diff --git "a/checkpoint-2400/trainer_state.json" "b/checkpoint-2400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2400/trainer_state.json"
@@ -0,0 +1,14416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.779600454766932,
+  "global_step": 2400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086024e-07,
+      "loss": 3.6948,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.301075268817205e-07,
+      "loss": 3.7648,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 2.8416,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.60215053763441e-07,
+      "loss": 2.7468,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0752688172043011e-06,
+      "loss": 2.7644,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 2.6625,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5053763440860217e-06,
+      "loss": 2.5855,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.720430107526882e-06,
+      "loss": 2.5464,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 2.4512,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 2.4721,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3655913978494625e-06,
+      "loss": 2.3981,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 2.3686,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7956989247311827e-06,
+      "loss": 2.3357,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0107526881720433e-06,
+      "loss": 2.3081,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 2.257,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440860215053764e-06,
+      "loss": 2.234,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.655913978494624e-06,
+      "loss": 2.2507,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 2.2335,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.086021505376344e-06,
+      "loss": 2.1748,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 2.1561,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 2.1426,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.731182795698925e-06,
+      "loss": 2.1667,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946236559139785e-06,
+      "loss": 2.1333,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 2.1497,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 2.1849,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.591397849462365e-06,
+      "loss": 2.0663,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 2.1544,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.021505376344087e-06,
+      "loss": 2.1504,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.236559139784947e-06,
+      "loss": 2.1158,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 2.1296,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.0637,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.881720430107528e-06,
+      "loss": 2.1321,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 2.0844,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.311827956989248e-06,
+      "loss": 2.0964,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 2.0819,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 2.1291,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.956989247311828e-06,
+      "loss": 1.9948,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.172043010752689e-06,
+      "loss": 2.0445,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 2.1166,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 2.0395,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.81720430107527e-06,
+      "loss": 2.1037,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 2.0654,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.24731182795699e-06,
+      "loss": 2.0031,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.46236559139785e-06,
+      "loss": 2.0557,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 2.0988,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.89247311827957e-06,
+      "loss": 2.0725,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0107526881720431e-05,
+      "loss": 2.0798,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 2.0399,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0537634408602151e-05,
+      "loss": 2.0845,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 2.0403,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 2.0101,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.118279569892473e-05,
+      "loss": 2.0589,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1397849462365593e-05,
+      "loss": 2.0408,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 2.0676,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 2.0267,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2043010752688173e-05,
+      "loss": 1.9908,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 2.0712,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2473118279569894e-05,
+      "loss": 2.0103,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2688172043010754e-05,
+      "loss": 2.0577,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 2.036,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3118279569892473e-05,
+      "loss": 2.0586,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.0466,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 2.06,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3763440860215056e-05,
+      "loss": 2.0954,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3978494623655916e-05,
+      "loss": 1.946,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 2.0347,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4408602150537636e-05,
+      "loss": 2.0207,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4623655913978497e-05,
+      "loss": 1.9838,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 2.0617,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 2.0208,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5268817204301076e-05,
+      "loss": 2.0643,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 1.9885,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5698924731182796e-05,
+      "loss": 2.0487,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5913978494623657e-05,
+      "loss": 2.0218,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 2.0455,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6344086021505377e-05,
+      "loss": 1.9669,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6559139784946237e-05,
+      "loss": 2.0081,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 1.9766,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6989247311827958e-05,
+      "loss": 1.9649,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 2.0655,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 2.0226,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.763440860215054e-05,
+      "loss": 2.0041,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.78494623655914e-05,
+      "loss": 1.9534,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 1.987,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 2.0174,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.849462365591398e-05,
+      "loss": 2.0071,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 1.9841,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.89247311827957e-05,
+      "loss": 2.0004,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.913978494623656e-05,
+      "loss": 2.0255,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 2.0358,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.956989247311828e-05,
+      "loss": 1.9901,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.978494623655914e-05,
+      "loss": 2.0196,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.967,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994461641828e-05,
+      "loss": 2.0473,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999977846573447e-05,
+      "loss": 1.9998,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999950154813253e-05,
+      "loss": 2.0183,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991138639193e-05,
+      "loss": 2.0359,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999861541352416e-05,
+      "loss": 2.0661,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999800619749922e-05,
+      "loss": 1.9936,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999728621651928e-05,
+      "loss": 1.9946,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999964554713819e-05,
+      "loss": 1.9967,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999551396300722e-05,
+      "loss": 2.0273,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999446169243816e-05,
+      "loss": 2.0144,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999329866084025e-05,
+      "loss": 2.018,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999202486950177e-05,
+      "loss": 1.9653,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999064031983366e-05,
+      "loss": 2.0853,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998914501336956e-05,
+      "loss": 2.0071,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998753895176576e-05,
+      "loss": 2.0163,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998582213680123e-05,
+      "loss": 2.0349,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839945703777e-05,
+      "loss": 2.0435,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998205625451942e-05,
+      "loss": 2.0066,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998000719137353e-05,
+      "loss": 2.0006,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999778473832096e-05,
+      "loss": 2.0133,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997557683242002e-05,
+      "loss": 2.0242,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997319554151988e-05,
+      "loss": 1.9894,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999707035131468e-05,
+      "loss": 2.0991,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996810075006118e-05,
+      "loss": 1.9687,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996538725514597e-05,
+      "loss": 1.9491,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996256303140688e-05,
+      "loss": 2.0519,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995962808197217e-05,
+      "loss": 2.0128,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995658241009287e-05,
+      "loss": 1.9924,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995342601914253e-05,
+      "loss": 2.005,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999501589126174e-05,
+      "loss": 2.0267,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994678109413638e-05,
+      "loss": 1.9778,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994329256744098e-05,
+      "loss": 1.9732,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993969333639533e-05,
+      "loss": 1.9977,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999359834049862e-05,
+      "loss": 1.9682,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993216277732302e-05,
+      "loss": 1.9855,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999282314576377e-05,
+      "loss": 2.0483,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992418945028494e-05,
+      "loss": 2.0315,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999200367597419e-05,
+      "loss": 2.0209,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991577339060842e-05,
+      "loss": 2.0238,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999113993476069e-05,
+      "loss": 1.9472,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990691463558235e-05,
+      "loss": 2.0619,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999023192595024e-05,
+      "loss": 1.994,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989761322445714e-05,
+      "loss": 2.0144,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989279653565937e-05,
+      "loss": 1.9915,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988786919844437e-05,
+      "loss": 2.0191,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988283121827004e-05,
+      "loss": 2.0329,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987768260071676e-05,
+      "loss": 1.953,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987242335148757e-05,
+      "loss": 2.0642,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986705347640797e-05,
+      "loss": 1.961,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986157298142595e-05,
+      "loss": 1.9796,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998559818726122e-05,
+      "loss": 1.9881,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985028015615977e-05,
+      "loss": 2.0152,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984446783838432e-05,
+      "loss": 2.0281,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983854492572394e-05,
+      "loss": 1.9869,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983251142473935e-05,
+      "loss": 2.0632,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982636734211363e-05,
+      "loss": 2.0451,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982011268465246e-05,
+      "loss": 1.997,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981374745928388e-05,
+      "loss": 2.02,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998072716730585e-05,
+      "loss": 1.9832,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980068533314937e-05,
+      "loss": 2.0539,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979398844685197e-05,
+      "loss": 2.0047,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978718102158426e-05,
+      "loss": 1.9962,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978026306488665e-05,
+      "loss": 1.9681,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99773234584422e-05,
+      "loss": 1.9586,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976609558797545e-05,
+      "loss": 1.9884,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975884608345476e-05,
+      "loss": 1.9971,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975148607888996e-05,
+      "loss": 1.9393,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974401558243355e-05,
+      "loss": 2.0442,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973643460236032e-05,
+      "loss": 1.9863,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972874314706755e-05,
+      "loss": 1.9771,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972094122507485e-05,
+      "loss": 1.9397,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997130288450242e-05,
+      "loss": 1.963,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997050060156799e-05,
+      "loss": 1.925,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969687274592863e-05,
+      "loss": 1.9448,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968862904477936e-05,
+      "loss": 2.0108,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968027492136337e-05,
+      "loss": 2.0254,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967181038493435e-05,
+      "loss": 2.0133,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966323544486817e-05,
+      "loss": 2.0061,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965455011066313e-05,
+      "loss": 1.9734,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964575439193966e-05,
+      "loss": 1.9657,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963684829844053e-05,
+      "loss": 1.9824,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996278318400308e-05,
+      "loss": 1.9854,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961870502669772e-05,
+      "loss": 2.0048,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996094678685508e-05,
+      "loss": 2.0018,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996001203758218e-05,
+      "loss": 1.9814,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959066255886465e-05,
+      "loss": 1.989,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958109442815553e-05,
+      "loss": 1.9706,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957141599429277e-05,
+      "loss": 1.9887,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995616272679969e-05,
+      "loss": 1.9887,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995517282601106e-05,
+      "loss": 1.9571,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954171898159876e-05,
+      "loss": 2.0014,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953159944354832e-05,
+      "loss": 1.9731,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952136965716846e-05,
+      "loss": 1.9569,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995110296337904e-05,
+      "loss": 1.9677,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950057938486745e-05,
+      "loss": 2.0321,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949001892197514e-05,
+      "loss": 1.975,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947934825681092e-05,
+      "loss": 2.0014,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946856740119442e-05,
+      "loss": 1.9938,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994576763670673e-05,
+      "loss": 1.9968,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994466751664932e-05,
+      "loss": 2.022,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943556381165785e-05,
+      "loss": 1.9445,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99424342314869e-05,
+      "loss": 1.9802,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994130106885564e-05,
+      "loss": 2.0178,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940156894527178e-05,
+      "loss": 1.9566,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993900170976888e-05,
+      "loss": 2.0086,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993783551586031e-05,
+      "loss": 2.0251,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993665831409323e-05,
+      "loss": 1.9763,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935470105771596e-05,
+      "loss": 1.9533,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934270892211548e-05,
+      "loss": 1.9889,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933060674741422e-05,
+      "loss": 1.9985,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931839454701744e-05,
+      "loss": 1.9696,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993060723344522e-05,
+      "loss": 1.9796,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992936401233675e-05,
+      "loss": 2.0065,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928109792753417e-05,
+      "loss": 1.9896,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926844576084483e-05,
+      "loss": 2.0173,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925568363731388e-05,
+      "loss": 2.0214,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924281157107762e-05,
+      "loss": 1.9646,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99229829576394e-05,
+      "loss": 1.9867,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921673766764292e-05,
+      "loss": 1.9837,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992035358593258e-05,
+      "loss": 1.9542,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919022416606596e-05,
+      "loss": 1.9279,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991768026026084e-05,
+      "loss": 2.0008,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991632711838198e-05,
+      "loss": 1.9406,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914962992468854e-05,
+      "loss": 1.9658,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991358788403246e-05,
+      "loss": 1.9964,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991220179459597e-05,
+      "loss": 1.9308,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991080472569472e-05,
+      "loss": 1.9985,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909396678876197e-05,
+      "loss": 1.9805,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907977655700056e-05,
+      "loss": 1.9433,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990654765773811e-05,
+      "loss": 1.9451,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990510668657433e-05,
+      "loss": 2.016,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903654743804833e-05,
+      "loss": 1.9562,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902191831037898e-05,
+      "loss": 1.9694,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900717949893952e-05,
+      "loss": 1.9679,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899233102005573e-05,
+      "loss": 1.965,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989773728901748e-05,
+      "loss": 1.983,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896230512586548e-05,
+      "loss": 1.9854,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894712774381787e-05,
+      "loss": 2.0205,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893184076084356e-05,
+      "loss": 1.9701,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891644419387545e-05,
+      "loss": 1.9555,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989009380599679e-05,
+      "loss": 1.9359,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988853223762967e-05,
+      "loss": 2.0219,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988695971601588e-05,
+      "loss": 1.9749,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885376242897257e-05,
+      "loss": 1.9565,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883781820027777e-05,
+      "loss": 1.9707,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988217644917353e-05,
+      "loss": 1.9683,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880560132112743e-05,
+      "loss": 1.9492,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878932870635766e-05,
+      "loss": 1.9065,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877294666545067e-05,
+      "loss": 1.9653,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987564552165524e-05,
+      "loss": 1.9764,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873985437792992e-05,
+      "loss": 1.9024,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987231441679716e-05,
+      "loss": 1.9696,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987063246051868e-05,
+      "loss": 1.9871,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868939570820608e-05,
+      "loss": 2.004,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867235749578108e-05,
+      "loss": 1.9137,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865520998678458e-05,
+      "loss": 1.9703,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986379532002104e-05,
+      "loss": 1.9999,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862058715517332e-05,
+      "loss": 1.9846,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860311187090927e-05,
+      "loss": 1.9709,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858552736677516e-05,
+      "loss": 1.9511,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985678336622488e-05,
+      "loss": 1.9799,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855003077692898e-05,
+      "loss": 1.9523,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853211873053547e-05,
+      "loss": 1.9784,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98514097542909e-05,
+      "loss": 1.9823,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984959672340111e-05,
+      "loss": 1.9523,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984777278239241e-05,
+      "loss": 1.9891,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984593793328514e-05,
+      "loss": 1.8587,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844092178111703e-05,
+      "loss": 1.9226,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842235518916592e-05,
+      "loss": 2.019,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984036795775638e-05,
+      "loss": 1.9858,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838489496699703e-05,
+      "loss": 2.0027,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983660013782729e-05,
+      "loss": 1.9435,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983469988323192e-05,
+      "loss": 1.9662,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832788735018452e-05,
+      "loss": 2.0029,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830866695303817e-05,
+      "loss": 1.9836,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828933766216998e-05,
+      "loss": 1.9697,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982698994989905e-05,
+      "loss": 1.9477,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982503524850308e-05,
+      "loss": 1.9539,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982306966419426e-05,
+      "loss": 1.9902,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821093199149806e-05,
+      "loss": 1.9304,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819105855558996e-05,
+      "loss": 2.0117,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981710763562315e-05,
+      "loss": 1.922,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815098541555645e-05,
+      "loss": 1.9506,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813078575581898e-05,
+      "loss": 1.939,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981104773993936e-05,
+      "loss": 2.0058,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809006036877537e-05,
+      "loss": 1.9797,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806953468657966e-05,
+      "loss": 1.9606,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804890037554215e-05,
+      "loss": 1.9986,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802815745851888e-05,
+      "loss": 1.9328,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980073059584862e-05,
+      "loss": 1.9669,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798634589854073e-05,
+      "loss": 1.9324,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796527730189936e-05,
+      "loss": 1.9421,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794410019189913e-05,
+      "loss": 1.9424,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792281459199737e-05,
+      "loss": 1.9675,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790142052577148e-05,
+      "loss": 2.0027,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787991801691907e-05,
+      "loss": 1.9261,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785830708925792e-05,
+      "loss": 1.9594,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978365877667258e-05,
+      "loss": 2.0118,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.951,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977928240334002e-05,
+      "loss": 1.9852,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777077967108254e-05,
+      "loss": 1.983,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774862701084555e-05,
+      "loss": 1.9346,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772636607722714e-05,
+      "loss": 1.9381,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770399689488506e-05,
+      "loss": 1.9785,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768151948859705e-05,
+      "loss": 1.9167,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765893388326064e-05,
+      "loss": 1.9357,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763624010389334e-05,
+      "loss": 1.9367,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761343817563235e-05,
+      "loss": 1.9841,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759052812373478e-05,
+      "loss": 1.9722,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756750997357738e-05,
+      "loss": 1.9868,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754438375065673e-05,
+      "loss": 2.0042,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752114948058908e-05,
+      "loss": 1.9658,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974978071891104e-05,
+      "loss": 1.9948,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747435690207625e-05,
+      "loss": 1.9643,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745079864546184e-05,
+      "loss": 1.9196,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742713244536205e-05,
+      "loss": 1.9424,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740335832799118e-05,
+      "loss": 1.9293,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973794763196832e-05,
+      "loss": 1.9349,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973554864468915e-05,
+      "loss": 1.9993,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97331388736189e-05,
+      "loss": 1.976,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730718321426804e-05,
+      "loss": 1.9633,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972828699079404e-05,
+      "loss": 1.9665,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972584488441372e-05,
+      "loss": 1.9345,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723392004990904e-05,
+      "loss": 1.9783,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972092835524257e-05,
+      "loss": 1.9494,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718453937897637e-05,
