
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.00953516090584e-07,
+      "loss": 1.4281,
+      "step": 10094
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.009137862534764e-07,
+      "loss": 1.4856,
+      "step": 10095
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.008740564163686e-07,
+      "loss": 1.6401,
+      "step": 10096
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.00834326579261e-07,
+      "loss": 1.4311,
+      "step": 10097
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.007945967421533e-07,
+      "loss": 1.597,
+      "step": 10098
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.007548669050456e-07,
+      "loss": 1.8021,
+      "step": 10099
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.00715137067938e-07,
+      "loss": 1.5582,
+      "step": 10100
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.006754072308303e-07,
+      "loss": 1.0641,
+      "step": 10101
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.006356773937227e-07,
+      "loss": 1.4146,
+      "step": 10102
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.00595947556615e-07,
+      "loss": 1.4186,
+      "step": 10103
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.005562177195074e-07,
+      "loss": 1.1837,
+      "step": 10104
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.005164878823996e-07,
+      "loss": 2.0027,
+      "step": 10105
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.00476758045292e-07,
+      "loss": 1.9214,
+      "step": 10106
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.004370282081844e-07,
+      "loss": 2.1616,
+      "step": 10107
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.003972983710767e-07,
+      "loss": 1.4602,
+      "step": 10108
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.00357568533969e-07,
+      "loss": 1.4084,
+      "step": 10109
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.003178386968613e-07,
+      "loss": 1.5956,
+      "step": 10110
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.002781088597536e-07,
+      "loss": 1.826,
+      "step": 10111
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.002383790226459e-07,
+      "loss": 1.9976,
+      "step": 10112
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.001986491855383e-07,
+      "loss": 1.7531,
+      "step": 10113
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.001589193484306e-07,
+      "loss": 1.6332,
+      "step": 10114
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.00119189511323e-07,
+      "loss": 1.7597,
+      "step": 10115
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.000794596742154e-07,
+      "loss": 2.3532,
+      "step": 10116
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.000397298371076e-07,
+      "loss": 1.2737,
+      "step": 10117
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6e-07,
+      "loss": 2.105,
+      "step": 10118
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.999602701628923e-07,
+      "loss": 1.7521,
+      "step": 10119
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.999205403257847e-07,
+      "loss": 1.907,
+      "step": 10120
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.99880810488677e-07,
+      "loss": 1.4673,
+      "step": 10121
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.998410806515694e-07,
+      "loss": 1.4664,
+      "step": 10122
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.998013508144616e-07,
+      "loss": 1.0685,
+      "step": 10123
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.997616209773539e-07,
+      "loss": 1.6761,
+      "step": 10124
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.997218911402463e-07,
+      "loss": 1.5112,
+      "step": 10125
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.996821613031386e-07,
+      "loss": 1.5151,
+      "step": 10126
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.99642431466031e-07,
+      "loss": 1.182,
+      "step": 10127
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.996027016289233e-07,
+      "loss": 1.6824,
+      "step": 10128
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.995629717918156e-07,
+      "loss": 1.7182,
+      "step": 10129
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.99523241954708e-07,
+      "loss": 1.4667,
+      "step": 10130
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.994835121176003e-07,
+      "loss": 1.482,
+      "step": 10131
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.994437822804927e-07,
+      "loss": 1.4115,
+      "step": 10132
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.99404052443385e-07,
+      "loss": 1.3186,
+      "step": 10133
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.993643226062774e-07,
+      "loss": 1.455,
+      "step": 10134
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.993245927691695e-07,
+      "loss": 1.6684,
+      "step": 10135
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.992848629320619e-07,
+      "loss": 1.6653,
+      "step": 10136
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.992451330949542e-07,
+      "loss": 1.7339,
+      "step": 10137
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.992054032578466e-07,
+      "loss": 1.8719,
+      "step": 10138
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.99165673420739e-07,
+      "loss": 1.6778,
+      "step": 10139
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.991259435836313e-07,
+      "loss": 1.6442,
+      "step": 10140
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.990862137465236e-07,
+      "loss": 1.7783,
+      "step": 10141
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.990464839094159e-07,
+      "loss": 1.8001,
+      "step": 10142
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.990067540723083e-07,
+      "loss": 1.6229,
+      "step": 10143
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.989670242352006e-07,
+      "loss": 1.421,
+      "step": 10144
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.98927294398093e-07,
+      "loss": 1.3114,
+      "step": 10145
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.988875645609854e-07,
+      "loss": 1.9117,
+      "step": 10146
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.988478347238776e-07,
+      "loss": 1.9962,
+      "step": 10147
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.9880810488677e-07,
+      "loss": 1.7217,
+      "step": 10148
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.987683750496622e-07,
+      "loss": 1.5042,
+      "step": 10149
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.987286452125546e-07,
+      "loss": 1.9795,
+      "step": 10150
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.986889153754469e-07,
+      "loss": 1.8151,
+      "step": 10151
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.986491855383393e-07,
+      "loss": 1.7088,
+      "step": 10152
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.986094557012315e-07,
+      "loss": 1.879,
+      "step": 10153
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.985697258641239e-07,
+      "loss": 1.5519,
+      "step": 10154
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.985299960270163e-07,
+      "loss": 1.6092,
+      "step": 10155
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.984902661899086e-07,
+      "loss": 1.8058,
+      "step": 10156
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.98450536352801e-07,
+      "loss": 1.6578,
+      "step": 10157
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.984108065156933e-07,
+      "loss": 1.3968,
+      "step": 10158
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.983710766785856e-07,
+      "loss": 1.7315,
+      "step": 10159
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.983313468414779e-07,
+      "loss": 1.3497,
+      "step": 10160
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.982916170043703e-07,
+      "loss": 1.4623,
+      "step": 10161
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.982518871672626e-07,
+      "loss": 1.8064,
+      "step": 10162
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.982121573301549e-07,
+      "loss": 1.2053,
+      "step": 10163
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.981724274930473e-07,
+      "loss": 1.6448,
+      "step": 10164
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.981326976559395e-07,
+      "loss": 2.2251,
+      "step": 10165
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.980929678188319e-07,
+      "loss": 1.8119,
+      "step": 10166
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.980532379817242e-07,
+      "loss": 1.4758,
+      "step": 10167
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.980135081446166e-07,
+      "loss": 1.7607,
+      "step": 10168
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.97973778307509e-07,
+      "loss": 1.5977,
+      "step": 10169
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.979340484704013e-07,
+      "loss": 1.6072,
+      "step": 10170
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.978943186332936e-07,
+      "loss": 1.2756,
+      "step": 10171
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.978545887961859e-07,
+      "loss": 1.6416,
+      "step": 10172
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.978148589590783e-07,
+      "loss": 1.5653,
+      "step": 10173
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.977751291219706e-07,
+      "loss": 1.5459,
+      "step": 10174
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.977353992848629e-07,
+      "loss": 1.3251,
+      "step": 10175
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.976956694477552e-07,
+      "loss": 1.9164,
+      "step": 10176
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.976559396106475e-07,
+      "loss": 1.0727,
+      "step": 10177
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.976162097735399e-07,
+      "loss": 1.3303,
+      "step": 10178
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.975764799364322e-07,
+      "loss": 1.8555,
+      "step": 10179
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.975367500993246e-07,
+      "loss": 1.7719,
+      "step": 10180
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.974970202622169e-07,
+      "loss": 1.5338,
+      "step": 10181
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.974572904251093e-07,
+      "loss": 1.4932,
+      "step": 10182
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.974175605880015e-07,
+      "loss": 1.8837,
+      "step": 10183
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.973778307508939e-07,
+      "loss": 1.6884,
+      "step": 10184
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.973381009137863e-07,
+      "loss": 2.1661,
+      "step": 10185
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.972983710766786e-07,
+      "loss": 1.3983,
+      "step": 10186
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.97258641239571e-07,
+      "loss": 1.5402,
+      "step": 10187
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.972189114024633e-07,
+      "loss": 1.6452,
+      "step": 10188
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.971791815653555e-07,
+      "loss": 1.6844,
+      "step": 10189
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.971394517282478e-07,
+      "loss": 1.869,
+      "step": 10190
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.970997218911402e-07,
+      "loss": 1.5472,
+      "step": 10191
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.970599920540325e-07,
+      "loss": 1.6274,
+      "step": 10192
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.970202622169249e-07,
+      "loss": 1.685,
+      "step": 10193
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.969805323798173e-07,
+      "loss": 1.5809,
+      "step": 10194
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.969408025427095e-07,
+      "loss": 1.6917,
+      "step": 10195
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.969010727056019e-07,
+      "loss": 1.8109,
+      "step": 10196
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.968613428684942e-07,
+      "loss": 1.6354,
+      "step": 10197
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.968216130313866e-07,
+      "loss": 1.7919,
+      "step": 10198
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.967818831942789e-07,
+      "loss": 1.6436,
+      "step": 10199
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.967421533571713e-07,
+      "loss": 1.725,
+      "step": 10200
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.967024235200635e-07,
+      "loss": 1.5562,
+      "step": 10201
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.966626936829558e-07,
+      "loss": 2.0155,
+      "step": 10202
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.966229638458482e-07,
+      "loss": 1.7638,
+      "step": 10203
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.965832340087405e-07,
+      "loss": 1.4805,
+      "step": 10204
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.965435041716329e-07,
+      "loss": 1.8281,
+      "step": 10205
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.965037743345252e-07,
+      "loss": 1.2242,
+      "step": 10206
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.964640444974175e-07,
+      "loss": 1.5115,
+      "step": 10207
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.964243146603098e-07,
+      "loss": 1.5883,
+      "step": 10208
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.963845848232022e-07,
+      "loss": 1.8163,
+      "step": 10209
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.963448549860946e-07,
+      "loss": 1.7194,
+      "step": 10210
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.963051251489869e-07,
+      "loss": 1.3907,
+      "step": 10211
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.962653953118793e-07,
+      "loss": 1.5608,
+      "step": 10212
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.962256654747715e-07,
+      "loss": 1.5625,
+      "step": 10213
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.961859356376639e-07,
+      "loss": 1.5099,
+      "step": 10214
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.961462058005561e-07,
+      "loss": 2.1996,
+      "step": 10215
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.961064759634485e-07,
+      "loss": 1.7058,
+      "step": 10216
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.960667461263409e-07,
+      "loss": 1.9479,
+      "step": 10217
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.960270162892332e-07,
+      "loss": 1.6868,
+      "step": 10218
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.959872864521255e-07,
+      "loss": 1.2066,
+      "step": 10219
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.959475566150178e-07,
+      "loss": 1.4387,
+      "step": 10220
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.959078267779102e-07,
+      "loss": 1.7255,
+      "step": 10221
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.958680969408025e-07,
+      "loss": 1.2211,
+      "step": 10222
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.958283671036949e-07,
+      "loss": 1.9183,
+      "step": 10223
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.957886372665873e-07,
+      "loss": 1.6296,
+      "step": 10224
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.957489074294795e-07,
+      "loss": 1.4902,
+      "step": 10225
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.957091775923719e-07,
+      "loss": 1.497,
+      "step": 10226
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.956694477552642e-07,
+      "loss": 1.7228,
+      "step": 10227
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.956297179181565e-07,
+      "loss": 1.3939,
+      "step": 10228
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.955899880810488e-07,
+      "loss": 1.868,
+      "step": 10229
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.955502582439412e-07,
+      "loss": 1.8604,
+      "step": 10230
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.955105284068334e-07,
+      "loss": 2.0046,
+      "step": 10231
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.954707985697258e-07,
+      "loss": 1.615,
+      "step": 10232
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.954310687326182e-07,
+      "loss": 1.932,
+      "step": 10233
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.953913388955105e-07,
+      "loss": 1.808,
+      "step": 10234
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.953516090584029e-07,
+      "loss": 1.5111,
+      "step": 10235
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.953118792212952e-07,
+      "loss": 1.1104,
+      "step": 10236
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.952721493841875e-07,
+      "loss": 1.5098,
+      "step": 10237
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.952324195470798e-07,
+      "loss": 1.7631,
+      "step": 10238
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.951926897099722e-07,
+      "loss": 1.3926,
+      "step": 10239
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.951529598728646e-07,
+      "loss": 1.6058,
+      "step": 10240
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.951132300357568e-07,
+      "loss": 1.7925,
+      "step": 10241
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.950735001986491e-07,
+      "loss": 1.8571,
+      "step": 10242
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.950337703615414e-07,
+      "loss": 1.65,
+      "step": 10243
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.949940405244338e-07,
+      "loss": 1.665,
+      "step": 10244
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.949543106873261e-07,
+      "loss": 1.5298,
+      "step": 10245
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.949145808502185e-07,
+      "loss": 1.7098,
+      "step": 10246
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.948748510131109e-07,
+      "loss": 1.328,
+      "step": 10247
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.948351211760031e-07,
+      "loss": 2.2841,
+      "step": 10248
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.947953913388955e-07,
+      "loss": 1.6756,
+      "step": 10249
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.947556615017878e-07,
+      "loss": 1.7079,
+      "step": 10250
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.947159316646802e-07,
+      "loss": 1.4724,
+      "step": 10251
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.946762018275725e-07,
+      "loss": 1.7012,
+      "step": 10252
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.946364719904649e-07,
+      "loss": 1.6197,
+      "step": 10253
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.94596742153357e-07,
+      "loss": 1.8438,
+      "step": 10254
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.945570123162494e-07,
+      "loss": 1.6231,
+      "step": 10255
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.945172824791418e-07,
+      "loss": 2.0031,
+      "step": 10256
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.944775526420341e-07,
+      "loss": 1.5785,
+      "step": 10257
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.944378228049265e-07,
+      "loss": 1.9239,
+      "step": 10258
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.943980929678188e-07,
+      "loss": 1.3685,
+      "step": 10259
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.943583631307111e-07,
+      "loss": 1.9404,
+      "step": 10260
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.943186332936034e-07,
+      "loss": 1.9714,
+      "step": 10261
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.942789034564958e-07,
+      "loss": 1.6335,
+      "step": 10262
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.942391736193882e-07,
+      "loss": 1.3223,
+      "step": 10263
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.941994437822805e-07,
+      "loss": 1.2881,
+      "step": 10264
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.941597139451729e-07,
+      "loss": 1.6251,
+      "step": 10265
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.941199841080651e-07,
+      "loss": 1.4535,
+      "step": 10266
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.940802542709574e-07,
+      "loss": 1.4751,
+      "step": 10267
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.940405244338497e-07,
+      "loss": 1.7633,
+      "step": 10268
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.940007945967421e-07,
+      "loss": 1.4182,
+      "step": 10269
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.939610647596344e-07,
+      "loss": 2.0885,
+      "step": 10270
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.939213349225268e-07,
+      "loss": 1.4359,
+      "step": 10271
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.938816050854191e-07,
+      "loss": 1.0918,
+      "step": 10272
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.938418752483114e-07,
+      "loss": 1.8543,
+      "step": 10273
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.938021454112038e-07,
+      "loss": 1.5871,
+      "step": 10274
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.937624155740961e-07,
+      "loss": 1.8044,
+      "step": 10275
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.937226857369885e-07,
+      "loss": 1.5073,
+      "step": 10276
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.936829558998808e-07,
+      "loss": 1.314,
+      "step": 10277
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.936432260627731e-07,
+      "loss": 1.8652,
+      "step": 10278
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.936034962256655e-07,
+      "loss": 1.5696,
+      "step": 10279
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.935637663885578e-07,
+      "loss": 1.4434,
+      "step": 10280
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.935240365514501e-07,
+      "loss": 1.2949,
+      "step": 10281
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.934843067143424e-07,
+      "loss": 1.6944,
+      "step": 10282
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.934445768772348e-07,
+      "loss": 1.8294,
+      "step": 10283
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.93404847040127e-07,
+      "loss": 1.9348,
+      "step": 10284
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.933651172030194e-07,
+      "loss": 1.5917,
+      "step": 10285
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.933253873659117e-07,
+      "loss": 1.7748,
+      "step": 10286
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.932856575288041e-07,
+      "loss": 1.5977,
+      "step": 10287
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.932459276916965e-07,
+      "loss": 2.0009,
+      "step": 10288
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.932061978545888e-07,
+      "loss": 2.2831,
+      "step": 10289
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.931664680174811e-07,
+      "loss": 1.6009,
+      "step": 10290
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.931267381803734e-07,
+      "loss": 1.5191,
+      "step": 10291
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.930870083432658e-07,
+      "loss": 2.2361,
+      "step": 10292
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.930472785061581e-07,
+      "loss": 1.7898,
+      "step": 10293
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.930075486690504e-07,
+      "loss": 1.8011,
+      "step": 10294
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.929678188319428e-07,
+      "loss": 1.5983,
+      "step": 10295
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.92928088994835e-07,
+      "loss": 1.6243,
+      "step": 10296
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.928883591577274e-07,
+      "loss": 1.2034,
+      "step": 10297
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.928486293206197e-07,
+      "loss": 1.5321,
+      "step": 10298
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.928088994835121e-07,
+      "loss": 1.4455,
+      "step": 10299
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.927691696464044e-07,
+      "loss": 1.6952,
+      "step": 10300
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.927294398092968e-07,
+      "loss": 1.2392,
+      "step": 10301
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.92689709972189e-07,
+      "loss": 1.8675,
+      "step": 10302
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.926499801350814e-07,
+      "loss": 1.5049,
+      "step": 10303
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.926102502979738e-07,
+      "loss": 1.5951,
+      "step": 10304
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.925705204608661e-07,
+      "loss": 1.8605,
+      "step": 10305
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.925307906237585e-07,
+      "loss": 1.8574,
+      "step": 10306
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.924910607866507e-07,
+      "loss": 2.0151,
+      "step": 10307
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.92451330949543e-07,
+      "loss": 1.5863,
+      "step": 10308
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.924116011124353e-07,
+      "loss": 1.3747,
+      "step": 10309
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.923718712753277e-07,
+      "loss": 1.762,
+      "step": 10310
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.923321414382201e-07,
+      "loss": 1.6654,
+      "step": 10311
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.922924116011124e-07,
+      "loss": 1.5292,
+      "step": 10312
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.922526817640048e-07,
+      "loss": 1.5984,
+      "step": 10313
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.92212951926897e-07,
+      "loss": 1.0798,
+      "step": 10314
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.921732220897894e-07,
+      "loss": 1.4989,
+      "step": 10315
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.921334922526817e-07,
+      "loss": 1.5099,
+      "step": 10316
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.920937624155741e-07,
+      "loss": 1.7166,
+      "step": 10317
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.920540325784665e-07,
+      "loss": 2.0289,
+      "step": 10318
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.920143027413588e-07,
+      "loss": 1.5256,
+      "step": 10319
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.91974572904251e-07,
+      "loss": 2.0993,
+      "step": 10320
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.919348430671433e-07,
+      "loss": 1.2301,
+      "step": 10321
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.918951132300357e-07,
+      "loss": 1.7303,
+      "step": 10322
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.91855383392928e-07,
+      "loss": 1.815,
+      "step": 10323
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.918156535558204e-07,
+      "loss": 1.6867,
+      "step": 10324
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.917759237187128e-07,
+      "loss": 1.5461,
+      "step": 10325
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.91736193881605e-07,
+      "loss": 1.8572,
+      "step": 10326
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.916964640444974e-07,
+      "loss": 2.1206,
+      "step": 10327
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.916567342073897e-07,
+      "loss": 1.3599,
+      "step": 10328
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.916170043702821e-07,
+      "loss": 1.7766,
+      "step": 10329
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.915772745331744e-07,
+      "loss": 1.5281,
+      "step": 10330
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.915375446960668e-07,
+      "loss": 1.6167,
+      "step": 10331
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.91497814858959e-07,
+      "loss": 1.8306,
+      "step": 10332
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.914580850218513e-07,
+      "loss": 1.3002,
+      "step": 10333
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.914183551847437e-07,
+      "loss": 1.8771,
+      "step": 10334
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.91378625347636e-07,
+      "loss": 2.1484,
+      "step": 10335
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.913388955105284e-07,
+      "loss": 1.3026,
+      "step": 10336
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.912991656734207e-07,
+      "loss": 2.0781,
+      "step": 10337
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.91259435836313e-07,
+      "loss": 1.8973,
+      "step": 10338
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.912197059992053e-07,
+      "loss": 1.9592,
+      "step": 10339
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.911799761620977e-07,
+      "loss": 1.8358,
+      "step": 10340
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.911402463249901e-07,
+      "loss": 2.352,
+      "step": 10341
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.911005164878824e-07,
+      "loss": 1.6768,
+      "step": 10342
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.910607866507748e-07,
+      "loss": 2.6256,
+      "step": 10343
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.91021056813667e-07,
+      "loss": 1.3647,
+      "step": 10344
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.909813269765594e-07,
+      "loss": 1.9835,
+      "step": 10345
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.909415971394517e-07,
+      "loss": 1.9545,
+      "step": 10346
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.90901867302344e-07,
+      "loss": 1.5541,
+      "step": 10347
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.908621374652363e-07,
+      "loss": 1.5178,
+      "step": 10348
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.908224076281287e-07,
+      "loss": 1.6364,
+      "step": 10349
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.90782677791021e-07,
+      "loss": 1.9626,
+      "step": 10350
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.907429479539133e-07,
+      "loss": 2.027,
+      "step": 10351
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.907032181168057e-07,
+      "loss": 1.6237,
+      "step": 10352
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.90663488279698e-07,
+      "loss": 1.6962,
+      "step": 10353
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.906237584425904e-07,
+      "loss": 1.8388,
+      "step": 10354
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.905840286054827e-07,
+      "loss": 1.4138,
+      "step": 10355
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.90544298768375e-07,
+      "loss": 1.6351,
+      "step": 10356
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.905045689312674e-07,
+      "loss": 1.5245,
+      "step": 10357
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.904648390941597e-07,
+      "loss": 2.0892,
+      "step": 10358
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.904251092570521e-07,
+      "loss": 1.691,
+      "step": 10359
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.903853794199443e-07,
+      "loss": 1.5492,
+      "step": 10360
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.903456495828367e-07,
+      "loss": 1.2227,
+      "step": 10361
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.903059197457289e-07,
+      "loss": 1.8058,
+      "step": 10362
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.902661899086213e-07,
+      "loss": 1.4378,
+      "step": 10363
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.902264600715137e-07,
+      "loss": 1.694,
+      "step": 10364
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.90186730234406e-07,
+      "loss": 1.514,
+      "step": 10365
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.901470003972984e-07,
+      "loss": 1.6491,
+      "step": 10366
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.901072705601907e-07,
+      "loss": 1.6328,
+      "step": 10367
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.90067540723083e-07,
+      "loss": 1.7282,
+      "step": 10368
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.900278108859753e-07,
+      "loss": 1.7978,
+      "step": 10369
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.899880810488677e-07,
+      "loss": 1.527,
+      "step": 10370
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.8994835121176e-07,
+      "loss": 1.7468,
+      "step": 10371
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.899086213746524e-07,
+      "loss": 1.3886,
+      "step": 10372
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.898688915375447e-07,
+      "loss": 1.9275,
+      "step": 10373
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.898291617004369e-07,
+      "loss": 1.7388,
+      "step": 10374
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.897894318633293e-07,
+      "loss": 2.067,
+      "step": 10375
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.897497020262216e-07,
+      "loss": 1.6478,
+      "step": 10376
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.89709972189114e-07,
+      "loss": 1.7864,
+      "step": 10377
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.896702423520063e-07,
+      "loss": 1.7622,
+      "step": 10378
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.896305125148987e-07,
+      "loss": 1.1279,
+      "step": 10379
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.89590782677791e-07,
+      "loss": 1.9107,
+      "step": 10380
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.895510528406833e-07,
+      "loss": 1.7319,
+      "step": 10381
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.895113230035757e-07,
+      "loss": 1.6079,
+      "step": 10382
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.89471593166468e-07,
+      "loss": 1.6072,
+      "step": 10383
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.894318633293604e-07,
+      "loss": 1.7936,
+      "step": 10384
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.893921334922527e-07,
+      "loss": 1.6389,
+      "step": 10385
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.893524036551449e-07,
+      "loss": 2.0148,
+      "step": 10386
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.893126738180372e-07,
+      "loss": 1.5878,
+      "step": 10387
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.892729439809296e-07,
+      "loss": 1.7912,
+      "step": 10388
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.89233214143822e-07,
+      "loss": 1.3874,
+      "step": 10389
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.891934843067143e-07,
+      "loss": 1.7601,
+      "step": 10390
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.891537544696067e-07,
+      "loss": 1.5303,
+      "step": 10391
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.891140246324989e-07,
+      "loss": 2.2649,
+      "step": 10392
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.890742947953913e-07,
+      "loss": 1.7816,
+      "step": 10393
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.890345649582836e-07,
+      "loss": 1.5495,
+      "step": 10394
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.88994835121176e-07,
+      "loss": 1.6856,
+      "step": 10395
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.889551052840684e-07,
+      "loss": 1.463,
+      "step": 10396
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.889153754469607e-07,
+      "loss": 1.514,
+      "step": 10397
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.88875645609853e-07,
+      "loss": 2.0077,
+      "step": 10398
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.888359157727452e-07,
+      "loss": 1.2812,
+      "step": 10399
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.887961859356376e-07,
+      "loss": 1.9422,
+      "step": 10400
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.887564560985299e-07,
+      "loss": 1.6909,
+      "step": 10401
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.887167262614223e-07,
+      "loss": 1.7341,
+      "step": 10402
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.886769964243147e-07,
+      "loss": 1.3166,
+      "step": 10403
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.886372665872069e-07,
+      "loss": 1.2155,
+      "step": 10404
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.885975367500993e-07,
+      "loss": 1.7038,
+      "step": 10405
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.885578069129916e-07,
+      "loss": 1.5463,
+      "step": 10406
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.88518077075884e-07,
+      "loss": 1.5652,
+      "step": 10407
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.884783472387763e-07,
+      "loss": 1.3368,
+      "step": 10408
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.884386174016687e-07,
