diff --git "a/checkpoint-4170/trainer_state.json" "b/checkpoint-4170/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4170/trainer_state.json"
@@ -0,0 +1,25096 @@
+{
+  "best_metric": 0.21844688057899475,
+  "best_model_checkpoint": "multiple_accent_classification/checkpoint-4170",
+  "epoch": 6.0,
+  "global_step": 4170,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.7809,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.7236,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.8205,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.7999,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.8185,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.7794,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.782,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.7723,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.8171,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.8095,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 1.7817,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.8079,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 1.8265,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.6441,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 1.8041,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.8014,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.7949,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.8021,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.9434,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.7863,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.7931,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 1.7781,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 1.7805,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.7841,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 1.7882,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2e-06,
+      "loss": 1.7881,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 1.7823,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.7948,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8e-06,
+      "loss": 1.7905,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 1.8093,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 1.7993,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.8261,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 1.8076,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.7184,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7e-06,
+      "loss": 1.7999,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.8002,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.4e-06,
+      "loss": 1.7983,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.7646,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 1.7974,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.7775,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.2e-06,
+      "loss": 1.7991,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.7888,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.6e-06,
+      "loss": 1.7765,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.8e-06,
+      "loss": 1.7836,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9e-06,
+      "loss": 1.7949,
+      "step": 45
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.7724,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.4e-06,
+      "loss": 1.8043,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.7796,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.7891,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 1.7836,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998550724637682e-06,
+      "loss": 1.8004,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997101449275363e-06,
+      "loss": 1.7932,
+      "step": 52
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.995652173913044e-06,
+      "loss": 1.8642,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.994202898550725e-06,
+      "loss": 1.7733,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992753623188408e-06,
+      "loss": 1.8387,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991304347826087e-06,
+      "loss": 1.8016,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98985507246377e-06,
+      "loss": 1.8028,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98840579710145e-06,
+      "loss": 1.7867,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986956521739132e-06,
+      "loss": 1.8025,
+      "step": 59
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.985507246376813e-06,
+      "loss": 1.8052,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.984057971014494e-06,
+      "loss": 1.7862,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982608695652175e-06,
+      "loss": 1.7925,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981159420289856e-06,
+      "loss": 1.8018,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979710144927537e-06,
+      "loss": 1.7766,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978260869565218e-06,
+      "loss": 1.7993,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9768115942029e-06,
+      "loss": 1.8006,
+      "step": 66
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.97536231884058e-06,
+      "loss": 1.789,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.973913043478262e-06,
+      "loss": 1.7795,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.972463768115943e-06,
+      "loss": 1.7781,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.971014492753624e-06,
+      "loss": 1.773,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.969565217391305e-06,
+      "loss": 1.7694,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.968115942028986e-06,
+      "loss": 1.7751,
+      "step": 72
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.966666666666667e-06,
+      "loss": 1.7843,
+      "step": 73
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.965217391304348e-06,
+      "loss": 1.7965,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.96376811594203e-06,
+      "loss": 1.9398,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.962318840579712e-06,
+      "loss": 1.7804,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.960869565217393e-06,
+      "loss": 1.7993,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.959420289855072e-06,
+      "loss": 1.8129,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.957971014492755e-06,
+      "loss": 1.7988,
+      "step": 79
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.956521739130436e-06,
+      "loss": 1.8938,
+      "step": 80
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.955072463768117e-06,
+      "loss": 1.7894,
+      "step": 81
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.953623188405798e-06,
+      "loss": 1.7984,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.952173913043479e-06,
+      "loss": 1.8354,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.95072463768116e-06,
+      "loss": 1.7562,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.949275362318841e-06,
+      "loss": 1.7773,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.947826086956522e-06,
+      "loss": 1.7774,
+      "step": 86
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.946376811594205e-06,
+      "loss": 1.7808,
+      "step": 87
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.944927536231886e-06,
+      "loss": 1.7769,
+      "step": 88
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.943478260869565e-06,
+      "loss": 1.8332,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.942028985507248e-06,
+      "loss": 1.7777,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.940579710144929e-06,
+      "loss": 1.8006,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.93913043478261e-06,
+      "loss": 1.8033,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.93768115942029e-06,
+      "loss": 1.7584,
+      "step": 93
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.936231884057972e-06,
+      "loss": 1.8261,
+      "step": 94
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.934782608695653e-06,
+      "loss": 1.8463,
+      "step": 95
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.933333333333334e-06,
+      "loss": 1.7987,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.931884057971015e-06,
+      "loss": 1.7971,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.930434782608697e-06,
+      "loss": 1.8483,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.928985507246378e-06,
+      "loss": 1.8237,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.927536231884058e-06,
+      "loss": 1.8163,
+      "step": 100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92608695652174e-06,
+      "loss": 1.7806,
+      "step": 101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924637681159421e-06,
+      "loss": 1.7938,
+      "step": 102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.923188405797102e-06,
+      "loss": 1.7786,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.921739130434783e-06,
+      "loss": 1.7904,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.920289855072464e-06,
+      "loss": 1.7967,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.918840579710145e-06,
+      "loss": 1.807,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.917391304347826e-06,
+      "loss": 1.8079,
+      "step": 107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.915942028985507e-06,
+      "loss": 1.8145,
+      "step": 108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91449275362319e-06,
+      "loss": 1.7979,
+      "step": 109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.913043478260871e-06,
+      "loss": 1.7861,
+      "step": 110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.911594202898552e-06,
+      "loss": 1.7905,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910144927536233e-06,
+      "loss": 1.8235,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.908695652173914e-06,
+      "loss": 1.8008,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.907246376811595e-06,
+      "loss": 1.7062,
+      "step": 114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.905797101449276e-06,
+      "loss": 1.7784,
+      "step": 115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.904347826086957e-06,
+      "loss": 1.7804,
+      "step": 116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.902898550724638e-06,
+      "loss": 1.7892,
+      "step": 117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.901449275362319e-06,
+      "loss": 1.7925,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.9e-06,
+      "loss": 1.8087,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.898550724637683e-06,
+      "loss": 1.7866,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.897101449275364e-06,
+      "loss": 1.77,
+      "step": 121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.895652173913045e-06,
+      "loss": 1.8127,
+      "step": 122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.894202898550726e-06,