+      "loss": 1.9616,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715968755696946e-05,
+      "loss": 1.977,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971347281139326e-05,
+      "loss": 1.9101,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971096610775127e-05,
+      "loss": 1.9798,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708448647547575e-05,
+      "loss": 1.9196,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705920433570696e-05,
+      "loss": 1.9767,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970338146862107e-05,
+      "loss": 1.9455,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700831755511028e-05,
+      "loss": 1.9169,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969827129706482e-05,
+      "loss": 2.0205,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695700096118594e-05,
+      "loss": 1.9768,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693118155520394e-05,
+      "loss": 1.9656,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690525478130162e-05,
+      "loss": 1.9702,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687922066819734e-05,
+      "loss": 1.9886,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685307924472835e-05,
+      "loss": 1.9488,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682683053985073e-05,
+      "loss": 1.9083,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968004745826395e-05,
+      "loss": 1.9531,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677401140228834e-05,
+      "loss": 1.9355,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967474410281098e-05,
+      "loss": 1.8934,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672076348953507e-05,
+      "loss": 1.9803,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966939788161142e-05,
+      "loss": 1.9229,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666708703751576e-05,
+      "loss": 1.9787,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96640088183527e-05,
+      "loss": 1.9516,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661298228405377e-05,
+      "loss": 1.9262,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658576936912057e-05,
+      "loss": 1.9484,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655844946887035e-05,
+      "loss": 1.9748,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653102261356456e-05,
+      "loss": 2.013,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650348883358313e-05,
+      "loss": 1.8889,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647584815942455e-05,
+      "loss": 1.9029,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964481006217055e-05,
+      "loss": 1.9733,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642024625116117e-05,
+      "loss": 1.9586,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963922850786451e-05,
+      "loss": 1.9424,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963642171351291e-05,
+      "loss": 1.9624,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963360424517031e-05,
+      "loss": 1.9854,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630776105957553e-05,
+      "loss": 1.9051,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627937299007286e-05,
+      "loss": 1.9653,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962508782746397e-05,
+      "loss": 1.9739,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622227694483893e-05,
+      "loss": 2.0066,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619356903235132e-05,
+      "loss": 1.9116,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961647545689759e-05,
+      "loss": 1.9127,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961358335866296e-05,
+      "loss": 2.0008,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610680611734737e-05,
+      "loss": 1.9408,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960776721932821e-05,
+      "loss": 1.9275,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604843184670462e-05,
+      "loss": 1.89,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601908511000362e-05,
+      "loss": 1.9549,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959896320156857e-05,
+      "loss": 1.9161,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959600725963752e-05,
+      "loss": 1.9198,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959304068848142e-05,
+      "loss": 1.9602,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590063491386262e-05,
+      "loss": 1.9412,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95870756716498e-05,
+      "loss": 1.935,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958407723258156e-05,
+      "loss": 1.9177,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581068177502826e-05,
+      "loss": 1.8578,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578048509746643e-05,
+      "loss": 1.9378,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575018232657806e-05,
+      "loss": 1.9268,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571977349592878e-05,
+      "loss": 1.9672,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568925863920155e-05,
+      "loss": 1.9358,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565863779019675e-05,
+      "loss": 1.929,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562791098283226e-05,
+      "loss": 1.8864,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559707825114334e-05,
+      "loss": 1.979,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556613962928244e-05,
+      "loss": 1.9211,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955350951515195e-05,
+      "loss": 1.9261,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550394485224154e-05,
+      "loss": 1.8456,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547268876595286e-05,
+      "loss": 1.9474,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544132692727497e-05,
+      "loss": 1.9438,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954098593709465e-05,
+      "loss": 1.9835,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537828613182314e-05,
+      "loss": 1.9218,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534660724487766e-05,
+      "loss": 1.9989,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953148227451999e-05,
+      "loss": 1.9649,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952829326679966e-05,
+      "loss": 1.9589,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525093704859156e-05,
+      "loss": 1.9345,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521883592242537e-05,
+      "loss": 1.9648,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518662932505554e-05,
+      "loss": 1.9249,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515431729215642e-05,
+      "loss": 1.9791,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951218998595192e-05,
+      "loss": 1.9128,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950893770630516e-05,
+      "loss": 1.9656,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950567489387783e-05,
+      "loss": 1.9639,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502401552284053e-05,
+      "loss": 1.963,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499117685149617e-05,
+      "loss": 1.9352,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495823296111968e-05,
+      "loss": 1.9781,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492518388820204e-05,
+      "loss": 1.9513,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489202966935084e-05,
+      "loss": 1.9514,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485877034129e-05,
+      "loss": 1.9353,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482540594086e-05,
+      "loss": 1.97,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479193650501756e-05,
+      "loss": 1.9289,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947583620708359e-05,
+      "loss": 1.9666,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947246826755044e-05,
+      "loss": 2.0028,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469089835632883e-05,
+      "loss": 1.986,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946570091507311e-05,
+      "loss": 1.9671,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462301509624933e-05,
+      "loss": 1.9104,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458891623053776e-05,
+      "loss": 1.9167,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945547125913667e-05,
+      "loss": 1.9819,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452040421662263e-05,
+      "loss": 1.9523,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448599114430793e-05,
+      "loss": 1.9586,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445147341254094e-05,
+      "loss": 1.9234,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441685105955605e-05,
+      "loss": 1.965,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943821241237034e-05,
+      "loss": 1.9468,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94347292643449e-05,
+      "loss": 1.9505,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943123566573748e-05,
+      "loss": 1.9066,
+      "step": 415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427731620417832e-05,
+      "loss": 1.9418,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424217132267287e-05,
+      "loss": 1.9694,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420692205178753e-05,
+      "loss": 1.9244,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941715684305668e-05,
+      "loss": 1.9081,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413611049817098e-05,
+      "loss": 1.9661,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410054829387577e-05,
+      "loss": 1.9313,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940648818570724e-05,
+      "loss": 1.9727,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402911122726756e-05,
+      "loss": 1.927,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399323644408343e-05,
+      "loss": 1.9475,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395725754725743e-05,
+      "loss": 1.9143,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939211745766424e-05,
+      "loss": 1.9241,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938849875722064e-05,
+      "loss": 1.994,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384869657403277e-05,
+      "loss": 1.9681,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381230162232e-05,
+      "loss": 1.9222,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377580275738173e-05,
+      "loss": 1.9131,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373920001964675e-05,
+      "loss": 1.9337,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370249344965884e-05,
+      "loss": 1.929,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366568308807685e-05,
+      "loss": 2.0023,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362876897567463e-05,
+      "loss": 1.9394,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359175115334077e-05,
+      "loss": 1.9062,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355462966207895e-05,
+      "loss": 1.9598,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351740454300757e-05,
+      "loss": 1.8951,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348007583735985e-05,
+      "loss": 1.9455,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934426435864837e-05,
+      "loss": 1.8948,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340510783184183e-05,
+      "loss": 1.9448,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336746861501147e-05,
+      "loss": 1.8841,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332972597768457e-05,
+      "loss": 1.9498,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329187996166747e-05,
+      "loss": 1.954,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325393060888124e-05,
+      "loss": 1.911,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932158779613613e-05,
+      "loss": 1.9387,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931777220612574e-05,
+      "loss": 1.9212,
+      "step": 446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313946295083383e-05,
+      "loss": 1.9618,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310110067246905e-05,
+      "loss": 1.9135,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306263526865593e-05,
+      "loss": 1.9191,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930240667820015e-05,
+      "loss": 1.9323,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298539525522698e-05,
+      "loss": 1.9412,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294662073116768e-05,
+      "loss": 1.9872,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290774325277305e-05,
+      "loss": 1.9422,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286876286310663e-05,
+      "loss": 1.9406,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282967960534584e-05,
+      "loss": 1.9842,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927904935227821e-05,
+      "loss": 1.9133,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927512046588207e-05,
+      "loss": 1.9322,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271181305698084e-05,
+      "loss": 1.9534,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926723187608955e-05,
+      "loss": 1.9387,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263272181431133e-05,
+      "loss": 1.9697,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259302226108878e-05,
+      "loss": 1.9832,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255322014520193e-05,
+      "loss": 1.8954,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251331551073843e-05,
+      "loss": 1.9132,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247330840189948e-05,
+      "loss": 1.944,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924331988629999e-05,
+      "loss": 1.9603,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239298693846783e-05,
+      "loss": 1.8851,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923526726728449e-05,
+      "loss": 1.9113,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923122561107861e-05,
+      "loss": 1.9339,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227173729705962e-05,
+      "loss": 1.9513,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223111627654713e-05,
+      "loss": 1.9417,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921903930942433e-05,
+      "loss": 1.9675,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214956779525605e-05,
+      "loss": 1.9727,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210864042480645e-05,
+      "loss": 1.9821,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920676110282285e-05,
+      "loss": 1.9565,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202647965096936e-05,
+      "loss": 1.968,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919852463385891e-05,
+      "loss": 1.9318,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194391113676066e-05,
+      "loss": 1.961,
+      "step": 477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190247409126993e-05,
+      "loss": 1.9386,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186093524801546e-05,
+      "loss": 1.9989,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918192946530087e-05,
+      "loss": 1.922,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917775523523737e-05,
+      "loss": 1.9078,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173570839234735e-05,
+      "loss": 1.9414,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916937628192789e-05,
+      "loss": 1.9433,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916517156796303e-05,
+      "loss": 1.9614,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91609567019976e-05,
+      "loss": 1.8821,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156731688700284e-05,
+      "loss": 1.8822,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152496532751014e-05,
+      "loss": 1.9666,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148251238840947e-05,
+      "loss": 1.8848,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143995811672477e-05,
+      "loss": 1.9026,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913973025595922e-05,
+      "loss": 1.9613,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.941,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131168777808898e-05,
+      "loss": 1.9299,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126872864855142e-05,
+      "loss": 1.8918,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91225668423232e-05,
+      "loss": 1.974,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118250714982733e-05,
+      "loss": 1.9927,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911392448761459e-05,
+      "loss": 1.9179,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910958816501081e-05,
+      "loss": 1.9498,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105241751974624e-05,
+      "loss": 1.9447,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910088525332042e-05,
+      "loss": 1.9915,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096518673873773e-05,
+      "loss": 1.9564,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092142018471415e-05,
+      "loss": 1.9729,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087755291961246e-05,
+      "loss": 1.9088,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083358499202323e-05,
+      "loss": 1.9213,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078951645064837e-05,
+      "loss": 1.9303,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074534734430147e-05,
+      "loss": 1.9201,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070107772190734e-05,
+      "loss": 1.9744,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906567076325022e-05,
+      "loss": 1.9186,
+      "step": 507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9061223712523352e-05,
+      "loss": 1.8502,
+      "step": 508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056766624936003e-05,
+      "loss": 1.9787,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052299505425164e-05,
+      "loss": 1.9335,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047822358938933e-05,
+      "loss": 1.9245,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904333519043652e-05,
+      "loss": 1.914,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903883800488824e-05,
+      "loss": 1.8894,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034330807275484e-05,
+      "loss": 1.967,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029813602590763e-05,
+      "loss": 1.9266,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025286395837647e-05,
+      "loss": 1.9478,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020749192030795e-05,
+      "loss": 1.9507,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016201996195943e-05,
+      "loss": 1.9251,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011644813369886e-05,
+      "loss": 1.896,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900707764860049e-05,
+      "loss": 1.9846,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900250050694667e-05,
+      "loss": 1.893,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997913393478404e-05,
+      "loss": 1.9632,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993316313276694e-05,
+      "loss": 1.9331,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988709271433607e-05,
+      "loss": 1.8653,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898409227305223e-05,
+      "loss": 1.9503,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979465323246676e-05,
+      "loss": 1.8925,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974828427142093e-05,
+      "loss": 2.03,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970181589874637e-05,
+      "loss": 1.9302,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965524816591477e-05,
+      "loss": 1.8439,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960858112450785e-05,
+      "loss": 1.9086,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956181482621746e-05,
+      "loss": 1.9106,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895149493228452e-05,
+      "loss": 1.9496,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894679846663027e-05,
+      "loss": 1.9624,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894209209086114e-05,
+      "loss": 1.8404,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937375810190247e-05,
+      "loss": 1.9297,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932649629841685e-05,
+      "loss": 1.8797,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927913555050503e-05,
+      "loss": 1.9574,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923167591062723e-05,
+      "loss": 1.9457,
+      "step": 538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891841174313531e-05,
+      "loss": 1.8838,
+      "step": 539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913646016536184e-05,
+      "loss": 1.8693,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908870416544202e-05,
+      "loss": 1.9274,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890408494844917e-05,
+      "loss": 1.9439,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899289617551803e-05,
+      "loss": 1.9711,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894484429163764e-05,
+      "loss": 1.8938,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889669388607615e-05,
+      "loss": 1.895,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884844501216845e-05,
+      "loss": 1.9271,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880009772335843e-05,
+      "loss": 1.9235,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875165207319902e-05,
+      "loss": 1.9984,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870310811535208e-05,
+      "loss": 1.93,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886544659035884e-05,
+      "loss": 1.9034,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860572549178755e-05,
+      "loss": 1.926,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855688693393788e-05,
+      "loss": 1.9737,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850795028413658e-05,
+      "loss": 1.9205,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845891559658925e-05,
+      "loss": 1.8709,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884097829256103e-05,
+      "loss": 1.947,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883605523256226e-05,
+      "loss": 1.9827,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883112238511575e-05,
+      "loss": 1.9662,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882617975568547e-05,
+      "loss": 1.9247,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821227349746235e-05,
+      "loss": 1.9162,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881626517278368e-05,
+      "loss": 1.9508,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881129323029427e-05,
+      "loss": 1.9203,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806311527785287e-05,
+      "loss": 1.8772,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880132007077482e-05,
+      "loss": 1.899,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796318864791763e-05,
+      "loss": 1.9686,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879130791537581e-05,
+      "loss": 1.8958,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786287228077453e-05,
+      "loss": 1.9144,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781256808457954e-05,
+      "loss": 1.96,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776216662089373e-05,
+      "loss": 1.8936,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771166794554544e-05,
+      "loss": 1.8663,
+      "step": 569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766107211447046e-05,
+      "loss": 1.8893,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761037918371248e-05,
+      "loss": 1.912,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875595892094226e-05,
+      "loss": 1.9463,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875087022478594e-05,
+      "loss": 2.0043,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745771835538893e-05,
+      "loss": 1.9508,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740663758848463e-05,
+      "loss": 1.9375,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873554600037272e-05,
+      "loss": 1.9604,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873041856578046e-05,
+      "loss": 1.9089,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725281460751198e-05,
+      "loss": 1.8638,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720134690975157e-05,
+      "loss": 1.9324,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714978262153268e-05,
+      "loss": 1.9731,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709812179997162e-05,
+      "loss": 1.8919,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704636450229164e-05,
+      "loss": 1.9082,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869945107858228e-05,
+      "loss": 1.9313,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694256070800202e-05,
+      "loss": 1.9543,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868905143263729e-05,
+      "loss": 1.9452,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683837169858573e-05,
+      "loss": 1.9352,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678613288239745e-05,
+      "loss": 1.8891,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867337979356715e-05,
+      "loss": 1.8977,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866813669163778e-05,
+      "loss": 1.9013,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662883988259276e-05,
+      "loss": 1.969,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86576216892499e-05,
+      "loss": 1.929,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652349800438558e-05,
+      "loss": 1.8832,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647068327664774e-05,
+      "loss": 1.855,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641777276778677e-05,
+      "loss": 2.0193,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636476653641015e-05,
+      "loss": 1.9424,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631166464123147e-05,
+      "loss": 1.9696,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625846714107013e-05,
+      "loss": 1.9399,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620517409485148e-05,
+      "loss": 1.8774,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615178556160675e-05,
+      "loss": 1.9191,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609830160047283e-05,
+      "loss": 1.8977,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604472227069253e-05,
+      "loss": 1.8954,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599104763161402e-05,
+      "loss": 1.9183,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593727774269122e-05,
+      "loss": 1.9417,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858834126634835e-05,
+      "loss": 1.8973,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582945245365574e-05,
+      "loss": 1.8593,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577539717297806e-05,
+      "loss": 1.9275,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85721246881326e-05,
+      "loss": 1.8956,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566700163868027e-05,
+      "loss": 1.8985,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856126615051268e-05,
+      "loss": 1.9632,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555822654085665e-05,
+      "loss": 1.9193,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550369680616584e-05,
+      "loss": 1.9248,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854490723614554e-05,
+      "loss": 1.9553,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539435326723135e-05,
+      "loss": 1.94,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853395395841044e-05,
+      "loss": 1.8915,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852846313727902e-05,
+      "loss": 1.9557,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852296286941089e-05,
+      "loss": 1.8679,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851745316089855e-05,
+      "loss": 1.9401,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851193401784495e-05,
+      "loss": 1.9942,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506405446363476e-05,
+      "loss": 1.9263,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500867452577978e-05,
+      "loss": 1.9126,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849532004262274e-05,
+      "loss": 1.9397,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489763222642457e-05,
+      "loss": 1.9212,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848419699879227e-05,
+      "loss": 1.8994,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478621377237724e-05,