+      "loss": 1.7584,
+      "step": 10409
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.883988875645609e-07,
+      "loss": 1.1805,
+      "step": 10410
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.883591577274533e-07,
+      "loss": 1.7,
+      "step": 10411
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.883194278903457e-07,
+      "loss": 1.8572,
+      "step": 10412
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.882796980532379e-07,
+      "loss": 1.6958,
+      "step": 10413
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.882399682161303e-07,
+      "loss": 1.3143,
+      "step": 10414
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.882002383790226e-07,
+      "loss": 1.8229,
+      "step": 10415
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.881605085419149e-07,
+      "loss": 1.6632,
+      "step": 10416
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.881207787048072e-07,
+      "loss": 1.4091,
+      "step": 10417
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.880810488676996e-07,
+      "loss": 1.6891,
+      "step": 10418
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.88041319030592e-07,
+      "loss": 1.3531,
+      "step": 10419
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.880015891934843e-07,
+      "loss": 1.5617,
+      "step": 10420
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.879618593563767e-07,
+      "loss": 1.4377,
+      "step": 10421
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.879221295192689e-07,
+      "loss": 2.0418,
+      "step": 10422
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.878823996821613e-07,
+      "loss": 1.3451,
+      "step": 10423
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.878426698450536e-07,
+      "loss": 1.7485,
+      "step": 10424
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.87802940007946e-07,
+      "loss": 1.4725,
+      "step": 10425
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.877632101708382e-07,
+      "loss": 1.8181,
+      "step": 10426
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.877234803337306e-07,
+      "loss": 1.2869,
+      "step": 10427
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.876837504966229e-07,
+      "loss": 1.785,
+      "step": 10428
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.876440206595152e-07,
+      "loss": 1.0809,
+      "step": 10429
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.876042908224076e-07,
+      "loss": 1.5938,
+      "step": 10430
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.875645609852999e-07,
+      "loss": 1.0958,
+      "step": 10431
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.875248311481923e-07,
+      "loss": 1.4798,
+      "step": 10432
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.874851013110846e-07,
+      "loss": 1.6356,
+      "step": 10433
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.874453714739769e-07,
+      "loss": 1.5007,
+      "step": 10434
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.874056416368693e-07,
+      "loss": 1.7135,
+      "step": 10435
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.873659117997616e-07,
+      "loss": 1.3351,
+      "step": 10436
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.87326181962654e-07,
+      "loss": 1.8779,
+      "step": 10437
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.872864521255463e-07,
+      "loss": 1.3801,
+      "step": 10438
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.872467222884386e-07,
+      "loss": 1.8884,
+      "step": 10439
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.872069924513308e-07,
+      "loss": 1.7728,
+      "step": 10440
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.871672626142232e-07,
+      "loss": 1.6427,
+      "step": 10441
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.871275327771156e-07,
+      "loss": 1.2481,
+      "step": 10442
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.870878029400079e-07,
+      "loss": 1.9773,
+      "step": 10443
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.870480731029003e-07,
+      "loss": 1.5746,
+      "step": 10444
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.870083432657926e-07,
+      "loss": 1.6391,
+      "step": 10445
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.869686134286849e-07,
+      "loss": 1.8421,
+      "step": 10446
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.869288835915772e-07,
+      "loss": 1.7095,
+      "step": 10447
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.868891537544696e-07,
+      "loss": 1.7608,
+      "step": 10448
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.86849423917362e-07,
+      "loss": 2.2394,
+      "step": 10449
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.868096940802543e-07,
+      "loss": 1.6121,
+      "step": 10450
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.867699642431467e-07,
+      "loss": 1.313,
+      "step": 10451
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.867302344060388e-07,
+      "loss": 1.4451,
+      "step": 10452
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.866905045689312e-07,
+      "loss": 1.959,
+      "step": 10453
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.866507747318235e-07,
+      "loss": 1.612,
+      "step": 10454
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.866110448947159e-07,
+      "loss": 1.7234,
+      "step": 10455
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.865713150576082e-07,
+      "loss": 1.3532,
+      "step": 10456
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.865315852205006e-07,
+      "loss": 1.5388,
+      "step": 10457
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.864918553833929e-07,
+      "loss": 1.4035,
+      "step": 10458
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.864521255462852e-07,
+      "loss": 1.3972,
+      "step": 10459
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.864123957091776e-07,
+      "loss": 1.962,
+      "step": 10460
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.863726658720699e-07,
+      "loss": 1.546,
+      "step": 10461
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.863329360349623e-07,
+      "loss": 1.5811,
+      "step": 10462
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.862932061978546e-07,
+      "loss": 1.4061,
+      "step": 10463
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.862534763607469e-07,
+      "loss": 1.6487,
+      "step": 10464
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.862137465236391e-07,
+      "loss": 1.6546,
+      "step": 10465
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.861740166865315e-07,
+      "loss": 1.7912,
+      "step": 10466
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.861342868494239e-07,
+      "loss": 1.5334,
+      "step": 10467
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.860945570123162e-07,
+      "loss": 1.3531,
+      "step": 10468
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.860548271752086e-07,
+      "loss": 1.452,
+      "step": 10469
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.860150973381008e-07,
+      "loss": 1.7362,
+      "step": 10470
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.859753675009932e-07,
+      "loss": 1.4472,
+      "step": 10471
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.859356376638855e-07,
+      "loss": 1.5284,
+      "step": 10472
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.858959078267779e-07,
+      "loss": 1.4151,
+      "step": 10473
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.858561779896703e-07,
+      "loss": 1.3763,
+      "step": 10474
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.858164481525626e-07,
+      "loss": 1.4916,
+      "step": 10475
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.857767183154549e-07,
+      "loss": 1.6383,
+      "step": 10476
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.857369884783472e-07,
+      "loss": 1.5739,
+      "step": 10477
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.856972586412396e-07,
+      "loss": 1.7215,
+      "step": 10478
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.856575288041318e-07,
+      "loss": 1.9997,
+      "step": 10479
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.856177989670242e-07,
+      "loss": 1.5296,
+      "step": 10480
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.855780691299166e-07,
+      "loss": 1.7014,
+      "step": 10481
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.855383392928088e-07,
+      "loss": 1.1882,
+      "step": 10482
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.854986094557012e-07,
+      "loss": 1.3716,
+      "step": 10483
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.854588796185935e-07,
+      "loss": 1.7407,
+      "step": 10484
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.854191497814859e-07,
+      "loss": 1.1435,
+      "step": 10485
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.853794199443782e-07,
+      "loss": 1.7816,
+      "step": 10486
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.853396901072706e-07,
+      "loss": 1.5743,
+      "step": 10487
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.852999602701628e-07,
+      "loss": 1.4294,
+      "step": 10488
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.852602304330552e-07,
+      "loss": 1.7531,
+      "step": 10489
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.852205005959476e-07,
+      "loss": 2.65,
+      "step": 10490
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.851807707588399e-07,
+      "loss": 1.3984,
+      "step": 10491
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.851410409217322e-07,
+      "loss": 1.4324,
+      "step": 10492
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.851013110846245e-07,
+      "loss": 2.0323,
+      "step": 10493
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.850615812475168e-07,
+      "loss": 1.6557,
+      "step": 10494
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.850218514104091e-07,
+      "loss": 1.88,
+      "step": 10495
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.849821215733015e-07,
+      "loss": 1.7838,
+      "step": 10496
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.849423917361939e-07,
+      "loss": 1.8158,
+      "step": 10497
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.849026618990862e-07,
+      "loss": 1.7991,
+      "step": 10498
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.848629320619786e-07,
+      "loss": 1.8926,
+      "step": 10499
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.848232022248708e-07,
+      "loss": 1.6911,
+      "step": 10500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.847834723877632e-07,
+      "loss": 1.23,
+      "step": 10501
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.847437425506555e-07,
+      "loss": 1.2827,
+      "step": 10502
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.847040127135479e-07,
+      "loss": 1.3561,
+      "step": 10503
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.846642828764403e-07,
+      "loss": 1.8407,
+      "step": 10504
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.846245530393325e-07,
+      "loss": 1.9302,
+      "step": 10505
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.845848232022248e-07,
+      "loss": 1.4809,
+      "step": 10506
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.845450933651171e-07,
+      "loss": 1.8745,
+      "step": 10507
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.845053635280095e-07,
+      "loss": 1.4713,
+      "step": 10508
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.844656336909018e-07,
+      "loss": 1.614,
+      "step": 10509
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.844259038537942e-07,
+      "loss": 1.0538,
+      "step": 10510
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.843861740166865e-07,
+      "loss": 1.5108,
+      "step": 10511
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.843464441795788e-07,
+      "loss": 1.4959,
+      "step": 10512
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.843067143424712e-07,
+      "loss": 1.915,
+      "step": 10513
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.842669845053635e-07,
+      "loss": 1.5451,
+      "step": 10514
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.842272546682559e-07,
+      "loss": 1.6017,
+      "step": 10515
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.841875248311482e-07,
+      "loss": 1.4345,
+      "step": 10516
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.841477949940406e-07,
+      "loss": 1.6195,
+      "step": 10517
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.841080651569327e-07,
+      "loss": 1.5187,
+      "step": 10518
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.840683353198251e-07,
+      "loss": 1.3985,
+      "step": 10519
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.840286054827175e-07,
+      "loss": 1.2802,
+      "step": 10520
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.839888756456098e-07,
+      "loss": 1.5944,
+      "step": 10521
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.839491458085022e-07,
+      "loss": 1.6795,
+      "step": 10522
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.839094159713945e-07,
+      "loss": 1.8596,
+      "step": 10523
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.838696861342868e-07,
+      "loss": 1.3047,
+      "step": 10524
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.838299562971791e-07,
+      "loss": 1.2612,
+      "step": 10525
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.837902264600715e-07,
+      "loss": 1.7312,
+      "step": 10526
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.837504966229638e-07,
+      "loss": 1.5108,
+      "step": 10527
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.837107667858562e-07,
+      "loss": 1.7982,
+      "step": 10528
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.836710369487486e-07,
+      "loss": 1.76,
+      "step": 10529
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.836313071116408e-07,
+      "loss": 1.8299,
+      "step": 10530
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.835915772745331e-07,
+      "loss": 1.7254,
+      "step": 10531
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.835518474374254e-07,
+      "loss": 1.6456,
+      "step": 10532
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.835121176003178e-07,
+      "loss": 1.4557,
+      "step": 10533
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.834723877632101e-07,
+      "loss": 1.6574,
+      "step": 10534
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.834326579261025e-07,
+      "loss": 1.3153,
+      "step": 10535
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.833929280889948e-07,
+      "loss": 1.8312,
+      "step": 10536
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.833531982518871e-07,
+      "loss": 1.6704,
+      "step": 10537
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.833134684147795e-07,
+      "loss": 1.2198,
+      "step": 10538
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.832737385776718e-07,
+      "loss": 1.7265,
+      "step": 10539
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.832340087405642e-07,
+      "loss": 1.559,
+      "step": 10540
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.831942789034565e-07,
+      "loss": 1.4254,
+      "step": 10541
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.831545490663488e-07,
+      "loss": 1.7841,
+      "step": 10542
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.831148192292412e-07,
+      "loss": 1.9077,
+      "step": 10543
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.830750893921335e-07,
+      "loss": 1.6344,
+      "step": 10544
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.830353595550258e-07,
+      "loss": 1.2708,
+      "step": 10545
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.829956297179181e-07,
+      "loss": 1.5425,
+      "step": 10546
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.829558998808105e-07,
+      "loss": 1.7421,
+      "step": 10547
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.829161700437027e-07,
+      "loss": 1.4887,
+      "step": 10548
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.828764402065951e-07,
+      "loss": 1.5604,
+      "step": 10549
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.828367103694874e-07,
+      "loss": 1.3033,
+      "step": 10550
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.827969805323798e-07,
+      "loss": 1.8597,
+      "step": 10551
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.827572506952722e-07,
+      "loss": 1.2885,
+      "step": 10552
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.827175208581645e-07,
+      "loss": 1.2861,
+      "step": 10553
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.826777910210568e-07,
+      "loss": 1.2443,
+      "step": 10554
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.826380611839491e-07,
+      "loss": 1.4439,
+      "step": 10555
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.825983313468415e-07,
+      "loss": 1.5753,
+      "step": 10556
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.825586015097338e-07,
+      "loss": 1.6031,
+      "step": 10557
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.825188716726261e-07,
+      "loss": 1.4091,
+      "step": 10558
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.824791418355185e-07,
+      "loss": 1.594,
+      "step": 10559
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.824394119984107e-07,
+      "loss": 1.81,
+      "step": 10560
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.823996821613031e-07,
+      "loss": 1.4275,
+      "step": 10561
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.823599523241954e-07,
+      "loss": 1.4989,
+      "step": 10562
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.823202224870878e-07,
+      "loss": 1.9044,
+      "step": 10563
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.822804926499801e-07,
+      "loss": 1.4169,
+      "step": 10564
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.822407628128725e-07,
+      "loss": 1.2991,
+      "step": 10565
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.822010329757647e-07,
+      "loss": 1.4582,
+      "step": 10566
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.821613031386571e-07,
+      "loss": 1.8581,
+      "step": 10567
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.821215733015495e-07,
+      "loss": 1.3695,
+      "step": 10568
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.820818434644418e-07,
+      "loss": 1.6894,
+      "step": 10569
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.820421136273342e-07,
+      "loss": 1.2098,
+      "step": 10570
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.820023837902264e-07,
+      "loss": 1.4803,
+      "step": 10571
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.819626539531187e-07,
+      "loss": 1.6538,
+      "step": 10572
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.81922924116011e-07,
+      "loss": 1.6695,
+      "step": 10573
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.818831942789034e-07,
+      "loss": 1.8001,
+      "step": 10574
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.818434644417958e-07,
+      "loss": 1.634,
+      "step": 10575
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.818037346046881e-07,
+      "loss": 1.3394,
+      "step": 10576
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.817640047675805e-07,
+      "loss": 1.6704,
+      "step": 10577
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.817242749304727e-07,
+      "loss": 1.5094,
+      "step": 10578
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.816845450933651e-07,
+      "loss": 1.7685,
+      "step": 10579
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.816448152562574e-07,
+      "loss": 1.6395,
+      "step": 10580
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.816050854191498e-07,
+      "loss": 1.5659,
+      "step": 10581
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.815653555820422e-07,
+      "loss": 1.8578,
+      "step": 10582
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.815256257449345e-07,
+      "loss": 1.4148,
+      "step": 10583
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.814858959078267e-07,
+      "loss": 1.4598,
+      "step": 10584
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.81446166070719e-07,
+      "loss": 1.6667,
+      "step": 10585
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.814064362336114e-07,
+      "loss": 1.7281,
+      "step": 10586
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.813667063965037e-07,
+      "loss": 2.0655,
+      "step": 10587
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.813269765593961e-07,
+      "loss": 1.9577,
+      "step": 10588
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.812872467222884e-07,
+      "loss": 1.8281,
+      "step": 10589
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.812475168851807e-07,
+      "loss": 1.5446,
+      "step": 10590
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.812077870480731e-07,
+      "loss": 1.4754,
+      "step": 10591
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.811680572109654e-07,
+      "loss": 1.5357,
+      "step": 10592
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.811283273738578e-07,
+      "loss": 1.9026,
+      "step": 10593
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.810885975367501e-07,
+      "loss": 1.7916,
+      "step": 10594
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.810488676996425e-07,
+      "loss": 1.9129,
+      "step": 10595
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.810091378625347e-07,
+      "loss": 1.7224,
+      "step": 10596
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.80969408025427e-07,
+      "loss": 1.9425,
+      "step": 10597
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.809296781883194e-07,
+      "loss": 1.9297,
+      "step": 10598
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.808899483512117e-07,
+      "loss": 1.2699,
+      "step": 10599
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.808502185141041e-07,
+      "loss": 1.7502,
+      "step": 10600
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.808104886769964e-07,
+      "loss": 1.1974,
+      "step": 10601
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.807707588398887e-07,
+      "loss": 1.6187,
+      "step": 10602
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.80731029002781e-07,
+      "loss": 1.866,
+      "step": 10603
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.806912991656734e-07,
+      "loss": 2.0134,
+      "step": 10604
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.806515693285657e-07,
+      "loss": 1.7486,
+      "step": 10605
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.806118394914581e-07,
+      "loss": 1.7394,
+      "step": 10606
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.805721096543505e-07,
+      "loss": 1.8773,
+      "step": 10607
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.805323798172427e-07,
+      "loss": 1.6612,
+      "step": 10608
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.804926499801351e-07,
+      "loss": 1.4279,
+      "step": 10609
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.804529201430274e-07,
+      "loss": 1.6332,
+      "step": 10610
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.804131903059197e-07,
+      "loss": 1.5316,
+      "step": 10611
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.80373460468812e-07,
+      "loss": 1.4274,
+      "step": 10612
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.803337306317044e-07,
+      "loss": 1.7949,
+      "step": 10613
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.802940007945967e-07,
+      "loss": 1.8493,
+      "step": 10614
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.80254270957489e-07,
+      "loss": 1.7535,
+      "step": 10615
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.802145411203814e-07,
+      "loss": 1.7567,
+      "step": 10616
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.801748112832737e-07,
+      "loss": 1.8372,
+      "step": 10617
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.801350814461661e-07,
+      "loss": 1.3358,
+      "step": 10618
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.800953516090584e-07,
+      "loss": 1.2598,
+      "step": 10619
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.800556217719507e-07,
+      "loss": 1.3161,
+      "step": 10620
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.80015891934843e-07,
+      "loss": 1.843,
+      "step": 10621
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.799761620977354e-07,
+      "loss": 2.0075,
+      "step": 10622
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.799364322606278e-07,
+      "loss": 1.7661,
+      "step": 10623
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.7989670242352e-07,
+      "loss": 1.4505,
+      "step": 10624
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.798569725864124e-07,
+      "loss": 1.5953,
+      "step": 10625
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.798172427493046e-07,
+      "loss": 1.1333,
+      "step": 10626
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.79777512912197e-07,
+      "loss": 1.5388,
+      "step": 10627
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.797377830750893e-07,
+      "loss": 1.8842,
+      "step": 10628
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.796980532379817e-07,
+      "loss": 1.4477,
+      "step": 10629
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.796583234008741e-07,
+      "loss": 1.9715,
+      "step": 10630
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.796185935637664e-07,
+      "loss": 1.5601,
+      "step": 10631
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.795788637266587e-07,
+      "loss": 1.4844,
+      "step": 10632
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.79539133889551e-07,
+      "loss": 1.6002,
+      "step": 10633
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.794994040524434e-07,
+      "loss": 1.5919,
+      "step": 10634
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.794596742153357e-07,
+      "loss": 1.6185,
+      "step": 10635
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.794199443782281e-07,
+      "loss": 1.3241,
+      "step": 10636
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.793802145411204e-07,
+      "loss": 1.1806,
+      "step": 10637
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.793404847040126e-07,
+      "loss": 1.7583,
+      "step": 10638
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.79300754866905e-07,
+      "loss": 1.3756,
+      "step": 10639
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.792610250297973e-07,
+      "loss": 1.2618,
+      "step": 10640
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.792212951926897e-07,
+      "loss": 1.5742,
+      "step": 10641
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.79181565355582e-07,
+      "loss": 1.4724,
+      "step": 10642
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.791418355184744e-07,
+      "loss": 1.4135,
+      "step": 10643
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.791021056813666e-07,
+      "loss": 1.5357,
+      "step": 10644
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.79062375844259e-07,
+      "loss": 1.9585,
+      "step": 10645
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.790226460071514e-07,
+      "loss": 1.5753,
+      "step": 10646
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.789829161700437e-07,
+      "loss": 2.0868,
+      "step": 10647
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.789431863329361e-07,
+      "loss": 1.8878,
+      "step": 10648
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.789034564958284e-07,
+      "loss": 1.6089,
+      "step": 10649
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.788637266587206e-07,
+      "loss": 1.4702,
+      "step": 10650
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.788239968216129e-07,
+      "loss": 1.3975,
+      "step": 10651
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.787842669845053e-07,
+      "loss": 1.4467,
+      "step": 10652
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.787445371473977e-07,
+      "loss": 2.0068,
+      "step": 10653
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.7870480731029e-07,
+      "loss": 1.6678,
+      "step": 10654
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.786650774731824e-07,
+      "loss": 1.6685,
+      "step": 10655
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.786253476360746e-07,
+      "loss": 1.7308,
+      "step": 10656
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.78585617798967e-07,
+      "loss": 2.2681,
+      "step": 10657
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.785458879618593e-07,
+      "loss": 1.5306,
+      "step": 10658
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.785061581247517e-07,
+      "loss": 1.6198,
+      "step": 10659
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.784664282876441e-07,
+      "loss": 2.0774,
+      "step": 10660
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.784266984505364e-07,
+      "loss": 1.5299,
+      "step": 10661
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.783869686134287e-07,
+      "loss": 1.4606,
+      "step": 10662
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.783472387763209e-07,
+      "loss": 1.5413,
+      "step": 10663
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.783075089392133e-07,
+      "loss": 1.2789,
+      "step": 10664
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.782677791021056e-07,
+      "loss": 1.4341,
+      "step": 10665
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.78228049264998e-07,
+      "loss": 1.7965,
+      "step": 10666
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.781883194278903e-07,
+      "loss": 1.4459,
+      "step": 10667
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.781485895907826e-07,
+      "loss": 1.8352,
+      "step": 10668
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.78108859753675e-07,
+      "loss": 1.678,
+      "step": 10669
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.780691299165673e-07,
+      "loss": 1.4673,
+      "step": 10670
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.780294000794597e-07,
+      "loss": 1.1817,
+      "step": 10671
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.77989670242352e-07,
+      "loss": 1.4511,
+      "step": 10672
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.779499404052444e-07,
+      "loss": 1.3509,
+      "step": 10673
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.779102105681366e-07,
+      "loss": 2.0322,
+      "step": 10674
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.77870480731029e-07,
+      "loss": 1.6005,
+      "step": 10675
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.778307508939213e-07,
+      "loss": 1.2668,
+      "step": 10676
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.777910210568136e-07,
+      "loss": 1.4589,
+      "step": 10677
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.77751291219706e-07,
+      "loss": 1.9638,
+      "step": 10678
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.777115613825983e-07,
+      "loss": 1.4662,
+      "step": 10679
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.776718315454906e-07,
+      "loss": 1.4694,
+      "step": 10680
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.776321017083829e-07,
+      "loss": 1.6946,
+      "step": 10681
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.775923718712753e-07,
+      "loss": 1.2914,
+      "step": 10682
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.775526420341677e-07,
+      "loss": 1.649,
+      "step": 10683
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.7751291219706e-07,
+      "loss": 1.4698,
+      "step": 10684
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.774731823599524e-07,
+      "loss": 1.4666,
+      "step": 10685
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.774334525228446e-07,
+      "loss": 1.862,
+      "step": 10686
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.77393722685737e-07,
+      "loss": 1.3283,
+      "step": 10687
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.773539928486293e-07,
+      "loss": 1.6043,
+      "step": 10688
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.773142630115217e-07,
+      "loss": 1.7314,
+      "step": 10689
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.772745331744139e-07,
+      "loss": 1.5166,
+      "step": 10690
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.772348033373063e-07,
+      "loss": 1.4497,
+      "step": 10691
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.771950735001986e-07,
+      "loss": 1.5461,
+      "step": 10692
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.771553436630909e-07,
+      "loss": 1.6498,
+      "step": 10693
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.771156138259833e-07,
+      "loss": 1.8076,
+      "step": 10694
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.770758839888756e-07,
+      "loss": 1.6131,
+      "step": 10695
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.77036154151768e-07,
+      "loss": 1.7447,
+      "step": 10696
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.769964243146603e-07,
+      "loss": 1.3847,
+      "step": 10697
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.769566944775526e-07,
+      "loss": 1.7744,
+      "step": 10698
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.76916964640445e-07,
+      "loss": 1.4623,
+      "step": 10699
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.768772348033373e-07,
+      "loss": 1.6915,
+      "step": 10700
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.768375049662297e-07,
+      "loss": 1.9357,
+      "step": 10701
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.76797775129122e-07,
+      "loss": 1.3116,
+      "step": 10702
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.767580452920143e-07,
+      "loss": 1.5731,
+      "step": 10703
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.767183154549065e-07,
+      "loss": 1.6515,
+      "step": 10704
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.766785856177989e-07,
+      "loss": 1.9832,
+      "step": 10705
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.766388557806912e-07,
+      "loss": 2.0005,
+      "step": 10706
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.765991259435836e-07,
+      "loss": 2.0823,
+      "step": 10707
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.76559396106476e-07,
+      "loss": 1.65,
+      "step": 10708
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.765196662693683e-07,
+      "loss": 1.573,
+      "step": 10709
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.764799364322606e-07,
+      "loss": 1.5563,
+      "step": 10710
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.764402065951529e-07,
+      "loss": 1.877,
+      "step": 10711
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.764004767580453e-07,
+      "loss": 1.3345,
+      "step": 10712
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.763607469209376e-07,
+      "loss": 1.8066,
+      "step": 10713
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.7632101708383e-07,
+      "loss": 1.6777,
+      "step": 10714
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.762812872467224e-07,
+      "loss": 1.6139,
+      "step": 10715
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.762415574096145e-07,
+      "loss": 1.4492,
+      "step": 10716
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.762018275725069e-07,
+      "loss": 1.5735,
+      "step": 10717
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.761620977353992e-07,
+      "loss": 1.9222,
+      "step": 10718
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.761223678982916e-07,
+      "loss": 1.4705,
+      "step": 10719
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.760826380611839e-07,
+      "loss": 1.7353,
+      "step": 10720
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.760429082240763e-07,
+      "loss": 1.6031,
+      "step": 10721
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.760031783869685e-07,
+      "loss": 2.1169,
+      "step": 10722
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.759634485498609e-07,
+      "loss": 1.7354,
+      "step": 10723
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.759237187127533e-07,
+      "loss": 2.1207,