+      "loss": 1.7999,
+      "step": 123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.892753623188407e-06,
+      "loss": 1.7879,
+      "step": 124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.891304347826088e-06,
+      "loss": 1.8021,
+      "step": 125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.889855072463769e-06,
+      "loss": 1.7853,
+      "step": 126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.88840579710145e-06,
+      "loss": 1.7957,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.886956521739132e-06,
+      "loss": 1.7958,
+      "step": 128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.885507246376812e-06,
+      "loss": 1.7858,
+      "step": 129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.884057971014493e-06,
+      "loss": 1.7753,
+      "step": 130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.882608695652175e-06,
+      "loss": 1.7965,
+      "step": 131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.881159420289856e-06,
+      "loss": 1.784,
+      "step": 132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.879710144927537e-06,
+      "loss": 1.7682,
+      "step": 133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.878260869565218e-06,
+      "loss": 1.7961,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.8768115942029e-06,
+      "loss": 1.7959,
+      "step": 135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.87536231884058e-06,
+      "loss": 1.7965,
+      "step": 136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.873913043478261e-06,
+      "loss": 1.7975,
+      "step": 137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.872463768115942e-06,
+      "loss": 1.7583,
+      "step": 138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.871014492753625e-06,
+      "loss": 1.8122,
+      "step": 139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.869565217391304e-06,
+      "loss": 1.773,
+      "step": 140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.868115942028986e-06,
+      "loss": 1.7903,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.7876,
+      "step": 142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.865217391304348e-06,
+      "loss": 1.7727,
+      "step": 143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.86376811594203e-06,
+      "loss": 1.7971,
+      "step": 144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.862318840579711e-06,
+      "loss": 1.7969,
+      "step": 145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.860869565217392e-06,
+      "loss": 1.7716,
+      "step": 146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.859420289855073e-06,
+      "loss": 1.7936,
+      "step": 147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.857971014492754e-06,
+      "loss": 1.8433,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.856521739130435e-06,
+      "loss": 1.7925,
+      "step": 149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.855072463768118e-06,
+      "loss": 1.7885,
+      "step": 150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.853623188405797e-06,
+      "loss": 1.7897,
+      "step": 151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.852173913043478e-06,
+      "loss": 1.7979,
+      "step": 152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.850724637681161e-06,
+      "loss": 1.7937,
+      "step": 153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.84927536231884e-06,
+      "loss": 1.8094,
+      "step": 154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.847826086956523e-06,
+      "loss": 1.7972,
+      "step": 155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.846376811594204e-06,
+      "loss": 1.7988,
+      "step": 156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.844927536231885e-06,
+      "loss": 1.8043,
+      "step": 157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.843478260869566e-06,
+      "loss": 1.7802,
+      "step": 158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.842028985507247e-06,
+      "loss": 1.8084,
+      "step": 159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.840579710144928e-06,
+      "loss": 1.7994,
+      "step": 160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.83913043478261e-06,
+      "loss": 1.7958,
+      "step": 161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.83768115942029e-06,
+      "loss": 1.799,
+      "step": 162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.836231884057973e-06,
+      "loss": 1.8047,
+      "step": 163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.834782608695654e-06,
+      "loss": 1.792,
+      "step": 164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 1.7957,
+      "step": 165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.831884057971016e-06,
+      "loss": 1.8003,
+      "step": 166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.830434782608697e-06,
+      "loss": 1.7996,
+      "step": 167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.828985507246378e-06,
+      "loss": 1.7828,
+      "step": 168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.827536231884059e-06,
+      "loss": 1.806,
+      "step": 169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.82608695652174e-06,
+      "loss": 1.7795,
+      "step": 170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.82463768115942e-06,
+      "loss": 1.7812,
+      "step": 171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.823188405797103e-06,
+      "loss": 1.7793,
+      "step": 172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.821739130434783e-06,
+      "loss": 1.7826,
+      "step": 173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.820289855072465e-06,
+      "loss": 1.794,
+      "step": 174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.818840579710146e-06,
+      "loss": 1.8022,
+      "step": 175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.817391304347826e-06,
+      "loss": 1.8022,
+      "step": 176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.815942028985508e-06,
+      "loss": 1.7854,
+      "step": 177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.81449275362319e-06,
+      "loss": 1.8071,
+      "step": 178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.81304347826087e-06,
+      "loss": 1.7822,
+      "step": 179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.811594202898551e-06,
+      "loss": 1.7871,
+      "step": 180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.810144927536232e-06,
+      "loss": 1.818,
+      "step": 181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.808695652173913e-06,
+      "loss": 1.7898,
+      "step": 182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.807246376811596e-06,
+      "loss": 1.7978,
+      "step": 183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.805797101449275e-06,
+      "loss": 1.7966,
+      "step": 184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.804347826086958e-06,
+      "loss": 1.7753,
+      "step": 185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.802898550724639e-06,
+      "loss": 1.7773,
+      "step": 186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.801449275362318e-06,
+      "loss": 1.7989,
+      "step": 187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.778,
+      "step": 188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.798550724637682e-06,
+      "loss": 1.7772,
+      "step": 189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.797101449275363e-06,
+      "loss": 1.7873,
+      "step": 190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.795652173913044e-06,
+      "loss": 1.784,
+      "step": 191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.794202898550725e-06,
+      "loss": 1.8003,
+      "step": 192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.792753623188406e-06,
+      "loss": 1.8156,
+      "step": 193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.791304347826089e-06,
+      "loss": 1.7934,
+      "step": 194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.789855072463768e-06,
+      "loss": 1.8066,
+      "step": 195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.78840579710145e-06,
+      "loss": 1.7851,
+      "step": 196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.786956521739132e-06,
+      "loss": 1.7973,
+      "step": 197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.785507246376813e-06,
+      "loss": 1.7799,
+      "step": 198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.784057971014494e-06,
+      "loss": 1.7895,
+      "step": 199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.782608695652175e-06,
+      "loss": 1.7787,
+      "step": 200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.781159420289856e-06,
+      "loss": 1.8015,
+      "step": 201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.779710144927537e-06,
+      "loss": 1.8044,
+      "step": 202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.778260869565218e-06,
+      "loss": 1.8018,
+      "step": 203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.776811594202899e-06,
+      "loss": 1.7978,
+      "step": 204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.77536231884058e-06,
+      "loss": 1.7901,
+      "step": 205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.77391304347826e-06,
+      "loss": 1.796,
+      "step": 206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.772463768115943e-06,
+      "loss": 1.7673,
+      "step": 207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.771014492753624e-06,
+      "loss": 1.7454,
+      "step": 208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.769565217391305e-06,
+      "loss": 1.7707,
+      "step": 209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.768115942028986e-06,
+      "loss": 1.8148,
+      "step": 210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.766666666666667e-06,
+      "loss": 1.7759,
+      "step": 211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.765217391304348e-06,
+      "loss": 1.7986,
+      "step": 212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.76376811594203e-06,
+      "loss": 1.7843,
+      "step": 213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.76231884057971e-06,
+      "loss": 1.7875,
+      "step": 214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.760869565217393e-06,
+      "loss": 1.7955,
+      "step": 215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.759420289855072e-06,
+      "loss": 1.7922,
+      "step": 216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.757971014492753e-06,
+      "loss": 1.7892,
+      "step": 217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.756521739130436e-06,
+      "loss": 1.7968,
+      "step": 218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.755072463768117e-06,
+      "loss": 1.7844,
+      "step": 219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.753623188405798e-06,