+      "loss": 1.9592,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847303636415478e-05,
+      "loss": 1.8893,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467441965729796e-05,
+      "loss": 1.8677,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846183818815953e-05,
+      "loss": 1.9112,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845622503765113e-05,
+      "loss": 1.8626,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450602520422113e-05,
+      "loss": 1.8774,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444970642700396e-05,
+      "loss": 1.9396,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843932941072424e-05,
+      "loss": 1.9553,
+      "step": 631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433678830742284e-05,
+      "loss": 1.9411,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842801890901351e-05,
+      "loss": 1.9565,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842234965180725e-05,
+      "loss": 1.975,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841667106540319e-05,
+      "loss": 1.9136,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410983156091334e-05,
+      "loss": 1.9266,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405285930172016e-05,
+      "loss": 1.8997,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399579393955893e-05,
+      "loss": 1.9078,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839386355376393e-05,
+      "loss": 1.9354,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838813841592741e-05,
+      "loss": 1.9249,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382403986787895e-05,
+      "loss": 1.9696,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376660272697257e-05,
+      "loss": 1.8899,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837090728001764e-05,
+      "loss": 1.927,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365145015121477e-05,
+      "loss": 1.8933,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835937348439146e-05,
+      "loss": 1.9198,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835359269422055e-05,
+      "loss": 1.8574,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347802651011965e-05,
+      "loss": 1.9011,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834200336117918e-05,
+      "loss": 1.9355,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336194831145888e-05,
+      "loss": 1.9302,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330377067346045e-05,
+      "loss": 1.9551,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324550076223818e-05,
+      "loss": 1.9158,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83187138642336e-05,
+      "loss": 1.915,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312868437840002e-05,
+      "loss": 1.9587,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307013803517834e-05,
+      "loss": 1.927,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301149967752104e-05,
+      "loss": 1.9229,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295276937038023e-05,
+      "loss": 1.9436,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289394717880977e-05,
+      "loss": 1.879,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283503316796536e-05,
+      "loss": 1.9004,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277602740310433e-05,
+      "loss": 1.9253,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271692994958576e-05,
+      "loss": 1.939,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265774087287016e-05,
+      "loss": 1.9354,
+      "step": 661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825984602385196e-05,
+      "loss": 1.9134,
+      "step": 662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253908811219764e-05,
+      "loss": 1.901,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82479624559669e-05,
+      "loss": 1.919,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242006964679978e-05,
+      "loss": 1.9249,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236042343955734e-05,
+      "loss": 1.8887,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230068600401e-05,
+      "loss": 1.8953,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822408574063273e-05,
+      "loss": 1.9099,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218093771277965e-05,
+      "loss": 1.9344,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212092698973837e-05,
+      "loss": 1.9326,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820608253036757e-05,
+      "loss": 1.9042,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820006327211645e-05,
+      "loss": 1.9163,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194034930887842e-05,
+      "loss": 1.9151,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818799751335917e-05,
+      "loss": 1.902,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181951026217908e-05,
+      "loss": 1.8608,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175895476161577e-05,
+      "loss": 1.9021,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169830869897743e-05,
+      "loss": 1.9812,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163757214143993e-05,
+      "loss": 1.9112,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157674515627945e-05,
+      "loss": 1.9221,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151582781087233e-05,
+      "loss": 1.8847,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81454820172695e-05,
+      "loss": 1.9001,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139372230932384e-05,
+      "loss": 1.9184,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133253428843524e-05,
+      "loss": 1.8944,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127125617780542e-05,
+      "loss": 1.9062,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812098880453105e-05,
+      "loss": 1.8913,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114842995892608e-05,
+      "loss": 1.9559,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108688198672766e-05,
+      "loss": 1.9295,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810252441968901e-05,
+      "loss": 1.8945,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096351665768782e-05,
+      "loss": 1.9288,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.9409,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80839792604784e-05,
+      "loss": 1.8964,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077779622812807e-05,
+      "loss": 1.8999,
+      "step": 692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071571037619856e-05,
+      "loss": 1.9495,
+      "step": 693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806535351177662e-05,
+      "loss": 1.8898,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059127052170072e-05,
+      "loss": 1.9427,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805289166569709e-05,
+      "loss": 1.9817,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046647359264437e-05,
+      "loss": 1.927,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804039413978875e-05,
+      "loss": 1.9765,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803413201419654e-05,
+      "loss": 1.8578,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027860989424195e-05,
+      "loss": 1.9545,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802158107241794e-05,
+      "loss": 1.9264,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015292270133872e-05,
+      "loss": 1.9729,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008994589537913e-05,
+      "loss": 1.9637,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002688037605824e-05,
+      "loss": 1.9628,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996372621323193e-05,
+      "loss": 1.9661,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990048347685432e-05,
+      "loss": 1.9594,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798371522369776e-05,
+      "loss": 1.8831,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977373256375194e-05,
+      "loss": 1.8327,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971022452742554e-05,
+      "loss": 1.9154,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964662819834445e-05,
+      "loss": 1.9342,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958294364695254e-05,
+      "loss": 1.968,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795191709437913e-05,
+      "loss": 1.9252,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945531015950008e-05,
+      "loss": 1.9489,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793913613648155e-05,
+      "loss": 1.8312,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932732463057194e-05,
+      "loss": 1.9221,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926320002770102e-05,
+      "loss": 1.9313,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791989876272318e-05,
+      "loss": 1.9091,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791346875002905e-05,
+      "loss": 1.936,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907029971810054e-05,
+      "loss": 1.8709,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790058243519824e-05,
+      "loss": 1.9522,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789412614733537e-05,
+      "loss": 1.9547,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788766111537289e-05,
+      "loss": 1.9416,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881187346471924e-05,
+      "loss": 1.9,
+      "step": 723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787470484780329e-05,
+      "loss": 1.914,
+      "step": 724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868213626547467e-05,
+      "loss": 1.9297,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861713689894594e-05,
+      "loss": 1.8897,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855205045044465e-05,
+      "loss": 1.8421,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784868769920653e-05,
+      "loss": 1.8969,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842161659599857e-05,
+      "loss": 1.926,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835626933453158e-05,
+      "loss": 1.9438,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782908352800477e-05,
+      "loss": 1.9491,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822531450502634e-05,
+      "loss": 1.8829,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815970708204296e-05,
+      "loss": 1.9879,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809401308376912e-05,
+      "loss": 1.9058,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802823258297213e-05,
+      "loss": 1.9837,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796236565251523e-05,
+      "loss": 1.9459,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778964123653573e-05,
+      "loss": 1.979,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77830372794553e-05,
+      "loss": 1.8978,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776424701325242e-05,
+      "loss": 1.9239,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769803509470127e-05,
+      "loss": 1.9578,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763173711224058e-05,
+      "loss": 1.8514,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756535313930676e-05,
+      "loss": 1.9521,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774988832494314e-05,
+      "loss": 1.9207,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743232751624136e-05,
+      "loss": 1.9137,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736568601345855e-05,
+      "loss": 1.9153,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729895881489984e-05,
+      "loss": 1.8879,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723214599447712e-05,
+      "loss": 1.9387,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716524762619695e-05,
+      "loss": 1.9404,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770982637841608e-05,
+      "loss": 1.8409,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703119454256484e-05,
+      "loss": 1.932,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696403997569967e-05,
+      "loss": 1.9102,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689680015795053e-05,
+      "loss": 1.9236,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682947516379706e-05,
+      "loss": 1.9697,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676206506781332e-05,
+      "loss": 1.9479,
+      "step": 754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669456994466744e-05,
+      "loss": 1.957,
+      "step": 755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766269898691219e-05,
+      "loss": 1.9087,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655932491603326e-05,
+      "loss": 1.9489,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649157516035205e-05,
+      "loss": 1.8715,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642374067712276e-05,
+      "loss": 1.8911,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763558215414837e-05,
+      "loss": 1.9517,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628781782866696e-05,
+      "loss": 1.8853,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762197296139984e-05,
+      "loss": 1.8795,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615155697289734e-05,
+      "loss": 1.874,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608329998087665e-05,
+      "loss": 1.8888,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601495871354272e-05,
+      "loss": 1.9105,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594653324659527e-05,
+      "loss": 1.8892,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587802365582717e-05,
+      "loss": 1.9633,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580943001712457e-05,
+      "loss": 1.9225,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757407524064667e-05,
+      "loss": 1.8957,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567199089992583e-05,
+      "loss": 1.8932,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756031455736671e-05,
+      "loss": 1.9661,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553421650394856e-05,
+      "loss": 1.8841,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546520376712093e-05,
+      "loss": 1.8979,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753961074396277e-05,
+      "loss": 1.9696,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753269275980049e-05,
+      "loss": 1.8969,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525766431888108e-05,
+      "loss": 1.8926,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518831767897722e-05,
+      "loss": 1.866,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511888775510662e-05,
+      "loss": 1.962,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504937462417482e-05,
+      "loss": 1.897,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497977836317958e-05,
+      "loss": 1.9368,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749100990492106e-05,
+      "loss": 1.9475,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484033675944988e-05,
+      "loss": 1.9219,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477049157117093e-05,
+      "loss": 1.9511,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470056356173936e-05,
+      "loss": 1.9351,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746305528086125e-05,
+      "loss": 1.9541,
+      "step": 785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745604593893392e-05,
+      "loss": 1.93,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449028338155995e-05,
+      "loss": 1.9189,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744200248630068e-05,
+      "loss": 1.9121,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434968391150302e-05,
+      "loss": 1.918,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427926060496334e-05,
+      "loss": 1.8509,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420875502139366e-05,
+      "loss": 1.9086,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74138167238891e-05,
+      "loss": 1.8901,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406749733564344e-05,
+      "loss": 1.939,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399674538993008e-05,
+      "loss": 1.9566,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392591148012078e-05,
+      "loss": 1.8899,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385499568467628e-05,
+      "loss": 1.9662,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73783998082148e-05,
+      "loss": 1.8755,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737129187511779e-05,
+      "loss": 1.8945,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364175777049866e-05,
+      "loss": 1.9259,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735705152189332e-05,
+      "loss": 1.9357,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734991911753949e-05,
+      "loss": 1.9056,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342778571888734e-05,
+      "loss": 1.8918,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335629892850436e-05,
+      "loss": 1.9719,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328473088342986e-05,
+      "loss": 1.9184,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321308166293772e-05,
+      "loss": 1.9212,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314135134639173e-05,
+      "loss": 1.9208,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306954001324552e-05,
+      "loss": 1.8852,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729976477430425e-05,
+      "loss": 1.896,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729256746154157e-05,
+      "loss": 1.8863,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728536207100877e-05,
+      "loss": 1.9156,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278148610687053e-05,
+      "loss": 1.9079,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270927088566573e-05,
+      "loss": 1.8712,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263697512646397e-05,
+      "loss": 1.9515,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256459890934524e-05,
+      "loss": 1.9214,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249214231447862e-05,
+      "loss": 1.8625,
+      "step": 815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7241960542212223e-05,
+      "loss": 1.8786,
+      "step": 816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234698831262316e-05,
+      "loss": 1.9442,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227429106641726e-05,
+      "loss": 1.8911,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220151376402923e-05,
+      "loss": 1.9004,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212865648607243e-05,
+      "loss": 1.9376,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205571931324883e-05,
+      "loss": 1.9627,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719827023263488e-05,
+      "loss": 1.8936,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190960560625127e-05,
+      "loss": 1.8739,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718364292339233e-05,
+      "loss": 1.9026,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717631732904204e-05,
+      "loss": 1.908,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168983785688603e-05,
+      "loss": 1.9055,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161642301455174e-05,
+      "loss": 1.898,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154292884473712e-05,
+      "loss": 1.9345,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714693554288496e-05,
+      "loss": 1.9254,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139570284838433e-05,
+      "loss": 1.953,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132197118492415e-05,
+      "loss": 1.8947,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712481605201396e-05,
+      "loss": 1.9105,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711742709357886e-05,
+      "loss": 1.9191,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110030251371656e-05,
+      "loss": 1.9481,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102625533585623e-05,
+      "loss": 1.9278,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709521294842276e-05,
+      "loss": 1.8643,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087792504093767e-05,
+      "loss": 1.9297,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708036420881807e-05,
+      "loss": 1.8685,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072928070823773e-05,
+      "loss": 1.9315,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065484098347678e-05,
+      "loss": 1.8661,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058032299635265e-05,
+      "loss": 1.843,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705057268294068e-05,
+      "loss": 1.8769,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043105256526723e-05,
+      "loss": 1.915,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035630028664854e-05,
+      "loss": 1.8773,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028147007635174e-05,
+      "loss": 1.8815,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020656201726406e-05,
+      "loss": 1.9472,
+      "step": 846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013157619235912e-05,
+      "loss": 1.9082,
+      "step": 847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005651268469652e-05,
+      "loss": 1.8576,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998137157742204e-05,
+      "loss": 1.8658,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990615295376724e-05,
+      "loss": 1.946,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698308568970498e-05,
+      "loss": 1.9061,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975548349067294e-05,
+      "loss": 1.8678,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968003281812563e-05,
+      "loss": 1.8639,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960450496298252e-05,
+      "loss": 1.9171,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952890000890358e-05,
+      "loss": 1.9056,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945321803963435e-05,
+      "loss": 1.8252,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937745913900556e-05,
+      "loss": 1.9545,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693016233909332e-05,
+      "loss": 1.9051,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922571087941837e-05,
+      "loss": 1.929,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914972168854724e-05,
+      "loss": 1.868,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907365590249083e-05,
+      "loss": 1.9548,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689975136055051e-05,
+      "loss": 1.9523,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689212948819307e-05,
+      "loss": 1.8893,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688449998161929e-05,
+      "loss": 1.9025,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876862849280167e-05,
+      "loss": 1.9525,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686921809963513e-05,
+      "loss": 1.9698,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861565741152052e-05,
+      "loss": 1.8342,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853905782307235e-05,
+      "loss": 1.8628,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846238231585396e-05,
+      "loss": 1.9004,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838563097479666e-05,
+      "loss": 1.8806,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830880388491567e-05,
+      "loss": 1.9082,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823190113131026e-05,
+      "loss": 1.8967,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681549227991634e-05,
+      "loss": 1.9242,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807786897374182e-05,
+      "loss": 1.8625,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800073974039576e-05,
+      "loss": 1.9272,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679235351845592e-05,
+      "loss": 1.9224,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784625539174943e-05,
+      "loss": 1.898,
+      "step": 877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67768900447567e-05,
+      "loss": 1.8679,
+      "step": 878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769147043769588e-05,
+      "loss": 1.892,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67613965447903e-05,
+      "loss": 1.9337,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753638556403857e-05,
+      "loss": 1.9704,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674587308720355e-05,
+      "loss": 1.9668,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738100145790977e-05,
+      "loss": 1.8748,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730319740776e-05,
+      "loss": 1.8932,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722531880776754e-05,
+      "loss": 1.9615,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671473657441963e-05,
+      "loss": 1.8689,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706933830339274e-05,
+      "loss": 1.8632,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699123657178553e-05,
+      "loss": 1.8513,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.8659,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668348105822869e-05,
+      "loss": 1.9186,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667564864976641e-05,
+      "loss": 1.8702,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666780884687747e-05,
+      "loss": 1.9326,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659961658245813e-05,
+      "loss": 1.8971,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665210709256354e-05,
+      "loss": 1.8918,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644245158530926e-05,
+      "loss": 1.9025,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663637586485642e-05,
+      "loss": 1.8808,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628499220256612e-05,
+      "loss": 1.8946,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620615233456235e-05,
+      "loss": 1.8693,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612723913188166e-05,
+      "loss": 1.9478,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604825268193388e-05,
+      "loss": 1.8843,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659691930722101e-05,
+      "loss": 1.922,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658900603902824e-05,
+      "loss": 1.9408,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658108547238038e-05,
+      "loss": 1.9207,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573157616050815e-05,
+      "loss": 1.9053,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565222478821012e-05,
+      "loss": 1.8904,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557280069480497e-05,
+      "loss": 1.8268,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549330396826845e-05,
+      "loss": 1.8789,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541373469665688e-05,
+      "loss": 1.9408,
+      "step": 908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6533409296810687e-05,
+      "loss": 1.9718,
+      "step": 909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525437887083532e-05,
+      "loss": 1.9126,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517459249313926e-05,
+      "loss": 1.9468,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509473392339584e-05,
+      "loss": 1.9548,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501480325006206e-05,
+      "loss": 1.9208,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649348005616749e-05,
+      "loss": 1.8278,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485472594685105e-05,
+      "loss": 1.9073,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647745794942869e-05,
+      "loss": 1.9057,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646943612927584e-05,
+      "loss": 1.8711,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64614071431121e-05,
+      "loss": 1.9307,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453370999830947e-05,
+      "loss": 1.8497,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644532770833379e-05,
+      "loss": 1.9017,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437277277529955e-05,
+      "loss": 1.9237,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642921971633668e-05,
+      "loss": 1.9298,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421155033679085e-05,
+      "loss": 1.8554,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413083238490204e-05,
+      "loss": 1.8653,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640500433971093e-05,
+      "loss": 1.8307,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396918346290025e-05,
+      "loss": 1.9543,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388825267184122e-05,
+      "loss": 1.9137,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380725111357693e-05,