+      "step": 10724
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.758839888756456e-07,
+      "loss": 1.7976,
+      "step": 10725
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.75844259038538e-07,
+      "loss": 2.0942,
+      "step": 10726
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.758045292014303e-07,
+      "loss": 1.5144,
+      "step": 10727
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.757647993643226e-07,
+      "loss": 1.3921,
+      "step": 10728
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.757250695272148e-07,
+      "loss": 1.541,
+      "step": 10729
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.756853396901072e-07,
+      "loss": 1.749,
+      "step": 10730
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.756456098529996e-07,
+      "loss": 1.6984,
+      "step": 10731
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.756058800158919e-07,
+      "loss": 1.5438,
+      "step": 10732
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.755661501787843e-07,
+      "loss": 1.5762,
+      "step": 10733
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.755264203416765e-07,
+      "loss": 1.9878,
+      "step": 10734
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.754866905045689e-07,
+      "loss": 1.6734,
+      "step": 10735
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.754469606674612e-07,
+      "loss": 1.2499,
+      "step": 10736
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.754072308303536e-07,
+      "loss": 1.5279,
+      "step": 10737
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.75367500993246e-07,
+      "loss": 1.5708,
+      "step": 10738
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.753277711561383e-07,
+      "loss": 1.5455,
+      "step": 10739
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.752880413190306e-07,
+      "loss": 1.1788,
+      "step": 10740
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.752483114819229e-07,
+      "loss": 1.6863,
+      "step": 10741
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.752085816448152e-07,
+      "loss": 1.24,
+      "step": 10742
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.751688518077075e-07,
+      "loss": 2.0248,
+      "step": 10743
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.751291219705999e-07,
+      "loss": 1.6772,
+      "step": 10744
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.750893921334922e-07,
+      "loss": 1.5363,
+      "step": 10745
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.750496622963845e-07,
+      "loss": 1.4496,
+      "step": 10746
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.750099324592769e-07,
+      "loss": 1.3381,
+      "step": 10747
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.749702026221692e-07,
+      "loss": 1.5342,
+      "step": 10748
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.749304727850616e-07,
+      "loss": 1.4617,
+      "step": 10749
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.748907429479539e-07,
+      "loss": 1.7496,
+      "step": 10750
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.748510131108463e-07,
+      "loss": 1.6639,
+      "step": 10751
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.748112832737385e-07,
+      "loss": 1.4958,
+      "step": 10752
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.747715534366309e-07,
+      "loss": 1.2286,
+      "step": 10753
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.747318235995233e-07,
+      "loss": 1.594,
+      "step": 10754
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.746920937624156e-07,
+      "loss": 1.8359,
+      "step": 10755
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.746523639253079e-07,
+      "loss": 1.4513,
+      "step": 10756
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.746126340882002e-07,
+      "loss": 1.6483,
+      "step": 10757
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.745729042510925e-07,
+      "loss": 1.6164,
+      "step": 10758
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.745331744139848e-07,
+      "loss": 1.9277,
+      "step": 10759
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.744934445768772e-07,
+      "loss": 1.4287,
+      "step": 10760
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.744537147397696e-07,
+      "loss": 1.25,
+      "step": 10761
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.744139849026619e-07,
+      "loss": 1.5659,
+      "step": 10762
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.743742550655543e-07,
+      "loss": 2.0213,
+      "step": 10763
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.743345252284465e-07,
+      "loss": 1.1991,
+      "step": 10764
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.742947953913389e-07,
+      "loss": 1.3985,
+      "step": 10765
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.742550655542312e-07,
+      "loss": 1.4508,
+      "step": 10766
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.742153357171236e-07,
+      "loss": 1.9323,
+      "step": 10767
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.74175605880016e-07,
+      "loss": 1.4944,
+      "step": 10768
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.741358760429082e-07,
+      "loss": 1.8052,
+      "step": 10769
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.740961462058005e-07,
+      "loss": 1.7481,
+      "step": 10770
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.740564163686928e-07,
+      "loss": 1.767,
+      "step": 10771
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.740166865315852e-07,
+      "loss": 1.3769,
+      "step": 10772
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.739769566944775e-07,
+      "loss": 1.3608,
+      "step": 10773
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.739372268573699e-07,
+      "loss": 1.416,
+      "step": 10774
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.738974970202622e-07,
+      "loss": 1.9163,
+      "step": 10775
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.738577671831545e-07,
+      "loss": 1.9461,
+      "step": 10776
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.738180373460469e-07,
+      "loss": 1.6973,
+      "step": 10777
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.737783075089392e-07,
+      "loss": 1.4208,
+      "step": 10778
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.737385776718316e-07,
+      "loss": 1.6181,
+      "step": 10779
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.736988478347239e-07,
+      "loss": 1.4729,
+      "step": 10780
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.736591179976163e-07,
+      "loss": 1.6147,
+      "step": 10781
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.736193881605084e-07,
+      "loss": 1.5973,
+      "step": 10782
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.735796583234008e-07,
+      "loss": 1.7439,
+      "step": 10783
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.735399284862931e-07,
+      "loss": 1.9782,
+      "step": 10784
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.735001986491855e-07,
+      "loss": 1.8285,
+      "step": 10785
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.734604688120779e-07,
+      "loss": 1.3627,
+      "step": 10786
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.734207389749702e-07,
+      "loss": 1.5969,
+      "step": 10787
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.733810091378625e-07,
+      "loss": 1.5044,
+      "step": 10788
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.733412793007548e-07,
+      "loss": 1.6221,
+      "step": 10789
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.733015494636472e-07,
+      "loss": 1.5797,
+      "step": 10790
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.732618196265395e-07,
+      "loss": 1.7884,
+      "step": 10791
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.732220897894319e-07,
+      "loss": 1.5932,
+      "step": 10792
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.731823599523243e-07,
+      "loss": 1.8573,
+      "step": 10793
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.731426301152165e-07,
+      "loss": 1.37,
+      "step": 10794
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.731029002781088e-07,
+      "loss": 1.4138,
+      "step": 10795
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.730631704410011e-07,
+      "loss": 1.6447,
+      "step": 10796
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.730234406038935e-07,
+      "loss": 1.3632,
+      "step": 10797
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.729837107667858e-07,
+      "loss": 1.6638,
+      "step": 10798
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.729439809296782e-07,
+      "loss": 1.5646,
+      "step": 10799
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.729042510925704e-07,
+      "loss": 1.9717,
+      "step": 10800
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.728645212554628e-07,
+      "loss": 1.2976,
+      "step": 10801
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.728247914183552e-07,
+      "loss": 1.9949,
+      "step": 10802
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.727850615812475e-07,
+      "loss": 1.4806,
+      "step": 10803
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.727453317441399e-07,
+      "loss": 1.5523,
+      "step": 10804
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.727056019070322e-07,
+      "loss": 1.3428,
+      "step": 10805
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.726658720699245e-07,
+      "loss": 1.7211,
+      "step": 10806
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.726261422328168e-07,
+      "loss": 2.182,
+      "step": 10807
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.725864123957091e-07,
+      "loss": 1.8077,
+      "step": 10808
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.725466825586015e-07,
+      "loss": 1.8267,
+      "step": 10809
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.725069527214938e-07,
+      "loss": 1.5312,
+      "step": 10810
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.724672228843862e-07,
+      "loss": 1.4201,
+      "step": 10811
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.724274930472784e-07,
+      "loss": 1.2912,
+      "step": 10812
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.723877632101708e-07,
+      "loss": 1.8215,
+      "step": 10813
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.723480333730631e-07,
+      "loss": 1.9649,
+      "step": 10814
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.723083035359555e-07,
+      "loss": 1.5998,
+      "step": 10815
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.722685736988479e-07,
+      "loss": 1.5751,
+      "step": 10816
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.722288438617402e-07,
+      "loss": 1.7158,
+      "step": 10817
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.721891140246325e-07,
+      "loss": 1.5455,
+      "step": 10818
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.721493841875248e-07,
+      "loss": 1.2441,
+      "step": 10819
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.721096543504172e-07,
+      "loss": 1.4149,
+      "step": 10820
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.720699245133095e-07,
+      "loss": 1.7351,
+      "step": 10821
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.720301946762018e-07,
+      "loss": 1.6978,
+      "step": 10822
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.719904648390941e-07,
+      "loss": 1.2862,
+      "step": 10823
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.719507350019864e-07,
+      "loss": 1.433,
+      "step": 10824
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.719110051648788e-07,
+      "loss": 2.1184,
+      "step": 10825
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.718712753277711e-07,
+      "loss": 1.1989,
+      "step": 10826
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.718315454906635e-07,
+      "loss": 1.4947,
+      "step": 10827
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.717918156535558e-07,
+      "loss": 1.5889,
+      "step": 10828
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.717520858164482e-07,
+      "loss": 1.5439,
+      "step": 10829
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.717123559793404e-07,
+      "loss": 1.9166,
+      "step": 10830
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.716726261422328e-07,
+      "loss": 1.283,
+      "step": 10831
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.716328963051252e-07,
+      "loss": 1.7286,
+      "step": 10832
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.715931664680175e-07,
+      "loss": 1.6838,
+      "step": 10833
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.715534366309099e-07,
+      "loss": 1.6569,
+      "step": 10834
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.715137067938021e-07,
+      "loss": 1.5572,
+      "step": 10835
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.714739769566944e-07,
+      "loss": 1.728,
+      "step": 10836
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.714342471195867e-07,
+      "loss": 1.5993,
+      "step": 10837
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.713945172824791e-07,
+      "loss": 1.5924,
+      "step": 10838
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.713547874453715e-07,
+      "loss": 1.6153,
+      "step": 10839
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.713150576082638e-07,
+      "loss": 1.4275,
+      "step": 10840
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.712753277711561e-07,
+      "loss": 1.185,
+      "step": 10841
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.712355979340484e-07,
+      "loss": 1.8973,
+      "step": 10842
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.711958680969408e-07,
+      "loss": 2.0674,
+      "step": 10843
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.711561382598331e-07,
+      "loss": 1.3845,
+      "step": 10844
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.711164084227255e-07,
+      "loss": 1.845,
+      "step": 10845
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.710766785856178e-07,
+      "loss": 1.3442,
+      "step": 10846
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.710369487485101e-07,
+      "loss": 1.7192,
+      "step": 10847
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.709972189114024e-07,
+      "loss": 1.6101,
+      "step": 10848
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.709574890742947e-07,
+      "loss": 1.8956,
+      "step": 10849
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.709177592371871e-07,
+      "loss": 1.6347,
+      "step": 10850
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.708780294000794e-07,
+      "loss": 1.8,
+      "step": 10851
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.708382995629718e-07,
+      "loss": 1.9376,
+      "step": 10852
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.70798569725864e-07,
+      "loss": 1.3091,
+      "step": 10853
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.707588398887564e-07,
+      "loss": 1.6433,
+      "step": 10854
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.707191100516488e-07,
+      "loss": 1.4727,
+      "step": 10855
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.706793802145411e-07,
+      "loss": 1.5573,
+      "step": 10856
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.706396503774335e-07,
+      "loss": 1.2855,
+      "step": 10857
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.705999205403258e-07,
+      "loss": 1.1902,
+      "step": 10858
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.705601907032181e-07,
+      "loss": 1.7039,
+      "step": 10859
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.705204608661104e-07,
+      "loss": 1.4416,
+      "step": 10860
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.704807310290027e-07,
+      "loss": 1.4375,
+      "step": 10861
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.70441001191895e-07,
+      "loss": 1.6695,
+      "step": 10862
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.704012713547874e-07,
+      "loss": 1.8687,
+      "step": 10863
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.703615415176798e-07,
+      "loss": 1.7237,
+      "step": 10864
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.70321811680572e-07,
+      "loss": 2.1979,
+      "step": 10865
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.702820818434644e-07,
+      "loss": 1.6349,
+      "step": 10866
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.702423520063567e-07,
+      "loss": 1.5681,
+      "step": 10867
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.702026221692491e-07,
+      "loss": 1.3771,
+      "step": 10868
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.701628923321414e-07,
+      "loss": 1.7527,
+      "step": 10869
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.701231624950338e-07,
+      "loss": 1.9168,
+      "step": 10870
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.700834326579261e-07,
+      "loss": 1.6235,
+      "step": 10871
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.700437028208184e-07,
+      "loss": 1.6715,
+      "step": 10872
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.700039729837108e-07,
+      "loss": 1.431,
+      "step": 10873
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.69964243146603e-07,
+      "loss": 1.7066,
+      "step": 10874
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.699245133094954e-07,
+      "loss": 1.831,
+      "step": 10875
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.698847834723877e-07,
+      "loss": 1.8939,
+      "step": 10876
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.6984505363528e-07,
+      "loss": 1.8092,
+      "step": 10877
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.698053237981723e-07,
+      "loss": 1.9342,
+      "step": 10878
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.697655939610647e-07,
+      "loss": 1.868,
+      "step": 10879
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.697258641239571e-07,
+      "loss": 1.6904,
+      "step": 10880
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.696861342868494e-07,
+      "loss": 1.4865,
+      "step": 10881
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.696464044497418e-07,
+      "loss": 1.8432,
+      "step": 10882
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.69606674612634e-07,
+      "loss": 2.0457,
+      "step": 10883
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.695669447755264e-07,
+      "loss": 1.7081,
+      "step": 10884
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.695272149384187e-07,
+      "loss": 1.6531,
+      "step": 10885
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.694874851013111e-07,
+      "loss": 1.7111,
+      "step": 10886
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.694477552642035e-07,
+      "loss": 1.3965,
+      "step": 10887
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.694080254270957e-07,
+      "loss": 1.8383,
+      "step": 10888
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.69368295589988e-07,
+      "loss": 1.6509,
+      "step": 10889
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.693285657528803e-07,
+      "loss": 1.4646,
+      "step": 10890
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.692888359157727e-07,
+      "loss": 1.629,
+      "step": 10891
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.69249106078665e-07,
+      "loss": 1.108,
+      "step": 10892
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.692093762415574e-07,
+      "loss": 2.2202,
+      "step": 10893
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.691696464044498e-07,
+      "loss": 1.7186,
+      "step": 10894
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.69129916567342e-07,
+      "loss": 1.2765,
+      "step": 10895
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.690901867302344e-07,
+      "loss": 1.6364,
+      "step": 10896
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.690504568931267e-07,
+      "loss": 1.5212,
+      "step": 10897
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.690107270560191e-07,
+      "loss": 1.8179,
+      "step": 10898
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.689709972189114e-07,
+      "loss": 1.4307,
+      "step": 10899
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.689312673818038e-07,
+      "loss": 1.6097,
+      "step": 10900
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.688915375446959e-07,
+      "loss": 1.5564,
+      "step": 10901
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.688518077075883e-07,
+      "loss": 1.7383,
+      "step": 10902
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.688120778704807e-07,
+      "loss": 1.5504,
+      "step": 10903
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.68772348033373e-07,
+      "loss": 1.8116,
+      "step": 10904
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.687326181962654e-07,
+      "loss": 1.9019,
+      "step": 10905
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.686928883591577e-07,
+      "loss": 1.7036,
+      "step": 10906
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.6865315852205e-07,
+      "loss": 1.3725,
+      "step": 10907
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.686134286849423e-07,
+      "loss": 1.4914,
+      "step": 10908
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.685736988478347e-07,
+      "loss": 2.2964,
+      "step": 10909
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.685339690107271e-07,
+      "loss": 1.5272,
+      "step": 10910
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.684942391736194e-07,
+      "loss": 1.9114,
+      "step": 10911
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.684545093365118e-07,
+      "loss": 1.7855,
+      "step": 10912
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.68414779499404e-07,
+      "loss": 1.5535,
+      "step": 10913
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.683750496622963e-07,
+      "loss": 1.4498,
+      "step": 10914
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.683353198251886e-07,
+      "loss": 1.4686,
+      "step": 10915
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.68295589988081e-07,
+      "loss": 1.5134,
+      "step": 10916
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.682558601509734e-07,
+      "loss": 1.8242,
+      "step": 10917
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.682161303138657e-07,
+      "loss": 1.375,
+      "step": 10918
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.68176400476758e-07,
+      "loss": 1.4822,
+      "step": 10919
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.681366706396503e-07,
+      "loss": 1.7258,
+      "step": 10920
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.680969408025427e-07,
+      "loss": 1.8196,
+      "step": 10921
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.68057210965435e-07,
+      "loss": 1.7979,
+      "step": 10922
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.680174811283274e-07,
+      "loss": 1.7931,
+      "step": 10923
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.679777512912197e-07,
+      "loss": 1.6696,
+      "step": 10924
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.67938021454112e-07,
+      "loss": 1.9305,
+      "step": 10925
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.678982916170044e-07,
+      "loss": 2.0026,
+      "step": 10926
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.678585617798966e-07,
+      "loss": 1.5506,
+      "step": 10927
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.67818831942789e-07,
+      "loss": 1.3182,
+      "step": 10928
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.677791021056813e-07,
+      "loss": 1.5906,
+      "step": 10929
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.677393722685737e-07,
+      "loss": 1.9676,
+      "step": 10930
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.676996424314659e-07,
+      "loss": 1.3468,
+      "step": 10931
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.676599125943583e-07,
+      "loss": 1.4053,
+      "step": 10932
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.676201827572507e-07,
+      "loss": 1.4426,
+      "step": 10933
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.67580452920143e-07,
+      "loss": 1.1751,
+      "step": 10934
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.675407230830354e-07,
+      "loss": 1.4249,
+      "step": 10935
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.675009932459277e-07,
+      "loss": 1.4435,
+      "step": 10936
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.6746126340882e-07,
+      "loss": 1.8771,
+      "step": 10937
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.674215335717123e-07,
+      "loss": 1.7007,
+      "step": 10938
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.673818037346047e-07,
+      "loss": 1.6523,
+      "step": 10939
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.673420738974969e-07,
+      "loss": 1.6475,
+      "step": 10940
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.673023440603893e-07,
+      "loss": 2.0377,
+      "step": 10941
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.672626142232817e-07,
+      "loss": 1.5675,
+      "step": 10942
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.672228843861739e-07,
+      "loss": 1.8174,
+      "step": 10943
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.671831545490663e-07,
+      "loss": 1.6567,
+      "step": 10944
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.671434247119586e-07,
+      "loss": 1.6972,
+      "step": 10945
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.67103694874851e-07,
+      "loss": 1.9475,
+      "step": 10946
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.670639650377433e-07,
+      "loss": 1.6516,
+      "step": 10947
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.670242352006357e-07,
+      "loss": 1.5532,
+      "step": 10948
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.66984505363528e-07,
+      "loss": 1.764,
+      "step": 10949
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.669447755264203e-07,
+      "loss": 1.8206,
+      "step": 10950
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.669050456893127e-07,
+      "loss": 1.5368,
+      "step": 10951
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.66865315852205e-07,
+      "loss": 1.9269,
+      "step": 10952
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.668255860150974e-07,
+      "loss": 1.0525,
+      "step": 10953
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.667858561779896e-07,
+      "loss": 1.4878,
+      "step": 10954
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.667461263408819e-07,
+      "loss": 2.0009,
+      "step": 10955
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.667063965037742e-07,
+      "loss": 1.3249,
+      "step": 10956
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.666666666666666e-07,
+      "loss": 1.1763,
+      "step": 10957
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.66626936829559e-07,
+      "loss": 1.6642,
+      "step": 10958
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.665872069924513e-07,
+      "loss": 1.6968,
+      "step": 10959
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.665474771553437e-07,
+      "loss": 1.5528,
+      "step": 10960
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.665077473182359e-07,
+      "loss": 1.4867,
+      "step": 10961
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.664680174811283e-07,
+      "loss": 1.4306,
+      "step": 10962
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.664282876440206e-07,
+      "loss": 1.3172,
+      "step": 10963
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.66388557806913e-07,
+      "loss": 1.3593,
+      "step": 10964
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.663488279698054e-07,
+      "loss": 1.3388,
+      "step": 10965
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.663090981326977e-07,
+      "loss": 1.6845,
+      "step": 10966
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.662693682955899e-07,
+      "loss": 1.4021,
+      "step": 10967
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.662296384584822e-07,
+      "loss": 1.6609,
+      "step": 10968
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.661899086213746e-07,
+      "loss": 2.2285,
+      "step": 10969
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.661501787842669e-07,
+      "loss": 1.6113,
+      "step": 10970
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.661104489471593e-07,
+      "loss": 1.9328,
+      "step": 10971
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.660707191100517e-07,
+      "loss": 1.6051,
+      "step": 10972
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.660309892729439e-07,
+      "loss": 1.2937,
+      "step": 10973
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.659912594358363e-07,
+      "loss": 2.0265,
+      "step": 10974
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.659515295987286e-07,
+      "loss": 1.9855,
+      "step": 10975
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.65911799761621e-07,
+      "loss": 1.4024,
+      "step": 10976
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.658720699245133e-07,
+      "loss": 1.7833,
+      "step": 10977
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.658323400874057e-07,
+      "loss": 1.8809,
+      "step": 10978
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.65792610250298e-07,
+      "loss": 1.9782,
+      "step": 10979
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.657528804131902e-07,
+      "loss": 1.3948,
+      "step": 10980
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.657131505760826e-07,
+      "loss": 1.7355,
+      "step": 10981
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.656734207389749e-07,
+      "loss": 1.5065,
+      "step": 10982
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.656336909018673e-07,
+      "loss": 1.7687,
+      "step": 10983
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.655939610647596e-07,
+      "loss": 1.9798,
+      "step": 10984
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.655542312276519e-07,
+      "loss": 1.4982,
+      "step": 10985
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.655145013905442e-07,
+      "loss": 1.6144,
+      "step": 10986
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.654747715534366e-07,
+      "loss": 1.3085,
+      "step": 10987
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.65435041716329e-07,
+      "loss": 1.4491,
+      "step": 10988
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.653953118792213e-07,
+      "loss": 1.7651,
+      "step": 10989
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.653555820421137e-07,
+      "loss": 1.6861,
+      "step": 10990
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.653158522050059e-07,
+      "loss": 1.7587,
+      "step": 10991
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.652761223678983e-07,
+      "loss": 1.4511,
+      "step": 10992
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.652363925307905e-07,
+      "loss": 1.8743,
+      "step": 10993
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.651966626936829e-07,
+      "loss": 1.6241,
+      "step": 10994
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.651569328565753e-07,
+      "loss": 1.6076,
+      "step": 10995
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.651172030194676e-07,
+      "loss": 1.4144,
+      "step": 10996
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.650774731823599e-07,
+      "loss": 1.3071,
+      "step": 10997
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.650377433452522e-07,
+      "loss": 1.7062,
+      "step": 10998
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.649980135081446e-07,
+      "loss": 1.2605,
+      "step": 10999
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.649582836710369e-07,
+      "loss": 1.3728,
+      "step": 11000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.649185538339293e-07,
+      "loss": 1.7584,
+      "step": 11001
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.648788239968216e-07,
+      "loss": 1.8641,
+      "step": 11002
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.648390941597139e-07,
+      "loss": 2.0414,
+      "step": 11003
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.647993643226063e-07,
+      "loss": 1.3702,
+      "step": 11004
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.647596344854986e-07,
+      "loss": 1.6415,
+      "step": 11005
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.647199046483909e-07,
+      "loss": 1.8041,
+      "step": 11006
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.646801748112832e-07,
+      "loss": 1.8888,
+      "step": 11007
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.646404449741756e-07,
+      "loss": 1.7319,
+      "step": 11008
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.646007151370678e-07,
+      "loss": 1.5799,
+      "step": 11009
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.645609852999602e-07,
+      "loss": 1.6863,
+      "step": 11010
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.645212554628526e-07,
+      "loss": 1.5819,
+      "step": 11011
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.644815256257449e-07,
+      "loss": 1.7641,
+      "step": 11012
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.644417957886373e-07,
+      "loss": 1.5113,
+      "step": 11013
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.644020659515296e-07,
+      "loss": 1.6628,
+      "step": 11014
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.643623361144219e-07,
+      "loss": 1.5186,
+      "step": 11015
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.643226062773142e-07,
+      "loss": 1.1942,
+      "step": 11016
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.642828764402066e-07,
+      "loss": 1.3931,
+      "step": 11017
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.64243146603099e-07,
+      "loss": 1.375,
+      "step": 11018
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.642034167659913e-07,
+      "loss": 1.9614,
+      "step": 11019
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.641636869288836e-07,
+      "loss": 1.437,
+      "step": 11020
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.641239570917758e-07,
+      "loss": 1.7672,
+      "step": 11021
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.640842272546682e-07,
+      "loss": 1.9057,
+      "step": 11022
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.640444974175605e-07,
+      "loss": 1.6742,
+      "step": 11023
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.640047675804529e-07,
+      "loss": 1.499,
+      "step": 11024
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.639650377433452e-07,
+      "loss": 1.5349,
+      "step": 11025
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.639253079062376e-07,
+      "loss": 1.5786,
+      "step": 11026
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.638855780691299e-07,
+      "loss": 1.4914,
+      "step": 11027
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.638458482320222e-07,
+      "loss": 1.5441,
+      "step": 11028
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.638061183949146e-07,
+      "loss": 0.9791,
+      "step": 11029
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.637663885578069e-07,
+      "loss": 1.8847,
+      "step": 11030
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.637266587206993e-07,
+      "loss": 1.5552,
+      "step": 11031
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.636869288835916e-07,
+      "loss": 1.6869,
+      "step": 11032
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.636471990464838e-07,
+      "loss": 1.4932,
+      "step": 11033
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.636074692093761e-07,
+      "loss": 1.943,
+      "step": 11034
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.635677393722685e-07,
+      "loss": 0.9796,
+      "step": 11035
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.635280095351609e-07,
+      "loss": 1.3704,
+      "step": 11036
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.634882796980532e-07,
+      "loss": 1.4556,
+      "step": 11037
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.634485498609456e-07,
+      "loss": 1.6652,
+      "step": 11038
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.634088200238378e-07,
+      "loss": 1.7443,
+      "step": 11039