+      "loss": 1.7768,
+      "step": 220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.75217391304348e-06,
+      "loss": 1.7621,
+      "step": 221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.75072463768116e-06,
+      "loss": 1.7873,
+      "step": 222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.749275362318841e-06,
+      "loss": 1.801,
+      "step": 223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.747826086956522e-06,
+      "loss": 1.7861,
+      "step": 224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.746376811594203e-06,
+      "loss": 1.7778,
+      "step": 225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.744927536231886e-06,
+      "loss": 1.773,
+      "step": 226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.743478260869565e-06,
+      "loss": 1.8384,
+      "step": 227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.742028985507246e-06,
+      "loss": 1.7723,
+      "step": 228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.740579710144929e-06,
+      "loss": 1.7699,
+      "step": 229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.73913043478261e-06,
+      "loss": 1.7861,
+      "step": 230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.737681159420291e-06,
+      "loss": 1.8091,
+      "step": 231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.736231884057972e-06,
+      "loss": 1.8037,
+      "step": 232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.734782608695653e-06,
+      "loss": 1.7961,
+      "step": 233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 1.7821,
+      "step": 234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.731884057971015e-06,
+      "loss": 1.7605,
+      "step": 235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.730434782608696e-06,
+      "loss": 1.7963,
+      "step": 236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.728985507246379e-06,
+      "loss": 1.8122,
+      "step": 237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.727536231884058e-06,
+      "loss": 1.7776,
+      "step": 238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.726086956521739e-06,
+      "loss": 1.7968,
+      "step": 239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.724637681159422e-06,
+      "loss": 1.8013,
+      "step": 240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.723188405797103e-06,
+      "loss": 1.7907,
+      "step": 241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.721739130434784e-06,
+      "loss": 1.7574,
+      "step": 242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.720289855072465e-06,
+      "loss": 1.7799,
+      "step": 243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.718840579710146e-06,
+      "loss": 1.7939,
+      "step": 244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.717391304347827e-06,
+      "loss": 1.7531,
+      "step": 245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.715942028985508e-06,
+      "loss": 1.7878,
+      "step": 246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.714492753623189e-06,
+      "loss": 1.7804,
+      "step": 247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.713043478260871e-06,
+      "loss": 1.79,
+      "step": 248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.71159420289855e-06,
+      "loss": 1.8005,
+      "step": 249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.710144927536233e-06,
+      "loss": 1.7805,
+      "step": 250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.708695652173914e-06,
+      "loss": 1.7825,
+      "step": 251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.707246376811595e-06,
+      "loss": 1.822,
+      "step": 252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.705797101449276e-06,
+      "loss": 1.7619,
+      "step": 253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.704347826086957e-06,
+      "loss": 1.7642,
+      "step": 254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.702898550724638e-06,
+      "loss": 1.7656,
+      "step": 255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.701449275362321e-06,
+      "loss": 1.7751,
+      "step": 256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.7e-06,
+      "loss": 1.7816,
+      "step": 257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.698550724637681e-06,
+      "loss": 1.752,
+      "step": 258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.697101449275364e-06,
+      "loss": 1.7537,
+      "step": 259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.695652173913043e-06,
+      "loss": 1.7902,
+      "step": 260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.694202898550726e-06,
+      "loss": 1.8062,
+      "step": 261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.692753623188407e-06,
+      "loss": 1.6422,
+      "step": 262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.691304347826088e-06,
+      "loss": 1.7912,
+      "step": 263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.689855072463769e-06,
+      "loss": 1.7721,
+      "step": 264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.68840579710145e-06,
+      "loss": 1.7612,
+      "step": 265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.686956521739131e-06,
+      "loss": 1.7566,
+      "step": 266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.685507246376812e-06,
+      "loss": 1.8152,
+      "step": 267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.684057971014493e-06,
+      "loss": 1.7972,
+      "step": 268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.682608695652174e-06,
+      "loss": 1.8002,
+      "step": 269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.681159420289857e-06,
+      "loss": 1.7958,
+      "step": 270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.679710144927536e-06,
+      "loss": 1.7978,
+      "step": 271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.678260869565219e-06,
+      "loss": 1.7986,
+      "step": 272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.6768115942029e-06,
+      "loss": 1.813,
+      "step": 273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.67536231884058e-06,
+      "loss": 1.7639,
+      "step": 274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.673913043478262e-06,
+      "loss": 1.7617,
+      "step": 275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.672463768115943e-06,
+      "loss": 1.794,
+      "step": 276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.671014492753624e-06,
+      "loss": 1.7577,
+      "step": 277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.669565217391305e-06,
+      "loss": 1.7905,
+      "step": 278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.668115942028986e-06,
+      "loss": 1.7782,
+      "step": 279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 1.7855,
+      "step": 280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.66521739130435e-06,
+      "loss": 1.7611,
+      "step": 281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.663768115942029e-06,
+      "loss": 1.7801,
+      "step": 282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.662318840579711e-06,
+      "loss": 1.8409,
+      "step": 283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.660869565217392e-06,
+      "loss": 1.7867,
+      "step": 284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.659420289855073e-06,
+      "loss": 1.7686,
+      "step": 285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.657971014492754e-06,
+      "loss": 1.7582,
+      "step": 286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.656521739130435e-06,
+      "loss": 1.7462,
+      "step": 287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.655072463768116e-06,
+      "loss": 1.7532,
+      "step": 288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.653623188405797e-06,
+      "loss": 1.7801,
+      "step": 289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.652173913043478e-06,
+      "loss": 1.7927,
+      "step": 290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.650724637681161e-06,
+      "loss": 1.7035,
+      "step": 291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.649275362318842e-06,
+      "loss": 1.8057,
+      "step": 292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.647826086956521e-06,
+      "loss": 1.8014,
+      "step": 293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.646376811594204e-06,
+      "loss": 1.7877,
+      "step": 294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.644927536231885e-06,
+      "loss": 1.7719,
+      "step": 295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.643478260869566e-06,
+      "loss": 1.8537,
+      "step": 296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.642028985507247e-06,
+      "loss": 1.8251,
+      "step": 297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.640579710144928e-06,
+      "loss": 1.7511,
+      "step": 298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.639130434782609e-06,
+      "loss": 1.7622,
+      "step": 299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.63768115942029e-06,
+      "loss": 1.8174,
+      "step": 300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.636231884057971e-06,
+      "loss": 1.8452,
+      "step": 301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.634782608695654e-06,
+      "loss": 1.7472,
+      "step": 302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.633333333333335e-06,
+      "loss": 1.8054,
+      "step": 303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.631884057971014e-06,
+      "loss": 1.7485,
+      "step": 304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.630434782608697e-06,
+      "loss": 1.804,
+      "step": 305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.628985507246378e-06,
+      "loss": 1.7905,
+      "step": 306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.627536231884059e-06,
+      "loss": 1.7839,
+      "step": 307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.62608695652174e-06,
+      "loss": 1.8166,
+      "step": 308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.624637681159421e-06,
+      "loss": 1.7956,
+      "step": 309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.623188405797102e-06,
+      "loss": 1.7903,
+      "step": 310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.621739130434783e-06,
+      "loss": 1.7935,
+      "step": 311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.620289855072464e-06,
+      "loss": 1.7804,
+      "step": 312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.618840579710147e-06,
+      "loss": 1.8251,
+      "step": 313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.617391304347828e-06,
+      "loss": 1.8077,
+      "step": 314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.615942028985507e-06,
+      "loss": 1.804,
+      "step": 315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.61449275362319e-06,
+      "loss": 1.8511,