+      "loss": 1.9187,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372617887783048e-05,
+      "loss": 1.8395,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636450360544033e-05,
+      "loss": 1.9105,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356382273317504e-05,
+      "loss": 1.923,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348253900410333e-05,
+      "loss": 1.8636,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634011849572239e-05,
+      "loss": 1.8717,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331976068265024e-05,
+      "loss": 1.8718,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323826627057382e-05,
+      "loss": 1.9008,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315670181126363e-05,
+      "loss": 1.8847,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630750673950663e-05,
+      "loss": 1.9071,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299336311240593e-05,
+      "loss": 1.9216,
+      "step": 938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6291158905378412e-05,
+      "loss": 1.9058,
+      "step": 939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282974530977963e-05,
+      "loss": 1.8698,
+      "step": 940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274783197104844e-05,
+      "loss": 1.9021,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266584912832365e-05,
+      "loss": 1.8975,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258379687241533e-05,
+      "loss": 1.9485,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250167529421042e-05,
+      "loss": 1.9176,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241948448467265e-05,
+      "loss": 1.897,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233722453484252e-05,
+      "loss": 1.8763,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225489553583698e-05,
+      "loss": 1.8949,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217249757884954e-05,
+      "loss": 1.9135,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209003075515008e-05,
+      "loss": 1.8707,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200749515608477e-05,
+      "loss": 1.8434,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192489087307594e-05,
+      "loss": 1.9283,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61842217997622e-05,
+      "loss": 1.936,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175947662129735e-05,
+      "loss": 1.8627,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167666683575234e-05,
+      "loss": 1.8869,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159378873271293e-05,
+      "loss": 1.9206,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615108424039809e-05,
+      "loss": 1.8535,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614278279414335e-05,
+      "loss": 1.9193,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134474543702353e-05,
+      "loss": 1.9355,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126159498277908e-05,
+      "loss": 1.8741,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611783766708036e-05,
+      "loss": 1.8788,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610950905932756e-05,
+      "loss": 1.8946,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610117368424488e-05,
+      "loss": 1.855,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609283155106517e-05,
+      "loss": 1.8763,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084482669028772e-05,
+      "loss": 1.8865,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607612704738352e-05,
+      "loss": 1.8844,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067764695384682e-05,
+      "loss": 1.8635,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059395622295005e-05,
+      "loss": 1.91,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605101983738468e-05,
+      "loss": 1.9293,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604263734993132e-05,
+      "loss": 1.8983,
+      "step": 969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034248169219965e-05,
+      "loss": 1.9168,
+      "step": 970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602585230454308e-05,
+      "loss": 1.9082,
+      "step": 971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017449765200526e-05,
+      "loss": 1.8846,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009040560499548e-05,
+      "loss": 1.9413,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000624699754795e-05,
+      "loss": 1.8723,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992202192288274e-05,
+      "loss": 1.9403,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983773047429353e-05,
+      "loss": 1.7869,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597533727451476e-05,
+      "loss": 1.8762,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596689488288856e-05,
+      "loss": 1.9158,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958445881902152e-05,
+      "loss": 1.9273,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594999028091426e-05,
+      "loss": 1.9386,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941528089290903e-05,
+      "loss": 1.8793,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933059316405416e-05,
+      "loss": 1.9154,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924583971638416e-05,
+      "loss": 1.9148,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916102064377808e-05,
+      "loss": 1.8898,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907613604018753e-05,
+      "loss": 1.8895,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899118599963683e-05,
+      "loss": 1.8975,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589061706162227e-05,
+      "loss": 1.946,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882108998411427e-05,
+      "loss": 1.8941,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58735944197553e-05,
+      "loss": 1.8588,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865073335085236e-05,
+      "loss": 1.7609,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5856545753839804e-05,
+      "loss": 1.8927,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848011685464765e-05,
+      "loss": 1.8962,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839471139413065e-05,
+      "loss": 1.8625,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830924125144822e-05,
+      "loss": 1.8895,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822370652127323e-05,
+      "loss": 1.9218,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813810729835004e-05,
+      "loss": 1.9233,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805244367749453e-05,
+      "loss": 1.8876,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796671575359382e-05,
+      "loss": 1.8682,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788092362160633e-05,
+      "loss": 1.8662,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577950673765616e-05,
+      "loss": 1.8561,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577091471135601e-05,
+      "loss": 1.9165,
+      "step": 1001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762316292777328e-05,
+      "loss": 1.8661,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753711491444336e-05,
+      "loss": 1.917,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745100316888337e-05,
+      "loss": 1.869,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736482778647676e-05,
+      "loss": 1.8593,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727858886267758e-05,
+      "loss": 1.9355,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719228649301027e-05,
+      "loss": 1.8768,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571059207730695e-05,
+      "loss": 1.8339,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570194917985201e-05,
+      "loss": 1.847,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693299966509704e-05,
+      "loss": 1.8962,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684644446860518e-05,
+      "loss": 1.9035,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675982630491926e-05,
+      "loss": 1.9353,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667314526998373e-05,
+      "loss": 1.901,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658640145981276e-05,
+      "loss": 1.9405,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649959497049e-05,
+      "loss": 1.8748,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641272589816853e-05,
+      "loss": 1.9247,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563257943390707e-05,
+      "loss": 1.8944,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623880038948828e-05,
+      "loss": 1.9026,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615174414578184e-05,
+      "loss": 1.8843,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560646257043812e-05,
+      "loss": 1.875,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597744516178495e-05,
+      "loss": 1.8589,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589020261456053e-05,
+      "loss": 1.9049,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55802898159344e-05,
+      "loss": 1.8515,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557155318928401e-05,
+      "loss": 1.8907,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556281039118219e-05,
+      "loss": 1.8551,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555406143131309e-05,
+      "loss": 1.8633,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554530631936769e-05,
+      "loss": 1.9164,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553654506504377e-05,
+      "loss": 1.915,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527777678045928e-05,
+      "loss": 1.8932,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519004168085553e-05,
+      "loss": 1.908,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510224544880806e-05,
+      "loss": 1.915,
+      "step": 1031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5501438818156636e-05,
+      "loss": 1.8594,
+      "step": 1032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492646997644737e-05,
+      "loss": 1.8768,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483849093083554e-05,
+      "loss": 1.894,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475045114218286e-05,
+      "loss": 1.8916,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466235070800848e-05,
+      "loss": 1.9903,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457418972589876e-05,
+      "loss": 1.895,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448596829350706e-05,
+      "loss": 1.9145,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439768650855383e-05,
+      "loss": 1.8422,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430934446882624e-05,
+      "loss": 1.8264,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542209422721783e-05,
+      "loss": 1.9358,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541324800165306e-05,
+      "loss": 1.9051,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540439577998703e-05,
+      "loss": 1.8743,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395537572025093e-05,
+      "loss": 1.903,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386673387579233e-05,
+      "loss": 1.8553,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377803236468057e-05,
+      "loss": 1.8656,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368927128516777e-05,
+      "loss": 1.9213,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360045073557214e-05,
+      "loss": 1.8879,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535115708142776e-05,
+      "loss": 1.8331,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5342263161973393e-05,
+      "loss": 1.8837,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533336332504566e-05,
+      "loss": 1.9188,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5324457580502654e-05,
+      "loss": 1.9485,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315545938209016e-05,
+      "loss": 1.8606,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306628408035923e-05,
+      "loss": 1.8911,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529770499986106e-05,
+      "loss": 1.9117,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528877572356865e-05,
+      "loss": 1.8909,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279840589049383e-05,
+      "loss": 1.9272,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527089960620046e-05,
+      "loss": 1.8625,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261952784925557e-05,
+      "loss": 1.8914,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253000135134813e-05,
+      "loss": 1.9188,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524404166674482e-05,
+      "loss": 1.8877,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235077389678625e-05,
+      "loss": 1.8794,
+      "step": 1062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226107313865701e-05,
+      "loss": 1.8499,
+      "step": 1063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217131449241944e-05,
+      "loss": 1.9349,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208149805749668e-05,
+      "loss": 1.8805,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199162393337584e-05,
+      "loss": 1.9202,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519016922196079e-05,
+      "loss": 1.8978,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181170301580776e-05,
+      "loss": 1.8367,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172165642165382e-05,
+      "loss": 1.9412,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163155253688815e-05,
+      "loss": 1.8638,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154139146131633e-05,
+      "loss": 1.9604,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145117329480717e-05,
+      "loss": 1.9271,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136089813729276e-05,
+      "loss": 1.9187,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127056608876838e-05,
+      "loss": 1.9027,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118017724929226e-05,
+      "loss": 1.8772,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108973171898554e-05,
+      "loss": 1.9014,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099922959803217e-05,
+      "loss": 1.9475,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509086709866788e-05,
+      "loss": 1.8236,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508180559852346e-05,
+      "loss": 1.8485,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507273846940713e-05,
+      "loss": 1.8602,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063665721362286e-05,
+      "loss": 1.9131,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054587364438556e-05,
+      "loss": 1.9233,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045503408691776e-05,
+      "loss": 1.8918,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503641386418399e-05,
+      "loss": 1.8603,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027318740983425e-05,
+      "loss": 1.8355,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018218049164495e-05,
+      "loss": 1.8697,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009111798807775e-05,
+      "loss": 1.8569,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.9383,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990882662834057e-05,
+      "loss": 1.8947,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498175979740895e-05,
+      "loss": 1.8708,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497263141382983e-05,
+      "loss": 1.905,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963497522207942e-05,
+      "loss": 1.8416,
+      "step": 1092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495435813266064e-05,
+      "loss": 1.8508,
+      "step": 1093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945213255311368e-05,
+      "loss": 1.9224,
+      "step": 1094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936062900289648e-05,
+      "loss": 1.9025,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926907077731063e-05,
+      "loss": 1.8566,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917745797777264e-05,
+      "loss": 1.9074,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908579070575936e-05,
+      "loss": 1.8603,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489940690628081e-05,
+      "loss": 1.8817,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890229315051627e-05,
+      "loss": 1.8517,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881046307054144e-05,
+      "loss": 1.8889,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487185789246012e-05,
+      "loss": 1.8795,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862664081447297e-05,
+      "loss": 1.9042,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853464884199409e-05,
+      "loss": 1.9234,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844260310906139e-05,
+      "loss": 1.8958,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835050371763126e-05,
+      "loss": 1.8264,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825835076971968e-05,
+      "loss": 1.8812,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816614436740184e-05,
+      "loss": 1.8965,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807388461281214e-05,
+      "loss": 1.8849,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798157160814407e-05,
+      "loss": 1.8994,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788920545565014e-05,
+      "loss": 1.9037,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779678625764174e-05,
+      "loss": 1.901,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770431411648898e-05,
+      "loss": 1.8872,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761178913462062e-05,
+      "loss": 1.9084,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751921141452397e-05,
+      "loss": 1.8786,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742658105874474e-05,
+      "loss": 1.9261,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733389816988694e-05,
+      "loss": 1.8642,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724116285061278e-05,
+      "loss": 1.9167,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714837520364259e-05,
+      "loss": 1.8721,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705553533175451e-05,
+      "loss": 1.8943,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696264333778471e-05,
+      "loss": 1.8771,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686969932462702e-05,
+      "loss": 1.9399,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677670339523285e-05,
+      "loss": 1.8757,
+      "step": 1123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668365565261118e-05,
+      "loss": 1.8735,
+      "step": 1124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659055619982835e-05,
+      "loss": 1.8455,
+      "step": 1125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46497405140008e-05,
+      "loss": 1.8902,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640420257633085e-05,
+      "loss": 1.8515,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631094861203478e-05,
+      "loss": 1.8271,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621764335041459e-05,
+      "loss": 1.9042,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612428689482184e-05,
+      "loss": 1.892,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603087934866485e-05,
+      "loss": 1.8823,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593742081540847e-05,
+      "loss": 1.909,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584391139857407e-05,
+      "loss": 1.9037,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575035120173942e-05,
+      "loss": 1.8673,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565674032853847e-05,
+      "loss": 1.9452,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556307888266132e-05,
+      "loss": 1.9312,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546936696785413e-05,
+      "loss": 1.8321,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537560468791889e-05,
+      "loss": 1.9213,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528179214671346e-05,
+      "loss": 1.9184,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518792944815127e-05,
+      "loss": 1.8586,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509401669620146e-05,
+      "loss": 1.9353,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500005399488841e-05,
+      "loss": 1.8842,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490604144829204e-05,
+      "loss": 1.89,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448119791605473e-05,
+      "loss": 1.9206,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471786723584439e-05,
+      "loss": 1.9127,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462370577842838e-05,
+      "loss": 1.9359,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452949489259923e-05,
+      "loss": 1.8699,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443523468271168e-05,
+      "loss": 1.8902,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443409252531751e-05,
+      "loss": 1.8491,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424656670845338e-05,
+      "loss": 1.942,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441521591530648e-05,
+      "loss": 1.8336,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440577026915819e-05,
+      "loss": 1.9081,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396319742863145e-05,
+      "loss": 1.9196,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386864346889424e-05,
+      "loss": 1.8965,
+      "step": 1154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377404091710503e-05,
+      "loss": 1.805,
+      "step": 1155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367938987805236e-05,
+      "loss": 1.8967,
+      "step": 1156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435846904565785e-05,
+      "loss": 1.8611,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348994275757933e-05,
+      "loss": 1.906,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339514688600416e-05,
+      "loss": 1.8247,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330030294685572e-05,
+      "loss": 1.8524,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320541104518992e-05,
+      "loss": 1.8835,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311047128611586e-05,
+      "loss": 1.8669,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301548377479562e-05,
+      "loss": 1.8524,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292044861644415e-05,
+      "loss": 1.8555,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282536591632918e-05,
+      "loss": 1.8715,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273023577977116e-05,
+      "loss": 1.8543,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263505831214302e-05,
+      "loss": 1.8687,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253983361887017e-05,
+      "loss": 1.8996,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244456180543026e-05,
+      "loss": 1.806,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234924297735322e-05,
+      "loss": 1.8845,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225387724022097e-05,
+      "loss": 1.8895,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215846469966745e-05,
+      "loss": 1.8499,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206300546137844e-05,
+      "loss": 1.8994,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196749963109143e-05,
+      "loss": 1.9468,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187194731459546e-05,
+      "loss": 1.8339,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177634861773117e-05,
+      "loss": 1.891,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168070364639055e-05,
+      "loss": 1.8961,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415850125065168e-05,
+      "loss": 1.9126,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148927530410425e-05,
+      "loss": 1.8671,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139349214519831e-05,
+      "loss": 1.9102,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129766313589526e-05,
+      "loss": 1.8652,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4120178838234222e-05,
+      "loss": 1.9391,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110586799073684e-05,
+      "loss": 1.84,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410099020673275e-05,
+      "loss": 1.8873,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091389071841291e-05,
+      "loss": 1.8828,
+      "step": 1185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081783405034208e-05,
+      "loss": 1.8102,
+      "step": 1186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072173216951433e-05,
+      "loss": 1.958,
+      "step": 1187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062558518237893e-05,
+      "loss": 1.8807,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052939319543517e-05,
+      "loss": 1.8872,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043315631523223e-05,
+      "loss": 1.9155,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403368746483689e-05,
+      "loss": 1.9117,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024054830149374e-05,
+      "loss": 1.8797,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014417738130464e-05,
+      "loss": 1.8836,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004776199454896e-05,
+      "loss": 1.9,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995130224802332e-05,
+      "loss": 1.8911,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398547982485734e-05,
+      "loss": 1.9067,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975825010309393e-05,
+      "loss": 1.8853,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966165791852862e-05,
+      "loss": 1.9275,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956502180186984e-05,
+      "loss": 1.8506,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394683418601587e-05,
+      "loss": 1.8552,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937161820048481e-05,
+      "loss": 1.8554,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927485092998623e-05,
+      "loss": 1.8553,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917804015584932e-05,
+      "loss": 1.8932,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908118598530864e-05,
+      "loss": 1.871,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898428852564678e-05,
+      "loss": 1.9064,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888734788419434e-05,
+      "loss": 1.8843,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879036416832974e-05,
+      "loss": 1.9091,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869333748547901e-05,
+      "loss": 1.9098,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859626794311592e-05,
+      "loss": 1.9451,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849915564876161e-05,
+      "loss": 1.8461,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840200070998466e-05,
+      "loss": 1.8589,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830480323440081e-05,
+      "loss": 1.9353,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820756332967294e-05,
+      "loss": 1.921,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811028110351092e-05,
+      "loss": 1.9017,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380129566636716e-05,
+      "loss": 1.8726,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791559011795838e-05,
+      "loss": 1.8513,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781818157422149e-05,
+      "loss": 1.9197,
+      "step": 1217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772073114035762e-05,
+      "loss": 1.9113,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762323892430982e-05,
+      "loss": 1.8979,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752570503406747e-05,
+      "loss": 1.8554,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742812957766607e-05,
+      "loss": 1.8327,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373305126631872e-05,
+      "loss": 1.8457,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723285439875836e-05,
+      "loss": 1.8971,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713515489255283e-05,
+      "loss": 1.895,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703741425278957e-05,
+      "loss": 1.861,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693963258773316e-05,
+      "loss": 1.8983,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368418100056935e-05,
+      "loss": 1.8874,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674394661502595e-05,
+      "loss": 1.8713,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664604252413098e-05,