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.633690901867302e-07,
+      "loss": 2.038,
+      "step": 11040
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.633293603496225e-07,
+      "loss": 1.6644,
+      "step": 11041
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.632896305125149e-07,
+      "loss": 1.4284,
+      "step": 11042
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.632499006754073e-07,
+      "loss": 1.5387,
+      "step": 11043
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.632101708382996e-07,
+      "loss": 2.2967,
+      "step": 11044
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.631704410011919e-07,
+      "loss": 1.241,
+      "step": 11045
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.631307111640841e-07,
+      "loss": 1.6162,
+      "step": 11046
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.630909813269765e-07,
+      "loss": 1.7214,
+      "step": 11047
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.630512514898688e-07,
+      "loss": 1.6171,
+      "step": 11048
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.630115216527612e-07,
+      "loss": 1.7247,
+      "step": 11049
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.629717918156536e-07,
+      "loss": 1.8063,
+      "step": 11050
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.629320619785458e-07,
+      "loss": 1.513,
+      "step": 11051
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.628923321414382e-07,
+      "loss": 1.598,
+      "step": 11052
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.628526023043305e-07,
+      "loss": 1.5886,
+      "step": 11053
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.628128724672229e-07,
+      "loss": 1.4511,
+      "step": 11054
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.627731426301152e-07,
+      "loss": 1.6294,
+      "step": 11055
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.627334127930076e-07,
+      "loss": 1.625,
+      "step": 11056
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.626936829558998e-07,
+      "loss": 1.4519,
+      "step": 11057
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.626539531187922e-07,
+      "loss": 1.179,
+      "step": 11058
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.626142232816845e-07,
+      "loss": 1.4041,
+      "step": 11059
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.625744934445768e-07,
+      "loss": 1.3485,
+      "step": 11060
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.625347636074692e-07,
+      "loss": 1.542,
+      "step": 11061
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.624950337703615e-07,
+      "loss": 1.7345,
+      "step": 11062
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.624553039332538e-07,
+      "loss": 2.0365,
+      "step": 11063
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.624155740961461e-07,
+      "loss": 1.7011,
+      "step": 11064
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.623758442590385e-07,
+      "loss": 1.7657,
+      "step": 11065
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.623361144219309e-07,
+      "loss": 1.656,
+      "step": 11066
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.622963845848232e-07,
+      "loss": 1.7924,
+      "step": 11067
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.622566547477156e-07,
+      "loss": 1.5298,
+      "step": 11068
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.622169249106078e-07,
+      "loss": 1.6754,
+      "step": 11069
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.621771950735002e-07,
+      "loss": 1.5566,
+      "step": 11070
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.621374652363925e-07,
+      "loss": 1.9754,
+      "step": 11071
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.620977353992848e-07,
+      "loss": 1.3668,
+      "step": 11072
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.620580055621772e-07,
+      "loss": 1.6484,
+      "step": 11073
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.620182757250695e-07,
+      "loss": 1.488,
+      "step": 11074
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.619785458879618e-07,
+      "loss": 2.1262,
+      "step": 11075
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.619388160508541e-07,
+      "loss": 1.8501,
+      "step": 11076
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.618990862137465e-07,
+      "loss": 1.6227,
+      "step": 11077
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.618593563766388e-07,
+      "loss": 1.5422,
+      "step": 11078
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.618196265395312e-07,
+      "loss": 1.2766,
+      "step": 11079
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.617798967024236e-07,
+      "loss": 1.4433,
+      "step": 11080
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.617401668653158e-07,
+      "loss": 1.4581,
+      "step": 11081
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.617004370282082e-07,
+      "loss": 1.9866,
+      "step": 11082
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.616607071911005e-07,
+      "loss": 1.9585,
+      "step": 11083
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.616209773539929e-07,
+      "loss": 2.0074,
+      "step": 11084
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.615812475168852e-07,
+      "loss": 1.5896,
+      "step": 11085
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.615415176797775e-07,
+      "loss": 1.9535,
+      "step": 11086
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.615017878426697e-07,
+      "loss": 1.5344,
+      "step": 11087
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.614620580055621e-07,
+      "loss": 1.6665,
+      "step": 11088
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.614223281684545e-07,
+      "loss": 1.501,
+      "step": 11089
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.613825983313468e-07,
+      "loss": 1.7192,
+      "step": 11090
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.613428684942392e-07,
+      "loss": 1.6379,
+      "step": 11091
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.613031386571315e-07,
+      "loss": 1.6244,
+      "step": 11092
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.612634088200238e-07,
+      "loss": 1.9759,
+      "step": 11093
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.612236789829161e-07,
+      "loss": 1.5616,
+      "step": 11094
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.611839491458085e-07,
+      "loss": 1.1459,
+      "step": 11095
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.611442193087009e-07,
+      "loss": 1.9244,
+      "step": 11096
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.611044894715932e-07,
+      "loss": 1.4969,
+      "step": 11097
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.610647596344856e-07,
+      "loss": 1.7399,
+      "step": 11098
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.610250297973777e-07,
+      "loss": 1.8199,
+      "step": 11099
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.609852999602701e-07,
+      "loss": 1.7403,
+      "step": 11100
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.609455701231624e-07,
+      "loss": 1.6915,
+      "step": 11101
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.609058402860548e-07,
+      "loss": 1.7055,
+      "step": 11102
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.608661104489471e-07,
+      "loss": 1.8178,
+      "step": 11103
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.608263806118395e-07,
+      "loss": 1.6133,
+      "step": 11104
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.607866507747318e-07,
+      "loss": 1.4518,
+      "step": 11105
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.607469209376241e-07,
+      "loss": 2.2171,
+      "step": 11106
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.607071911005165e-07,
+      "loss": 1.3732,
+      "step": 11107
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.606674612634088e-07,
+      "loss": 1.5682,
+      "step": 11108
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.606277314263012e-07,
+      "loss": 1.2586,
+      "step": 11109
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.605880015891935e-07,
+      "loss": 1.3345,
+      "step": 11110
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.605482717520858e-07,
+      "loss": 1.4756,
+      "step": 11111
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.60508541914978e-07,
+      "loss": 1.8367,
+      "step": 11112
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.604688120778704e-07,
+      "loss": 1.3765,
+      "step": 11113
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.604290822407628e-07,
+      "loss": 1.8009,
+      "step": 11114
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.603893524036551e-07,
+      "loss": 1.7881,
+      "step": 11115
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.603496225665475e-07,
+      "loss": 1.8601,
+      "step": 11116
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.603098927294397e-07,
+      "loss": 1.7415,
+      "step": 11117
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.602701628923321e-07,
+      "loss": 0.9483,
+      "step": 11118
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.602304330552244e-07,
+      "loss": 1.4316,
+      "step": 11119
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.601907032181168e-07,
+      "loss": 1.3331,
+      "step": 11120
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.601509733810092e-07,
+      "loss": 1.2011,
+      "step": 11121
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.601112435439015e-07,
+      "loss": 1.6011,
+      "step": 11122
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.600715137067938e-07,
+      "loss": 1.5263,
+      "step": 11123
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.600317838696861e-07,
+      "loss": 1.9122,
+      "step": 11124
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.599920540325784e-07,
+      "loss": 1.6842,
+      "step": 11125
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.599523241954707e-07,
+      "loss": 1.6502,
+      "step": 11126
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.599125943583631e-07,
+      "loss": 1.5532,
+      "step": 11127
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.598728645212555e-07,
+      "loss": 1.5569,
+      "step": 11128
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.598331346841477e-07,
+      "loss": 1.7021,
+      "step": 11129
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.597934048470401e-07,
+      "loss": 1.8229,
+      "step": 11130
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.597536750099324e-07,
+      "loss": 1.7532,
+      "step": 11131
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.597139451728248e-07,
+      "loss": 1.2908,
+      "step": 11132
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.596742153357171e-07,
+      "loss": 1.5497,
+      "step": 11133
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.596344854986095e-07,
+      "loss": 1.5367,
+      "step": 11134
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.595947556615017e-07,
+      "loss": 1.5079,
+      "step": 11135
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.595550258243941e-07,
+      "loss": 1.9283,
+      "step": 11136
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.595152959872865e-07,
+      "loss": 1.9276,
+      "step": 11137
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.594755661501787e-07,
+      "loss": 1.5053,
+      "step": 11138
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.594358363130711e-07,
+      "loss": 1.5254,
+      "step": 11139
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.593961064759634e-07,
+      "loss": 1.3926,
+      "step": 11140
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.593563766388557e-07,
+      "loss": 1.4933,
+      "step": 11141
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.59316646801748e-07,
+      "loss": 1.9428,
+      "step": 11142
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.592769169646404e-07,
+      "loss": 1.5576,
+      "step": 11143
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.592371871275328e-07,
+      "loss": 1.6354,
+      "step": 11144
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.591974572904251e-07,
+      "loss": 1.3849,
+      "step": 11145
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.591577274533175e-07,
+      "loss": 1.487,
+      "step": 11146
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.591179976162097e-07,
+      "loss": 1.5314,
+      "step": 11147
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.590782677791021e-07,
+      "loss": 1.5405,
+      "step": 11148
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.590385379419944e-07,
+      "loss": 1.9158,
+      "step": 11149
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.589988081048868e-07,
+      "loss": 1.8498,
+      "step": 11150
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.589590782677792e-07,
+      "loss": 1.1707,
+      "step": 11151
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.589193484306714e-07,
+      "loss": 1.9285,
+      "step": 11152
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.588796185935637e-07,
+      "loss": 1.6579,
+      "step": 11153
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.58839888756456e-07,
+      "loss": 1.7334,
+      "step": 11154
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.588001589193484e-07,
+      "loss": 1.3809,
+      "step": 11155
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.587604290822407e-07,
+      "loss": 1.3597,
+      "step": 11156
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.587206992451331e-07,
+      "loss": 1.2335,
+      "step": 11157
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.586809694080255e-07,
+      "loss": 0.974,
+      "step": 11158
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.586412395709177e-07,
+      "loss": 1.6821,
+      "step": 11159
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.586015097338101e-07,
+      "loss": 1.8424,
+      "step": 11160
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.585617798967024e-07,
+      "loss": 1.4571,
+      "step": 11161
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.585220500595948e-07,
+      "loss": 1.8576,
+      "step": 11162
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.584823202224871e-07,
+      "loss": 2.0081,
+      "step": 11163
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.584425903853795e-07,
+      "loss": 1.9196,
+      "step": 11164
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.584028605482716e-07,
+      "loss": 1.6073,
+      "step": 11165
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.58363130711164e-07,
+      "loss": 1.3738,
+      "step": 11166
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.583234008740564e-07,
+      "loss": 1.422,
+      "step": 11167
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.582836710369487e-07,
+      "loss": 1.5875,
+      "step": 11168
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.582439411998411e-07,
+      "loss": 1.6285,
+      "step": 11169
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.582042113627334e-07,
+      "loss": 1.6703,
+      "step": 11170
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.581644815256257e-07,
+      "loss": 1.4278,
+      "step": 11171
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.58124751688518e-07,
+      "loss": 1.109,
+      "step": 11172
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.580850218514104e-07,
+      "loss": 1.8959,
+      "step": 11173
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.580452920143028e-07,
+      "loss": 1.5679,
+      "step": 11174
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.580055621771951e-07,
+      "loss": 1.8614,
+      "step": 11175
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.579658323400875e-07,
+      "loss": 1.6833,
+      "step": 11176
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.579261025029797e-07,
+      "loss": 1.3946,
+      "step": 11177
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.57886372665872e-07,
+      "loss": 1.5118,
+      "step": 11178
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.578466428287643e-07,
+      "loss": 2.0769,
+      "step": 11179
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.578069129916567e-07,
+      "loss": 1.6698,
+      "step": 11180
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.57767183154549e-07,
+      "loss": 1.5888,
+      "step": 11181
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.577274533174414e-07,
+      "loss": 1.3548,
+      "step": 11182
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.576877234803337e-07,
+      "loss": 1.1722,
+      "step": 11183
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.57647993643226e-07,
+      "loss": 1.5496,
+      "step": 11184
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.576082638061184e-07,
+      "loss": 1.3875,
+      "step": 11185
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.575685339690107e-07,
+      "loss": 1.2646,
+      "step": 11186
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.575288041319031e-07,
+      "loss": 1.7548,
+      "step": 11187
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.574890742947954e-07,
+      "loss": 1.4513,
+      "step": 11188
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.574493444576877e-07,
+      "loss": 1.751,
+      "step": 11189
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.574096146205801e-07,
+      "loss": 1.7109,
+      "step": 11190
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.573698847834723e-07,
+      "loss": 1.8573,
+      "step": 11191
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.573301549463647e-07,
+      "loss": 1.7646,
+      "step": 11192
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.57290425109257e-07,
+      "loss": 2.0152,
+      "step": 11193
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.572506952721494e-07,
+      "loss": 1.5266,
+      "step": 11194
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.572109654350416e-07,
+      "loss": 1.2601,
+      "step": 11195
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.57171235597934e-07,
+      "loss": 1.4164,
+      "step": 11196
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.571315057608263e-07,
+      "loss": 1.3598,
+      "step": 11197
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.570917759237187e-07,
+      "loss": 1.4538,
+      "step": 11198
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.570520460866111e-07,
+      "loss": 1.9254,
+      "step": 11199
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.570123162495034e-07,
+      "loss": 1.8265,
+      "step": 11200
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.569725864123957e-07,
+      "loss": 1.6848,
+      "step": 11201
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.56932856575288e-07,
+      "loss": 1.4649,
+      "step": 11202
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.568931267381804e-07,
+      "loss": 1.5074,
+      "step": 11203
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.568533969010726e-07,
+      "loss": 1.8092,
+      "step": 11204
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.56813667063965e-07,
+      "loss": 1.4424,
+      "step": 11205
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.567739372268574e-07,
+      "loss": 1.3375,
+      "step": 11206
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.567342073897496e-07,
+      "loss": 1.5965,
+      "step": 11207
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.56694477552642e-07,
+      "loss": 1.5954,
+      "step": 11208
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.566547477155343e-07,
+      "loss": 1.6609,
+      "step": 11209
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.566150178784267e-07,
+      "loss": 1.9926,
+      "step": 11210
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.56575288041319e-07,
+      "loss": 1.1825,
+      "step": 11211
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.565355582042114e-07,
+      "loss": 1.3866,
+      "step": 11212
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.564958283671036e-07,
+      "loss": 1.7611,
+      "step": 11213
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.56456098529996e-07,
+      "loss": 1.5491,
+      "step": 11214
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.564163686928884e-07,
+      "loss": 1.4757,
+      "step": 11215
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.563766388557807e-07,
+      "loss": 2.1385,
+      "step": 11216
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.563369090186731e-07,
+      "loss": 1.2769,
+      "step": 11217
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.562971791815653e-07,
+      "loss": 1.3523,
+      "step": 11218
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.562574493444576e-07,
+      "loss": 1.8229,
+      "step": 11219
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.562177195073499e-07,
+      "loss": 1.7936,
+      "step": 11220
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.561779896702423e-07,
+      "loss": 1.4704,
+      "step": 11221
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.561382598331347e-07,
+      "loss": 1.6054,
+      "step": 11222
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.56098529996027e-07,
+      "loss": 1.797,
+      "step": 11223
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.560588001589194e-07,
+      "loss": 1.4095,
+      "step": 11224
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.560190703218116e-07,
+      "loss": 1.7589,
+      "step": 11225
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.55979340484704e-07,
+      "loss": 1.774,
+      "step": 11226
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.559396106475963e-07,
+      "loss": 1.7201,
+      "step": 11227
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.558998808104887e-07,
+      "loss": 1.307,
+      "step": 11228
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.558601509733811e-07,
+      "loss": 1.4544,
+      "step": 11229
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.558204211362734e-07,
+      "loss": 1.4518,
+      "step": 11230
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.557806912991656e-07,
+      "loss": 2.201,
+      "step": 11231
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.557409614620579e-07,
+      "loss": 1.7422,
+      "step": 11232
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.557012316249503e-07,
+      "loss": 1.7551,
+      "step": 11233
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.556615017878426e-07,
+      "loss": 1.3376,
+      "step": 11234
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.55621771950735e-07,
+      "loss": 1.9173,
+      "step": 11235
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.555820421136274e-07,
+      "loss": 1.5289,
+      "step": 11236
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.555423122765196e-07,
+      "loss": 1.6871,
+      "step": 11237
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.55502582439412e-07,
+      "loss": 1.6564,
+      "step": 11238
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.554628526023043e-07,
+      "loss": 2.045,
+      "step": 11239
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.554231227651967e-07,
+      "loss": 1.6297,
+      "step": 11240
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.55383392928089e-07,
+      "loss": 1.6842,
+      "step": 11241
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.553436630909814e-07,
+      "loss": 1.7364,
+      "step": 11242
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.553039332538736e-07,
+      "loss": 1.9727,
+      "step": 11243
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.552642034167659e-07,
+      "loss": 1.3699,
+      "step": 11244
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.552244735796583e-07,
+      "loss": 1.6396,
+      "step": 11245
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.551847437425506e-07,
+      "loss": 1.8434,
+      "step": 11246
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.55145013905443e-07,
+      "loss": 1.162,
+      "step": 11247
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.551052840683353e-07,
+      "loss": 1.5115,
+      "step": 11248
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.550655542312276e-07,
+      "loss": 1.7842,
+      "step": 11249
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.550258243941199e-07,
+      "loss": 1.7994,
+      "step": 11250
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.549860945570123e-07,
+      "loss": 1.5659,
+      "step": 11251
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.549463647199047e-07,
+      "loss": 1.3705,
+      "step": 11252
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.54906634882797e-07,
+      "loss": 2.198,
+      "step": 11253
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.548669050456894e-07,
+      "loss": 1.6948,
+      "step": 11254
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.548271752085816e-07,
+      "loss": 1.4491,
+      "step": 11255
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.54787445371474e-07,
+      "loss": 1.4538,
+      "step": 11256
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.547477155343662e-07,
+      "loss": 1.374,
+      "step": 11257
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.547079856972586e-07,
+      "loss": 1.4609,
+      "step": 11258
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.546682558601509e-07,
+      "loss": 1.4243,
+      "step": 11259
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.546285260230433e-07,
+      "loss": 1.3356,
+      "step": 11260
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.545887961859356e-07,
+      "loss": 1.5411,
+      "step": 11261
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.545490663488279e-07,
+      "loss": 1.675,
+      "step": 11262
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.545093365117203e-07,
+      "loss": 1.5174,
+      "step": 11263
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.544696066746126e-07,
+      "loss": 1.3114,
+      "step": 11264
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.54429876837505e-07,
+      "loss": 1.6944,
+      "step": 11265
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.543901470003973e-07,
+      "loss": 1.6701,
+      "step": 11266
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.543504171632896e-07,
+      "loss": 1.4692,
+      "step": 11267
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.54310687326182e-07,
+      "loss": 1.1857,
+      "step": 11268
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.542709574890743e-07,
+      "loss": 1.6776,
+      "step": 11269
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.542312276519666e-07,
+      "loss": 1.7782,
+      "step": 11270
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.541914978148589e-07,
+      "loss": 1.483,
+      "step": 11271
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.541517679777513e-07,
+      "loss": 1.2993,
+      "step": 11272
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.541120381406435e-07,
+      "loss": 1.6029,
+      "step": 11273
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.540723083035359e-07,
+      "loss": 1.2624,
+      "step": 11274
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.540325784664282e-07,
+      "loss": 1.3917,
+      "step": 11275
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.539928486293206e-07,
+      "loss": 1.6385,
+      "step": 11276
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.53953118792213e-07,
+      "loss": 1.8547,
+      "step": 11277
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.539133889551053e-07,
+      "loss": 1.4302,
+      "step": 11278
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.538736591179976e-07,
+      "loss": 1.8431,
+      "step": 11279
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.538339292808899e-07,
+      "loss": 1.3647,
+      "step": 11280
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.537941994437823e-07,
+      "loss": 1.6103,
+      "step": 11281
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.537544696066746e-07,
+      "loss": 1.9227,
+      "step": 11282
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.53714739769567e-07,
+      "loss": 1.7106,
+      "step": 11283
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.536750099324593e-07,
+      "loss": 1.5303,
+      "step": 11284
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.536352800953515e-07,
+      "loss": 1.7935,
+      "step": 11285
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.535955502582439e-07,
+      "loss": 1.9981,
+      "step": 11286
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.535558204211362e-07,
+      "loss": 1.6748,
+      "step": 11287
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.535160905840286e-07,
+      "loss": 1.7445,
+      "step": 11288
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.534763607469209e-07,
+      "loss": 1.8643,
+      "step": 11289
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.534366309098133e-07,
+      "loss": 1.4782,
+      "step": 11290
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.533969010727055e-07,
+      "loss": 1.6855,
+      "step": 11291
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.533571712355979e-07,
+      "loss": 1.1368,
+      "step": 11292
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.533174413984903e-07,
+      "loss": 1.4136,
+      "step": 11293
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.532777115613826e-07,
+      "loss": 1.9979,
+      "step": 11294
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.53237981724275e-07,
+      "loss": 1.7506,
+      "step": 11295
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.531982518871673e-07,
+      "loss": 1.2688,
+      "step": 11296
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.531585220500595e-07,
+      "loss": 1.562,
+      "step": 11297
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.531187922129518e-07,
+      "loss": 1.3519,
+      "step": 11298
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.530790623758442e-07,
+      "loss": 1.2487,
+      "step": 11299
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.530393325387366e-07,
+      "loss": 1.773,
+      "step": 11300
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.529996027016289e-07,
+      "loss": 1.4903,
+      "step": 11301
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.529598728645213e-07,
+      "loss": 1.8483,
+      "step": 11302
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.529201430274135e-07,
+      "loss": 1.552,
+      "step": 11303
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.528804131903059e-07,
+      "loss": 1.3026,
+      "step": 11304
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.528406833531982e-07,
+      "loss": 1.3433,
+      "step": 11305
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.528009535160906e-07,
+      "loss": 1.4324,
+      "step": 11306
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.52761223678983e-07,
+      "loss": 1.6335,
+      "step": 11307
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.527214938418753e-07,
+      "loss": 1.7067,
+      "step": 11308
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.526817640047675e-07,
+      "loss": 2.0931,
+      "step": 11309
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.526420341676598e-07,
+      "loss": 1.7375,
+      "step": 11310
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.526023043305522e-07,
+      "loss": 1.4968,
+      "step": 11311
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.525625744934445e-07,
+      "loss": 1.9459,
+      "step": 11312
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.525228446563369e-07,
+      "loss": 1.9199,
+      "step": 11313
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.524831148192293e-07,
+      "loss": 1.7654,
+      "step": 11314
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.524433849821215e-07,
+      "loss": 2.034,
+      "step": 11315
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.524036551450139e-07,
+      "loss": 1.9008,
+      "step": 11316
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.523639253079062e-07,
+      "loss": 1.2611,
+      "step": 11317
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.523241954707986e-07,
+      "loss": 1.528,
+      "step": 11318
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.522844656336909e-07,
+      "loss": 1.355,
+      "step": 11319
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.522447357965833e-07,
+      "loss": 2.0431,
+      "step": 11320
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.522050059594755e-07,
+      "loss": 1.5033,
+      "step": 11321
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.521652761223679e-07,
+      "loss": 1.4136,
+      "step": 11322
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.521255462852602e-07,
+      "loss": 1.5247,
+      "step": 11323
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.520858164481525e-07,
+      "loss": 2.1938,
+      "step": 11324
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.520460866110449e-07,
+      "loss": 1.3638,
+      "step": 11325
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.520063567739372e-07,
+      "loss": 1.8269,
+      "step": 11326
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.519666269368295e-07,
+      "loss": 1.7634,
+      "step": 11327
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.519268970997218e-07,
+      "loss": 1.4591,
+      "step": 11328
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.518871672626142e-07,
+      "loss": 1.6079,
+      "step": 11329
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.518474374255066e-07,
+      "loss": 1.5931,
+      "step": 11330
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.518077075883989e-07,
+      "loss": 1.8608,
+      "step": 11331
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.517679777512913e-07,
+      "loss": 1.6082,
+      "step": 11332
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.517282479141835e-07,
+      "loss": 1.819,
+      "step": 11333
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.516885180770759e-07,
+      "loss": 1.4592,
+      "step": 11334
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.516487882399682e-07,
+      "loss": 1.9485,
+      "step": 11335
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.516090584028605e-07,
+      "loss": 1.7282,
+      "step": 11336
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.515693285657528e-07,
+      "loss": 1.7376,
+      "step": 11337
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.515295987286452e-07,
+      "loss": 1.9597,
+      "step": 11338
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.514898688915375e-07,
+      "loss": 1.5209,
+      "step": 11339
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.514501390544298e-07,
+      "loss": 2.0333,
+      "step": 11340
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.514104092173222e-07,
+      "loss": 1.62,
+      "step": 11341
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.513706793802145e-07,
+      "loss": 1.3091,
+      "step": 11342
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.513309495431069e-07,
+      "loss": 1.7867,
+      "step": 11343
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.512912197059992e-07,
+      "loss": 1.9294,
+      "step": 11344
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.512514898688915e-07,
+      "loss": 1.0979,
+      "step": 11345
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.512117600317839e-07,
+      "loss": 1.4032,
+      "step": 11346
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.511720301946762e-07,
+      "loss": 1.7871,
+      "step": 11347
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.511323003575686e-07,
+      "loss": 1.633,
+      "step": 11348
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.510925705204608e-07,
+      "loss": 1.5417,
+      "step": 11349
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.35804916732751785,
+      "eval_loss": 1.7287479639053345,
+      "eval_runtime": 6.4514,
+      "eval_samples_per_second": 390.92,
+      "eval_steps_per_second": 6.2,
+      "step": 11349
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.510528406833532e-07,
+      "loss": 1.6555,
+      "step": 11350
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.510131108462454e-07,
+      "loss": 1.7126,
+      "step": 11351
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.509733810091378e-07,
+      "loss": 1.5168,
+      "step": 11352
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.509336511720301e-07,