+      "step": 316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.61304347826087e-06,
+      "loss": 1.7939,
+      "step": 317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.611594202898552e-06,
+      "loss": 1.8087,
+      "step": 318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.610144927536233e-06,
+      "loss": 1.808,
+      "step": 319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.608695652173914e-06,
+      "loss": 1.828,
+      "step": 320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.607246376811595e-06,
+      "loss": 1.7458,
+      "step": 321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.605797101449276e-06,
+      "loss": 1.8063,
+      "step": 322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.604347826086957e-06,
+      "loss": 1.7583,
+      "step": 323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.60289855072464e-06,
+      "loss": 1.7572,
+      "step": 324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.60144927536232e-06,
+      "loss": 1.7583,
+      "step": 325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.7465,
+      "step": 326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.598550724637682e-06,
+      "loss": 1.7986,
+      "step": 327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.597101449275363e-06,
+      "loss": 1.7838,
+      "step": 328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.595652173913044e-06,
+      "loss": 1.8339,
+      "step": 329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.594202898550725e-06,
+      "loss": 1.7616,
+      "step": 330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.592753623188406e-06,
+      "loss": 1.8175,
+      "step": 331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.591304347826087e-06,
+      "loss": 1.7892,
+      "step": 332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.589855072463768e-06,
+      "loss": 1.8266,
+      "step": 333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.58840579710145e-06,
+      "loss": 1.7261,
+      "step": 334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.586956521739132e-06,
+      "loss": 1.7498,
+      "step": 335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.585507246376813e-06,
+      "loss": 1.7778,
+      "step": 336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.584057971014494e-06,
+      "loss": 1.8263,
+      "step": 337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.582608695652175e-06,
+      "loss": 1.9147,
+      "step": 338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.581159420289856e-06,
+      "loss": 1.7922,
+      "step": 339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.579710144927537e-06,
+      "loss": 1.8063,
+      "step": 340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.578260869565218e-06,
+      "loss": 1.7929,
+      "step": 341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.576811594202899e-06,
+      "loss": 1.788,
+      "step": 342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.575362318840582e-06,
+      "loss": 1.7505,
+      "step": 343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.573913043478261e-06,
+      "loss": 1.7945,
+      "step": 344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.572463768115942e-06,
+      "loss": 1.8055,
+      "step": 345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.571014492753625e-06,
+      "loss": 1.7758,
+      "step": 346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.569565217391306e-06,
+      "loss": 1.7891,
+      "step": 347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.568115942028987e-06,
+      "loss": 1.7642,
+      "step": 348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.566666666666668e-06,
+      "loss": 1.8232,
+      "step": 349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 1.802,
+      "step": 350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56376811594203e-06,
+      "loss": 1.7589,
+      "step": 351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56231884057971e-06,
+      "loss": 1.7147,
+      "step": 352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.560869565217392e-06,
+      "loss": 1.7982,
+      "step": 353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.559420289855074e-06,
+      "loss": 1.7565,
+      "step": 354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.557971014492754e-06,
+      "loss": 1.7405,
+      "step": 355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.556521739130435e-06,
+      "loss": 1.7587,
+      "step": 356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.555072463768117e-06,
+      "loss": 1.7871,
+      "step": 357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.553623188405798e-06,
+      "loss": 1.7503,
+      "step": 358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.55217391304348e-06,
+      "loss": 1.7435,
+      "step": 359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.55072463768116e-06,
+      "loss": 1.7851,
+      "step": 360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.549275362318841e-06,
+      "loss": 1.7478,
+      "step": 361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.547826086956522e-06,
+      "loss": 1.82,
+      "step": 362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.546376811594203e-06,
+      "loss": 1.7634,
+      "step": 363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.544927536231884e-06,
+      "loss": 1.8127,
+      "step": 364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543478260869567e-06,
+      "loss": 1.7303,
+      "step": 365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542028985507246e-06,
+      "loss": 1.8087,
+      "step": 366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540579710144927e-06,
+      "loss": 1.8442,
+      "step": 367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53913043478261e-06,
+      "loss": 1.7595,
+      "step": 368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537681159420291e-06,
+      "loss": 1.7638,
+      "step": 369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536231884057972e-06,
+      "loss": 1.8031,
+      "step": 370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534782608695653e-06,
+      "loss": 1.793,
+      "step": 371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.533333333333334e-06,
+      "loss": 1.7709,
+      "step": 372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.531884057971015e-06,
+      "loss": 1.7846,
+      "step": 373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.530434782608696e-06,
+      "loss": 1.7332,
+      "step": 374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.528985507246377e-06,
+      "loss": 1.7891,
+      "step": 375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.52753623188406e-06,
+      "loss": 1.7546,
+      "step": 376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.526086956521739e-06,
+      "loss": 1.7659,
+      "step": 377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.524637681159422e-06,
+      "loss": 1.8536,
+      "step": 378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.523188405797103e-06,
+      "loss": 1.7542,
+      "step": 379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.521739130434784e-06,
+      "loss": 1.8001,
+      "step": 380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.520289855072465e-06,
+      "loss": 1.7604,
+      "step": 381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.518840579710146e-06,
+      "loss": 1.7424,
+      "step": 382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.517391304347827e-06,
+      "loss": 1.778,
+      "step": 383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.515942028985508e-06,
+      "loss": 1.8946,
+      "step": 384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.514492753623189e-06,
+      "loss": 1.7935,
+      "step": 385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.51304347826087e-06,
+      "loss": 1.817,
+      "step": 386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.511594202898553e-06,
+      "loss": 1.754,
+      "step": 387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.510144927536232e-06,
+      "loss": 1.7728,
+      "step": 388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.508695652173915e-06,
+      "loss": 1.7988,
+      "step": 389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.507246376811596e-06,
+      "loss": 1.7612,
+      "step": 390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.505797101449275e-06,
+      "loss": 1.7944,
+      "step": 391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.504347826086958e-06,
+      "loss": 1.7624,
+      "step": 392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.502898550724639e-06,
+      "loss": 1.7135,
+      "step": 393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.50144927536232e-06,
+      "loss": 1.8313,
+      "step": 394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.5e-06,
+      "loss": 1.7822,
+      "step": 395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.498550724637682e-06,
+      "loss": 1.7591,
+      "step": 396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.497101449275363e-06,
+      "loss": 1.7854,
+      "step": 397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.495652173913045e-06,
+      "loss": 1.7905,
+      "step": 398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.494202898550725e-06,
+      "loss": 1.7267,
+      "step": 399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.492753623188407e-06,
+      "loss": 1.7959,
+      "step": 400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.491304347826088e-06,
+      "loss": 1.7904,
+      "step": 401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.489855072463768e-06,
+      "loss": 1.798,
+      "step": 402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.48840579710145e-06,
+      "loss": 1.7562,
+      "step": 403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.486956521739131e-06,
+      "loss": 1.7432,
+      "step": 404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.485507246376812e-06,
+      "loss": 1.7399,
+      "step": 405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.484057971014493e-06,
+      "loss": 1.834,
+      "step": 406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.482608695652174e-06,
+      "loss": 1.7601,
+      "step": 407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.481159420289855e-06,
+      "loss": 1.7044,
+      "step": 408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.479710144927538e-06,
+      "loss": 1.7983,
+      "step": 409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.478260869565217e-06,
+      "loss": 1.7584,
+      "step": 410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.4768115942029e-06,
+      "loss": 1.7515,
+      "step": 411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.475362318840581e-06,
+      "loss": 1.7249,