+      "loss": 1.8544,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365480978414542e-05,
+      "loss": 1.9473,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645011267548612e-05,
+      "loss": 1.8682,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363520871347622e-05,
+      "loss": 1.8989,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625402132786247e-05,
+      "loss": 1.844,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615591536341168e-05,
+      "loss": 1.8504,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605776935007902e-05,
+      "loss": 1.8504,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595958339657805e-05,
+      "loss": 1.8877,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586135761166655e-05,
+      "loss": 1.9057,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576309210414646e-05,
+      "loss": 1.8499,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566478698286367e-05,
+      "loss": 1.8891,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556644235670799e-05,
+      "loss": 1.8741,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546805833461298e-05,
+      "loss": 1.9016,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536963502555581e-05,
+      "loss": 1.8684,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352711725385572e-05,
+      "loss": 1.9111,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517267098268126e-05,
+      "loss": 1.887,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507413046703534e-05,
+      "loss": 1.8694,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497555110077001e-05,
+      "loss": 1.9166,
+      "step": 1246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348769329930788e-05,
+      "loss": 1.9126,
+      "step": 1247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477827625319826e-05,
+      "loss": 1.9084,
+      "step": 1248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346795809904076e-05,
+      "loss": 1.8251,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458084731402874e-05,
+      "loss": 1.8245,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448207533342625e-05,
+      "loss": 1.8719,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438326515800699e-05,
+      "loss": 1.8561,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428441689722023e-05,
+      "loss": 1.8783,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418553066055734e-05,
+      "loss": 1.979,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408660655755186e-05,
+      "loss": 1.8714,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398764469777914e-05,
+      "loss": 1.8919,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388864519085649e-05,
+      "loss": 1.8677,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378960814644283e-05,
+      "loss": 1.8855,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369053367423868e-05,
+      "loss": 1.8829,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33591421883986e-05,
+      "loss": 1.8265,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349227288546815e-05,
+      "loss": 1.883,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339308678850964e-05,
+      "loss": 1.8695,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329386370297615e-05,
+      "loss": 1.9203,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319460373877418e-05,
+      "loss": 1.8781,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309530700585125e-05,
+      "loss": 1.8535,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299597361419549e-05,
+      "loss": 1.8698,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289660367383573e-05,
+      "loss": 1.8422,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279719729484117e-05,
+      "loss": 1.897,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269775458732153e-05,
+      "loss": 1.8511,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259827566142658e-05,
+      "loss": 1.8947,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249876062734634e-05,
+      "loss": 1.9449,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323992095953108e-05,
+      "loss": 1.8347,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229962267558982e-05,
+      "loss": 1.8923,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219999997849299e-05,
+      "loss": 1.8669,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210034161436955e-05,
+      "loss": 1.9105,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200064769360828e-05,
+      "loss": 1.9169,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190091832663722e-05,
+      "loss": 1.8425,
+      "step": 1277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180115362392383e-05,
+      "loss": 1.8681,
+      "step": 1278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170135369597463e-05,
+      "loss": 1.8083,
+      "step": 1279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160151865333515e-05,
+      "loss": 1.8905,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150164860658986e-05,
+      "loss": 1.8731,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140174366636198e-05,
+      "loss": 1.8927,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130180394331335e-05,
+      "loss": 1.8975,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120182954814438e-05,
+      "loss": 1.8892,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311018205915939e-05,
+      "loss": 1.8931,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100177718443892e-05,
+      "loss": 1.8636,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.8524,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080158746161468e-05,
+      "loss": 1.9418,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070144136768984e-05,
+      "loss": 1.8689,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060126126664929e-05,
+      "loss": 1.8814,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050104726945962e-05,
+      "loss": 1.8447,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040079948712507e-05,
+      "loss": 1.8551,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030051803068729e-05,
+      "loss": 1.8455,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020020301122517e-05,
+      "loss": 1.8668,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300998545398548e-05,
+      "loss": 1.7992,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999947272772935e-05,
+      "loss": 1.8085,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989905768603893e-05,
+      "loss": 1.8747,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979860952601038e-05,
+      "loss": 1.9297,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969812835890732e-05,
+      "loss": 1.8471,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959761429602986e-05,
+      "loss": 1.9146,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949706744871464e-05,
+      "loss": 1.8847,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939648792833449e-05,
+      "loss": 1.8439,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929587584629845e-05,
+      "loss": 1.8407,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919523131405178e-05,
+      "loss": 1.9125,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290945544430755e-05,
+      "loss": 1.8428,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899384534488656e-05,
+      "loss": 1.8461,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889310413103755e-05,
+      "loss": 1.8447,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879233091311667e-05,
+      "loss": 1.9166,
+      "step": 1308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869152580274751e-05,
+      "loss": 1.8482,
+      "step": 1309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285906889115891e-05,
+      "loss": 1.8942,
+      "step": 1310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848982035133555e-05,
+      "loss": 1.8732,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838892023371612e-05,
+      "loss": 1.9449,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828798867049504e-05,
+      "loss": 1.8537,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818702577347129e-05,
+      "loss": 1.8296,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808603165447862e-05,
+      "loss": 1.907,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798500642538536e-05,
+      "loss": 1.8675,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278839501980943e-05,
+      "loss": 1.8972,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778286308454255e-05,
+      "loss": 1.851,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768174519670144e-05,
+      "loss": 1.9134,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758059664657636e-05,
+      "loss": 1.8747,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747941754620674e-05,
+      "loss": 1.8235,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737820800766574e-05,
+      "loss": 1.8957,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727696814306034e-05,
+      "loss": 1.8578,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717569806453103e-05,
+      "loss": 1.8341,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707439788425184e-05,
+      "loss": 1.8416,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269730677144301e-05,
+      "loss": 1.8665,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687170766730635e-05,
+      "loss": 1.9373,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677031785515423e-05,
+      "loss": 1.8808,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666889839028038e-05,
+      "loss": 1.8765,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656744938502425e-05,
+      "loss": 1.9203,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646597095175805e-05,
+      "loss": 1.8674,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636446320288655e-05,
+      "loss": 1.8887,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26262926250847e-05,
+      "loss": 1.8197,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616136020810898e-05,
+      "loss": 1.9001,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605976518717436e-05,
+      "loss": 1.9105,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595814130057702e-05,
+      "loss": 1.8836,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585648866088287e-05,
+      "loss": 1.7655,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575480738068971e-05,
+      "loss": 1.8202,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565309757262689e-05,
+      "loss": 1.8541,
+      "step": 1339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555135934935558e-05,
+      "loss": 1.9313,
+      "step": 1340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254495928235683e-05,
+      "loss": 1.9387,
+      "step": 1341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534779810798894e-05,
+      "loss": 1.9144,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524597531537261e-05,
+      "loss": 1.8637,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2514412455850553e-05,
+      "loss": 1.8639,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504224595020493e-05,
+      "loss": 1.9231,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494033960331879e-05,
+      "loss": 1.8937,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483840563072594e-05,
+      "loss": 1.9417,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473644414533573e-05,
+      "loss": 1.8628,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463445526008797e-05,
+      "loss": 1.8562,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453243908795289e-05,
+      "loss": 1.7993,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443039574193094e-05,
+      "loss": 1.8082,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432832533505257e-05,
+      "loss": 1.8469,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422622798037833e-05,
+      "loss": 1.9123,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412410379099855e-05,
+      "loss": 1.9099,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402195288003327e-05,
+      "loss": 1.9331,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391977536063218e-05,
+      "loss": 1.8379,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381757134597443e-05,
+      "loss": 1.8645,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371534094926852e-05,
+      "loss": 1.8641,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361308428375209e-05,
+      "loss": 1.8402,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351080146269201e-05,
+      "loss": 1.903,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340849259938404e-05,
+      "loss": 1.8005,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233061578071528e-05,
+      "loss": 1.88,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232037971993517e-05,
+      "loss": 1.8288,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310141088936259e-05,
+      "loss": 1.9196,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299899899059589e-05,
+      "loss": 1.8824,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289656161649041e-05,
+      "loss": 1.8789,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227940988805131e-05,
+      "loss": 1.8552,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269161089615902e-05,
+      "loss": 1.8527,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258909777695123e-05,
+      "loss": 1.8641,
+      "step": 1369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248655963644056e-05,
+      "loss": 1.8541,
+      "step": 1370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238399658820563e-05,
+      "loss": 1.8819,
+      "step": 1371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228140874585262e-05,
+      "loss": 1.8892,
+      "step": 1372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217879622301514e-05,
+      "loss": 1.8991,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220761591333542e-05,
+      "loss": 1.8927,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197349759055799e-05,
+      "loss": 1.8677,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218708117083418e-05,
+      "loss": 1.8694,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176810160044786e-05,
+      "loss": 1.8737,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166536738064523e-05,
+      "loss": 1.8938,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156260916272969e-05,
+      "loss": 1.8493,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145982706052361e-05,
+      "loss": 1.9012,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135702118787582e-05,
+      "loss": 1.889,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212541916586615e-05,
+      "loss": 1.8142,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115133858678192e-05,
+      "loss": 1.889,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104846208616457e-05,
+      "loss": 1.833,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209455622707628e-05,
+      "loss": 1.9354,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084263925455583e-05,
+      "loss": 1.8516,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207396931515486e-05,
+      "loss": 1.8559,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063672407577154e-05,
+      "loss": 1.8926,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053373214128057e-05,
+      "loss": 1.8588,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043071746215698e-05,
+      "loss": 1.8964,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203276801525072e-05,
+      "loss": 1.8179,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202246203264627e-05,
+      "loss": 1.8323,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012153809817992e-05,
+      "loss": 1.9167,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001843358184016e-05,
+      "loss": 1.8693,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991530689164938e-05,
+      "loss": 1.8635,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981215814183804e-05,
+      "loss": 1.8725,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197089874466611e-05,
+      "loss": 1.9263,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960579492039783e-05,
+      "loss": 1.8589,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950258067735163e-05,
+      "loss": 1.812,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939934483185005e-05,
+      "loss": 1.8422,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929608749824446e-05,
+      "loss": 1.8494,
+      "step": 1401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1919280879091008e-05,
+      "loss": 1.9088,
+      "step": 1402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908950882424581e-05,
+      "loss": 1.8704,
+      "step": 1403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898618771267411e-05,
+      "loss": 1.8673,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888284557064083e-05,
+      "loss": 1.8348,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877948251261512e-05,
+      "loss": 1.9081,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867609865308936e-05,
+      "loss": 1.9333,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857269410657883e-05,
+      "loss": 1.871,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846926898762187e-05,
+      "loss": 1.8109,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836582341077956e-05,
+      "loss": 1.8738,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182623574906356e-05,
+      "loss": 1.808,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815887134179627e-05,
+      "loss": 1.8679,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805536507889021e-05,
+      "loss": 1.8822,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795183881656846e-05,
+      "loss": 1.8604,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784829266950401e-05,
+      "loss": 1.9078,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774472675239208e-05,
+      "loss": 1.8374,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764114117994964e-05,
+      "loss": 1.8725,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753753606691554e-05,
+      "loss": 1.7883,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743391152805017e-05,
+      "loss": 1.9123,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733026767813556e-05,
+      "loss": 1.854,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17226604631975e-05,
+      "loss": 1.8792,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171229225043931e-05,
+      "loss": 1.8625,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701922141023566e-05,
+      "loss": 1.88,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691550146436948e-05,
+      "loss": 1.8261,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681176278168207e-05,
+      "loss": 1.8936,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670800547708192e-05,
+      "loss": 1.8865,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660422966549802e-05,
+      "loss": 1.8234,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650043546187994e-05,
+      "loss": 1.8889,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639662298119753e-05,
+      "loss": 1.8211,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629279233844092e-05,
+      "loss": 1.868,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161889436486204e-05,
+      "loss": 1.8019,
+      "step": 1431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608507702676624e-05,
+      "loss": 1.8182,
+      "step": 1432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598119258792848e-05,
+      "loss": 1.9148,
+      "step": 1433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587729044717702e-05,
+      "loss": 1.8885,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157733707196013e-05,
+      "loss": 1.8677,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566943352031022e-05,
+      "loss": 1.9414,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155654789644321e-05,
+      "loss": 1.8383,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546150716711448e-05,
+      "loss": 1.8761,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535751824352395e-05,
+      "loss": 1.9044,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525351230884607e-05,
+      "loss": 1.8237,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514948947828528e-05,
+      "loss": 1.8612,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504544986706471e-05,
+      "loss": 1.9355,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494139359042612e-05,
+      "loss": 1.8577,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483732076362964e-05,
+      "loss": 1.8723,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473323150195383e-05,
+      "loss": 1.8542,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146291259206954e-05,
+      "loss": 1.8551,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452500413516913e-05,
+      "loss": 1.8262,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442086626070781e-05,
+      "loss": 1.8737,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431671241266198e-05,
+      "loss": 1.8741,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421254270639989e-05,
+      "loss": 1.8358,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410835725730745e-05,
+      "loss": 1.9003,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400415618078782e-05,
+      "loss": 1.8933,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389993959226163e-05,
+      "loss": 1.8702,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379570760716661e-05,
+      "loss": 1.8969,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369146034095761e-05,
+      "loss": 1.8567,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358719790910635e-05,
+      "loss": 1.86,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348292042710141e-05,
+      "loss": 1.9081,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337862801044792e-05,
+      "loss": 1.9302,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1327432077466769e-05,
+      "loss": 1.8354,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316999883529887e-05,
+      "loss": 1.8824,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306566230789593e-05,
+      "loss": 1.9315,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296131130802943e-05,
+      "loss": 1.8669,
+      "step": 1462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285694595128606e-05,
+      "loss": 1.8281,
+      "step": 1463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275256635326837e-05,
+      "loss": 1.8353,
+      "step": 1464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264817262959465e-05,
+      "loss": 1.8849,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254376489589886e-05,
+      "loss": 1.8516,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243934326783052e-05,
+      "loss": 1.9098,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123349078610545e-05,
+      "loss": 1.8846,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122304587912509e-05,
+      "loss": 1.8862,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212599617411506e-05,
+      "loss": 1.8747,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202152012535718e-05,
+      "loss": 1.8777,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191703076070248e-05,
+      "loss": 1.897,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181252819589081e-05,
+      "loss": 1.8842,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170801254667673e-05,
+      "loss": 1.8502,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160348392882928e-05,
+      "loss": 1.9165,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149894245813181e-05,
+      "loss": 1.8742,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139438825038196e-05,
+      "loss": 1.9206,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128982142139142e-05,
+      "loss": 1.8597,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118524208698596e-05,
+      "loss": 1.8436,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108065036300509e-05,
+      "loss": 1.8559,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097604636530213e-05,
+      "loss": 1.8452,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087143020974396e-05,
+      "loss": 1.8681,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076680201221093e-05,
+      "loss": 1.8069,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066216188859671e-05,
+      "loss": 1.9387,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055750995480818e-05,
+      "loss": 1.825,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.8721,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034817112040116e-05,
+      "loss": 1.8788,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024348445166133e-05,
+      "loss": 1.923,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013878643650436e-05,
+      "loss": 1.866,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003407719090123e-05,
+      "loss": 1.8835,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099293568308354e-05,
+      "loss": 1.8576,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982462547230269e-05,
+      "loss": 1.9054,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971988323131099e-05,
+      "loss": 1.8676,
+      "step": 1493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096151302238804e-05,
+      "loss": 1.8593,
+      "step": 1494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951036656604275e-05,
+      "loss": 1.8518,
+      "step": 1495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940559237384183e-05,
+      "loss": 1.8286,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930080776333304e-05,
+      "loss": 1.8646,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091960128505833e-05,
+      "loss": 1.7971,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09091207751671e-05,
+      "loss": 1.9319,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898639258268571e-05,
+      "loss": 1.8405,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088815674597283e-05,
+      "loss": 1.7849,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877673249891052e-05,
+      "loss": 1.799,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086718878163551e-05,
+      "loss": 1.8537,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856703352819554e-05,
+      "loss": 1.8979,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846216975057593e-05,
+      "loss": 1.9087,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835729659965095e-05,
+      "loss": 1.8351,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825241419158557e-05,
+      "loss": 1.8234,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814752264255508e-05,
+      "loss": 1.8637,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804262206874486e-05,
+      "loss": 1.9007,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793771258635031e-05,
+      "loss": 1.9025,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783279431157668e-05,
+      "loss": 1.8902,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772786736063896e-05,
+      "loss": 1.8773,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762293184976178e-05,
+      "loss": 1.8782,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751798789517925e-05,
+      "loss": 1.9113,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741303561313474e-05,
+      "loss": 1.9096,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0730807511988097e-05,
+      "loss": 1.8693,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720310653167967e-05,
+      "loss": 1.8445,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070981299648016e-05,
+      "loss": 1.836,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069931455355263e-05,
+      "loss": 1.8558,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688815336014209e-05,
+      "loss": 1.8744,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678315355494575e-05,
+      "loss": 1.8675,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667814623624262e-05,
+      "loss": 1.8839,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657313152034634e-05,
+      "loss": 1.9155,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646810952357874e-05,
+      "loss": 1.8047,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636308036226968e-05,
+      "loss": 1.8475,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06258044152757e-05,
+      "loss": 1.8428,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615300101138632e-05,
+      "loss": 1.8824,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604795105451096e-05,
+      "loss": 1.8992,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594289439849174e-05,
+      "loss": 1.9398,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583783115969699e-05,