+      "loss": 1.1636,
+      "step": 11353
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.508939213349225e-07,
+      "loss": 1.7347,
+      "step": 11354
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.508541914978149e-07,
+      "loss": 1.9785,
+      "step": 11355
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.508144616607072e-07,
+      "loss": 1.4322,
+      "step": 11356
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.507747318235995e-07,
+      "loss": 1.1581,
+      "step": 11357
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.507350019864918e-07,
+      "loss": 1.2983,
+      "step": 11358
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.506952721493842e-07,
+      "loss": 1.2724,
+      "step": 11359
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.506555423122765e-07,
+      "loss": 1.6866,
+      "step": 11360
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.506158124751689e-07,
+      "loss": 1.3122,
+      "step": 11361
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.505760826380613e-07,
+      "loss": 1.6965,
+      "step": 11362
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.505363528009534e-07,
+      "loss": 1.5912,
+      "step": 11363
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.504966229638458e-07,
+      "loss": 1.5733,
+      "step": 11364
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.504568931267381e-07,
+      "loss": 1.5113,
+      "step": 11365
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.504171632896305e-07,
+      "loss": 1.8662,
+      "step": 11366
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.503774334525228e-07,
+      "loss": 1.3816,
+      "step": 11367
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.503377036154152e-07,
+      "loss": 1.5845,
+      "step": 11368
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.502979737783075e-07,
+      "loss": 1.66,
+      "step": 11369
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.502582439411998e-07,
+      "loss": 1.9815,
+      "step": 11370
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.502185141040922e-07,
+      "loss": 1.4716,
+      "step": 11371
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.501787842669845e-07,
+      "loss": 1.7452,
+      "step": 11372
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.501390544298769e-07,
+      "loss": 2.0569,
+      "step": 11373
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.500993245927692e-07,
+      "loss": 1.1676,
+      "step": 11374
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.500595947556614e-07,
+      "loss": 2.26,
+      "step": 11375
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.500198649185537e-07,
+      "loss": 1.1697,
+      "step": 11376
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.499801350814461e-07,
+      "loss": 1.3667,
+      "step": 11377
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.499404052443385e-07,
+      "loss": 1.977,
+      "step": 11378
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.499006754072308e-07,
+      "loss": 2.0119,
+      "step": 11379
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.498609455701232e-07,
+      "loss": 1.2592,
+      "step": 11380
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.498212157330154e-07,
+      "loss": 2.2275,
+      "step": 11381
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.497814858959078e-07,
+      "loss": 1.5571,
+      "step": 11382
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.497417560588001e-07,
+      "loss": 1.3706,
+      "step": 11383
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.497020262216925e-07,
+      "loss": 1.4725,
+      "step": 11384
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.496622963845849e-07,
+      "loss": 1.5602,
+      "step": 11385
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.496225665474772e-07,
+      "loss": 1.5777,
+      "step": 11386
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.495828367103695e-07,
+      "loss": 1.4847,
+      "step": 11387
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.495431068732618e-07,
+      "loss": 1.7528,
+      "step": 11388
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.495033770361541e-07,
+      "loss": 1.3599,
+      "step": 11389
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.494636471990464e-07,
+      "loss": 2.0536,
+      "step": 11390
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.494239173619388e-07,
+      "loss": 1.6147,
+      "step": 11391
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.493841875248312e-07,
+      "loss": 1.6811,
+      "step": 11392
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.493444576877234e-07,
+      "loss": 1.5599,
+      "step": 11393
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.493047278506158e-07,
+      "loss": 1.6112,
+      "step": 11394
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.492649980135081e-07,
+      "loss": 1.4129,
+      "step": 11395
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.492252681764005e-07,
+      "loss": 1.7365,
+      "step": 11396
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.491855383392928e-07,
+      "loss": 1.7056,
+      "step": 11397
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.491458085021852e-07,
+      "loss": 1.7744,
+      "step": 11398
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.491060786650774e-07,
+      "loss": 1.7078,
+      "step": 11399
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.490663488279698e-07,
+      "loss": 1.8734,
+      "step": 11400
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.490266189908622e-07,
+      "loss": 2.3841,
+      "step": 11401
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.489868891537544e-07,
+      "loss": 1.0531,
+      "step": 11402
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.489471593166468e-07,
+      "loss": 1.8747,
+      "step": 11403
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.489074294795391e-07,
+      "loss": 1.5865,
+      "step": 11404
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.488676996424314e-07,
+      "loss": 1.882,
+      "step": 11405
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.488279698053237e-07,
+      "loss": 1.9326,
+      "step": 11406
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.487882399682161e-07,
+      "loss": 1.7091,
+      "step": 11407
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.487485101311085e-07,
+      "loss": 1.3561,
+      "step": 11408
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.487087802940008e-07,
+      "loss": 1.6017,
+      "step": 11409
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.486690504568932e-07,
+      "loss": 1.3532,
+      "step": 11410
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.486293206197854e-07,
+      "loss": 1.6381,
+      "step": 11411
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.485895907826778e-07,
+      "loss": 1.6213,
+      "step": 11412
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.485498609455701e-07,
+      "loss": 1.8887,
+      "step": 11413
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.485101311084625e-07,
+      "loss": 1.4985,
+      "step": 11414
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.484704012713547e-07,
+      "loss": 1.6609,
+      "step": 11415
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.484306714342471e-07,
+      "loss": 1.6209,
+      "step": 11416
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.483909415971394e-07,
+      "loss": 1.8844,
+      "step": 11417
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.483512117600317e-07,
+      "loss": 1.7471,
+      "step": 11418
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.483114819229241e-07,
+      "loss": 1.7498,
+      "step": 11419
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.482717520858164e-07,
+      "loss": 1.6026,
+      "step": 11420
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.482320222487088e-07,
+      "loss": 1.6944,
+      "step": 11421
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.481922924116011e-07,
+      "loss": 1.3401,
+      "step": 11422
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.481525625744934e-07,
+      "loss": 2.0435,
+      "step": 11423
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.481128327373858e-07,
+      "loss": 1.8588,
+      "step": 11424
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.480731029002781e-07,
+      "loss": 1.7041,
+      "step": 11425
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.480333730631705e-07,
+      "loss": 1.9882,
+      "step": 11426
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.479936432260628e-07,
+      "loss": 1.9253,
+      "step": 11427
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.479539133889552e-07,
+      "loss": 1.3095,
+      "step": 11428
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.479141835518473e-07,
+      "loss": 1.6006,
+      "step": 11429
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.478744537147397e-07,
+      "loss": 1.5085,
+      "step": 11430
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.47834723877632e-07,
+      "loss": 1.3126,
+      "step": 11431
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.477949940405244e-07,
+      "loss": 1.5233,
+      "step": 11432
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.477552642034168e-07,
+      "loss": 1.6084,
+      "step": 11433
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.477155343663091e-07,
+      "loss": 1.4442,
+      "step": 11434
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.476758045292014e-07,
+      "loss": 1.9844,
+      "step": 11435
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.476360746920937e-07,
+      "loss": 1.2598,
+      "step": 11436
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.475963448549861e-07,
+      "loss": 1.4387,
+      "step": 11437
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.475566150178784e-07,
+      "loss": 1.5659,
+      "step": 11438
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.475168851807708e-07,
+      "loss": 1.3909,
+      "step": 11439
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.474771553436631e-07,
+      "loss": 1.3829,
+      "step": 11440
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.474374255065553e-07,
+      "loss": 1.9714,
+      "step": 11441
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.473976956694477e-07,
+      "loss": 1.2261,
+      "step": 11442
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.4735796583234e-07,
+      "loss": 1.7317,
+      "step": 11443
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.473182359952324e-07,
+      "loss": 1.5251,
+      "step": 11444
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.472785061581247e-07,
+      "loss": 1.172,
+      "step": 11445
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.47238776321017e-07,
+      "loss": 2.3948,
+      "step": 11446
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.471990464839094e-07,
+      "loss": 1.5029,
+      "step": 11447
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.471593166468017e-07,
+      "loss": 1.8482,
+      "step": 11448
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.471195868096941e-07,
+      "loss": 1.7093,
+      "step": 11449
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.470798569725864e-07,
+      "loss": 1.4673,
+      "step": 11450
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.470401271354788e-07,
+      "loss": 1.6517,
+      "step": 11451
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.47000397298371e-07,
+      "loss": 1.7088,
+      "step": 11452
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.469606674612634e-07,
+      "loss": 1.3569,
+      "step": 11453
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.469209376241557e-07,
+      "loss": 1.5825,
+      "step": 11454
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.46881207787048e-07,
+      "loss": 1.5116,
+      "step": 11455
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.468414779499404e-07,
+      "loss": 1.6181,
+      "step": 11456
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.468017481128327e-07,
+      "loss": 2.2525,
+      "step": 11457
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.46762018275725e-07,
+      "loss": 1.5198,
+      "step": 11458
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.467222884386173e-07,
+      "loss": 1.4003,
+      "step": 11459
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.466825586015097e-07,
+      "loss": 1.46,
+      "step": 11460
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.46642828764402e-07,
+      "loss": 1.6311,
+      "step": 11461
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.466030989272944e-07,
+      "loss": 1.896,
+      "step": 11462
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.465633690901868e-07,
+      "loss": 1.2884,
+      "step": 11463
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.46523639253079e-07,
+      "loss": 1.7258,
+      "step": 11464
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.464839094159714e-07,
+      "loss": 2.0136,
+      "step": 11465
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.464441795788637e-07,
+      "loss": 1.3323,
+      "step": 11466
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.464044497417561e-07,
+      "loss": 1.2719,
+      "step": 11467
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.463647199046483e-07,
+      "loss": 1.9216,
+      "step": 11468
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.463249900675407e-07,
+      "loss": 1.5789,
+      "step": 11469
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.462852602304329e-07,
+      "loss": 1.3313,
+      "step": 11470
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.462455303933253e-07,
+      "loss": 1.1723,
+      "step": 11471
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.462058005562177e-07,
+      "loss": 1.6365,
+      "step": 11472
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.4616607071911e-07,
+      "loss": 1.5569,
+      "step": 11473
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.461263408820024e-07,
+      "loss": 1.3533,
+      "step": 11474
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.460866110448947e-07,
+      "loss": 1.3785,
+      "step": 11475
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.46046881207787e-07,
+      "loss": 1.5439,
+      "step": 11476
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.460071513706793e-07,
+      "loss": 2.2725,
+      "step": 11477
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.459674215335717e-07,
+      "loss": 1.4653,
+      "step": 11478
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.459276916964641e-07,
+      "loss": 2.1484,
+      "step": 11479
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.458879618593564e-07,
+      "loss": 1.7318,
+      "step": 11480
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.458482320222487e-07,
+      "loss": 1.6457,
+      "step": 11481
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.458085021851409e-07,
+      "loss": 1.6155,
+      "step": 11482
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.457687723480333e-07,
+      "loss": 1.4888,
+      "step": 11483
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.457290425109256e-07,
+      "loss": 1.5705,
+      "step": 11484
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.45689312673818e-07,
+      "loss": 1.4212,
+      "step": 11485
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.456495828367104e-07,
+      "loss": 1.5079,
+      "step": 11486
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.456098529996027e-07,
+      "loss": 1.8198,
+      "step": 11487
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.45570123162495e-07,
+      "loss": 1.4194,
+      "step": 11488
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.455303933253873e-07,
+      "loss": 1.0349,
+      "step": 11489
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.454906634882797e-07,
+      "loss": 1.7374,
+      "step": 11490
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.45450933651172e-07,
+      "loss": 1.6021,
+      "step": 11491
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.454112038140644e-07,
+      "loss": 1.5277,
+      "step": 11492
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.453714739769568e-07,
+      "loss": 1.8378,
+      "step": 11493
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.453317441398489e-07,
+      "loss": 1.46,
+      "step": 11494
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.452920143027413e-07,
+      "loss": 1.0874,
+      "step": 11495
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.452522844656336e-07,
+      "loss": 1.4947,
+      "step": 11496
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.45212554628526e-07,
+      "loss": 1.7633,
+      "step": 11497
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.451728247914183e-07,
+      "loss": 1.7185,
+      "step": 11498
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.451330949543107e-07,
+      "loss": 1.2239,
+      "step": 11499
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.450933651172029e-07,
+      "loss": 1.8514,
+      "step": 11500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.450536352800953e-07,
+      "loss": 1.2483,
+      "step": 11501
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.450139054429877e-07,
+      "loss": 1.4413,
+      "step": 11502
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.4497417560588e-07,
+      "loss": 1.5226,
+      "step": 11503
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.449344457687724e-07,
+      "loss": 1.6032,
+      "step": 11504
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.448947159316647e-07,
+      "loss": 1.5241,
+      "step": 11505
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.44854986094557e-07,
+      "loss": 1.4713,
+      "step": 11506
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.448152562574492e-07,
+      "loss": 1.721,
+      "step": 11507
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.447755264203416e-07,
+      "loss": 1.914,
+      "step": 11508
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.44735796583234e-07,
+      "loss": 1.3571,
+      "step": 11509
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.446960667461263e-07,
+      "loss": 1.7509,
+      "step": 11510
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.446563369090187e-07,
+      "loss": 1.8812,
+      "step": 11511
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.446166070719109e-07,
+      "loss": 1.6847,
+      "step": 11512
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.445768772348033e-07,
+      "loss": 1.5925,
+      "step": 11513
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.445371473976956e-07,
+      "loss": 2.0705,
+      "step": 11514
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.44497417560588e-07,
+      "loss": 1.2631,
+      "step": 11515
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.444576877234803e-07,
+      "loss": 1.9293,
+      "step": 11516
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.444179578863727e-07,
+      "loss": 1.4126,
+      "step": 11517
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.44378228049265e-07,
+      "loss": 1.4041,
+      "step": 11518
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.443384982121573e-07,
+      "loss": 1.3984,
+      "step": 11519
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.442987683750497e-07,
+      "loss": 1.5072,
+      "step": 11520
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.442590385379419e-07,
+      "loss": 1.7383,
+      "step": 11521
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.442193087008343e-07,
+      "loss": 1.6304,
+      "step": 11522
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.441795788637266e-07,
+      "loss": 1.5047,
+      "step": 11523
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.441398490266189e-07,
+      "loss": 1.7029,
+      "step": 11524
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.441001191895113e-07,
+      "loss": 1.8743,
+      "step": 11525
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.440603893524036e-07,
+      "loss": 1.516,
+      "step": 11526
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.44020659515296e-07,
+      "loss": 1.9579,
+      "step": 11527
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.439809296781883e-07,
+      "loss": 1.4966,
+      "step": 11528
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.439411998410807e-07,
+      "loss": 1.6566,
+      "step": 11529
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.439014700039729e-07,
+      "loss": 2.1386,
+      "step": 11530
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.438617401668653e-07,
+      "loss": 1.4254,
+      "step": 11531
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.438220103297576e-07,
+      "loss": 1.6014,
+      "step": 11532
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.4378228049265e-07,
+      "loss": 1.2439,
+      "step": 11533
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.437425506555423e-07,
+      "loss": 1.4054,
+      "step": 11534
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.437028208184346e-07,
+      "loss": 1.523,
+      "step": 11535
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.436630909813269e-07,
+      "loss": 1.4248,
+      "step": 11536
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.436233611442192e-07,
+      "loss": 1.5464,
+      "step": 11537
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.435836313071116e-07,
+      "loss": 1.7181,
+      "step": 11538
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.435439014700039e-07,
+      "loss": 1.802,
+      "step": 11539
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.435041716328963e-07,
+      "loss": 2.083,
+      "step": 11540
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.434644417957887e-07,
+      "loss": 1.1367,
+      "step": 11541
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.434247119586809e-07,
+      "loss": 1.545,
+      "step": 11542
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.433849821215733e-07,
+      "loss": 1.4273,
+      "step": 11543
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.433452522844656e-07,
+      "loss": 1.5345,
+      "step": 11544
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.43305522447358e-07,
+      "loss": 1.6198,
+      "step": 11545
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.432657926102503e-07,
+      "loss": 2.0526,
+      "step": 11546
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.432260627731426e-07,
+      "loss": 1.6149,
+      "step": 11547
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.431863329360348e-07,
+      "loss": 1.9127,
+      "step": 11548
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.431466030989272e-07,
+      "loss": 1.4386,
+      "step": 11549
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.431068732618196e-07,
+      "loss": 1.8485,
+      "step": 11550
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.430671434247119e-07,
+      "loss": 1.7205,
+      "step": 11551
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.430274135876043e-07,
+      "loss": 1.3817,
+      "step": 11552
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.429876837504966e-07,
+      "loss": 1.7505,
+      "step": 11553
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.429479539133889e-07,
+      "loss": 1.8435,
+      "step": 11554
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.429082240762812e-07,
+      "loss": 1.7017,
+      "step": 11555
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.428684942391736e-07,
+      "loss": 1.8159,
+      "step": 11556
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.42828764402066e-07,
+      "loss": 2.1943,
+      "step": 11557
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.427890345649583e-07,
+      "loss": 1.7011,
+      "step": 11558
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.427493047278507e-07,
+      "loss": 1.7759,
+      "step": 11559
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.427095748907428e-07,
+      "loss": 1.8643,
+      "step": 11560
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.426698450536352e-07,
+      "loss": 1.811,
+      "step": 11561
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.426301152165275e-07,
+      "loss": 1.7661,
+      "step": 11562
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.425903853794199e-07,
+      "loss": 1.8019,
+      "step": 11563
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.425506555423123e-07,
+      "loss": 1.5605,
+      "step": 11564
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.425109257052046e-07,
+      "loss": 1.61,
+      "step": 11565
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.424711958680969e-07,
+      "loss": 1.8904,
+      "step": 11566
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.424314660309892e-07,
+      "loss": 1.7198,
+      "step": 11567
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.423917361938816e-07,
+      "loss": 1.6694,
+      "step": 11568
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.423520063567739e-07,
+      "loss": 1.2214,
+      "step": 11569
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.423122765196663e-07,
+      "loss": 1.4387,
+      "step": 11570
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.422725466825587e-07,
+      "loss": 1.4052,
+      "step": 11571
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.422328168454509e-07,
+      "loss": 1.5857,
+      "step": 11572
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.421930870083432e-07,
+      "loss": 1.571,
+      "step": 11573
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.421533571712355e-07,
+      "loss": 2.1369,
+      "step": 11574
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.421136273341279e-07,
+      "loss": 1.801,
+      "step": 11575
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.420738974970202e-07,
+      "loss": 1.6233,
+      "step": 11576
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.420341676599126e-07,
+      "loss": 1.7993,
+      "step": 11577
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.419944378228048e-07,
+      "loss": 1.6095,
+      "step": 11578
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.419547079856972e-07,
+      "loss": 1.4172,
+      "step": 11579
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.419149781485896e-07,
+      "loss": 1.5192,
+      "step": 11580
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.418752483114819e-07,
+      "loss": 1.4248,
+      "step": 11581
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.418355184743743e-07,
+      "loss": 1.4322,
+      "step": 11582
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.417957886372666e-07,
+      "loss": 1.4676,
+      "step": 11583
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.417560588001589e-07,
+      "loss": 1.2794,
+      "step": 11584
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.417163289630512e-07,
+      "loss": 1.8578,
+      "step": 11585
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.416765991259436e-07,
+      "loss": 1.6481,
+      "step": 11586
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.416368692888358e-07,
+      "loss": 2.082,
+      "step": 11587
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.415971394517282e-07,
+      "loss": 1.5662,
+      "step": 11588
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.415574096146206e-07,
+      "loss": 1.3232,
+      "step": 11589
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.415176797775128e-07,
+      "loss": 1.5674,
+      "step": 11590
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.414779499404052e-07,
+      "loss": 1.458,
+      "step": 11591
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.414382201032975e-07,
+      "loss": 1.7195,
+      "step": 11592
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.413984902661899e-07,
+      "loss": 1.8889,
+      "step": 11593
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.413587604290822e-07,
+      "loss": 1.796,
+      "step": 11594
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.413190305919746e-07,
+      "loss": 1.5017,
+      "step": 11595
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.412793007548669e-07,
+      "loss": 1.7077,
+      "step": 11596
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.412395709177592e-07,
+      "loss": 1.3327,
+      "step": 11597
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.411998410806516e-07,
+      "loss": 1.2822,
+      "step": 11598
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.411601112435439e-07,
+      "loss": 1.9596,
+      "step": 11599
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.411203814064362e-07,
+      "loss": 1.6874,
+      "step": 11600
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.410806515693285e-07,
+      "loss": 1.542,
+      "step": 11601
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.410409217322208e-07,
+      "loss": 2.0316,
+      "step": 11602
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.410011918951132e-07,
+      "loss": 2.0789,
+      "step": 11603
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.409614620580055e-07,
+      "loss": 1.3579,
+      "step": 11604
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.409217322208979e-07,
+      "loss": 1.719,
+      "step": 11605
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.408820023837902e-07,
+      "loss": 1.8658,
+      "step": 11606
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.408422725466826e-07,
+      "loss": 1.7315,
+      "step": 11607
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.408025427095748e-07,
+      "loss": 1.9096,
+      "step": 11608
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.407628128724672e-07,
+      "loss": 1.4439,
+      "step": 11609
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.407230830353595e-07,
+      "loss": 1.829,
+      "step": 11610
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.406833531982519e-07,
+      "loss": 1.3906,
+      "step": 11611
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.406436233611443e-07,
+      "loss": 1.4402,
+      "step": 11612
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.406038935240365e-07,
+      "loss": 1.2313,
+      "step": 11613
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.405641636869288e-07,
+      "loss": 1.3803,
+      "step": 11614
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.405244338498211e-07,
+      "loss": 1.9143,
+      "step": 11615
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.404847040127135e-07,
+      "loss": 2.0078,
+      "step": 11616
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.404449741756058e-07,
+      "loss": 1.7218,
+      "step": 11617
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.404052443384982e-07,
+      "loss": 1.2327,
+      "step": 11618
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.403655145013906e-07,
+      "loss": 1.7385,
+      "step": 11619
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.403257846642828e-07,
+      "loss": 1.7029,
+      "step": 11620
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.402860548271752e-07,
+      "loss": 1.84,
+      "step": 11621
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.402463249900675e-07,
+      "loss": 1.4257,
+      "step": 11622
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.402065951529599e-07,
+      "loss": 1.8433,
+      "step": 11623
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.401668653158522e-07,
+      "loss": 1.5213,
+      "step": 11624
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.401271354787446e-07,
+      "loss": 1.6006,
+      "step": 11625
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.400874056416367e-07,
+      "loss": 2.2526,
+      "step": 11626
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.400476758045291e-07,
+      "loss": 2.0554,
+      "step": 11627
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.400079459674215e-07,
+      "loss": 1.649,
+      "step": 11628
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.399682161303138e-07,
+      "loss": 1.4604,
+      "step": 11629
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.399284862932062e-07,
+      "loss": 1.4358,
+      "step": 11630
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.398887564560985e-07,
+      "loss": 1.719,
+      "step": 11631
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.398490266189908e-07,
+      "loss": 1.5998,
+      "step": 11632
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.398092967818831e-07,
+      "loss": 1.3077,
+      "step": 11633
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.397695669447755e-07,
+      "loss": 1.3053,
+      "step": 11634
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.397298371076679e-07,
+      "loss": 1.5206,
+      "step": 11635
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.396901072705602e-07,
+      "loss": 1.8496,
+      "step": 11636
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.396503774334526e-07,
+      "loss": 1.9133,
+      "step": 11637
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.396106475963448e-07,
+      "loss": 1.6827,
+      "step": 11638
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.395709177592371e-07,
+      "loss": 1.9152,
+      "step": 11639
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.395311879221294e-07,
+      "loss": 1.6226,
+      "step": 11640
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.394914580850218e-07,
+      "loss": 1.5426,
+      "step": 11641
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.394517282479142e-07,
+      "loss": 1.7243,
+      "step": 11642
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.394119984108065e-07,
+      "loss": 1.4169,
+      "step": 11643
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.393722685736988e-07,
+      "loss": 1.6751,
+      "step": 11644
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.393325387365911e-07,
+      "loss": 1.4279,
+      "step": 11645
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.392928088994835e-07,
+      "loss": 1.8417,
+      "step": 11646
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.392530790623758e-07,
+      "loss": 1.4613,
+      "step": 11647
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.392133492252682e-07,
+      "loss": 1.2601,
+      "step": 11648
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.391736193881606e-07,
+      "loss": 1.3475,
+      "step": 11649
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.391338895510528e-07,
+      "loss": 1.5778,
+      "step": 11650
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.390941597139452e-07,
+      "loss": 1.9033,
+      "step": 11651
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.390544298768375e-07,
+      "loss": 1.1617,
+      "step": 11652
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.390147000397298e-07,
+      "loss": 1.4954,
+      "step": 11653
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.389749702026221e-07,
+      "loss": 2.0197,
+      "step": 11654
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.389352403655145e-07,
+      "loss": 1.588,
+      "step": 11655
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.388955105284067e-07,
+      "loss": 1.4989,
+      "step": 11656
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.388557806912991e-07,
+      "loss": 1.8262,
+      "step": 11657
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.388160508541915e-07,
+      "loss": 1.8237,
+      "step": 11658
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.387763210170838e-07,
+      "loss": 1.5907,
+      "step": 11659
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.387365911799762e-07,
+      "loss": 1.2965,
+      "step": 11660
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.386968613428685e-07,
+      "loss": 1.4156,
+      "step": 11661
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.386571315057608e-07,
+      "loss": 1.1734,
+      "step": 11662
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.386174016686531e-07,
+      "loss": 1.7756,
+      "step": 11663
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.385776718315455e-07,
+      "loss": 1.2518,
+      "step": 11664
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.385379419944379e-07,
+      "loss": 1.4385,
+      "step": 11665
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.384982121573301e-07,
+      "loss": 1.6288,
+      "step": 11666
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.384584823202225e-07,
+      "loss": 1.5913,
+      "step": 11667
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.384187524831147e-07,
+      "loss": 1.7003,