+      "step": 412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.473913043478262e-06,
+      "loss": 1.7837,
+      "step": 413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.472463768115943e-06,
+      "loss": 1.7713,
+      "step": 414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.471014492753624e-06,
+      "loss": 1.7265,
+      "step": 415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.469565217391305e-06,
+      "loss": 1.7714,
+      "step": 416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.468115942028986e-06,
+      "loss": 1.6978,
+      "step": 417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 1.7585,
+      "step": 418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.46521739130435e-06,
+      "loss": 1.7676,
+      "step": 419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.46376811594203e-06,
+      "loss": 1.7695,
+      "step": 420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.46231884057971e-06,
+      "loss": 1.7101,
+      "step": 421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.460869565217393e-06,
+      "loss": 1.7459,
+      "step": 422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.459420289855074e-06,
+      "loss": 1.759,
+      "step": 423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.457971014492755e-06,
+      "loss": 1.7008,
+      "step": 424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.456521739130436e-06,
+      "loss": 1.7623,
+      "step": 425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.455072463768117e-06,
+      "loss": 1.7589,
+      "step": 426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.453623188405798e-06,
+      "loss": 1.6821,
+      "step": 427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.452173913043479e-06,
+      "loss": 1.7601,
+      "step": 428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.45072463768116e-06,
+      "loss": 1.6942,
+      "step": 429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.449275362318842e-06,
+      "loss": 1.6735,
+      "step": 430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.447826086956523e-06,
+      "loss": 1.6923,
+      "step": 431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.446376811594203e-06,
+      "loss": 1.643,
+      "step": 432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.444927536231885e-06,
+      "loss": 1.7404,
+      "step": 433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.443478260869566e-06,
+      "loss": 1.7223,
+      "step": 434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.442028985507247e-06,
+      "loss": 1.7114,
+      "step": 435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.440579710144928e-06,
+      "loss": 1.6646,
+      "step": 436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.43913043478261e-06,
+      "loss": 1.8709,
+      "step": 437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.43768115942029e-06,
+      "loss": 1.6551,
+      "step": 438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.436231884057971e-06,
+      "loss": 1.803,
+      "step": 439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.434782608695652e-06,
+      "loss": 1.7061,
+      "step": 440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.433333333333335e-06,
+      "loss": 1.7442,
+      "step": 441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.431884057971016e-06,
+      "loss": 1.5954,
+      "step": 442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.430434782608695e-06,
+      "loss": 1.873,
+      "step": 443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.428985507246378e-06,
+      "loss": 1.744,
+      "step": 444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.427536231884059e-06,
+      "loss": 1.7217,
+      "step": 445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.42608695652174e-06,
+      "loss": 1.8362,
+      "step": 446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.424637681159421e-06,
+      "loss": 1.5976,
+      "step": 447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.423188405797102e-06,
+      "loss": 1.7857,
+      "step": 448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.421739130434783e-06,
+      "loss": 1.8564,
+      "step": 449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.420289855072464e-06,
+      "loss": 1.6733,
+      "step": 450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.418840579710145e-06,
+      "loss": 1.7016,
+      "step": 451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.417391304347828e-06,
+      "loss": 1.6946,
+      "step": 452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.415942028985507e-06,
+      "loss": 1.7693,
+      "step": 453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.41449275362319e-06,
+      "loss": 1.9134,
+      "step": 454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.41304347826087e-06,
+      "loss": 1.8033,
+      "step": 455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.411594202898552e-06,
+      "loss": 1.8767,
+      "step": 456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.410144927536233e-06,
+      "loss": 1.8345,
+      "step": 457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.408695652173914e-06,
+      "loss": 1.7909,
+      "step": 458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.407246376811595e-06,
+      "loss": 1.7373,
+      "step": 459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.405797101449276e-06,
+      "loss": 1.7614,
+      "step": 460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.404347826086957e-06,
+      "loss": 1.7336,
+      "step": 461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.402898550724638e-06,
+      "loss": 1.7612,
+      "step": 462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.40144927536232e-06,
+      "loss": 1.6613,
+      "step": 463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.4e-06,
+      "loss": 1.6221,
+      "step": 464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.398550724637682e-06,
+      "loss": 1.6708,
+      "step": 465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.397101449275363e-06,
+      "loss": 1.7306,
+      "step": 466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.395652173913044e-06,
+      "loss": 1.6663,
+      "step": 467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.394202898550725e-06,
+      "loss": 1.6109,
+      "step": 468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.392753623188406e-06,
+      "loss": 1.7376,
+      "step": 469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.391304347826087e-06,
+      "loss": 1.6614,
+      "step": 470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.38985507246377e-06,
+      "loss": 1.7024,
+      "step": 471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.38840579710145e-06,
+      "loss": 1.7758,
+      "step": 472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.38695652173913e-06,
+      "loss": 1.8538,
+      "step": 473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.385507246376813e-06,
+      "loss": 1.7568,
+      "step": 474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.384057971014492e-06,
+      "loss": 1.671,
+      "step": 475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.382608695652175e-06,
+      "loss": 1.7339,
+      "step": 476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.381159420289856e-06,
+      "loss": 1.8321,
+      "step": 477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.379710144927537e-06,
+      "loss": 1.7613,
+      "step": 478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.378260869565218e-06,
+      "loss": 1.7558,
+      "step": 479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.3768115942029e-06,
+      "loss": 1.7601,
+      "step": 480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.37536231884058e-06,
+      "loss": 1.6581,
+      "step": 481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.373913043478263e-06,
+      "loss": 1.7532,
+      "step": 482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.372463768115942e-06,
+      "loss": 1.8625,
+      "step": 483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.371014492753623e-06,
+      "loss": 1.7552,
+      "step": 484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.369565217391306e-06,
+      "loss": 1.6593,
+      "step": 485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.368115942028985e-06,
+      "loss": 1.7272,
+      "step": 486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.366666666666668e-06,
+      "loss": 1.6921,
+      "step": 487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.365217391304349e-06,
+      "loss": 1.9005,
+      "step": 488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.36376811594203e-06,
+      "loss": 1.6615,
+      "step": 489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.362318840579711e-06,
+      "loss": 1.6557,
+      "step": 490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.360869565217392e-06,
+      "loss": 1.7617,
+      "step": 491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.359420289855073e-06,
+      "loss": 1.6262,
+      "step": 492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.357971014492756e-06,
+      "loss": 1.6038,
+      "step": 493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.356521739130435e-06,
+      "loss": 1.7017,
+      "step": 494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.355072463768116e-06,
+      "loss": 1.6771,
+      "step": 495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.353623188405799e-06,
+      "loss": 1.7926,
+      "step": 496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.352173913043478e-06,
+      "loss": 1.7765,
+      "step": 497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.35072463768116e-06,
+      "loss": 1.6132,
+      "step": 498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.349275362318842e-06,
+      "loss": 1.7929,
+      "step": 499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.347826086956523e-06,
+      "loss": 1.7747,
+      "step": 500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.346376811594204e-06,
+      "loss": 1.7371,
+      "step": 501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.344927536231885e-06,
+      "loss": 1.7765,
+      "step": 502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.343478260869566e-06,
+      "loss": 1.7343,
+      "step": 503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.342028985507248e-06,
+      "loss": 1.7641,
+      "step": 504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.340579710144928e-06,
+      "loss": 1.66,
+      "step": 505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.33913043478261e-06,
+      "loss": 1.4669,
+      "step": 506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.337681159420291e-06,
+      "loss": 1.6771,
+      "step": 507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.33623188405797e-06,
+      "loss": 1.7796,