+      "loss": 1.8717,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573276145450228e-05,
+      "loss": 1.8791,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562768539929029e-05,
+      "loss": 1.8668,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552260311045082e-05,
+      "loss": 1.8112,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541751470438056e-05,
+      "loss": 1.881,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053124202974829e-05,
+      "loss": 1.8927,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520732000616797e-05,
+      "loss": 1.824,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510221394685238e-05,
+      "loss": 1.864,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499710223595913e-05,
+      "loss": 1.9126,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048919849899175e-05,
+      "loss": 1.8507,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478686232516284e-05,
+      "loss": 1.9001,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468173435813656e-05,
+      "loss": 1.9045,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457660120528593e-05,
+      "loss": 1.9349,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447146298306394e-05,
+      "loss": 1.8028,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436631980792927e-05,
+      "loss": 1.8831,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04261171796346e-05,
+      "loss": 1.8697,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415601906478362e-05,
+      "loss": 1.8619,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405086172971676e-05,
+      "loss": 1.8648,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394569990762528e-05,
+      "loss": 1.8122,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384053371499397e-05,
+      "loss": 1.8528,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373536326831235e-05,
+      "loss": 1.8245,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363018868407481e-05,
+      "loss": 1.8738,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352501007878029e-05,
+      "loss": 1.8662,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341982756893203e-05,
+      "loss": 1.889,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331464127103781e-05,
+      "loss": 1.8569,
+      "step": 1554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320945130160944e-05,
+      "loss": 1.8519,
+      "step": 1555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310425777716293e-05,
+      "loss": 1.8538,
+      "step": 1556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299906081421814e-05,
+      "loss": 1.8588,
+      "step": 1557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289386052929874e-05,
+      "loss": 1.8768,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278865703893212e-05,
+      "loss": 1.8577,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026834504596492e-05,
+      "loss": 1.883,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257824090798431e-05,
+      "loss": 1.8479,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247302850047512e-05,
+      "loss": 1.8782,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236781335366239e-05,
+      "loss": 1.9348,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0226259558409002e-05,
+      "loss": 1.8422,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215737530830465e-05,
+      "loss": 1.8113,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205215264285587e-05,
+      "loss": 1.9179,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019469277042958e-05,
+      "loss": 1.8637,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184170060917914e-05,
+      "loss": 1.8665,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0173647147406298e-05,
+      "loss": 1.8138,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163124041550657e-05,
+      "loss": 1.8438,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015260075500715e-05,
+      "loss": 1.811,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142077299432111e-05,
+      "loss": 1.8434,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131553686482077e-05,
+      "loss": 1.8599,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121029927813757e-05,
+      "loss": 1.8399,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110506035084017e-05,
+      "loss": 1.8683,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009998201994988e-05,
+      "loss": 1.8984,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089457894068493e-05,
+      "loss": 1.8506,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078933669097135e-05,
+      "loss": 1.8322,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006840935669319e-05,
+      "loss": 1.8544,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057884968514142e-05,
+      "loss": 1.8481,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047360516217555e-05,
+      "loss": 1.8735,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003683601146107e-05,
+      "loss": 1.7826,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002631146590238e-05,
+      "loss": 1.8636,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015786891199222e-05,
+      "loss": 1.8565,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005262299009374e-05,
+      "loss": 1.9033,
+      "step": 1585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994737700990627e-06,
+      "loss": 1.9699,
+      "step": 1586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98421310880078e-06,
+      "loss": 1.8793,
+      "step": 1587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973688534097624e-06,
+      "loss": 1.8485,
+      "step": 1588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963163988538932e-06,
+      "loss": 1.8677,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952639483782445e-06,
+      "loss": 1.8138,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942115031485858e-06,
+      "loss": 1.8398,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931590643306815e-06,
+      "loss": 1.8763,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92106633090287e-06,
+      "loss": 1.844,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910542105931512e-06,
+      "loss": 1.8899,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900017980050125e-06,
+      "loss": 1.8732,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889493964915986e-06,
+      "loss": 1.8456,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878970072186246e-06,
+      "loss": 1.8708,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868446313517927e-06,
+      "loss": 1.8545,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857922700567892e-06,
+      "loss": 1.8856,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847399244992852e-06,
+      "loss": 1.8199,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836875958449344e-06,
+      "loss": 1.8982,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826352852593705e-06,
+      "loss": 1.8388,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815829939082087e-06,
+      "loss": 1.8888,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805307229570421e-06,
+      "loss": 1.85,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794784735714418e-06,
+      "loss": 1.8547,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784262469169537e-06,
+      "loss": 1.873,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773740441591002e-06,
+      "loss": 1.8479,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763218664633763e-06,
+      "loss": 1.8931,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752697149952491e-06,
+      "loss": 1.9004,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74217590920157e-06,
+      "loss": 1.7639,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731654954035082e-06,
+      "loss": 1.8816,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72113429610679e-06,
+      "loss": 1.8934,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710613947070127e-06,
+      "loss": 1.8384,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700093918578188e-06,
+      "loss": 1.8634,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689574222283707e-06,
+      "loss": 1.8974,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679054869839055e-06,
+      "loss": 1.8192,
+      "step": 1616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668535872896226e-06,
+      "loss": 1.8255,
+      "step": 1617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658017243106802e-06,
+      "loss": 1.8351,
+      "step": 1618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647498992121978e-06,
+      "loss": 1.893,
+      "step": 1619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63698113159252e-06,
+      "loss": 1.8219,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626463673168768e-06,
+      "loss": 1.9007,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615946628500608e-06,
+      "loss": 1.8566,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605430009237474e-06,
+      "loss": 1.8333,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594913827028327e-06,
+      "loss": 1.8138,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584398093521643e-06,
+      "loss": 1.8832,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573882820365401e-06,
+      "loss": 1.8236,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563368019207075e-06,
+      "loss": 1.8865,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552853701693606e-06,
+      "loss": 1.8443,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542339879471409e-06,
+      "loss": 1.9053,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53182656418635e-06,
+      "loss": 1.9051,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521313767483721e-06,
+      "loss": 1.8314,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510801501008255e-06,
+      "loss": 1.8891,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50028977640409e-06,
+      "loss": 1.8798,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489778605314764e-06,
+      "loss": 1.9031,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479267999383204e-06,
+      "loss": 1.8924,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468757970251712e-06,
+      "loss": 1.8711,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458248529561946e-06,
+      "loss": 1.8725,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44773968895492e-06,
+      "loss": 1.892,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437231460070973e-06,
+      "loss": 1.8593,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426723854549774e-06,
+      "loss": 1.8927,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4162168840303e-06,
+      "loss": 1.8747,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405710560150828e-06,
+      "loss": 1.8624,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395204894548907e-06,
+      "loss": 1.8122,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384699898861373e-06,
+      "loss": 1.8622,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374195584724303e-06,
+      "loss": 1.8613,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363691963773036e-06,
+      "loss": 1.8417,
+      "step": 1646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35318904764213e-06,
+      "loss": 1.8831,
+      "step": 1647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342686847965367e-06,
+      "loss": 1.8801,
+      "step": 1648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33218537637574e-06,
+      "loss": 1.8725,
+      "step": 1649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321684644505429e-06,
+      "loss": 1.8862,
+      "step": 1650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311184663985795e-06,
+      "loss": 1.7662,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300685446447371e-06,
+      "loss": 1.8329,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290187003519841e-06,
+      "loss": 1.8779,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279689346832033e-06,
+      "loss": 1.8508,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26919248801191e-06,
+      "loss": 1.8053,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258696438686531e-06,
+      "loss": 1.8594,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24820121048208e-06,
+      "loss": 1.8105,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237706815023824e-06,
+      "loss": 1.9023,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227213263936106e-06,
+      "loss": 1.8395,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216720568842336e-06,
+      "loss": 1.9072,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206228741364972e-06,
+      "loss": 1.8297,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195737793125516e-06,
+      "loss": 1.9096,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185247735744495e-06,
+      "loss": 1.8912,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174758580841445e-06,
+      "loss": 1.7967,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164270340034906e-06,
+      "loss": 1.8584,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153783024942407e-06,
+      "loss": 1.8749,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14329664718045e-06,
+      "loss": 1.8598,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132811218364494e-06,
+      "loss": 1.8378,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122326750108953e-06,
+      "loss": 1.8839,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111843254027176e-06,
+      "loss": 1.8675,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101360741731432e-06,
+      "loss": 1.8386,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090879224832904e-06,
+      "loss": 1.8436,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080398714941672e-06,
+      "loss": 1.8513,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0699192236667e-06,
+      "loss": 1.8902,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05944076261582e-06,
+      "loss": 1.8677,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048963343395729e-06,
+      "loss": 1.8405,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038486977611965e-06,
+      "loss": 1.85,
+      "step": 1677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028011676868901e-06,
+      "loss": 1.8439,
+      "step": 1678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017537452769733e-06,
+      "loss": 1.861,
+      "step": 1679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007064316916461e-06,
+      "loss": 1.8708,
+      "step": 1680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996592280909882e-06,
+      "loss": 1.8939,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986121356349569e-06,
+      "loss": 1.8691,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975651554833869e-06,
+      "loss": 1.785,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965182887959888e-06,
+      "loss": 1.9032,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 1.8421,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944249004519184e-06,
+      "loss": 1.8431,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933783811140334e-06,
+      "loss": 1.8134,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92331979877891e-06,
+      "loss": 1.8815,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912856979025606e-06,
+      "loss": 1.8563,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902395363469787e-06,
+      "loss": 1.8497,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891934963699491e-06,
+      "loss": 1.841,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881475791301406e-06,
+      "loss": 1.8441,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871017857860863e-06,
+      "loss": 1.869,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86056117496181e-06,
+      "loss": 1.8513,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850105754186824e-06,
+      "loss": 1.8525,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839651607117076e-06,
+      "loss": 1.8351,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829198745332328e-06,
+      "loss": 1.8746,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81874718041092e-06,
+      "loss": 1.8943,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808296923929755e-06,
+      "loss": 1.8485,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797847987464284e-06,
+      "loss": 1.8658,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787400382588497e-06,
+      "loss": 1.8075,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77695412087491e-06,
+      "loss": 1.8067,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766509213894552e-06,
+      "loss": 1.8589,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75606567321695e-06,
+      "loss": 1.865,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745623510410119e-06,
+      "loss": 1.8571,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73518273704054e-06,
+      "loss": 1.9166,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724743364673168e-06,
+      "loss": 1.7649,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714305404871397e-06,
+      "loss": 1.878,
+      "step": 1708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703868869197059e-06,
+      "loss": 1.9007,
+      "step": 1709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69343376921041e-06,
+      "loss": 1.8917,
+      "step": 1710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683000116470117e-06,
+      "loss": 1.8446,
+      "step": 1711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672567922533234e-06,
+      "loss": 1.8382,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662137198955211e-06,
+      "loss": 1.8285,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651707957289862e-06,
+      "loss": 1.86,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641280209089365e-06,
+      "loss": 1.8994,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63085396590424e-06,
+      "loss": 1.8397,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62042923928334e-06,
+      "loss": 1.8452,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610006040773844e-06,
+      "loss": 1.882,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599584381921224e-06,
+      "loss": 1.848,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589164274269262e-06,
+      "loss": 1.833,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578745729360013e-06,
+      "loss": 1.8563,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568328758733807e-06,
+      "loss": 1.8038,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557913373929222e-06,
+      "loss": 1.8828,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547499586483088e-06,
+      "loss": 1.8718,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537087407930464e-06,
+      "loss": 1.8999,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526676849804619e-06,
+      "loss": 1.8346,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516267923637038e-06,
+      "loss": 1.8796,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50586064095739e-06,
+      "loss": 1.9053,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495455013293529e-06,
+      "loss": 1.7929,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485051052171477e-06,
+      "loss": 1.9155,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474648769115396e-06,
+      "loss": 1.8562,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464248175647609e-06,
+      "loss": 1.8769,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453849283288554e-06,
+      "loss": 1.8219,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443452103556791e-06,
+      "loss": 1.9065,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433056647968982e-06,
+      "loss": 1.8712,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422662928039876e-06,
+      "loss": 1.7926,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412270955282303e-06,
+      "loss": 1.7673,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401880741207155e-06,
+      "loss": 1.8805,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39149229732338e-06,
+      "loss": 1.8358,
+      "step": 1739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381105635137959e-06,
+      "loss": 1.828,
+      "step": 1740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37072076615591e-06,
+      "loss": 1.8584,
+      "step": 1741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36033770188025e-06,
+      "loss": 1.738,
+      "step": 1742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349956453812009e-06,
+      "loss": 1.858,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3395770334502e-06,
+      "loss": 1.8412,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329199452291811e-06,
+      "loss": 1.8481,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318823721831796e-06,
+      "loss": 1.8872,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308449853563057e-06,
+      "loss": 1.8304,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298077858976435e-06,
+      "loss": 1.8364,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287707749560691e-06,
+      "loss": 1.8767,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277339536802505e-06,
+      "loss": 1.8422,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266973232186449e-06,
+      "loss": 1.8676,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256608847194983e-06,
+      "loss": 1.887,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246246393308448e-06,
+      "loss": 1.8551,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235885882005036e-06,
+      "loss": 1.8159,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225527324760797e-06,
+      "loss": 1.8173,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215170733049602e-06,
+      "loss": 1.9344,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20481611834316e-06,
+      "loss": 1.8341,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194463492110982e-06,
+      "loss": 1.8644,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184112865820377e-06,
+      "loss": 1.8809,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173764250936444e-06,
+      "loss": 1.8629,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163417658922048e-06,
+      "loss": 1.8506,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153073101237814e-06,
+      "loss": 1.9074,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142730589342119e-06,
+      "loss": 1.8702,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132390134691069e-06,
+      "loss": 1.8844,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12205174873849e-06,
+      "loss": 1.8029,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111715442935919e-06,
+      "loss": 1.8613,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10138122873259e-06,
+      "loss": 1.9026,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091049117575424e-06,
+      "loss": 1.921,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080719120908997e-06,
+      "loss": 1.8875,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070391250175559e-06,
+      "loss": 1.8664,
+      "step": 1770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060065516814998e-06,
+      "loss": 1.8274,
+      "step": 1771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049741932264839e-06,
+      "loss": 1.8425,
+      "step": 1772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03942050796022e-06,
+      "loss": 1.8284,
+      "step": 1773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029101255333893e-06,
+      "loss": 1.8159,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0187841858162e-06,
+      "loss": 1.8874,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008469310835066e-06,
+      "loss": 1.8395,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998156641815985e-06,
+      "loss": 1.8447,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98784619018201e-06,
+      "loss": 1.8242,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977537967353734e-06,
+      "loss": 1.8349,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967231984749287e-06,
+      "loss": 1.8495,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956928253784305e-06,
+      "loss": 1.8549,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946626785871945e-06,
+      "loss": 1.8413,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93632759242285e-06,
+      "loss": 1.9071,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926030684845146e-06,
+      "loss": 1.835,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91573607454442e-06,
+      "loss": 1.8124,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905443772923724e-06,
+      "loss": 1.8867,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895153791383547e-06,
+      "loss": 1.8784,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884866141321811e-06,
+      "loss": 1.8487,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874580834133853e-06,
+      "loss": 1.794,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864297881212418e-06,
+      "loss": 1.7862,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854017293947639e-06,
+      "loss": 1.8748,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84373908372703e-06,
+      "loss": 1.8336,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833463261935482e-06,
+      "loss": 1.8238,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82318983995522e-06,
+      "loss": 1.839,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812918829165823e-06,
+      "loss": 1.8467,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802650240944203e-06,
+      "loss": 1.8462,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792384086664582e-06,
+      "loss": 1.8256,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782120377698489e-06,
+      "loss": 1.8883,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771859125414743e-06,
+      "loss": 1.8534,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761600341179439e-06,
+      "loss": 1.877,
+      "step": 1800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751344036355945e-06,
+      "loss": 1.8447,
+      "step": 1801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741090222304878e-06,
+      "loss": 1.8263,
+      "step": 1802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730838910384098e-06,
+      "loss": 1.8992,
+      "step": 1803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72059011194869e-06,
+      "loss": 1.8247,
+      "step": 1804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710343838350964e-06,
+      "loss": 1.8776,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700100100940416e-06,
+      "loss": 1.8137,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689858911063748e-06,
+      "loss": 1.8428,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679620280064837e-06,
+      "loss": 1.8112,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669384219284722e-06,
+      "loss": 1.8896,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6591507400616e-06,
+      "loss": 1.7778,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648919853730804e-06,
+      "loss": 1.8188,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638691571624795e-06,
+      "loss": 1.8167,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6284659050731525e-06,
+      "loss": 1.836,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61824286540256e-06,
+      "loss": 1.7829,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608022463936784e-06,
+      "loss": 1.7489,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597804711996675e-06,
+      "loss": 1.7644,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587589620900148e-06,
+      "loss": 1.8498,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57737720196217e-06,
+      "loss": 1.8802,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567167466494745e-06,
+      "loss": 1.8365,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5569604258069096e-06,
+      "loss": 1.8393,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546756091204713e-06,
+      "loss": 1.8434,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536554473991207e-06,
+      "loss": 1.7779,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526355585466432e-06,
+      "loss": 1.8497,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516159436927408e-06,
+      "loss": 1.8355,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505966039668123e-06,
+      "loss": 1.8834,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495775404979509e-06,
+      "loss": 1.846,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485587544149448e-06,
+      "loss": 1.8241,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4754024684627405e-06,
+      "loss": 1.8426,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465220189201106e-06,