+      "step": 11668
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.383790226460071e-07,
+      "loss": 1.6693,
+      "step": 11669
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.383392928088994e-07,
+      "loss": 1.5193,
+      "step": 11670
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.382995629717918e-07,
+      "loss": 1.3917,
+      "step": 11671
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.382598331346841e-07,
+      "loss": 1.6646,
+      "step": 11672
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.382201032975765e-07,
+      "loss": 1.3383,
+      "step": 11673
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.381803734604688e-07,
+      "loss": 1.7328,
+      "step": 11674
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.381406436233611e-07,
+      "loss": 1.9174,
+      "step": 11675
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.381009137862535e-07,
+      "loss": 1.4423,
+      "step": 11676
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.380611839491458e-07,
+      "loss": 1.4943,
+      "step": 11677
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.380214541120382e-07,
+      "loss": 1.5277,
+      "step": 11678
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.379817242749304e-07,
+      "loss": 1.6743,
+      "step": 11679
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.379419944378227e-07,
+      "loss": 1.6665,
+      "step": 11680
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.37902264600715e-07,
+      "loss": 1.6626,
+      "step": 11681
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.378625347636074e-07,
+      "loss": 1.665,
+      "step": 11682
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.378228049264998e-07,
+      "loss": 1.7488,
+      "step": 11683
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.377830750893921e-07,
+      "loss": 1.9705,
+      "step": 11684
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.377433452522845e-07,
+      "loss": 1.8948,
+      "step": 11685
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.377036154151767e-07,
+      "loss": 1.6759,
+      "step": 11686
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.376638855780691e-07,
+      "loss": 1.6038,
+      "step": 11687
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.376241557409615e-07,
+      "loss": 1.7567,
+      "step": 11688
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.375844259038538e-07,
+      "loss": 1.2916,
+      "step": 11689
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.375446960667462e-07,
+      "loss": 1.7113,
+      "step": 11690
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.375049662296385e-07,
+      "loss": 1.5371,
+      "step": 11691
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.374652363925307e-07,
+      "loss": 1.3679,
+      "step": 11692
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.37425506555423e-07,
+      "loss": 1.6909,
+      "step": 11693
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.373857767183154e-07,
+      "loss": 1.166,
+      "step": 11694
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.373460468812077e-07,
+      "loss": 1.7222,
+      "step": 11695
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.373063170441001e-07,
+      "loss": 1.5685,
+      "step": 11696
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.372665872069925e-07,
+      "loss": 1.3236,
+      "step": 11697
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.372268573698847e-07,
+      "loss": 1.5375,
+      "step": 11698
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.371871275327771e-07,
+      "loss": 1.617,
+      "step": 11699
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.371473976956694e-07,
+      "loss": 1.59,
+      "step": 11700
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.371076678585618e-07,
+      "loss": 1.5165,
+      "step": 11701
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.370679380214541e-07,
+      "loss": 1.6937,
+      "step": 11702
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.370282081843465e-07,
+      "loss": 1.685,
+      "step": 11703
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.369884783472388e-07,
+      "loss": 1.7232,
+      "step": 11704
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.36948748510131e-07,
+      "loss": 1.9433,
+      "step": 11705
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.369090186730234e-07,
+      "loss": 1.2734,
+      "step": 11706
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.368692888359157e-07,
+      "loss": 1.7798,
+      "step": 11707
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.368295589988081e-07,
+      "loss": 1.6009,
+      "step": 11708
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.367898291617004e-07,
+      "loss": 1.4597,
+      "step": 11709
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.367500993245927e-07,
+      "loss": 1.7463,
+      "step": 11710
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.36710369487485e-07,
+      "loss": 1.4349,
+      "step": 11711
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.366706396503774e-07,
+      "loss": 1.4835,
+      "step": 11712
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.366309098132698e-07,
+      "loss": 1.2225,
+      "step": 11713
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.365911799761621e-07,
+      "loss": 1.6822,
+      "step": 11714
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.365514501390545e-07,
+      "loss": 1.8056,
+      "step": 11715
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.365117203019467e-07,
+      "loss": 1.6337,
+      "step": 11716
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.364719904648391e-07,
+      "loss": 1.7804,
+      "step": 11717
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.364322606277314e-07,
+      "loss": 1.6514,
+      "step": 11718
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.363925307906237e-07,
+      "loss": 1.6829,
+      "step": 11719
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.363528009535161e-07,
+      "loss": 1.7147,
+      "step": 11720
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.363130711164084e-07,
+      "loss": 1.5903,
+      "step": 11721
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.362733412793007e-07,
+      "loss": 1.8974,
+      "step": 11722
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.36233611442193e-07,
+      "loss": 1.8218,
+      "step": 11723
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.361938816050854e-07,
+      "loss": 1.3385,
+      "step": 11724
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.361541517679777e-07,
+      "loss": 1.3872,
+      "step": 11725
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.361144219308701e-07,
+      "loss": 1.5752,
+      "step": 11726
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.360746920937625e-07,
+      "loss": 1.7339,
+      "step": 11727
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.360349622566547e-07,
+      "loss": 1.5462,
+      "step": 11728
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.359952324195471e-07,
+      "loss": 1.6985,
+      "step": 11729
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.359555025824394e-07,
+      "loss": 1.8431,
+      "step": 11730
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.359157727453318e-07,
+      "loss": 1.5285,
+      "step": 11731
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.35876042908224e-07,
+      "loss": 1.828,
+      "step": 11732
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.358363130711164e-07,
+      "loss": 1.4393,
+      "step": 11733
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.357965832340086e-07,
+      "loss": 1.2428,
+      "step": 11734
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.35756853396901e-07,
+      "loss": 2.0039,
+      "step": 11735
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.357171235597934e-07,
+      "loss": 1.7178,
+      "step": 11736
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.356773937226857e-07,
+      "loss": 1.8404,
+      "step": 11737
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.356376638855781e-07,
+      "loss": 1.47,
+      "step": 11738
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.355979340484704e-07,
+      "loss": 1.4282,
+      "step": 11739
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.355582042113627e-07,
+      "loss": 1.5453,
+      "step": 11740
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.35518474374255e-07,
+      "loss": 1.688,
+      "step": 11741
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.354787445371474e-07,
+      "loss": 1.3953,
+      "step": 11742
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.354390147000398e-07,
+      "loss": 1.3235,
+      "step": 11743
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.353992848629321e-07,
+      "loss": 1.7952,
+      "step": 11744
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.353595550258244e-07,
+      "loss": 1.8086,
+      "step": 11745
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.353198251887166e-07,
+      "loss": 1.3393,
+      "step": 11746
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.35280095351609e-07,
+      "loss": 1.9239,
+      "step": 11747
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.352403655145013e-07,
+      "loss": 1.3491,
+      "step": 11748
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.352006356773937e-07,
+      "loss": 1.8432,
+      "step": 11749
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.35160905840286e-07,
+      "loss": 2.0458,
+      "step": 11750
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.351211760031784e-07,
+      "loss": 1.7113,
+      "step": 11751
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.350814461660707e-07,
+      "loss": 1.7627,
+      "step": 11752
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.35041716328963e-07,
+      "loss": 1.5496,
+      "step": 11753
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.350019864918554e-07,
+      "loss": 1.6218,
+      "step": 11754
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.349622566547477e-07,
+      "loss": 1.5492,
+      "step": 11755
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.349225268176401e-07,
+      "loss": 1.5634,
+      "step": 11756
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.348827969805324e-07,
+      "loss": 0.9918,
+      "step": 11757
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.348430671434246e-07,
+      "loss": 1.7254,
+      "step": 11758
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.34803337306317e-07,
+      "loss": 1.471,
+      "step": 11759
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.347636074692093e-07,
+      "loss": 1.5737,
+      "step": 11760
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.347238776321017e-07,
+      "loss": 1.8167,
+      "step": 11761
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.34684147794994e-07,
+      "loss": 2.0973,
+      "step": 11762
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.346444179578864e-07,
+      "loss": 1.4189,
+      "step": 11763
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.346046881207786e-07,
+      "loss": 1.3299,
+      "step": 11764
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.34564958283671e-07,
+      "loss": 1.3884,
+      "step": 11765
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.345252284465634e-07,
+      "loss": 1.4191,
+      "step": 11766
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.344854986094557e-07,
+      "loss": 1.3502,
+      "step": 11767
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.344457687723481e-07,
+      "loss": 1.8026,
+      "step": 11768
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.344060389352404e-07,
+      "loss": 1.5632,
+      "step": 11769
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.343663090981327e-07,
+      "loss": 1.8573,
+      "step": 11770
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.343265792610249e-07,
+      "loss": 1.4796,
+      "step": 11771
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.342868494239173e-07,
+      "loss": 1.0928,
+      "step": 11772
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.342471195868096e-07,
+      "loss": 2.0494,
+      "step": 11773
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.34207389749702e-07,
+      "loss": 1.7624,
+      "step": 11774
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.341676599125944e-07,
+      "loss": 1.8526,
+      "step": 11775
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.341279300754866e-07,
+      "loss": 2.1323,
+      "step": 11776
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.34088200238379e-07,
+      "loss": 1.7698,
+      "step": 11777
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.340484704012713e-07,
+      "loss": 1.8708,
+      "step": 11778
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.340087405641637e-07,
+      "loss": 1.3405,
+      "step": 11779
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.33969010727056e-07,
+      "loss": 1.6158,
+      "step": 11780
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.339292808899484e-07,
+      "loss": 1.4848,
+      "step": 11781
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.338895510528407e-07,
+      "loss": 1.7236,
+      "step": 11782
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.33849821215733e-07,
+      "loss": 2.3405,
+      "step": 11783
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.338100913786254e-07,
+      "loss": 1.1473,
+      "step": 11784
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.337703615415176e-07,
+      "loss": 1.7467,
+      "step": 11785
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.3373063170441e-07,
+      "loss": 1.7297,
+      "step": 11786
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.336909018673023e-07,
+      "loss": 1.6037,
+      "step": 11787
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.336511720301946e-07,
+      "loss": 1.481,
+      "step": 11788
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.336114421930869e-07,
+      "loss": 1.8943,
+      "step": 11789
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.335717123559793e-07,
+      "loss": 1.1065,
+      "step": 11790
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.335319825188717e-07,
+      "loss": 1.9204,
+      "step": 11791
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.33492252681764e-07,
+      "loss": 1.7328,
+      "step": 11792
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.334525228446564e-07,
+      "loss": 1.481,
+      "step": 11793
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.334127930075486e-07,
+      "loss": 1.0853,
+      "step": 11794
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.33373063170441e-07,
+      "loss": 1.4856,
+      "step": 11795
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.333333333333333e-07,
+      "loss": 1.5392,
+      "step": 11796
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.332936034962257e-07,
+      "loss": 1.5273,
+      "step": 11797
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.33253873659118e-07,
+      "loss": 1.5008,
+      "step": 11798
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.332141438220103e-07,
+      "loss": 1.5787,
+      "step": 11799
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.331744139849026e-07,
+      "loss": 1.2994,
+      "step": 11800
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.331346841477949e-07,
+      "loss": 1.2737,
+      "step": 11801
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.330949543106873e-07,
+      "loss": 1.4489,
+      "step": 11802
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.330552244735796e-07,
+      "loss": 1.9201,
+      "step": 11803
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.33015494636472e-07,
+      "loss": 1.6858,
+      "step": 11804
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.329757647993644e-07,
+      "loss": 1.6319,
+      "step": 11805
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.329360349622566e-07,
+      "loss": 1.6433,
+      "step": 11806
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.32896305125149e-07,
+      "loss": 1.6338,
+      "step": 11807
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.328565752880413e-07,
+      "loss": 1.8682,
+      "step": 11808
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.328168454509337e-07,
+      "loss": 1.6099,
+      "step": 11809
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.32777115613826e-07,
+      "loss": 1.8695,
+      "step": 11810
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.327373857767183e-07,
+      "loss": 1.2865,
+      "step": 11811
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.326976559396105e-07,
+      "loss": 1.7722,
+      "step": 11812
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.326579261025029e-07,
+      "loss": 2.0683,
+      "step": 11813
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.326181962653953e-07,
+      "loss": 1.4736,
+      "step": 11814
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.325784664282876e-07,
+      "loss": 1.7426,
+      "step": 11815
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.3253873659118e-07,
+      "loss": 2.0229,
+      "step": 11816
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.324990067540723e-07,
+      "loss": 2.0366,
+      "step": 11817
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.324592769169646e-07,
+      "loss": 1.6039,
+      "step": 11818
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.324195470798569e-07,
+      "loss": 1.5401,
+      "step": 11819
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.323798172427493e-07,
+      "loss": 1.3387,
+      "step": 11820
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.323400874056417e-07,
+      "loss": 1.4374,
+      "step": 11821
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.32300357568534e-07,
+      "loss": 1.7177,
+      "step": 11822
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.322606277314264e-07,
+      "loss": 1.1349,
+      "step": 11823
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.322208978943185e-07,
+      "loss": 1.9697,
+      "step": 11824
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.321811680572109e-07,
+      "loss": 1.6975,
+      "step": 11825
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.321414382201032e-07,
+      "loss": 1.6698,
+      "step": 11826
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.321017083829956e-07,
+      "loss": 1.5579,
+      "step": 11827
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.32061978545888e-07,
+      "loss": 1.896,
+      "step": 11828
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.320222487087803e-07,
+      "loss": 1.8446,
+      "step": 11829
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.319825188716726e-07,
+      "loss": 1.8081,
+      "step": 11830
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.319427890345649e-07,
+      "loss": 2.0424,
+      "step": 11831
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.319030591974573e-07,
+      "loss": 1.2546,
+      "step": 11832
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.318633293603496e-07,
+      "loss": 1.9807,
+      "step": 11833
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.31823599523242e-07,
+      "loss": 1.4618,
+      "step": 11834
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.317838696861343e-07,
+      "loss": 1.5562,
+      "step": 11835
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.317441398490266e-07,
+      "loss": 1.8749,
+      "step": 11836
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.317044100119189e-07,
+      "loss": 1.8051,
+      "step": 11837
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.316646801748112e-07,
+      "loss": 1.7581,
+      "step": 11838
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.316249503377036e-07,
+      "loss": 1.4004,
+      "step": 11839
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.315852205005959e-07,
+      "loss": 1.7429,
+      "step": 11840
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.315454906634883e-07,
+      "loss": 1.7222,
+      "step": 11841
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.315057608263805e-07,
+      "loss": 1.8865,
+      "step": 11842
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.314660309892729e-07,
+      "loss": 1.4448,
+      "step": 11843
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.314263011521653e-07,
+      "loss": 1.5724,
+      "step": 11844
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.313865713150576e-07,
+      "loss": 1.7407,
+      "step": 11845
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.3134684147795e-07,
+      "loss": 1.5487,
+      "step": 11846
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.313071116408423e-07,
+      "loss": 1.5269,
+      "step": 11847
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.312673818037346e-07,
+      "loss": 1.4747,
+      "step": 11848
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.312276519666269e-07,
+      "loss": 2.0499,
+      "step": 11849
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.311879221295193e-07,
+      "loss": 1.4658,
+      "step": 11850
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.311481922924115e-07,
+      "loss": 1.3178,
+      "step": 11851
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.311084624553039e-07,
+      "loss": 1.4313,
+      "step": 11852
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.310687326181963e-07,
+      "loss": 1.6468,
+      "step": 11853
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.310290027810885e-07,
+      "loss": 1.2031,
+      "step": 11854
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.309892729439809e-07,
+      "loss": 1.2731,
+      "step": 11855
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.309495431068732e-07,
+      "loss": 1.6622,
+      "step": 11856
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.309098132697656e-07,
+      "loss": 1.2904,
+      "step": 11857
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.308700834326579e-07,
+      "loss": 1.5954,
+      "step": 11858
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.308303535955503e-07,
+      "loss": 1.1426,
+      "step": 11859
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.307906237584426e-07,
+      "loss": 1.9936,
+      "step": 11860
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.307508939213349e-07,
+      "loss": 1.8614,
+      "step": 11861
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.307111640842273e-07,
+      "loss": 1.527,
+      "step": 11862
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.306714342471196e-07,
+      "loss": 1.5303,
+      "step": 11863
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.306317044100119e-07,
+      "loss": 1.4357,
+      "step": 11864
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.305919745729042e-07,
+      "loss": 1.4639,
+      "step": 11865
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.305522447357965e-07,
+      "loss": 1.3256,
+      "step": 11866
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.305125148986888e-07,
+      "loss": 1.3743,
+      "step": 11867
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.304727850615812e-07,
+      "loss": 1.7748,
+      "step": 11868
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.304330552244736e-07,
+      "loss": 1.8687,
+      "step": 11869
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.303933253873659e-07,
+      "loss": 1.7821,
+      "step": 11870
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.303535955502583e-07,
+      "loss": 1.381,
+      "step": 11871
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.303138657131505e-07,
+      "loss": 1.4502,
+      "step": 11872
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.302741358760429e-07,
+      "loss": 2.1168,
+      "step": 11873
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.302344060389352e-07,
+      "loss": 1.9161,
+      "step": 11874
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.301946762018276e-07,
+      "loss": 1.5331,
+      "step": 11875
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.3015494636472e-07,
+      "loss": 1.4304,
+      "step": 11876
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.301152165276122e-07,
+      "loss": 1.2949,
+      "step": 11877
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.300754866905045e-07,
+      "loss": 1.9531,
+      "step": 11878
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.300357568533968e-07,
+      "loss": 2.2722,
+      "step": 11879
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.299960270162892e-07,
+      "loss": 1.4991,
+      "step": 11880
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.299562971791815e-07,
+      "loss": 1.5529,
+      "step": 11881
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.299165673420739e-07,
+      "loss": 1.6886,
+      "step": 11882
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.298768375049663e-07,
+      "loss": 1.5473,
+      "step": 11883
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.298371076678585e-07,
+      "loss": 1.3448,
+      "step": 11884
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.297973778307509e-07,
+      "loss": 1.6453,
+      "step": 11885
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.297576479936432e-07,
+      "loss": 1.5524,
+      "step": 11886
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.297179181565356e-07,
+      "loss": 1.1168,
+      "step": 11887
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.296781883194279e-07,
+      "loss": 1.5108,
+      "step": 11888
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.296384584823203e-07,
+      "loss": 2.0675,
+      "step": 11889
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.295987286452124e-07,
+      "loss": 1.7207,
+      "step": 11890
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.295589988081048e-07,
+      "loss": 1.2129,
+      "step": 11891
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.295192689709972e-07,
+      "loss": 1.2876,
+      "step": 11892
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.294795391338895e-07,
+      "loss": 1.6622,
+      "step": 11893
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.294398092967819e-07,
+      "loss": 1.6004,
+      "step": 11894
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.294000794596742e-07,
+      "loss": 1.8924,
+      "step": 11895
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.293603496225665e-07,
+      "loss": 1.4439,
+      "step": 11896
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.293206197854588e-07,
+      "loss": 1.1368,
+      "step": 11897
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.292808899483512e-07,
+      "loss": 2.0646,
+      "step": 11898
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.292411601112436e-07,
+      "loss": 1.7977,
+      "step": 11899
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.292014302741359e-07,
+      "loss": 1.3856,
+      "step": 11900
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.291617004370283e-07,
+      "loss": 1.7065,
+      "step": 11901
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.291219705999205e-07,
+      "loss": 1.859,
+      "step": 11902
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.290822407628128e-07,
+      "loss": 1.6352,
+      "step": 11903
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.290425109257051e-07,
+      "loss": 1.5778,
+      "step": 11904
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.290027810885975e-07,
+      "loss": 1.535,
+      "step": 11905
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.289630512514898e-07,
+      "loss": 1.0393,
+      "step": 11906
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.289233214143822e-07,
+      "loss": 1.6637,
+      "step": 11907
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.288835915772745e-07,
+      "loss": 1.5659,
+      "step": 11908
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.288438617401668e-07,
+      "loss": 1.9918,
+      "step": 11909
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.288041319030592e-07,
+      "loss": 1.0867,
+      "step": 11910
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.287644020659515e-07,
+      "loss": 1.6142,
+      "step": 11911
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.287246722288439e-07,
+      "loss": 1.6568,
+      "step": 11912
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.286849423917362e-07,
+      "loss": 1.622,
+      "step": 11913
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.286452125546285e-07,
+      "loss": 1.6612,
+      "step": 11914
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.286054827175209e-07,
+      "loss": 1.3587,
+      "step": 11915
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.285657528804132e-07,
+      "loss": 1.5051,
+      "step": 11916
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.285260230433055e-07,
+      "loss": 1.7192,
+      "step": 11917
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.284862932061978e-07,
+      "loss": 1.3946,
+      "step": 11918
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.284465633690902e-07,
+      "loss": 2.0052,
+      "step": 11919
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.284068335319824e-07,
+      "loss": 1.5422,
+      "step": 11920
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.283671036948748e-07,
+      "loss": 1.4921,
+      "step": 11921
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.283273738577672e-07,
+      "loss": 1.7228,
+      "step": 11922
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.282876440206595e-07,
+      "loss": 1.8149,
+      "step": 11923
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.282479141835519e-07,
+      "loss": 1.4956,
+      "step": 11924
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.282081843464442e-07,
+      "loss": 1.7441,
+      "step": 11925
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.281684545093365e-07,
+      "loss": 1.7257,
+      "step": 11926
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.281287246722288e-07,
+      "loss": 1.4587,
+      "step": 11927
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.280889948351212e-07,
+      "loss": 1.4573,
+      "step": 11928
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.280492649980135e-07,
+      "loss": 1.4675,
+      "step": 11929
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.280095351609058e-07,
+      "loss": 1.8066,
+      "step": 11930
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.279698053237982e-07,
+      "loss": 1.529,
+      "step": 11931
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.279300754866904e-07,
+      "loss": 1.2714,
+      "step": 11932
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.278903456495828e-07,
+      "loss": 1.4243,
+      "step": 11933
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.278506158124751e-07,
+      "loss": 1.5745,
+      "step": 11934
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.278108859753675e-07,
+      "loss": 1.4277,
+      "step": 11935
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.277711561382598e-07,
+      "loss": 1.5432,
+      "step": 11936
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.277314263011522e-07,
+      "loss": 1.6911,
+      "step": 11937
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.276916964640445e-07,
+      "loss": 1.2424,
+      "step": 11938
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.276519666269368e-07,
+      "loss": 1.8213,
+      "step": 11939
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.276122367898292e-07,
+      "loss": 1.7788,
+      "step": 11940
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.275725069527215e-07,
+      "loss": 1.6926,
+      "step": 11941
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.275327771156139e-07,
+      "loss": 1.0253,
+      "step": 11942
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.274930472785061e-07,
+      "loss": 1.4403,
+      "step": 11943
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.274533174413984e-07,
+      "loss": 1.3377,
+      "step": 11944
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.274135876042907e-07,
+      "loss": 1.5561,
+      "step": 11945
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.273738577671831e-07,
+      "loss": 1.4819,
+      "step": 11946
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.273341279300755e-07,
+      "loss": 1.7325,
+      "step": 11947
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.272943980929678e-07,
+      "loss": 1.5214,
+      "step": 11948
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.272546682558602e-07,
+      "loss": 1.9299,
+      "step": 11949
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.272149384187524e-07,
+      "loss": 1.5916,
+      "step": 11950
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.271752085816448e-07,
+      "loss": 1.3635,
+      "step": 11951
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.271354787445371e-07,
+      "loss": 1.8904,
+      "step": 11952
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.270957489074295e-07,
+      "loss": 1.6717,
+      "step": 11953
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.270560190703219e-07,
+      "loss": 1.6531,
+      "step": 11954
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.270162892332142e-07,
+      "loss": 1.3863,
+      "step": 11955
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.269765593961064e-07,
+      "loss": 1.5116,
+      "step": 11956
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.269368295589987e-07,
+      "loss": 1.5968,
+      "step": 11957
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.268970997218911e-07,
+      "loss": 1.8914,
+      "step": 11958
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.268573698847834e-07,
+      "loss": 1.3707,
+      "step": 11959
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.268176400476758e-07,
+      "loss": 1.0102,
+      "step": 11960
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.267779102105682e-07,
+      "loss": 1.4039,
+      "step": 11961
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.267381803734604e-07,
+      "loss": 1.7572,
+      "step": 11962
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.266984505363528e-07,
+      "loss": 1.2527,
+      "step": 11963
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.266587206992451e-07,
+      "loss": 1.6488,
+      "step": 11964
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.266189908621375e-07,
+      "loss": 1.8255,
+      "step": 11965
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.265792610250298e-07,
+      "loss": 2.1739,
+      "step": 11966
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.265395311879222e-07,
+      "loss": 1.7482,
+      "step": 11967
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.264998013508144e-07,
+      "loss": 2.1485,
+      "step": 11968
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.264600715137067e-07,
+      "loss": 1.8009,
+      "step": 11969
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.264203416765991e-07,
+      "loss": 1.1748,
+      "step": 11970
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.263806118394914e-07,
+      "loss": 1.7132,
+      "step": 11971
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.263408820023838e-07,
+      "loss": 1.4553,
+      "step": 11972
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.263011521652761e-07,
+      "loss": 1.052,
+      "step": 11973
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.262614223281684e-07,
+      "loss": 1.6124,
+      "step": 11974
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.262216924910607e-07,
+      "loss": 1.8412,
+      "step": 11975
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.261819626539531e-07,
+      "loss": 1.7729,
+      "step": 11976
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.261422328168455e-07,
+      "loss": 1.7401,
+      "step": 11977
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.261025029797378e-07,
+      "loss": 1.7257,
+      "step": 11978
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.260627731426302e-07,
+      "loss": 2.3427,
+      "step": 11979
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.260230433055224e-07,
+      "loss": 1.8934,
+      "step": 11980
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.259833134684148e-07,
+      "loss": 1.5967,
+      "step": 11981
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.25943583631307e-07,
+      "loss": 1.9518,
+      "step": 11982
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.259038537941994e-07,
+      "loss": 1.7428,
+      "step": 11983