+      "step": 508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.334782608695653e-06,
+      "loss": 1.6644,
+      "step": 509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.6729,
+      "step": 510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.331884057971015e-06,
+      "loss": 1.5036,
+      "step": 511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.330434782608696e-06,
+      "loss": 1.7212,
+      "step": 512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.328985507246377e-06,
+      "loss": 1.7362,
+      "step": 513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.327536231884058e-06,
+      "loss": 1.7772,
+      "step": 514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.32608695652174e-06,
+      "loss": 1.6804,
+      "step": 515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.32463768115942e-06,
+      "loss": 1.6967,
+      "step": 516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.323188405797103e-06,
+      "loss": 1.9051,
+      "step": 517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.321739130434784e-06,
+      "loss": 1.7633,
+      "step": 518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.320289855072463e-06,
+      "loss": 1.5182,
+      "step": 519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.318840579710146e-06,
+      "loss": 1.7184,
+      "step": 520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.317391304347827e-06,
+      "loss": 2.2586,
+      "step": 521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.315942028985508e-06,
+      "loss": 1.5746,
+      "step": 522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.314492753623189e-06,
+      "loss": 1.8775,
+      "step": 523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.31304347826087e-06,
+      "loss": 1.607,
+      "step": 524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.311594202898551e-06,
+      "loss": 1.8892,
+      "step": 525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.310144927536232e-06,
+      "loss": 1.7163,
+      "step": 526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.308695652173913e-06,
+      "loss": 1.7406,
+      "step": 527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.307246376811596e-06,
+      "loss": 1.8209,
+      "step": 528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.305797101449277e-06,
+      "loss": 2.2019,
+      "step": 529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.304347826086956e-06,
+      "loss": 1.6532,
+      "step": 530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.302898550724639e-06,
+      "loss": 1.7642,
+      "step": 531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.30144927536232e-06,
+      "loss": 1.734,
+      "step": 532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.3e-06,
+      "loss": 1.7189,
+      "step": 533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.298550724637682e-06,
+      "loss": 1.6603,
+      "step": 534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.297101449275363e-06,
+      "loss": 1.8562,
+      "step": 535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.295652173913044e-06,
+      "loss": 1.5545,
+      "step": 536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.294202898550725e-06,
+      "loss": 1.7548,
+      "step": 537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.292753623188406e-06,
+      "loss": 1.6227,
+      "step": 538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.291304347826088e-06,
+      "loss": 1.5726,
+      "step": 539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.28985507246377e-06,
+      "loss": 1.8771,
+      "step": 540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.28840579710145e-06,
+      "loss": 1.773,
+      "step": 541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.286956521739131e-06,
+      "loss": 1.7309,
+      "step": 542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.285507246376812e-06,
+      "loss": 1.5423,
+      "step": 543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.284057971014493e-06,
+      "loss": 1.7325,
+      "step": 544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.282608695652174e-06,
+      "loss": 1.8245,
+      "step": 545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.281159420289855e-06,
+      "loss": 1.8336,
+      "step": 546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.279710144927536e-06,
+      "loss": 1.6275,
+      "step": 547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.278260869565217e-06,
+      "loss": 1.7493,
+      "step": 548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.276811594202898e-06,
+      "loss": 1.7878,
+      "step": 549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.275362318840581e-06,
+      "loss": 1.8985,
+      "step": 550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.273913043478262e-06,
+      "loss": 1.5902,
+      "step": 551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.272463768115943e-06,
+      "loss": 1.6755,
+      "step": 552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.271014492753624e-06,
+      "loss": 1.4542,
+      "step": 553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.269565217391305e-06,
+      "loss": 2.0015,
+      "step": 554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.268115942028986e-06,
+      "loss": 1.6533,
+      "step": 555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.266666666666667e-06,
+      "loss": 1.3897,
+      "step": 556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.265217391304348e-06,
+      "loss": 1.5342,
+      "step": 557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.263768115942031e-06,
+      "loss": 1.5419,
+      "step": 558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.26231884057971e-06,
+      "loss": 1.6568,
+      "step": 559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.260869565217391e-06,
+      "loss": 1.6842,
+      "step": 560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.259420289855074e-06,
+      "loss": 1.6922,
+      "step": 561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.257971014492755e-06,
+      "loss": 1.7124,
+      "step": 562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.256521739130436e-06,
+      "loss": 1.6837,
+      "step": 563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.255072463768117e-06,
+      "loss": 1.6708,
+      "step": 564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.253623188405798e-06,
+      "loss": 1.6237,
+      "step": 565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.252173913043479e-06,
+      "loss": 1.6164,
+      "step": 566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.25072463768116e-06,
+      "loss": 2.0735,
+      "step": 567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.249275362318841e-06,
+      "loss": 1.7095,
+      "step": 568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.247826086956524e-06,
+      "loss": 1.6885,
+      "step": 569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.246376811594203e-06,
+      "loss": 1.6518,
+      "step": 570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.244927536231884e-06,
+      "loss": 1.6327,
+      "step": 571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.243478260869567e-06,
+      "loss": 1.6598,
+      "step": 572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.242028985507248e-06,
+      "loss": 1.6891,
+      "step": 573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.240579710144929e-06,
+      "loss": 1.6103,
+      "step": 574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.23913043478261e-06,
+      "loss": 1.5785,
+      "step": 575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.23768115942029e-06,
+      "loss": 1.9188,
+      "step": 576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.236231884057972e-06,
+      "loss": 1.575,
+      "step": 577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.234782608695653e-06,
+      "loss": 2.0226,
+      "step": 578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.233333333333334e-06,
+      "loss": 1.6416,
+      "step": 579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.231884057971016e-06,
+      "loss": 1.7377,
+      "step": 580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.230434782608696e-06,
+      "loss": 1.5323,
+      "step": 581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.228985507246377e-06,
+      "loss": 1.6688,
+      "step": 582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.22753623188406e-06,
+      "loss": 1.6499,
+      "step": 583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.22608695652174e-06,
+      "loss": 1.6245,
+      "step": 584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.224637681159421e-06,
+      "loss": 1.7317,
+      "step": 585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.223188405797102e-06,
+      "loss": 1.6601,
+      "step": 586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.221739130434783e-06,
+      "loss": 1.6956,
+      "step": 587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.220289855072464e-06,
+      "loss": 1.6712,
+      "step": 588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.218840579710145e-06,
+      "loss": 1.6764,
+      "step": 589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.217391304347826e-06,
+      "loss": 1.7504,
+      "step": 590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.215942028985509e-06,
+      "loss": 1.6428,
+      "step": 591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.214492753623188e-06,
+      "loss": 1.9458,
+      "step": 592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.213043478260871e-06,
+      "loss": 1.6619,
+      "step": 593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.211594202898552e-06,
+      "loss": 1.5026,
+      "step": 594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.210144927536233e-06,
+      "loss": 1.9632,
+      "step": 595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.208695652173914e-06,
+      "loss": 1.8338,
+      "step": 596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.207246376811595e-06,
+      "loss": 1.5195,
+      "step": 597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.205797101449276e-06,
+      "loss": 1.9442,
+      "step": 598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.204347826086957e-06,
+      "loss": 1.6933,
+      "step": 599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.202898550724638e-06,
+      "loss": 1.6615,
+      "step": 600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.201449275362319e-06,
+      "loss": 1.3086,
+      "step": 601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.6395,
+      "step": 602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.198550724637681e-06,
+      "loss": 1.692,
+      "step": 603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.197101449275364e-06,
+      "loss": 1.3012,