+      "loss": 1.8373,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455040717643169e-06,
+      "loss": 1.8188,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444864065064446e-06,
+      "loss": 1.875,
+      "step": 1831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434690242737314e-06,
+      "loss": 1.8876,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424519261931036e-06,
+      "loss": 1.8438,
+      "step": 1833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414351133911716e-06,
+      "loss": 1.8334,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404185869942302e-06,
+      "loss": 1.8257,
+      "step": 1835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394023481282568e-06,
+      "loss": 1.8635,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383863979189104e-06,
+      "loss": 1.831,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373707374915303e-06,
+      "loss": 1.8654,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363553679711347e-06,
+      "loss": 1.864,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353402904824196e-06,
+      "loss": 1.7773,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343255061497574e-06,
+      "loss": 1.8897,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333110160971962e-06,
+      "loss": 1.8168,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322968214484583e-06,
+      "loss": 1.8981,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312829233269371e-06,
+      "loss": 1.8357,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3026932285569944e-06,
+      "loss": 1.8693,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2925602115748195e-06,
+      "loss": 1.8798,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2824301935469e-06,
+      "loss": 1.8228,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27230318569397e-06,
+      "loss": 1.8063,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262179199233429e-06,
+      "loss": 1.8436,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25205824537933e-06,
+      "loss": 1.8772,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241940335342367e-06,
+      "loss": 1.8465,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231825480329859e-06,
+      "loss": 1.8766,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221713691545746e-06,
+      "loss": 1.8253,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211604980190571e-06,
+      "loss": 1.8696,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201499357461464e-06,
+      "loss": 1.8352,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19139683455214e-06,
+      "loss": 1.8958,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181297422652874e-06,
+      "loss": 1.8305,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171201132950502e-06,
+      "loss": 1.8309,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161107976628391e-06,
+      "loss": 1.8599,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151017964866449e-06,
+      "loss": 1.8357,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140931108841094e-06,
+      "loss": 1.9014,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1308474197252505e-06,
+      "loss": 1.8397,
+      "step": 1862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1207669086883366e-06,
+      "loss": 1.8612,
+      "step": 1863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110689586896247e-06,
+      "loss": 1.8127,
+      "step": 1864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100615465511344e-06,
+      "loss": 1.8135,
+      "step": 1865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090544555692449e-06,
+      "loss": 1.838,
+      "step": 1866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080476868594822e-06,
+      "loss": 1.8886,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070412415370158e-06,
+      "loss": 1.8503,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060351207166558e-06,
+      "loss": 1.8013,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050293255128542e-06,
+      "loss": 1.8661,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040238570397015e-06,
+      "loss": 1.8422,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030187164109272e-06,
+      "loss": 1.8288,
+      "step": 1872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020139047398966e-06,
+      "loss": 1.7828,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0100942313961115e-06,
+      "loss": 1.8434,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000052727227068e-06,
+      "loss": 1.8705,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990014546014523e-06,
+      "loss": 1.8033,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979979698877486e-06,
+      "loss": 1.8696,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969948196931272e-06,
+      "loss": 1.8264,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959920051287492e-06,
+      "loss": 1.8959,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949895273054037e-06,
+      "loss": 1.8667,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939873873335077e-06,
+      "loss": 1.8698,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929855863231019e-06,
+      "loss": 1.8427,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919841253838537e-06,
+      "loss": 1.8637,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 1.886,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89982228155611e-06,
+      "loss": 1.8659,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889817940840614e-06,
+      "loss": 1.8738,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879817045185565e-06,
+      "loss": 1.8443,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869819605668669e-06,
+      "loss": 1.9025,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859825633363805e-06,
+      "loss": 1.8347,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849835139341014e-06,
+      "loss": 1.8535,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839848134666485e-06,
+      "loss": 1.8129,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829864630402538e-06,
+      "loss": 1.8994,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819884637607619e-06,
+      "loss": 1.8041,
+      "step": 1893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809908167336283e-06,
+      "loss": 1.784,
+      "step": 1894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799935230639179e-06,
+      "loss": 1.7883,
+      "step": 1895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7899658385630475e-06,
+      "loss": 1.8537,
+      "step": 1896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780000002150704e-06,
+      "loss": 1.8188,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770037732441019e-06,
+      "loss": 1.8857,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760079040468921e-06,
+      "loss": 1.8109,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750123937265367e-06,
+      "loss": 1.8177,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740172433857345e-06,
+      "loss": 1.8503,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73022454126785e-06,
+      "loss": 1.8095,
+      "step": 1902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720280270515882e-06,
+      "loss": 1.8734,
+      "step": 1903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71033963261643e-06,
+      "loss": 1.8327,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700402638580452e-06,
+      "loss": 1.8647,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690469299414881e-06,
+      "loss": 1.8481,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680539626122586e-06,
+      "loss": 1.8816,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670613629702391e-06,
+      "loss": 1.8275,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660691321149038e-06,
+      "loss": 1.8948,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650772711453189e-06,
+      "loss": 1.8094,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6408578116014025e-06,
+      "loss": 1.8841,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630946632576137e-06,
+      "loss": 1.8591,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62103918535572e-06,
+      "loss": 1.8591,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611135480914353e-06,
+      "loss": 1.9003,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601235530222086e-06,
+      "loss": 1.8569,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591339344244816e-06,
+      "loss": 1.9103,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5814469339442665e-06,
+      "loss": 1.8383,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5715583102779815e-06,
+      "loss": 1.8109,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5616734841993045e-06,
+      "loss": 1.8083,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551792466657378e-06,
+      "loss": 1.8419,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541915268597128e-06,
+      "loss": 1.8684,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532041900959245e-06,
+      "loss": 1.872,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522172374680177e-06,
+      "loss": 1.8801,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512306700692121e-06,
+      "loss": 1.8743,
+      "step": 1924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502444889923002e-06,
+      "loss": 1.8121,
+      "step": 1925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492586953296469e-06,
+      "loss": 1.8642,
+      "step": 1926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4827329017318776e-06,
+      "loss": 1.8269,
+      "step": 1927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472882746144282e-06,
+      "loss": 1.7762,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463036497444421e-06,
+      "loss": 1.8942,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453194166538703e-06,
+      "loss": 1.9285,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443355764329205e-06,
+      "loss": 1.85,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433521301713637e-06,
+      "loss": 1.8582,
+      "step": 1932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423690789585359e-06,
+      "loss": 1.8125,
+      "step": 1933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41386423883335e-06,
+      "loss": 1.8668,
+      "step": 1934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4040416603421996e-06,
+      "loss": 1.8102,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394223064992102e-06,
+      "loss": 1.8168,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384408463658836e-06,
+      "loss": 1.849,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374597867213756e-06,
+      "loss": 1.8895,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3647912865237835e-06,
+      "loss": 1.8972,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3549887324513884e-06,
+      "loss": 1.7977,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345190215854582e-06,
+      "loss": 1.8131,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3353957475869035e-06,
+      "loss": 1.8977,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3256053384974105e-06,
+      "loss": 1.8468,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315818999430654e-06,
+      "loss": 1.8163,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30603674122669e-06,
+      "loss": 1.8813,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296258574721046e-06,
+      "loss": 1.841,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2864845107447205e-06,
+      "loss": 1.84,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276714560124166e-06,
+      "loss": 1.9512,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266948733681282e-06,
+      "loss": 1.7848,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257187042233396e-06,
+      "loss": 1.8638,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247429496593256e-06,
+      "loss": 1.8383,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237676107569019e-06,
+      "loss": 1.8325,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2279268859642396e-06,
+      "loss": 1.8809,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218181842577853e-06,
+      "loss": 1.8663,
+      "step": 1954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208440988204164e-06,
+      "loss": 1.892,
+      "step": 1955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198704333632845e-06,
+      "loss": 1.7931,
+      "step": 1956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188971889648909e-06,
+      "loss": 1.8259,
+      "step": 1957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179243667032709e-06,
+      "loss": 1.837,
+      "step": 1958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169519676559922e-06,
+      "loss": 1.8028,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159799929001536e-06,
+      "loss": 1.8783,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150084435123842e-06,
+      "loss": 1.8986,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140373205688411e-06,
+      "loss": 1.8166,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130666251452102e-06,
+      "loss": 1.8614,
+      "step": 1963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12096358316703e-06,
+      "loss": 1.8783,
+      "step": 1964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111265211580566e-06,
+      "loss": 1.8702,
+      "step": 1965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101571147435321e-06,
+      "loss": 1.8625,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0918814014691375e-06,
+      "loss": 1.8709,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082195984415069e-06,
+      "loss": 1.8567,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072514907001381e-06,
+      "loss": 1.8849,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062838179951523e-06,
+      "loss": 1.846,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053165813984134e-06,
+      "loss": 1.9019,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043497819813019e-06,
+      "loss": 1.7993,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03383420814714e-06,
+      "loss": 1.8666,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024174989690608e-06,
+      "loss": 1.8269,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014520175142664e-06,
+      "loss": 1.8683,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0048697751976704e-06,
+      "loss": 1.8434,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995223800545104e-06,
+      "loss": 1.8544,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9855822618695385e-06,
+      "loss": 1.86,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975945169850627e-06,
+      "loss": 1.8332,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96631253516311e-06,
+      "loss": 1.8478,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956684368476784e-06,
+      "loss": 1.8439,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947060680456487e-06,
+      "loss": 1.8382,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937441481762112e-06,
+      "loss": 1.8336,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927826783048571e-06,
+      "loss": 1.8349,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918216594965793e-06,
+      "loss": 1.883,
+      "step": 1985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908610928158713e-06,
+      "loss": 1.8569,
+      "step": 1986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899009793267252e-06,
+      "loss": 1.805,
+      "step": 1987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889413200926317e-06,
+      "loss": 1.836,
+      "step": 1988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879821161765782e-06,
+      "loss": 1.8582,
+      "step": 1989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870233686410474e-06,
+      "loss": 1.8791,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860650785480171e-06,
+      "loss": 1.8557,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851072469589578e-06,
+      "loss": 1.8743,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841498749348322e-06,
+      "loss": 1.809,
+      "step": 1993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831929635360949e-06,
+      "loss": 1.8545,
+      "step": 1994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822365138226886e-06,
+      "loss": 1.8086,
+      "step": 1995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8128052685404575e-06,
+      "loss": 1.8567,
+      "step": 1996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803250036890864e-06,
+      "loss": 1.8017,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793699453862161e-06,
+      "loss": 1.8606,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784153530033257e-06,
+      "loss": 1.8614,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774612275977907e-06,
+      "loss": 1.81,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765075702264681e-06,
+      "loss": 1.8387,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755543819456974e-06,
+      "loss": 1.8439,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746016638112986e-06,
+      "loss": 1.8064,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736494168785698e-06,
+      "loss": 1.7739,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726976422022885e-06,
+      "loss": 1.869,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7174634083670864e-06,
+      "loss": 1.884,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707955138355591e-06,
+      "loss": 1.8463,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698451622520442e-06,
+      "loss": 1.8618,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688952871388417e-06,
+      "loss": 1.8503,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679458895481011e-06,
+      "loss": 1.852,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669969705314432e-06,
+      "loss": 1.8223,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660485311399587e-06,
+      "loss": 1.8366,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651005724242072e-06,
+      "loss": 1.8012,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6415309543421526e-06,
+      "loss": 1.8599,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632061012194767e-06,
+      "loss": 1.8388,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622595908289499e-06,
+      "loss": 1.8235,
+      "step": 2016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613135653110576e-06,
+      "loss": 1.8657,
+      "step": 2017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603680257136857e-06,
+      "loss": 1.852,
+      "step": 2018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594229730841815e-06,
+      "loss": 1.8101,
+      "step": 2019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584784084693525e-06,
+      "loss": 1.8706,
+      "step": 2020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575343329154665e-06,
+      "loss": 1.8091,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565907474682493e-06,
+      "loss": 1.8997,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556476531728836e-06,
+      "loss": 1.8799,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5470505107400815e-06,
+      "loss": 1.8105,
+      "step": 2024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537629422157167e-06,
+      "loss": 1.8061,
+      "step": 2025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528213276415564e-06,
+      "loss": 1.8335,
+      "step": 2026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518802083945271e-06,
+      "loss": 1.8417,
+      "step": 2027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509395855170798e-06,
+      "loss": 1.8436,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499994600511158e-06,
+      "loss": 1.8459,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4905983303798545e-06,
+      "loss": 1.9174,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481207055184874e-06,
+      "loss": 1.8257,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471820785328657e-06,
+      "loss": 1.9232,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4624395312081125e-06,
+      "loss": 1.8409,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453063303214589e-06,
+      "loss": 1.8171,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4436921117338684e-06,
+      "loss": 1.8454,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434325967146154e-06,
+      "loss": 1.8904,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424964879826058e-06,
+      "loss": 1.8448,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415608860142593e-06,
+      "loss": 1.8138,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406257918459155e-06,
+      "loss": 1.9096,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396912065133516e-06,
+      "loss": 1.8599,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387571310517815e-06,
+      "loss": 1.8434,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378235664958543e-06,
+      "loss": 1.8733,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368905138796523e-06,
+      "loss": 1.8122,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3595797423669185e-06,
+      "loss": 1.8659,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3502594859992055e-06,
+      "loss": 1.8116,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340944380017166e-06,
+      "loss": 1.8973,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331634434738884e-06,
+      "loss": 1.826,
+      "step": 2047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322329660476715e-06,
+      "loss": 1.857,
+      "step": 2048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313030067537304e-06,
+      "loss": 1.8586,
+      "step": 2049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303735666221532e-06,
+      "loss": 1.8581,
+      "step": 2050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294446466824553e-06,
+      "loss": 1.8621,
+      "step": 2051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285162479635748e-06,
+      "loss": 1.843,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275883714938726e-06,
+      "loss": 1.7977,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2666101830113095e-06,
+      "loss": 1.8499,
+      "step": 2054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25734189412553e-06,
+      "loss": 1.8215,
+      "step": 2055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248078858547609e-06,
+      "loss": 1.7957,
+      "step": 2056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238821086537944e-06,
+      "loss": 1.8083,
+      "step": 2057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2295685883511086e-06,
+      "loss": 1.7536,
+      "step": 2058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2203213742358305e-06,
+      "loss": 1.8587,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2110794544349906e-06,
+      "loss": 1.895,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201842839185598e-06,
+      "loss": 1.8288,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1926115387187915e-06,
+      "loss": 1.829,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183385563259819e-06,
+      "loss": 1.7792,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174164923028034e-06,
+      "loss": 1.8712,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164949628236875e-06,
+      "loss": 1.8037,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155739689093866e-06,
+      "loss": 1.803,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146535115800593e-06,
+      "loss": 1.8436,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137335918552702e-06,
+      "loss": 1.8981,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1281421075398866e-06,
+      "loss": 1.864,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118953692945862e-06,
+      "loss": 1.836,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1097706849483785e-06,
+      "loss": 1.7541,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100593093719193e-06,
+      "loss": 1.7219,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091420929424065e-06,
+      "loss": 1.757,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082254202222739e-06,
+      "loss": 1.8186,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07309292226894e-06,
+      "loss": 1.8003,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0639370997103565e-06,
+      "loss": 1.903,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054786744688633e-06,
+      "loss": 1.8761,
+      "step": 2077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045641867339361e-06,
+      "loss": 1.8398,
+      "step": 2078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03650247779206e-06,
+      "loss": 1.7801,
+      "step": 2079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027368586170171e-06,
+      "loss": 1.8322,
+      "step": 2080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018240202591054e-06,
+      "loss": 1.8431,
+      "step": 2081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0091173371659496e-06,
+      "loss": 1.8478,
+      "step": 2082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.8334,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990888201192228e-06,
+      "loss": 1.8715,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9817819508355085e-06,
+      "loss": 1.8641,
+      "step": 2085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972681259016576e-06,
+      "loss": 1.8876,
+      "step": 2086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963586135816013e-06,
+      "loss": 1.8763,
+      "step": 2087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954496591308227e-06,
+      "loss": 1.8488,
+      "step": 2088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945412635561448e-06,
+      "loss": 1.8474,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936334278637716e-06,
+      "loss": 1.8234,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9272615305928725e-06,
+      "loss": 1.7488,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91819440147654e-06,
+      "loss": 1.8304,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909132901332122e-06,
+      "loss": 1.8532,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900077040196788e-06,
+      "loss": 1.8375,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8910268281014514e-06,
+      "loss": 1.8675,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881982275070779e-06,
+      "loss": 1.8007,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872943391123165e-06,
+      "loss": 1.7792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863910186270726e-06,
+      "loss": 1.8452,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854882670519286e-06,
+      "loss": 1.8148,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8458608538683695e-06,
+      "loss": 1.8297,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836844746311184e-06,
+      "loss": 1.8232,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827834357834619e-06,
+      "loss": 1.7717,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818829698419225e-06,
+      "loss": 1.8415,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809830778039207e-06,
+      "loss": 1.8954,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800837606662416e-06,
+      "loss": 1.8673,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791850194250335e-06,
+      "loss": 1.8185,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782868550758059e-06,
+      "loss": 1.8467,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773892686134301e-06,
+      "loss": 1.8279,
+      "step": 2108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7649226103213765e-06,
+      "loss": 1.7981,
+      "step": 2109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75595833325518e-06,
+      "loss": 1.8037,
+      "step": 2110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746999864865188e-06,
+      "loss": 1.8587,
+      "step": 2111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7380472150744416e-06,
+      "loss": 1.8425,
+      "step": 2112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729100393799538e-06,
+      "loss": 1.8991,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7201594109506175e-06,
+      "loss": 1.8002,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711224276431352e-06,
+      "loss": 1.8894,
+      "step": 2115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702295000138941e-06,
+      "loss": 1.8417,
+      "step": 2116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693371591964082e-06,
+      "loss": 1.8747,
+      "step": 2117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684454061790987e-06,
+      "loss": 1.9056,
+      "step": 2118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67554241949735e-06,