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.258641239570917e-07,
+      "loss": 1.9325,
+      "step": 11984
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.258243941199841e-07,
+      "loss": 1.7158,
+      "step": 11985
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.257846642828764e-07,
+      "loss": 1.532,
+      "step": 11986
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.257449344457687e-07,
+      "loss": 1.7779,
+      "step": 11987
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.257052046086611e-07,
+      "loss": 2.1346,
+      "step": 11988
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.256654747715534e-07,
+      "loss": 1.5575,
+      "step": 11989
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.256257449344458e-07,
+      "loss": 1.5985,
+      "step": 11990
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.255860150973381e-07,
+      "loss": 1.611,
+      "step": 11991
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.255462852602304e-07,
+      "loss": 2.0303,
+      "step": 11992
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.255065554231228e-07,
+      "loss": 1.9982,
+      "step": 11993
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.254668255860151e-07,
+      "loss": 1.5083,
+      "step": 11994
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.254270957489075e-07,
+      "loss": 1.688,
+      "step": 11995
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.253873659117997e-07,
+      "loss": 1.5533,
+      "step": 11996
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.253476360746921e-07,
+      "loss": 1.621,
+      "step": 11997
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.253079062375843e-07,
+      "loss": 1.21,
+      "step": 11998
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.252681764004767e-07,
+      "loss": 1.7705,
+      "step": 11999
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.25228446563369e-07,
+      "loss": 1.634,
+      "step": 12000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.251887167262614e-07,
+      "loss": 1.9064,
+      "step": 12001
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.251489868891538e-07,
+      "loss": 1.4547,
+      "step": 12002
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.251092570520461e-07,
+      "loss": 2.0374,
+      "step": 12003
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.250695272149384e-07,
+      "loss": 1.5953,
+      "step": 12004
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.250297973778307e-07,
+      "loss": 1.0955,
+      "step": 12005
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.249900675407231e-07,
+      "loss": 1.4166,
+      "step": 12006
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.249503377036155e-07,
+      "loss": 1.8394,
+      "step": 12007
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.249106078665078e-07,
+      "loss": 1.9831,
+      "step": 12008
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.248708780294001e-07,
+      "loss": 1.1815,
+      "step": 12009
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.248311481922923e-07,
+      "loss": 1.7444,
+      "step": 12010
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.247914183551847e-07,
+      "loss": 1.5578,
+      "step": 12011
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.24751688518077e-07,
+      "loss": 1.9236,
+      "step": 12012
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.247119586809694e-07,
+      "loss": 1.6615,
+      "step": 12013
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.246722288438617e-07,
+      "loss": 1.3665,
+      "step": 12014
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.246324990067541e-07,
+      "loss": 1.9851,
+      "step": 12015
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.245927691696464e-07,
+      "loss": 2.255,
+      "step": 12016
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.245530393325387e-07,
+      "loss": 1.5534,
+      "step": 12017
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.245133094954311e-07,
+      "loss": 1.5272,
+      "step": 12018
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.244735796583234e-07,
+      "loss": 1.51,
+      "step": 12019
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.244338498212158e-07,
+      "loss": 1.5219,
+      "step": 12020
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.243941199841081e-07,
+      "loss": 1.5601,
+      "step": 12021
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.243543901470003e-07,
+      "loss": 1.8601,
+      "step": 12022
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.243146603098926e-07,
+      "loss": 1.9084,
+      "step": 12023
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.24274930472785e-07,
+      "loss": 1.6042,
+      "step": 12024
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.242352006356774e-07,
+      "loss": 1.6587,
+      "step": 12025
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.241954707985697e-07,
+      "loss": 1.8601,
+      "step": 12026
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.241557409614621e-07,
+      "loss": 1.7272,
+      "step": 12027
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.241160111243543e-07,
+      "loss": 1.4648,
+      "step": 12028
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.240762812872467e-07,
+      "loss": 1.5359,
+      "step": 12029
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.24036551450139e-07,
+      "loss": 1.8242,
+      "step": 12030
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.239968216130314e-07,
+      "loss": 1.7444,
+      "step": 12031
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.239570917759238e-07,
+      "loss": 1.5578,
+      "step": 12032
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.239173619388161e-07,
+      "loss": 1.4898,
+      "step": 12033
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.238776321017084e-07,
+      "loss": 1.4199,
+      "step": 12034
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.238379022646006e-07,
+      "loss": 1.528,
+      "step": 12035
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.23798172427493e-07,
+      "loss": 1.3423,
+      "step": 12036
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.237584425903853e-07,
+      "loss": 2.1284,
+      "step": 12037
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.237187127532777e-07,
+      "loss": 2.1594,
+      "step": 12038
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.2367898291617e-07,
+      "loss": 1.5971,
+      "step": 12039
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.236392530790623e-07,
+      "loss": 1.4648,
+      "step": 12040
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.235995232419547e-07,
+      "loss": 1.9453,
+      "step": 12041
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.23559793404847e-07,
+      "loss": 1.4356,
+      "step": 12042
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.235200635677394e-07,
+      "loss": 1.9235,
+      "step": 12043
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.234803337306317e-07,
+      "loss": 1.4236,
+      "step": 12044
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.23440603893524e-07,
+      "loss": 1.692,
+      "step": 12045
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.234008740564163e-07,
+      "loss": 1.4266,
+      "step": 12046
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.233611442193087e-07,
+      "loss": 1.765,
+      "step": 12047
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.23321414382201e-07,
+      "loss": 1.6334,
+      "step": 12048
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.232816845450933e-07,
+      "loss": 1.3005,
+      "step": 12049
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.232419547079857e-07,
+      "loss": 1.6623,
+      "step": 12050
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.232022248708779e-07,
+      "loss": 1.095,
+      "step": 12051
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.231624950337703e-07,
+      "loss": 1.5424,
+      "step": 12052
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.231227651966626e-07,
+      "loss": 1.7325,
+      "step": 12053
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.23083035359555e-07,
+      "loss": 2.0729,
+      "step": 12054
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.230433055224474e-07,
+      "loss": 1.8413,
+      "step": 12055
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.230035756853397e-07,
+      "loss": 1.4539,
+      "step": 12056
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.22963845848232e-07,
+      "loss": 1.7856,
+      "step": 12057
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.229241160111243e-07,
+      "loss": 1.5957,
+      "step": 12058
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.228843861740167e-07,
+      "loss": 1.3459,
+      "step": 12059
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.22844656336909e-07,
+      "loss": 1.399,
+      "step": 12060
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.228049264998014e-07,
+      "loss": 1.4767,
+      "step": 12061
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.227651966626936e-07,
+      "loss": 1.8695,
+      "step": 12062
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.227254668255859e-07,
+      "loss": 1.7093,
+      "step": 12063
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.226857369884783e-07,
+      "loss": 1.5985,
+      "step": 12064
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.226460071513706e-07,
+      "loss": 1.5218,
+      "step": 12065
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.22606277314263e-07,
+      "loss": 1.7328,
+      "step": 12066
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.225665474771553e-07,
+      "loss": 1.7609,
+      "step": 12067
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.225268176400477e-07,
+      "loss": 1.2246,
+      "step": 12068
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.224870878029399e-07,
+      "loss": 1.3935,
+      "step": 12069
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.224473579658323e-07,
+      "loss": 1.8382,
+      "step": 12070
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.224076281287247e-07,
+      "loss": 1.5027,
+      "step": 12071
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.22367898291617e-07,
+      "loss": 1.4609,
+      "step": 12072
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.223281684545094e-07,
+      "loss": 1.6294,
+      "step": 12073
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.222884386174017e-07,
+      "loss": 1.5413,
+      "step": 12074
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.222487087802939e-07,
+      "loss": 1.8469,
+      "step": 12075
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.222089789431862e-07,
+      "loss": 1.3632,
+      "step": 12076
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.221692491060786e-07,
+      "loss": 1.7165,
+      "step": 12077
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.22129519268971e-07,
+      "loss": 1.472,
+      "step": 12078
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.220897894318633e-07,
+      "loss": 2.211,
+      "step": 12079
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.220500595947557e-07,
+      "loss": 1.8613,
+      "step": 12080
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.220103297576479e-07,
+      "loss": 1.7733,
+      "step": 12081
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.219705999205403e-07,
+      "loss": 1.4624,
+      "step": 12082
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.219308700834326e-07,
+      "loss": 1.5983,
+      "step": 12083
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.21891140246325e-07,
+      "loss": 1.7128,
+      "step": 12084
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.218514104092174e-07,
+      "loss": 1.2437,
+      "step": 12085
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.218116805721097e-07,
+      "loss": 1.3479,
+      "step": 12086
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.21771950735002e-07,
+      "loss": 1.4159,
+      "step": 12087
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.217322208978942e-07,
+      "loss": 1.7462,
+      "step": 12088
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.216924910607866e-07,
+      "loss": 1.5359,
+      "step": 12089
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.216527612236789e-07,
+      "loss": 1.2425,
+      "step": 12090
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.216130313865713e-07,
+      "loss": 1.9781,
+      "step": 12091
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.215733015494636e-07,
+      "loss": 1.1974,
+      "step": 12092
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.215335717123559e-07,
+      "loss": 1.5263,
+      "step": 12093
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.214938418752483e-07,
+      "loss": 1.7464,
+      "step": 12094
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.214541120381406e-07,
+      "loss": 1.7109,
+      "step": 12095
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.21414382201033e-07,
+      "loss": 1.1834,
+      "step": 12096
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.213746523639253e-07,
+      "loss": 1.539,
+      "step": 12097
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.213349225268177e-07,
+      "loss": 1.1611,
+      "step": 12098
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.212951926897099e-07,
+      "loss": 1.6879,
+      "step": 12099
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.212554628526023e-07,
+      "loss": 1.6172,
+      "step": 12100
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.212157330154945e-07,
+      "loss": 1.8354,
+      "step": 12101
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.211760031783869e-07,
+      "loss": 1.596,
+      "step": 12102
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.211362733412793e-07,
+      "loss": 1.0826,
+      "step": 12103
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.210965435041716e-07,
+      "loss": 1.917,
+      "step": 12104
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.210568136670639e-07,
+      "loss": 1.5265,
+      "step": 12105
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.210170838299562e-07,
+      "loss": 1.8433,
+      "step": 12106
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.209773539928486e-07,
+      "loss": 1.9302,
+      "step": 12107
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.209376241557409e-07,
+      "loss": 1.6606,
+      "step": 12108
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.208978943186333e-07,
+      "loss": 1.7907,
+      "step": 12109
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.208581644815257e-07,
+      "loss": 2.0415,
+      "step": 12110
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.208184346444179e-07,
+      "loss": 1.1924,
+      "step": 12111
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.207787048073103e-07,
+      "loss": 1.6456,
+      "step": 12112
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.207389749702026e-07,
+      "loss": 1.0981,
+      "step": 12113
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.206992451330949e-07,
+      "loss": 1.9493,
+      "step": 12114
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.206595152959872e-07,
+      "loss": 1.6528,
+      "step": 12115
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.206197854588796e-07,
+      "loss": 1.8579,
+      "step": 12116
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.205800556217718e-07,
+      "loss": 1.5874,
+      "step": 12117
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.205403257846642e-07,
+      "loss": 1.696,
+      "step": 12118
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.205005959475566e-07,
+      "loss": 1.1406,
+      "step": 12119
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.204608661104489e-07,
+      "loss": 1.136,
+      "step": 12120
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.204211362733413e-07,
+      "loss": 1.7619,
+      "step": 12121
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.203814064362336e-07,
+      "loss": 1.8602,
+      "step": 12122
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.203416765991259e-07,
+      "loss": 1.4615,
+      "step": 12123
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.203019467620182e-07,
+      "loss": 1.331,
+      "step": 12124
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.202622169249106e-07,
+      "loss": 1.5489,
+      "step": 12125
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.20222487087803e-07,
+      "loss": 1.2909,
+      "step": 12126
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.201827572506953e-07,
+      "loss": 1.7793,
+      "step": 12127
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.201430274135876e-07,
+      "loss": 1.6683,
+      "step": 12128
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.201032975764798e-07,
+      "loss": 1.6737,
+      "step": 12129
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.200635677393722e-07,
+      "loss": 1.6788,
+      "step": 12130
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.200238379022645e-07,
+      "loss": 1.2356,
+      "step": 12131
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.199841080651569e-07,
+      "loss": 1.9631,
+      "step": 12132
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.199443782280493e-07,
+      "loss": 1.4826,
+      "step": 12133
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.199046483909416e-07,
+      "loss": 1.0311,
+      "step": 12134
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.198649185538339e-07,
+      "loss": 1.9031,
+      "step": 12135
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.198251887167262e-07,
+      "loss": 1.5198,
+      "step": 12136
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.197854588796186e-07,
+      "loss": 1.1499,
+      "step": 12137
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.197457290425109e-07,
+      "loss": 1.2585,
+      "step": 12138
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.197059992054033e-07,
+      "loss": 1.8732,
+      "step": 12139
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.196662693682957e-07,
+      "loss": 1.7193,
+      "step": 12140
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.196265395311878e-07,
+      "loss": 1.844,
+      "step": 12141
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.195868096940802e-07,
+      "loss": 1.4994,
+      "step": 12142
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.195470798569725e-07,
+      "loss": 1.8446,
+      "step": 12143
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.195073500198649e-07,
+      "loss": 1.3341,
+      "step": 12144
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.194676201827572e-07,
+      "loss": 1.238,
+      "step": 12145
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.194278903456496e-07,
+      "loss": 1.1197,
+      "step": 12146
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.193881605085418e-07,
+      "loss": 1.5951,
+      "step": 12147
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.193484306714342e-07,
+      "loss": 1.7715,
+      "step": 12148
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.193087008343266e-07,
+      "loss": 1.2348,
+      "step": 12149
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.192689709972189e-07,
+      "loss": 1.7872,
+      "step": 12150
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.192292411601113e-07,
+      "loss": 1.2781,
+      "step": 12151
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.191895113230036e-07,
+      "loss": 1.9796,
+      "step": 12152
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.191497814858959e-07,
+      "loss": 1.3057,
+      "step": 12153
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.191100516487881e-07,
+      "loss": 1.3506,
+      "step": 12154
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.190703218116805e-07,
+      "loss": 1.7091,
+      "step": 12155
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.190305919745729e-07,
+      "loss": 1.523,
+      "step": 12156
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.189908621374652e-07,
+      "loss": 1.4999,
+      "step": 12157
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.189511323003576e-07,
+      "loss": 1.2631,
+      "step": 12158
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.189114024632498e-07,
+      "loss": 1.506,
+      "step": 12159
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.188716726261422e-07,
+      "loss": 1.5776,
+      "step": 12160
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.188319427890345e-07,
+      "loss": 1.5222,
+      "step": 12161
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.187922129519269e-07,
+      "loss": 1.8235,
+      "step": 12162
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.187524831148193e-07,
+      "loss": 1.2018,
+      "step": 12163
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.187127532777116e-07,
+      "loss": 1.7259,
+      "step": 12164
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.186730234406039e-07,
+      "loss": 1.9914,
+      "step": 12165
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.186332936034962e-07,
+      "loss": 2.1045,
+      "step": 12166
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.185935637663885e-07,
+      "loss": 1.8451,
+      "step": 12167
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.185538339292808e-07,
+      "loss": 1.6064,
+      "step": 12168
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.185141040921732e-07,
+      "loss": 1.6008,
+      "step": 12169
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.184743742550655e-07,
+      "loss": 1.7866,
+      "step": 12170
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.184346444179578e-07,
+      "loss": 1.4718,
+      "step": 12171
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.183949145808502e-07,
+      "loss": 1.7152,
+      "step": 12172
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.183551847437425e-07,
+      "loss": 1.4363,
+      "step": 12173
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.183154549066349e-07,
+      "loss": 1.9469,
+      "step": 12174
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.182757250695272e-07,
+      "loss": 1.8678,
+      "step": 12175
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.182359952324196e-07,
+      "loss": 1.9084,
+      "step": 12176
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.181962653953118e-07,
+      "loss": 1.441,
+      "step": 12177
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.181565355582042e-07,
+      "loss": 1.7453,
+      "step": 12178
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.181168057210966e-07,
+      "loss": 1.7554,
+      "step": 12179
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.180770758839888e-07,
+      "loss": 1.3734,
+      "step": 12180
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.180373460468812e-07,
+      "loss": 1.4162,
+      "step": 12181
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.179976162097735e-07,
+      "loss": 1.6467,
+      "step": 12182
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.179578863726658e-07,
+      "loss": 1.4163,
+      "step": 12183
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.179181565355581e-07,
+      "loss": 1.8567,
+      "step": 12184
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.178784266984505e-07,
+      "loss": 1.542,
+      "step": 12185
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.178386968613428e-07,
+      "loss": 1.6641,
+      "step": 12186
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.177989670242352e-07,
+      "loss": 2.1047,
+      "step": 12187
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.177592371871276e-07,
+      "loss": 1.835,
+      "step": 12188
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.177195073500198e-07,
+      "loss": 1.3456,
+      "step": 12189
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.176797775129122e-07,
+      "loss": 1.6649,
+      "step": 12190
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.176400476758045e-07,
+      "loss": 1.6813,
+      "step": 12191
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.176003178386969e-07,
+      "loss": 1.0074,
+      "step": 12192
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.175605880015892e-07,
+      "loss": 1.3134,
+      "step": 12193
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.175208581644815e-07,
+      "loss": 1.7917,
+      "step": 12194
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.174811283273737e-07,
+      "loss": 1.342,
+      "step": 12195
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.174413984902661e-07,
+      "loss": 1.5766,
+      "step": 12196
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.174016686531585e-07,
+      "loss": 1.3873,
+      "step": 12197
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.173619388160508e-07,
+      "loss": 1.6553,
+      "step": 12198
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.173222089789432e-07,
+      "loss": 1.5519,
+      "step": 12199
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.172824791418355e-07,
+      "loss": 1.9234,
+      "step": 12200
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.172427493047278e-07,
+      "loss": 1.9747,
+      "step": 12201
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.172030194676201e-07,
+      "loss": 1.6054,
+      "step": 12202
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.171632896305125e-07,
+      "loss": 1.7621,
+      "step": 12203
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.171235597934049e-07,
+      "loss": 2.137,
+      "step": 12204
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.170838299562972e-07,
+      "loss": 1.5838,
+      "step": 12205
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.170441001191896e-07,
+      "loss": 1.6198,
+      "step": 12206
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.170043702820817e-07,
+      "loss": 1.395,
+      "step": 12207
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.169646404449741e-07,
+      "loss": 1.7288,
+      "step": 12208
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.169249106078664e-07,
+      "loss": 1.5188,
+      "step": 12209
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.168851807707588e-07,
+      "loss": 2.1146,
+      "step": 12210
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.168454509336512e-07,
+      "loss": 1.4575,
+      "step": 12211
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.168057210965435e-07,
+      "loss": 2.4636,
+      "step": 12212
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.167659912594358e-07,
+      "loss": 1.7621,
+      "step": 12213
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.167262614223281e-07,
+      "loss": 1.1249,
+      "step": 12214
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.166865315852205e-07,
+      "loss": 1.4899,
+      "step": 12215
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.166468017481128e-07,
+      "loss": 1.3124,
+      "step": 12216
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.166070719110052e-07,
+      "loss": 1.4836,
+      "step": 12217
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.165673420738976e-07,
+      "loss": 1.6021,
+      "step": 12218
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.165276122367898e-07,
+      "loss": 1.9869,
+      "step": 12219
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.164878823996821e-07,
+      "loss": 1.5079,
+      "step": 12220
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.164481525625744e-07,
+      "loss": 1.636,
+      "step": 12221
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.164084227254668e-07,
+      "loss": 1.5316,
+      "step": 12222
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.163686928883591e-07,
+      "loss": 1.585,
+      "step": 12223
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.163289630512515e-07,
+      "loss": 1.4856,
+      "step": 12224
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.162892332141437e-07,
+      "loss": 1.5107,
+      "step": 12225
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.162495033770361e-07,
+      "loss": 1.7316,
+      "step": 12226
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.162097735399285e-07,
+      "loss": 2.1691,
+      "step": 12227
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.161700437028208e-07,
+      "loss": 1.4733,
+      "step": 12228
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.161303138657132e-07,
+      "loss": 2.0123,
+      "step": 12229
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.160905840286055e-07,
+      "loss": 1.7896,
+      "step": 12230
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.160508541914978e-07,
+      "loss": 1.7483,
+      "step": 12231
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.160111243543901e-07,
+      "loss": 1.6902,
+      "step": 12232
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.159713945172824e-07,
+      "loss": 1.68,
+      "step": 12233
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.159316646801748e-07,
+      "loss": 1.6236,
+      "step": 12234
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.158919348430671e-07,
+      "loss": 1.3336,
+      "step": 12235
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.158522050059595e-07,
+      "loss": 0.9674,
+      "step": 12236
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.158124751688517e-07,
+      "loss": 1.6042,
+      "step": 12237
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.157727453317441e-07,
+      "loss": 1.8123,
+      "step": 12238
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.157330154946364e-07,
+      "loss": 1.7283,
+      "step": 12239
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.156932856575288e-07,
+      "loss": 1.5819,
+      "step": 12240
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.156535558204212e-07,
+      "loss": 1.944,
+      "step": 12241
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.156138259833135e-07,
+      "loss": 1.3458,
+      "step": 12242
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.155740961462058e-07,
+      "loss": 1.6284,
+      "step": 12243
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.155343663090981e-07,
+      "loss": 1.0706,
+      "step": 12244
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.154946364719905e-07,
+      "loss": 1.3465,
+      "step": 12245
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.154549066348827e-07,
+      "loss": 1.5,
+      "step": 12246
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.154151767977751e-07,
+      "loss": 1.5322,
+      "step": 12247
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.153754469606674e-07,
+      "loss": 1.4192,
+      "step": 12248
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.153357171235597e-07,
+      "loss": 1.3321,
+      "step": 12249
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.152959872864521e-07,
+      "loss": 1.5398,
+      "step": 12250
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.152562574493444e-07,
+      "loss": 1.908,
+      "step": 12251
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.152165276122368e-07,
+      "loss": 1.505,
+      "step": 12252
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.151767977751291e-07,
+      "loss": 1.275,
+      "step": 12253
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.151370679380215e-07,
+      "loss": 1.5251,
+      "step": 12254
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.150973381009137e-07,
+      "loss": 1.6213,
+      "step": 12255
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.150576082638061e-07,
+      "loss": 1.6734,
+      "step": 12256
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.150178784266985e-07,
+      "loss": 1.3482,
+      "step": 12257
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.149781485895908e-07,
+      "loss": 1.5939,
+      "step": 12258
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.149384187524832e-07,
+      "loss": 1.5338,
+      "step": 12259
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.148986889153754e-07,
+      "loss": 1.5223,
+      "step": 12260
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.148589590782677e-07,
+      "loss": 1.8805,
+      "step": 12261
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.1481922924116e-07,
+      "loss": 1.864,
+      "step": 12262
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.147794994040524e-07,
+      "loss": 1.4956,
+      "step": 12263
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.147397695669447e-07,
+      "loss": 1.8319,
+      "step": 12264
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.147000397298371e-07,
+      "loss": 1.6721,
+      "step": 12265
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.146603098927295e-07,
+      "loss": 1.6571,
+      "step": 12266
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.146205800556217e-07,
+      "loss": 1.5465,
+      "step": 12267
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.145808502185141e-07,
+      "loss": 1.705,
+      "step": 12268
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.145411203814064e-07,
+      "loss": 1.5051,
+      "step": 12269
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.145013905442988e-07,
+      "loss": 1.436,
+      "step": 12270
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.144616607071911e-07,
+      "loss": 1.703,
+      "step": 12271
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.144219308700835e-07,
+      "loss": 1.4986,
+      "step": 12272
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.143822010329756e-07,
+      "loss": 1.5311,
+      "step": 12273
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.14342471195868e-07,
+      "loss": 2.0367,
+      "step": 12274
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.143027413587604e-07,
+      "loss": 1.4074,
+      "step": 12275
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.142630115216527e-07,
+      "loss": 1.9177,
+      "step": 12276
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.142232816845451e-07,
+      "loss": 1.5084,
+      "step": 12277
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.141835518474374e-07,
+      "loss": 1.8813,
+      "step": 12278
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.141438220103297e-07,
+      "loss": 1.7829,
+      "step": 12279
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.14104092173222e-07,
+      "loss": 1.6924,
+      "step": 12280
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.140643623361144e-07,
+      "loss": 1.3677,
+      "step": 12281
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.140246324990068e-07,
+      "loss": 1.5932,
+      "step": 12282
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.139849026618991e-07,
+      "loss": 1.5335,
+      "step": 12283
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.139451728247915e-07,
+      "loss": 1.3703,
+      "step": 12284
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.139054429876837e-07,
+      "loss": 1.4177,
+      "step": 12285
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.13865713150576e-07,
+      "loss": 1.877,
+      "step": 12286
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.138259833134683e-07,
+      "loss": 1.6215,
+      "step": 12287
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.137862534763607e-07,
+      "loss": 1.8112,
+      "step": 12288
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.137465236392531e-07,
+      "loss": 1.2455,
+      "step": 12289
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.137067938021454e-07,
+      "loss": 1.6044,
+      "step": 12290
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.136670639650377e-07,
+      "loss": 1.7676,
+      "step": 12291
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.1362733412793e-07,
+      "loss": 1.8091,
+      "step": 12292
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.135876042908224e-07,
+      "loss": 1.3575,
+      "step": 12293
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.135478744537147e-07,
+      "loss": 1.7696,
+      "step": 12294
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.135081446166071e-07,
+      "loss": 1.6134,
+      "step": 12295
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.134684147794995e-07,
+      "loss": 2.0323,
+      "step": 12296
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.134286849423917e-07,
+      "loss": 1.6601,
+      "step": 12297
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.133889551052841e-07,
+      "loss": 1.7389,
+      "step": 12298
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.133492252681763e-07,