+      "step": 604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.195652173913045e-06,
+      "loss": 1.4745,
+      "step": 605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.194202898550726e-06,
+      "loss": 1.665,
+      "step": 606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.192753623188407e-06,
+      "loss": 1.9095,
+      "step": 607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.191304347826088e-06,
+      "loss": 1.9521,
+      "step": 608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.189855072463769e-06,
+      "loss": 1.7353,
+      "step": 609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.18840579710145e-06,
+      "loss": 1.6643,
+      "step": 610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.18695652173913e-06,
+      "loss": 1.4785,
+      "step": 611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.185507246376812e-06,
+      "loss": 1.6211,
+      "step": 612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.184057971014494e-06,
+      "loss": 1.9069,
+      "step": 613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.182608695652174e-06,
+      "loss": 1.7239,
+      "step": 614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.181159420289856e-06,
+      "loss": 1.918,
+      "step": 615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.179710144927537e-06,
+      "loss": 1.3954,
+      "step": 616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.178260869565218e-06,
+      "loss": 1.6497,
+      "step": 617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.1768115942029e-06,
+      "loss": 1.7112,
+      "step": 618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.17536231884058e-06,
+      "loss": 1.3031,
+      "step": 619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.173913043478261e-06,
+      "loss": 1.6014,
+      "step": 620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.172463768115942e-06,
+      "loss": 1.5809,
+      "step": 621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.171014492753623e-06,
+      "loss": 1.6354,
+      "step": 622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.169565217391304e-06,
+      "loss": 1.5894,
+      "step": 623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.168115942028987e-06,
+      "loss": 1.5135,
+      "step": 624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.802,
+      "step": 625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.165217391304349e-06,
+      "loss": 1.8681,
+      "step": 626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.16376811594203e-06,
+      "loss": 1.3531,
+      "step": 627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.162318840579711e-06,
+      "loss": 1.4468,
+      "step": 628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.160869565217392e-06,
+      "loss": 1.6508,
+      "step": 629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.159420289855073e-06,
+      "loss": 1.2979,
+      "step": 630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.157971014492754e-06,
+      "loss": 1.7596,
+      "step": 631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.156521739130435e-06,
+      "loss": 1.7862,
+      "step": 632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.155072463768116e-06,
+      "loss": 1.7375,
+      "step": 633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.153623188405799e-06,
+      "loss": 1.6296,
+      "step": 634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.15217391304348e-06,
+      "loss": 1.8474,
+      "step": 635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.150724637681159e-06,
+      "loss": 1.5052,
+      "step": 636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.149275362318842e-06,
+      "loss": 1.6135,
+      "step": 637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.147826086956523e-06,
+      "loss": 1.7252,
+      "step": 638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.146376811594204e-06,
+      "loss": 1.7536,
+      "step": 639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.144927536231885e-06,
+      "loss": 1.469,
+      "step": 640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.143478260869566e-06,
+      "loss": 1.6801,
+      "step": 641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.142028985507247e-06,
+      "loss": 1.6273,
+      "step": 642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.140579710144928e-06,
+      "loss": 1.6722,
+      "step": 643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.139130434782609e-06,
+      "loss": 1.7045,
+      "step": 644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.137681159420292e-06,
+      "loss": 1.4982,
+      "step": 645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.136231884057973e-06,
+      "loss": 1.6666,
+      "step": 646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.134782608695652e-06,
+      "loss": 1.4154,
+      "step": 647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.133333333333335e-06,
+      "loss": 1.4674,
+      "step": 648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.131884057971016e-06,
+      "loss": 1.6787,
+      "step": 649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 1.7374,
+      "step": 650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.128985507246378e-06,
+      "loss": 2.0465,
+      "step": 651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.127536231884059e-06,
+      "loss": 1.6424,
+      "step": 652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.12608695652174e-06,
+      "loss": 1.6249,
+      "step": 653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.12463768115942e-06,
+      "loss": 1.6703,
+      "step": 654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.123188405797102e-06,
+      "loss": 1.8832,
+      "step": 655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.121739130434784e-06,
+      "loss": 1.6985,
+      "step": 656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.120289855072465e-06,
+      "loss": 1.9327,
+      "step": 657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.118840579710145e-06,
+      "loss": 1.6431,
+      "step": 658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.117391304347827e-06,
+      "loss": 1.3707,
+      "step": 659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.115942028985508e-06,
+      "loss": 1.671,
+      "step": 660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.11449275362319e-06,
+      "loss": 1.6511,
+      "step": 661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.11304347826087e-06,
+      "loss": 2.3661,
+      "step": 662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.111594202898551e-06,
+      "loss": 1.4736,
+      "step": 663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.110144927536232e-06,
+      "loss": 1.702,
+      "step": 664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.108695652173913e-06,
+      "loss": 1.6404,
+      "step": 665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.107246376811594e-06,
+      "loss": 1.7918,
+      "step": 666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.105797101449277e-06,
+      "loss": 1.6485,
+      "step": 667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.104347826086958e-06,
+      "loss": 1.6458,
+      "step": 668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.102898550724639e-06,
+      "loss": 1.73,
+      "step": 669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.10144927536232e-06,
+      "loss": 1.6779,
+      "step": 670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 1.7368,
+      "step": 671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.098550724637682e-06,
+      "loss": 1.6192,
+      "step": 672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.097101449275363e-06,
+      "loss": 1.6573,
+      "step": 673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.095652173913044e-06,
+      "loss": 1.6685,
+      "step": 674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.094202898550725e-06,
+      "loss": 1.6459,
+      "step": 675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.092753623188406e-06,
+      "loss": 1.96,
+      "step": 676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.091304347826087e-06,
+      "loss": 1.2422,
+      "step": 677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.08985507246377e-06,
+      "loss": 1.2432,
+      "step": 678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.08840579710145e-06,
+      "loss": 2.4282,
+      "step": 679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.086956521739132e-06,
+      "loss": 1.651,
+      "step": 680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.085507246376813e-06,
+      "loss": 1.6602,
+      "step": 681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.084057971014494e-06,
+      "loss": 1.4327,
+      "step": 682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.082608695652175e-06,
+      "loss": 1.4728,
+      "step": 683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.081159420289856e-06,
+      "loss": 1.677,
+      "step": 684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.079710144927537e-06,
+      "loss": 1.7016,
+      "step": 685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.07826086956522e-06,
+      "loss": 1.7507,
+      "step": 686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.076811594202899e-06,
+      "loss": 1.5511,
+      "step": 687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.07536231884058e-06,
+      "loss": 1.4525,
+      "step": 688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.073913043478262e-06,
+      "loss": 1.6997,
+      "step": 689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.072463768115943e-06,
+      "loss": 1.6878,
+      "step": 690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.071014492753624e-06,
+      "loss": 1.6826,
+      "step": 691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.069565217391305e-06,
+      "loss": 1.4774,
+      "step": 692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.068115942028986e-06,
+      "loss": 1.7614,
+      "step": 693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.6491,
+      "step": 694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.065217391304348e-06,
+      "loss": 1.3832,
+      "step": 695
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.3107913669064748,
+      "eval_loss": 1.560390591621399,
+      "eval_roc_auc": 0.7616650252630864,
+      "eval_runtime": 84.9159,
+      "eval_samples_per_second": 8.185,
+      "eval_steps_per_second": 2.049,
+      "step": 695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.06376811594203e-06,
+      "loss": 1.6292,
+      "step": 696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.062318840579712e-06,
+      "loss": 1.4408,
+      "step": 697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.060869565217391e-06,
+      "loss": 1.6335,
+      "step": 698
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.059420289855072e-06,