+      "loss": 1.8132,
+      "step": 2119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666636674954344e-06,
+      "loss": 1.8267,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657736838026608e-06,
+      "loss": 1.8542,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648842918572247e-06,
+      "loss": 1.8317,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639954926442792e-06,
+      "loss": 1.8561,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631072871483226e-06,
+      "loss": 1.875,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622196763531948e-06,
+      "loss": 1.8922,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613326612420772e-06,
+      "loss": 1.8523,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60446242797491e-06,
+      "loss": 1.8762,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5956042200129725e-06,
+      "loss": 1.8891,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586751998346941e-06,
+      "loss": 1.8509,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577905772782172e-06,
+      "loss": 1.8747,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5690655531173775e-06,
+      "loss": 1.8236,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560231349144623e-06,
+      "loss": 1.8687,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551403170649299e-06,
+      "loss": 1.8551,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54258102741013e-06,
+      "loss": 1.8335,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533764929199156e-06,
+      "loss": 1.8352,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524954885781717e-06,
+      "loss": 1.8361,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516150906916449e-06,
+      "loss": 1.8354,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507353002355269e-06,
+      "loss": 1.8411,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498561181843368e-06,
+      "loss": 1.8163,
+      "step": 2139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489775455119194e-06,
+      "loss": 1.8254,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48099583191445e-06,
+      "loss": 1.8805,
+      "step": 2141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472222321954073e-06,
+      "loss": 1.8313,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4634549349562315e-06,
+      "loss": 1.8535,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454693680632317e-06,
+      "loss": 1.8195,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445938568686914e-06,
+      "loss": 1.8366,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437189608817814e-06,
+      "loss": 1.8061,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428446810715993e-06,
+      "loss": 1.8653,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4197101840656e-06,
+      "loss": 1.8476,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410979738543949e-06,
+      "loss": 1.7975,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402255483821507e-06,
+      "loss": 1.8522,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393537429561882e-06,
+      "loss": 1.8528,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384825585421818e-06,
+      "loss": 1.8511,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376119961051175e-06,
+      "loss": 1.8878,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367420566092928e-06,
+      "loss": 1.8784,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358727410183149e-06,
+      "loss": 1.8611,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350040502951001e-06,
+      "loss": 1.8327,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341359854018728e-06,
+      "loss": 1.8579,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33268547300163e-06,
+      "loss": 1.8536,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3240173695080786e-06,
+      "loss": 1.8488,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315355553139485e-06,
+      "loss": 1.8855,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3067000334902985e-06,
+      "loss": 1.8026,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298050820147992e-06,
+      "loss": 1.8256,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289407922693053e-06,
+      "loss": 1.8114,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280771350698974e-06,
+      "loss": 1.8508,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272141113732242e-06,
+      "loss": 1.8808,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2635172213523255e-06,
+      "loss": 1.8364,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254899683111665e-06,
+      "loss": 1.8496,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2462885085556635e-06,
+      "loss": 1.8479,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237683707222677e-06,
+      "loss": 1.8224,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229085288643995e-06,
+      "loss": 1.8772,
+      "step": 2170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220493262343843e-06,
+      "loss": 1.8187,
+      "step": 2171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211907637839367e-06,
+      "loss": 1.8578,
+      "step": 2172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203328424640619e-06,
+      "loss": 1.788,
+      "step": 2173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19475563225055e-06,
+      "loss": 1.8297,
+      "step": 2174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186189270164998e-06,
+      "loss": 1.8171,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1776293478726806e-06,
+      "loss": 1.8604,
+      "step": 2176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1690758748551805e-06,
+      "loss": 1.8286,
+      "step": 2177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1605288605869365e-06,
+      "loss": 1.819,
+      "step": 2178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151988314535235e-06,
+      "loss": 1.831,
+      "step": 2179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143454246160196e-06,
+      "loss": 1.8801,
+      "step": 2180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134926664914766e-06,
+      "loss": 1.8389,
+      "step": 2181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126405580244706e-06,
+      "loss": 1.8564,
+      "step": 2182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117891001588574e-06,
+      "loss": 1.9077,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1093829383777315e-06,
+      "loss": 1.823,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1008814000363185e-06,
+      "loss": 1.8053,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0923863959812486e-06,
+      "loss": 1.8268,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083897935622194e-06,
+      "loss": 1.8236,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075416028361584e-06,
+      "loss": 1.8137,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0669406835945895e-06,
+      "loss": 1.8034,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058471910709102e-06,
+      "loss": 1.8227,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0500097190857455e-06,
+      "loss": 1.7875,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0415541180978494e-06,
+      "loss": 1.7833,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033105117111441e-06,
+      "loss": 1.8447,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024662725485242e-06,
+      "loss": 1.8866,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01622695257065e-06,
+      "loss": 1.8226,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0077978077117315e-06,
+      "loss": 1.7976,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999375300245208e-06,
+      "loss": 1.8325,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9909594395004545e-06,
+      "loss": 1.7804,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982550234799479e-06,
+      "loss": 1.7787,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974147695456922e-06,
+      "loss": 1.8095,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965751830780038e-06,
+      "loss": 1.8377,
+      "step": 2201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9573626500686845e-06,
+      "loss": 1.8412,
+      "step": 2202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948980162615323e-06,
+      "loss": 1.7662,
+      "step": 2203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940604377704995e-06,
+      "loss": 1.8236,
+      "step": 2204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932235304615321e-06,
+      "loss": 1.8176,
+      "step": 2205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923872952616486e-06,
+      "loss": 1.7934,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91551733097123e-06,
+      "loss": 1.8185,
+      "step": 2207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907168448934836e-06,
+      "loss": 1.8787,
+      "step": 2208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898826315755125e-06,
+      "loss": 1.8863,
+      "step": 2209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8904909406724425e-06,
+      "loss": 1.8216,
+      "step": 2210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882162332919645e-06,
+      "loss": 1.8026,
+      "step": 2211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873840501722096e-06,
+      "loss": 1.8619,
+      "step": 2212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865525456297652e-06,
+      "loss": 1.8022,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857217205856654e-06,
+      "loss": 1.7777,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848915759601913e-06,
+      "loss": 1.8189,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840621126728707e-06,
+      "loss": 1.8794,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8323333164247666e-06,
+      "loss": 1.8711,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824052337870263e-06,
+      "loss": 1.8559,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815778200237804e-06,
+      "loss": 1.824,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8075109126924115e-06,
+      "loss": 1.8372,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799250484391528e-06,
+      "loss": 1.8333,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7909969244849963e-06,
+      "loss": 1.8769,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7827502421150497e-06,
+      "loss": 1.8351,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7745104464163053e-06,
+      "loss": 1.8394,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7662775465157507e-06,
+      "loss": 1.829,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.758051551532735e-06,
+      "loss": 1.8091,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7498324705789613e-06,
+      "loss": 1.8209,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741620312758469e-06,
+      "loss": 1.8261,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7334150871676366e-06,
+      "loss": 1.8001,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7252168028951572e-06,
+      "loss": 1.8211,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7170254690220386e-06,
+      "loss": 1.8378,
+      "step": 2231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7088410946215915e-06,
+      "loss": 1.8394,
+      "step": 2232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7006636887594095e-06,
+      "loss": 1.8246,
+      "step": 2233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692493260493375e-06,
+      "loss": 1.8344,
+      "step": 2234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6843298188736408e-06,
+      "loss": 1.8101,
+      "step": 2235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6761733729426206e-06,
+      "loss": 1.8425,
+      "step": 2236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6680239317349764e-06,
+      "loss": 1.7822,
+      "step": 2237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6598815042776135e-06,
+      "loss": 1.9142,
+      "step": 2238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651746099589668e-06,
+      "loss": 1.7907,
+      "step": 2239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643617726682498e-06,
+      "loss": 1.8819,
+      "step": 2240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63549639455967e-06,
+      "loss": 1.8487,
+      "step": 2241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627382112216954e-06,
+      "loss": 1.8384,
+      "step": 2242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619274888642309e-06,
+      "loss": 1.8604,
+      "step": 2243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611174732815883e-06,
+      "loss": 1.8441,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603081653709979e-06,
+      "loss": 1.8097,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594995660289077e-06,
+      "loss": 1.7858,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5869167615098e-06,
+      "loss": 1.834,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578844966320917e-06,
+      "loss": 1.7749,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5707802836633254e-06,
+      "loss": 1.8215,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.562722722470047e-06,
+      "loss": 1.8652,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5546722916662113e-06,
+      "loss": 1.7953,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5466290001690553e-06,
+      "loss": 1.8316,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5385928568879012e-06,
+      "loss": 1.8357,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5305638707241595e-06,
+      "loss": 1.8367,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5225420505713136e-06,
+      "loss": 1.7973,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514527405314899e-06,
+      "loss": 1.8001,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5065199438325146e-06,
+      "loss": 1.8394,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4985196749937976e-06,
+      "loss": 1.8124,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4905266076604192e-06,
+      "loss": 1.8733,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482540750686074e-06,
+      "loss": 1.8238,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4745621129164676e-06,
+      "loss": 1.8425,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4665907031893163e-06,
+      "loss": 1.8464,
+      "step": 2262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458626530334316e-06,
+      "loss": 1.8124,
+      "step": 2263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450669603173159e-06,
+      "loss": 1.791,
+      "step": 2264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442719930519508e-06,
+      "loss": 1.8721,
+      "step": 2265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43477752117899e-06,
+      "loss": 1.7933,
+      "step": 2266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4268423839491873e-06,
+      "loss": 1.8125,
+      "step": 2267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4189145276196244e-06,
+      "loss": 1.8923,
+      "step": 2268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4109939609717634e-06,
+      "loss": 1.8623,
+      "step": 2269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4030806927789962e-06,
+      "loss": 1.8227,
+      "step": 2270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3951747318066177e-06,
+      "loss": 1.8022,
+      "step": 2271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387276086811839e-06,
+      "loss": 1.8402,
+      "step": 2272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3793847665437674e-06,
+      "loss": 1.8294,
+      "step": 2273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3715007797433928e-06,
+      "loss": 1.8971,
+      "step": 2274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363624135143584e-06,
+      "loss": 1.7921,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3557548414690768e-06,
+      "loss": 1.8146,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347892907436465e-06,
+      "loss": 1.7742,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340038341754189e-06,
+      "loss": 1.8565,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3321911531225306e-06,
+      "loss": 1.8395,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3243513502335954e-06,
+      "loss": 1.8322,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3165189417713128e-06,
+      "loss": 1.8674,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 1.7524,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300876342821451e-06,
+      "loss": 1.8215,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2930661696607326e-06,
+      "loss": 1.8293,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285263425580373e-06,
+      "loss": 1.839,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2774681192232504e-06,
+      "loss": 1.8124,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269680259224004e-06,
+      "loss": 1.768,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2618998542090263e-06,
+      "loss": 1.8261,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2541269127964516e-06,
+      "loss": 1.7979,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2463614435961456e-06,
+      "loss": 1.7929,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238603455209698e-06,
+      "loss": 1.8595,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2308529562304136e-06,
+      "loss": 1.8254,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2231099552433e-06,
+      "loss": 1.8047,
+      "step": 2293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2153744608250582e-06,
+      "loss": 1.816,
+      "step": 2294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2076464815440822e-06,
+      "loss": 1.8188,
+      "step": 2295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199926025960426e-06,
+      "loss": 1.8359,
+      "step": 2296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192213102625825e-06,
+      "loss": 1.7765,
+      "step": 2297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1845077200836638e-06,
+      "loss": 1.8353,
+      "step": 2298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1768098868689756e-06,
+      "loss": 1.8198,
+      "step": 2299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1691196115084344e-06,
+      "loss": 1.8802,
+      "step": 2300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1614369025203384e-06,
+      "loss": 1.8727,
+      "step": 2301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153761768414606e-06,
+      "loss": 1.7708,
+      "step": 2302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1460942176927666e-06,
+      "loss": 1.7682,
+      "step": 2303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1384342588479476e-06,
+      "loss": 1.8279,
+      "step": 2304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1307819003648698e-06,
+      "loss": 1.8366,
+      "step": 2305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123137150719833e-06,
+      "loss": 1.8857,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1155000183807103e-06,
+      "loss": 1.8051,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107870511806934e-06,
+      "loss": 1.8361,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1002486394494924e-06,
+      "loss": 1.7966,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0926344097509185e-06,
+      "loss": 1.8238,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0850278311452797e-06,
+      "loss": 1.8793,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0774289120581647e-06,
+      "loss": 1.8491,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0698376609066828e-06,
+      "loss": 1.904,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0622540860994463e-06,
+      "loss": 1.8341,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0546781960365667e-06,
+      "loss": 1.8415,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047109999109642e-06,
+      "loss": 1.8492,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03954950370175e-06,
+      "loss": 1.8913,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0319967181874366e-06,
+      "loss": 1.8189,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024451650932707e-06,
+      "loss": 1.8135,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0169143102950236e-06,
+      "loss": 1.8755,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0093847046232773e-06,
+      "loss": 1.9015,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001862842257801e-06,
+      "loss": 1.818,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9943487315303486e-06,
+      "loss": 1.8094,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.986842380764089e-06,
+      "loss": 1.8621,
+      "step": 2324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9793437982735928e-06,
+      "loss": 1.8412,
+      "step": 2325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9718529923648264e-06,
+      "loss": 1.9163,
+      "step": 2326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9643699713351446e-06,
+      "loss": 1.8667,
+      "step": 2327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9568947434732777e-06,
+      "loss": 1.8638,
+      "step": 2328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9494273170593245e-06,
+      "loss": 1.864,
+      "step": 2329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941967700364737e-06,
+      "loss": 1.7987,
+      "step": 2330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934515901652324e-06,
+      "loss": 1.7864,
+      "step": 2331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9270719291762307e-06,
+      "loss": 1.8073,
+      "step": 2332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.919635791181934e-06,
+      "loss": 1.8025,
+      "step": 2333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912207495906235e-06,
+      "loss": 1.8283,
+      "step": 2334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9047870515772435e-06,
+      "loss": 1.7927,
+      "step": 2335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89737446641438e-06,
+      "loss": 1.868,
+      "step": 2336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8899697486283475e-06,
+      "loss": 1.8344,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882572906421145e-06,
+      "loss": 1.785,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8751839479860454e-06,
+      "loss": 1.798,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867802881507589e-06,
+      "loss": 1.7761,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8604297151615725e-06,
+      "loss": 1.8067,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8530644571150423e-06,
+      "loss": 1.7472,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8457071155262885e-06,
+      "loss": 1.7954,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8383576985448267e-06,
+      "loss": 1.8045,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831016214311403e-06,
+      "loss": 1.8539,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8236826709579644e-06,
+      "loss": 1.8843,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8163570766076707e-06,
+      "loss": 1.8098,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809039439374878e-06,
+      "loss": 1.8182,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8017297673651223e-06,
+      "loss": 1.864,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7944280686751213e-06,
+      "loss": 1.7693,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7871343513927596e-06,
+      "loss": 1.8422,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7798486235970802e-06,
+      "loss": 1.8581,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7725708933582785e-06,
+      "loss": 1.8119,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7653011687376886e-06,
+      "loss": 1.8221,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7580394577877787e-06,
+      "loss": 1.8086,
+      "step": 2355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75078576855214e-06,
+      "loss": 1.8313,
+      "step": 2356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743540109065482e-06,
+      "loss": 1.8248,
+      "step": 2357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7363024873536093e-06,
+      "loss": 1.784,
+      "step": 2358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7290729114334326e-06,
+      "loss": 1.8776,
+      "step": 2359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7218513893129493e-06,
+      "loss": 1.8475,
+      "step": 2360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7146379289912337e-06,
+      "loss": 1.8655,
+      "step": 2361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707432538458433e-06,
+      "loss": 1.8305,
+      "step": 2362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.700235225695752e-06,
+      "loss": 1.8748,
+      "step": 2363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69304599867545e-06,
+      "loss": 1.7986,
+      "step": 2364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.685864865360831e-06,
+      "loss": 1.8278,
+      "step": 2365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6786918337062304e-06,
+      "loss": 1.8112,
+      "step": 2366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671526911657015e-06,
+      "loss": 1.8427,
+      "step": 2367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6643701071495644e-06,
+      "loss": 1.7568,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6572214281112675e-06,
+      "loss": 1.8359,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6500808824605163e-06,
+      "loss": 1.7825,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642948478106685e-06,
+      "loss": 1.8198,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6358242229501387e-06,
+      "loss": 1.8355,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628708124882212e-06,
+      "loss": 1.892,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621600191785205e-06,
+      "loss": 1.8045,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614500431532375e-06,
+      "loss": 1.8071,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6074088519879236e-06,
+      "loss": 1.8416,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6003254610069938e-06,
+      "loss": 1.8253,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5932502664356553e-06,
+      "loss": 1.7942,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5861832761108997e-06,
+      "loss": 1.8693,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5791244978606344e-06,
+      "loss": 1.8513,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5720739395036665e-06,
+      "loss": 1.8426,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565031608849702e-06,
+      "loss": 1.8,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5579975136993253e-06,
+      "loss": 1.8567,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5509716618440063e-06,
+      "loss": 1.8396,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5439540610660828e-06,
+      "loss": 1.8115,
+      "step": 2385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536944719138752e-06,
+      "loss": 1.8551,
+      "step": 2386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.529943643826064e-06,
+      "loss": 1.819,
+      "step": 2387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52295084288291e-06,
+      "loss": 1.811,
+      "step": 2388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515966324055016e-06,
+      "loss": 1.9167,
+      "step": 2389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508990095078938e-06,
+      "loss": 1.8136,
+      "step": 2390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502022163682046e-06,
+      "loss": 1.8337,
+      "step": 2391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49506253758252e-06,
+      "loss": 1.8422,
+      "step": 2392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4881112244893403e-06,
+      "loss": 1.8394,
+      "step": 2393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481168232102279e-06,
+      "loss": 1.8288,
+      "step": 2394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4742335681118945e-06,
+      "loss": 1.8454,
+      "step": 2395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4673072401995123e-06,
+      "loss": 1.862,
+      "step": 2396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4603892560372324e-06,
+      "loss": 1.8162,
+      "step": 2397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453479623287909e-06,
+      "loss": 1.8345,
+      "step": 2398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4465783496051463e-06,
+      "loss": 1.7662,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4396854426332906e-06,
+      "loss": 1.7555,
+      "step": 2400
+    }
+  ],
+  "max_steps": 3078,
+  "num_train_epochs": 1,
+  "total_flos": 7.410096532986266e+17,
+  "trial_name": null,
+  "trial_params": null
+}