+      "loss": 2.0262,
+      "step": 12299
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.133094954310687e-07,
+      "loss": 1.8886,
+      "step": 12300
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.13269765593961e-07,
+      "loss": 1.6639,
+      "step": 12301
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.132300357568534e-07,
+      "loss": 1.7484,
+      "step": 12302
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.131903059197456e-07,
+      "loss": 1.4104,
+      "step": 12303
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.13150576082638e-07,
+      "loss": 1.3596,
+      "step": 12304
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.131108462455304e-07,
+      "loss": 1.4751,
+      "step": 12305
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.130711164084227e-07,
+      "loss": 1.5807,
+      "step": 12306
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.130313865713151e-07,
+      "loss": 1.0435,
+      "step": 12307
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.129916567342074e-07,
+      "loss": 1.7639,
+      "step": 12308
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.129519268970997e-07,
+      "loss": 1.6397,
+      "step": 12309
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.12912197059992e-07,
+      "loss": 1.5354,
+      "step": 12310
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.128724672228844e-07,
+      "loss": 1.7454,
+      "step": 12311
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.128327373857767e-07,
+      "loss": 1.4963,
+      "step": 12312
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.12793007548669e-07,
+      "loss": 1.3079,
+      "step": 12313
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.127532777115614e-07,
+      "loss": 1.3656,
+      "step": 12314
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.127135478744536e-07,
+      "loss": 1.6977,
+      "step": 12315
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.12673818037346e-07,
+      "loss": 1.2158,
+      "step": 12316
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.126340882002383e-07,
+      "loss": 1.4308,
+      "step": 12317
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.125943583631307e-07,
+      "loss": 1.4668,
+      "step": 12318
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.12554628526023e-07,
+      "loss": 2.1036,
+      "step": 12319
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.125148986889154e-07,
+      "loss": 1.3968,
+      "step": 12320
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.124751688518077e-07,
+      "loss": 1.445,
+      "step": 12321
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.124354390147e-07,
+      "loss": 1.8466,
+      "step": 12322
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.123957091775924e-07,
+      "loss": 1.5516,
+      "step": 12323
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.123559793404847e-07,
+      "loss": 1.4108,
+      "step": 12324
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.123162495033771e-07,
+      "loss": 1.5875,
+      "step": 12325
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.122765196662693e-07,
+      "loss": 1.9017,
+      "step": 12326
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.122367898291616e-07,
+      "loss": 1.916,
+      "step": 12327
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.12197059992054e-07,
+      "loss": 1.2768,
+      "step": 12328
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.121573301549463e-07,
+      "loss": 1.5147,
+      "step": 12329
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.121176003178387e-07,
+      "loss": 1.1895,
+      "step": 12330
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.12077870480731e-07,
+      "loss": 1.2998,
+      "step": 12331
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.120381406436234e-07,
+      "loss": 1.2211,
+      "step": 12332
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.119984108065156e-07,
+      "loss": 1.5946,
+      "step": 12333
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.11958680969408e-07,
+      "loss": 1.8076,
+      "step": 12334
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.119189511323004e-07,
+      "loss": 2.0246,
+      "step": 12335
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.118792212951927e-07,
+      "loss": 1.5689,
+      "step": 12336
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.118394914580851e-07,
+      "loss": 1.5133,
+      "step": 12337
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.117997616209774e-07,
+      "loss": 2.1212,
+      "step": 12338
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.117600317838696e-07,
+      "loss": 1.8938,
+      "step": 12339
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.117203019467619e-07,
+      "loss": 2.0141,
+      "step": 12340
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.116805721096543e-07,
+      "loss": 1.4841,
+      "step": 12341
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.116408422725466e-07,
+      "loss": 1.6615,
+      "step": 12342
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.11601112435439e-07,
+      "loss": 1.1852,
+      "step": 12343
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.115613825983314e-07,
+      "loss": 1.5625,
+      "step": 12344
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.115216527612236e-07,
+      "loss": 1.3787,
+      "step": 12345
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.11481922924116e-07,
+      "loss": 1.9226,
+      "step": 12346
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.114421930870083e-07,
+      "loss": 1.216,
+      "step": 12347
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.114024632499007e-07,
+      "loss": 2.2844,
+      "step": 12348
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.11362733412793e-07,
+      "loss": 1.4666,
+      "step": 12349
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.113230035756854e-07,
+      "loss": 1.7069,
+      "step": 12350
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.112832737385777e-07,
+      "loss": 1.5042,
+      "step": 12351
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.112435439014699e-07,
+      "loss": 1.091,
+      "step": 12352
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.112038140643623e-07,
+      "loss": 1.5301,
+      "step": 12353
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.111640842272546e-07,
+      "loss": 1.5464,
+      "step": 12354
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.11124354390147e-07,
+      "loss": 1.7565,
+      "step": 12355
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.110846245530393e-07,
+      "loss": 1.6458,
+      "step": 12356
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.110448947159316e-07,
+      "loss": 1.7131,
+      "step": 12357
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.11005164878824e-07,
+      "loss": 1.8054,
+      "step": 12358
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.109654350417163e-07,
+      "loss": 1.7458,
+      "step": 12359
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.109257052046087e-07,
+      "loss": 1.6838,
+      "step": 12360
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.10885975367501e-07,
+      "loss": 1.7093,
+      "step": 12361
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.108462455303934e-07,
+      "loss": 1.4046,
+      "step": 12362
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.108065156932856e-07,
+      "loss": 1.5179,
+      "step": 12363
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.10766785856178e-07,
+      "loss": 1.7406,
+      "step": 12364
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.107270560190702e-07,
+      "loss": 1.4881,
+      "step": 12365
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.106873261819626e-07,
+      "loss": 1.6831,
+      "step": 12366
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.10647596344855e-07,
+      "loss": 1.4196,
+      "step": 12367
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.106078665077473e-07,
+      "loss": 1.1915,
+      "step": 12368
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.105681366706396e-07,
+      "loss": 1.7349,
+      "step": 12369
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.105284068335319e-07,
+      "loss": 2.2359,
+      "step": 12370
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.104886769964243e-07,
+      "loss": 1.2482,
+      "step": 12371
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.104489471593166e-07,
+      "loss": 1.4651,
+      "step": 12372
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.10409217322209e-07,
+      "loss": 2.4124,
+      "step": 12373
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.103694874851014e-07,
+      "loss": 1.5631,
+      "step": 12374
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.103297576479936e-07,
+      "loss": 1.9688,
+      "step": 12375
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.10290027810886e-07,
+      "loss": 1.5551,
+      "step": 12376
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.102502979737783e-07,
+      "loss": 1.1357,
+      "step": 12377
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.102105681366706e-07,
+      "loss": 1.8779,
+      "step": 12378
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.101708382995629e-07,
+      "loss": 1.2928,
+      "step": 12379
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.101311084624553e-07,
+      "loss": 1.7448,
+      "step": 12380
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.100913786253475e-07,
+      "loss": 1.6482,
+      "step": 12381
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.100516487882399e-07,
+      "loss": 1.8623,
+      "step": 12382
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.100119189511323e-07,
+      "loss": 1.98,
+      "step": 12383
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.099721891140246e-07,
+      "loss": 1.1322,
+      "step": 12384
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.09932459276917e-07,
+      "loss": 0.9331,
+      "step": 12385
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.098927294398093e-07,
+      "loss": 1.6628,
+      "step": 12386
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.098529996027016e-07,
+      "loss": 1.6356,
+      "step": 12387
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.098132697655939e-07,
+      "loss": 1.6889,
+      "step": 12388
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.097735399284863e-07,
+      "loss": 1.6061,
+      "step": 12389
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.097338100913787e-07,
+      "loss": 1.3658,
+      "step": 12390
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.09694080254271e-07,
+      "loss": 1.6896,
+      "step": 12391
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.096543504171633e-07,
+      "loss": 1.7413,
+      "step": 12392
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.096146205800555e-07,
+      "loss": 1.5721,
+      "step": 12393
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.095748907429479e-07,
+      "loss": 1.4741,
+      "step": 12394
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.095351609058402e-07,
+      "loss": 1.6661,
+      "step": 12395
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.094954310687326e-07,
+      "loss": 1.5794,
+      "step": 12396
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.09455701231625e-07,
+      "loss": 1.7698,
+      "step": 12397
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.094159713945173e-07,
+      "loss": 1.4908,
+      "step": 12398
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.093762415574096e-07,
+      "loss": 1.6168,
+      "step": 12399
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.093365117203019e-07,
+      "loss": 1.8131,
+      "step": 12400
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.092967818831943e-07,
+      "loss": 1.7556,
+      "step": 12401
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.092570520460866e-07,
+      "loss": 1.795,
+      "step": 12402
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.09217322208979e-07,
+      "loss": 1.7956,
+      "step": 12403
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.091775923718714e-07,
+      "loss": 1.4131,
+      "step": 12404
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.091378625347635e-07,
+      "loss": 1.2888,
+      "step": 12405
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.090981326976559e-07,
+      "loss": 1.706,
+      "step": 12406
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.090584028605482e-07,
+      "loss": 1.5962,
+      "step": 12407
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.090186730234406e-07,
+      "loss": 1.8601,
+      "step": 12408
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.089789431863329e-07,
+      "loss": 1.9132,
+      "step": 12409
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.089392133492253e-07,
+      "loss": 1.5403,
+      "step": 12410
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.088994835121175e-07,
+      "loss": 1.3805,
+      "step": 12411
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.088597536750099e-07,
+      "loss": 1.3387,
+      "step": 12412
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.088200238379023e-07,
+      "loss": 1.4719,
+      "step": 12413
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.087802940007946e-07,
+      "loss": 1.4533,
+      "step": 12414
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.08740564163687e-07,
+      "loss": 1.7672,
+      "step": 12415
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.087008343265793e-07,
+      "loss": 2.0205,
+      "step": 12416
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.086611044894716e-07,
+      "loss": 1.2635,
+      "step": 12417
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.086213746523638e-07,
+      "loss": 1.5604,
+      "step": 12418
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.085816448152562e-07,
+      "loss": 1.7726,
+      "step": 12419
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.085419149781485e-07,
+      "loss": 1.3361,
+      "step": 12420
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.085021851410409e-07,
+      "loss": 1.7882,
+      "step": 12421
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.084624553039333e-07,
+      "loss": 1.7247,
+      "step": 12422
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.084227254668255e-07,
+      "loss": 1.7652,
+      "step": 12423
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.083829956297179e-07,
+      "loss": 1.9796,
+      "step": 12424
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.083432657926102e-07,
+      "loss": 1.4425,
+      "step": 12425
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.083035359555026e-07,
+      "loss": 1.9083,
+      "step": 12426
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.082638061183949e-07,
+      "loss": 1.6779,
+      "step": 12427
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.082240762812873e-07,
+      "loss": 1.6188,
+      "step": 12428
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.081843464441796e-07,
+      "loss": 1.9369,
+      "step": 12429
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.081446166070719e-07,
+      "loss": 1.5747,
+      "step": 12430
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.081048867699642e-07,
+      "loss": 1.485,
+      "step": 12431
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.080651569328565e-07,
+      "loss": 1.3644,
+      "step": 12432
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.080254270957489e-07,
+      "loss": 1.8648,
+      "step": 12433
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.079856972586412e-07,
+      "loss": 1.3757,
+      "step": 12434
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.079459674215335e-07,
+      "loss": 2.0003,
+      "step": 12435
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.079062375844258e-07,
+      "loss": 1.4317,
+      "step": 12436
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.078665077473182e-07,
+      "loss": 1.6533,
+      "step": 12437
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.078267779102106e-07,
+      "loss": 1.9994,
+      "step": 12438
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.077870480731029e-07,
+      "loss": 1.6244,
+      "step": 12439
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.077473182359953e-07,
+      "loss": 1.2611,
+      "step": 12440
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.077075883988875e-07,
+      "loss": 1.6867,
+      "step": 12441
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.076678585617799e-07,
+      "loss": 1.9706,
+      "step": 12442
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.076281287246722e-07,
+      "loss": 1.6054,
+      "step": 12443
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.075883988875645e-07,
+      "loss": 1.4906,
+      "step": 12444
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.075486690504569e-07,
+      "loss": 1.8571,
+      "step": 12445
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.075089392133492e-07,
+      "loss": 1.8858,
+      "step": 12446
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.074692093762415e-07,
+      "loss": 1.7367,
+      "step": 12447
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.074294795391338e-07,
+      "loss": 0.7174,
+      "step": 12448
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.073897497020262e-07,
+      "loss": 1.7739,
+      "step": 12449
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.073500198649185e-07,
+      "loss": 1.3908,
+      "step": 12450
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.073102900278109e-07,
+      "loss": 1.0573,
+      "step": 12451
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.072705601907033e-07,
+      "loss": 1.8094,
+      "step": 12452
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.072308303535955e-07,
+      "loss": 1.5115,
+      "step": 12453
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.071911005164879e-07,
+      "loss": 1.2749,
+      "step": 12454
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.071513706793802e-07,
+      "loss": 1.3248,
+      "step": 12455
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.071116408422726e-07,
+      "loss": 2.131,
+      "step": 12456
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.070719110051649e-07,
+      "loss": 1.7584,
+      "step": 12457
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.070321811680572e-07,
+      "loss": 1.3286,
+      "step": 12458
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.069924513309494e-07,
+      "loss": 1.6107,
+      "step": 12459
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.069527214938418e-07,
+      "loss": 1.4643,
+      "step": 12460
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.069129916567342e-07,
+      "loss": 1.3531,
+      "step": 12461
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.068732618196265e-07,
+      "loss": 1.7094,
+      "step": 12462
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.068335319825189e-07,
+      "loss": 1.4643,
+      "step": 12463
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.067938021454112e-07,
+      "loss": 1.8187,
+      "step": 12464
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.067540723083035e-07,
+      "loss": 1.4947,
+      "step": 12465
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.067143424711958e-07,
+      "loss": 2.0225,
+      "step": 12466
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.066746126340882e-07,
+      "loss": 1.5293,
+      "step": 12467
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.066348827969806e-07,
+      "loss": 1.4168,
+      "step": 12468
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.065951529598729e-07,
+      "loss": 1.3192,
+      "step": 12469
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.065554231227653e-07,
+      "loss": 1.2924,
+      "step": 12470
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.065156932856574e-07,
+      "loss": 1.86,
+      "step": 12471
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.064759634485498e-07,
+      "loss": 1.4103,
+      "step": 12472
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.064362336114421e-07,
+      "loss": 1.6374,
+      "step": 12473
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.063965037743345e-07,
+      "loss": 1.5576,
+      "step": 12474
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.063567739372269e-07,
+      "loss": 1.617,
+      "step": 12475
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.063170441001192e-07,
+      "loss": 1.4976,
+      "step": 12476
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.062773142630115e-07,
+      "loss": 2.1534,
+      "step": 12477
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.062375844259038e-07,
+      "loss": 1.6392,
+      "step": 12478
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.061978545887962e-07,
+      "loss": 2.1593,
+      "step": 12479
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.061581247516885e-07,
+      "loss": 1.9486,
+      "step": 12480
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.061183949145809e-07,
+      "loss": 1.3306,
+      "step": 12481
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.060786650774733e-07,
+      "loss": 1.808,
+      "step": 12482
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.060389352403655e-07,
+      "loss": 1.7556,
+      "step": 12483
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.059992054032578e-07,
+      "loss": 1.8158,
+      "step": 12484
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.059594755661501e-07,
+      "loss": 1.926,
+      "step": 12485
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.059197457290425e-07,
+      "loss": 1.6151,
+      "step": 12486
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.058800158919348e-07,
+      "loss": 2.0045,
+      "step": 12487
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.058402860548272e-07,
+      "loss": 1.4968,
+      "step": 12488
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.058005562177194e-07,
+      "loss": 1.7039,
+      "step": 12489
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.057608263806118e-07,
+      "loss": 1.8815,
+      "step": 12490
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.057210965435042e-07,
+      "loss": 1.6654,
+      "step": 12491
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.056813667063965e-07,
+      "loss": 1.6043,
+      "step": 12492
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.056416368692889e-07,
+      "loss": 1.6695,
+      "step": 12493
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.056019070321812e-07,
+      "loss": 1.2109,
+      "step": 12494
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.055621771950735e-07,
+      "loss": 1.7982,
+      "step": 12495
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.055224473579658e-07,
+      "loss": 1.4182,
+      "step": 12496
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.054827175208581e-07,
+      "loss": 1.1071,
+      "step": 12497
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.054429876837504e-07,
+      "loss": 1.4534,
+      "step": 12498
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.054032578466428e-07,
+      "loss": 1.3856,
+      "step": 12499
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.053635280095352e-07,
+      "loss": 1.4951,
+      "step": 12500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.053237981724274e-07,
+      "loss": 1.2622,
+      "step": 12501
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.052840683353198e-07,
+      "loss": 1.243,
+      "step": 12502
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.052443384982121e-07,
+      "loss": 1.8868,
+      "step": 12503
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.052046086611045e-07,
+      "loss": 1.0924,
+      "step": 12504
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.051648788239968e-07,
+      "loss": 2.0987,
+      "step": 12505
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.051251489868892e-07,
+      "loss": 2.1169,
+      "step": 12506
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.050854191497815e-07,
+      "loss": 1.2963,
+      "step": 12507
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.050456893126738e-07,
+      "loss": 1.6444,
+      "step": 12508
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.050059594755662e-07,
+      "loss": 1.4035,
+      "step": 12509
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.049662296384584e-07,
+      "loss": 2.2926,
+      "step": 12510
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.049264998013508e-07,
+      "loss": 1.6283,
+      "step": 12511
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.048867699642431e-07,
+      "loss": 1.3157,
+      "step": 12512
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.048470401271354e-07,
+      "loss": 1.7258,
+      "step": 12513
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.048073102900277e-07,
+      "loss": 1.6559,
+      "step": 12514
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.047675804529201e-07,
+      "loss": 1.9411,
+      "step": 12515
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.047278506158125e-07,
+      "loss": 1.4338,
+      "step": 12516
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.046881207787048e-07,
+      "loss": 1.8036,
+      "step": 12517
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.046483909415972e-07,
+      "loss": 1.347,
+      "step": 12518
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.046086611044894e-07,
+      "loss": 1.621,
+      "step": 12519
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.045689312673818e-07,
+      "loss": 1.6255,
+      "step": 12520
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.045292014302741e-07,
+      "loss": 1.828,
+      "step": 12521
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.044894715931665e-07,
+      "loss": 1.9204,
+      "step": 12522
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.044497417560589e-07,
+      "loss": 1.6845,
+      "step": 12523
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.044100119189511e-07,
+      "loss": 1.5976,
+      "step": 12524
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.043702820818434e-07,
+      "loss": 1.4839,
+      "step": 12525
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.043305522447357e-07,
+      "loss": 1.3935,
+      "step": 12526
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.042908224076281e-07,
+      "loss": 1.0985,
+      "step": 12527
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.042510925705204e-07,
+      "loss": 1.4526,
+      "step": 12528
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.042113627334128e-07,
+      "loss": 1.3889,
+      "step": 12529
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.041716328963052e-07,
+      "loss": 1.4656,
+      "step": 12530
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.041319030591974e-07,
+      "loss": 2.3145,
+      "step": 12531
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.040921732220898e-07,
+      "loss": 1.7347,
+      "step": 12532
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.040524433849821e-07,
+      "loss": 1.2681,
+      "step": 12533
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.040127135478745e-07,
+      "loss": 1.6624,
+      "step": 12534
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.039729837107668e-07,
+      "loss": 1.2283,
+      "step": 12535
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.039332538736592e-07,
+      "loss": 1.4791,
+      "step": 12536
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.038935240365513e-07,
+      "loss": 1.9095,
+      "step": 12537
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.038537941994437e-07,
+      "loss": 1.7555,
+      "step": 12538
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.038140643623361e-07,
+      "loss": 1.6134,
+      "step": 12539
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.037743345252284e-07,
+      "loss": 1.7463,
+      "step": 12540
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.037346046881208e-07,
+      "loss": 1.5172,
+      "step": 12541
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.036948748510131e-07,
+      "loss": 1.4454,
+      "step": 12542
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.036551450139054e-07,
+      "loss": 1.38,
+      "step": 12543
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.036154151767977e-07,
+      "loss": 1.7098,
+      "step": 12544
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.035756853396901e-07,
+      "loss": 1.5754,
+      "step": 12545
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.035359555025825e-07,
+      "loss": 1.9965,
+      "step": 12546
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.034962256654748e-07,
+      "loss": 1.6652,
+      "step": 12547
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.034564958283672e-07,
+      "loss": 1.2896,
+      "step": 12548
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.034167659912594e-07,
+      "loss": 1.7121,
+      "step": 12549
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.033770361541517e-07,
+      "loss": 1.7233,
+      "step": 12550
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.03337306317044e-07,
+      "loss": 1.3585,
+      "step": 12551
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.032975764799364e-07,
+      "loss": 1.6799,
+      "step": 12552
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.032578466428288e-07,
+      "loss": 1.3126,
+      "step": 12553
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.032181168057211e-07,
+      "loss": 1.2152,
+      "step": 12554
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.031783869686134e-07,
+      "loss": 1.4224,
+      "step": 12555
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.031386571315057e-07,
+      "loss": 2.0498,
+      "step": 12556
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.030989272943981e-07,
+      "loss": 1.3454,
+      "step": 12557
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.030591974572904e-07,
+      "loss": 1.6712,
+      "step": 12558
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.030194676201828e-07,
+      "loss": 2.0375,
+      "step": 12559
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.029797377830752e-07,
+      "loss": 1.7764,
+      "step": 12560
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.029400079459674e-07,
+      "loss": 1.6123,
+      "step": 12561
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.029002781088598e-07,
+      "loss": 1.7081,
+      "step": 12562
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.02860548271752e-07,
+      "loss": 1.5462,
+      "step": 12563
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.028208184346444e-07,
+      "loss": 2.043,
+      "step": 12564
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.027810885975367e-07,
+      "loss": 0.9775,
+      "step": 12565
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.027413587604291e-07,
+      "loss": 2.0643,
+      "step": 12566
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.027016289233213e-07,
+      "loss": 1.3776,
+      "step": 12567
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.026618990862137e-07,
+      "loss": 1.3327,
+      "step": 12568
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.026221692491061e-07,
+      "loss": 1.4396,
+      "step": 12569
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.025824394119984e-07,
+      "loss": 1.7866,
+      "step": 12570
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.025427095748908e-07,
+      "loss": 1.678,
+      "step": 12571
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.025029797377831e-07,
+      "loss": 1.1291,
+      "step": 12572
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.024632499006754e-07,
+      "loss": 1.6104,
+      "step": 12573
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.024235200635677e-07,
+      "loss": 1.8963,
+      "step": 12574
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.023837902264601e-07,
+      "loss": 2.0016,
+      "step": 12575
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.023440603893523e-07,
+      "loss": 1.7367,
+      "step": 12576
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.023043305522447e-07,
+      "loss": 1.4526,
+      "step": 12577
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.022646007151371e-07,
+      "loss": 1.712,
+      "step": 12578
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.022248708780293e-07,
+      "loss": 1.9084,
+      "step": 12579
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.021851410409217e-07,
+      "loss": 1.9179,
+      "step": 12580
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.02145411203814e-07,
+      "loss": 1.3352,
+      "step": 12581
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.021056813667064e-07,
+      "loss": 1.9794,
+      "step": 12582
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.020659515295987e-07,
+      "loss": 1.534,
+      "step": 12583
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.020262216924911e-07,
+      "loss": 1.1048,
+      "step": 12584
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.019864918553834e-07,
+      "loss": 1.8544,
+      "step": 12585
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.019467620182757e-07,
+      "loss": 1.9912,
+      "step": 12586
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.019070321811681e-07,
+      "loss": 1.8784,
+      "step": 12587
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.018673023440604e-07,
+      "loss": 1.7876,
+      "step": 12588
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.018275725069528e-07,
+      "loss": 1.8566,
+      "step": 12589
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.01787842669845e-07,
+      "loss": 1.468,
+      "step": 12590
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.017481128327373e-07,
+      "loss": 1.9331,
+      "step": 12591
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.017083829956296e-07,
+      "loss": 1.4916,
+      "step": 12592
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.01668653158522e-07,
+      "loss": 1.1837,
+      "step": 12593
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.016289233214144e-07,
+      "loss": 1.7903,
+      "step": 12594
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.015891934843067e-07,
+      "loss": 1.5963,
+      "step": 12595
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.015494636471991e-07,
+      "loss": 1.6806,
+      "step": 12596
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.015097338100913e-07,
+      "loss": 2.1757,
+      "step": 12597
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.014700039729837e-07,
+      "loss": 1.6708,
+      "step": 12598
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.01430274135876e-07,
+      "loss": 1.3703,
+      "step": 12599
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.013905442987684e-07,
+      "loss": 1.7245,
+      "step": 12600
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.013508144616608e-07,
+      "loss": 1.28,
+      "step": 12601
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.013110846245531e-07,
+      "loss": 1.4269,
+      "step": 12602
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.012713547874453e-07,
+      "loss": 1.3802,
+      "step": 12603
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.012316249503376e-07,
+      "loss": 1.4065,
+      "step": 12604
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0119189511323e-07,
+      "loss": 1.6335,
+      "step": 12605
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.011521652761223e-07,
+      "loss": 1.5769,
+      "step": 12606
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.011124354390147e-07,
+      "loss": 1.9318,
+      "step": 12607
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.010727056019071e-07,
+      "loss": 1.5057,
+      "step": 12608
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.010329757647993e-07,
+      "loss": 1.4641,
+      "step": 12609
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.009932459276917e-07,
+      "loss": 1.3151,
+      "step": 12610
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.36518636003172084,
+      "eval_loss": 1.7273346185684204,
+      "eval_runtime": 6.4154,
+      "eval_samples_per_second": 393.114,
+      "eval_steps_per_second": 6.235,
+      "step": 12610
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 25220,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "total_flos": 2291512382676000.0,
+  "trial_name": null,
+  "trial_params": null
+}