+      "loss": 1.6124,
+      "step": 699
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.057971014492755e-06,
+      "loss": 1.4772,
+      "step": 700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.056521739130434e-06,
+      "loss": 1.682,
+      "step": 701
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.055072463768117e-06,
+      "loss": 1.6434,
+      "step": 702
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.053623188405798e-06,
+      "loss": 2.1587,
+      "step": 703
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.052173913043479e-06,
+      "loss": 1.5141,
+      "step": 704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.05072463768116e-06,
+      "loss": 1.6807,
+      "step": 705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.049275362318841e-06,
+      "loss": 1.391,
+      "step": 706
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.047826086956522e-06,
+      "loss": 1.649,
+      "step": 707
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.046376811594205e-06,
+      "loss": 1.1985,
+      "step": 708
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.044927536231884e-06,
+      "loss": 2.0363,
+      "step": 709
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.043478260869565e-06,
+      "loss": 1.6333,
+      "step": 710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.042028985507248e-06,
+      "loss": 1.6959,
+      "step": 711
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.040579710144927e-06,
+      "loss": 1.1806,
+      "step": 712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.03913043478261e-06,
+      "loss": 1.982,
+      "step": 713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.03768115942029e-06,
+      "loss": 1.5944,
+      "step": 714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.036231884057972e-06,
+      "loss": 1.5748,
+      "step": 715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.034782608695653e-06,
+      "loss": 1.6111,
+      "step": 716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.033333333333334e-06,
+      "loss": 1.6622,
+      "step": 717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.031884057971015e-06,
+      "loss": 1.6149,
+      "step": 718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.030434782608698e-06,
+      "loss": 1.3978,
+      "step": 719
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.028985507246377e-06,
+      "loss": 1.6497,
+      "step": 720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.02753623188406e-06,
+      "loss": 1.7629,
+      "step": 721
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.02608695652174e-06,
+      "loss": 1.7114,
+      "step": 722
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.02463768115942e-06,
+      "loss": 1.4262,
+      "step": 723
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.023188405797103e-06,
+      "loss": 2.0046,
+      "step": 724
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.021739130434784e-06,
+      "loss": 1.6204,
+      "step": 725
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.020289855072465e-06,
+      "loss": 1.7414,
+      "step": 726
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.018840579710146e-06,
+      "loss": 1.3943,
+      "step": 727
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.017391304347827e-06,
+      "loss": 1.166,
+      "step": 728
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.015942028985508e-06,
+      "loss": 1.4277,
+      "step": 729
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.01449275362319e-06,
+      "loss": 1.3278,
+      "step": 730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.01304347826087e-06,
+      "loss": 1.8223,
+      "step": 731
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.011594202898552e-06,
+      "loss": 1.7558,
+      "step": 732
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.010144927536233e-06,
+      "loss": 1.5626,
+      "step": 733
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.008695652173913e-06,
+      "loss": 1.572,
+      "step": 734
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.007246376811595e-06,
+      "loss": 1.639,
+      "step": 735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.005797101449276e-06,
+      "loss": 1.6352,
+      "step": 736
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.004347826086957e-06,
+      "loss": 1.5911,
+      "step": 737
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.002898550724638e-06,
+      "loss": 1.4206,
+      "step": 738
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.00144927536232e-06,
+      "loss": 1.6128,
+      "step": 739
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9e-06,
+      "loss": 1.477,
+      "step": 740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.998550724637683e-06,
+      "loss": 1.4794,
+      "step": 741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.997101449275362e-06,
+      "loss": 1.5863,
+      "step": 742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.995652173913045e-06,
+      "loss": 1.8247,
+      "step": 743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.994202898550726e-06,
+      "loss": 1.6878,
+      "step": 744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.992753623188405e-06,
+      "loss": 1.7672,
+      "step": 745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.991304347826088e-06,
+      "loss": 1.6637,
+      "step": 746
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.989855072463769e-06,
+      "loss": 1.6579,
+      "step": 747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.98840579710145e-06,
+      "loss": 1.8682,
+      "step": 748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.986956521739131e-06,
+      "loss": 1.596,
+      "step": 749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.985507246376812e-06,
+      "loss": 1.4322,
+      "step": 750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.984057971014493e-06,
+      "loss": 1.6116,
+      "step": 751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.982608695652176e-06,
+      "loss": 1.647,
+      "step": 752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.981159420289855e-06,
+      "loss": 1.6839,
+      "step": 753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.979710144927538e-06,
+      "loss": 1.4523,
+      "step": 754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.978260869565219e-06,
+      "loss": 2.0168,
+      "step": 755
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.9768115942029e-06,
+      "loss": 1.6499,
+      "step": 756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97536231884058e-06,
+      "loss": 1.5751,
+      "step": 757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.973913043478262e-06,
+      "loss": 1.4304,
+      "step": 758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.972463768115943e-06,
+      "loss": 1.4293,
+      "step": 759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.971014492753624e-06,
+      "loss": 1.8518,
+      "step": 760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.969565217391305e-06,
+      "loss": 1.6377,
+      "step": 761
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.968115942028987e-06,
+      "loss": 1.6637,
+      "step": 762
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.966666666666667e-06,
+      "loss": 1.5302,
+      "step": 763
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.965217391304348e-06,
+      "loss": 1.6774,
+      "step": 764
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.96376811594203e-06,
+      "loss": 1.8129,
+      "step": 765
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.962318840579711e-06,
+      "loss": 1.3452,
+      "step": 766
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.960869565217392e-06,
+      "loss": 1.6534,
+      "step": 767
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.959420289855073e-06,
+      "loss": 1.6257,
+      "step": 768
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.957971014492754e-06,
+      "loss": 1.4351,
+      "step": 769
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.956521739130435e-06,
+      "loss": 1.6867,
+      "step": 770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.955072463768116e-06,
+      "loss": 1.647,
+      "step": 771
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.953623188405797e-06,
+      "loss": 1.5673,
+      "step": 772
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.95217391304348e-06,
+      "loss": 0.889,
+      "step": 773
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.95072463768116e-06,
+      "loss": 2.0211,
+      "step": 774
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.94927536231884e-06,
+      "loss": 2.0851,
+      "step": 775
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.947826086956523e-06,
+      "loss": 1.6561,
+      "step": 776
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.946376811594204e-06,
+      "loss": 1.6377,
+      "step": 777
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.944927536231885e-06,
+      "loss": 1.672,
+      "step": 778
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.943478260869566e-06,
+      "loss": 1.6637,
+      "step": 779
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.942028985507247e-06,
+      "loss": 1.3829,
+      "step": 780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.940579710144928e-06,
+      "loss": 2.0199,
+      "step": 781
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.939130434782609e-06,
+      "loss": 1.6445,
+      "step": 782
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.93768115942029e-06,
+      "loss": 1.6485,
+      "step": 783
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.936231884057973e-06,
+      "loss": 1.6675,
+      "step": 784
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.934782608695652e-06,
+      "loss": 1.828,
+      "step": 785
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 1.4132,
+      "step": 786
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.931884057971016e-06,
+      "loss": 1.8105,
+      "step": 787
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.930434782608697e-06,
+      "loss": 1.622,
+      "step": 788
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.928985507246378e-06,
+      "loss": 1.6458,
+      "step": 789
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.927536231884059e-06,
+      "loss": 1.424,
+      "step": 790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.92608695652174e-06,
+      "loss": 1.4497,
+      "step": 791
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.92463768115942e-06,
+      "loss": 1.6232,
+      "step": 792
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.923188405797102e-06,
+      "loss": 1.7147,
+      "step": 793
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.921739130434783e-06,
+      "loss": 1.7364,
+      "step": 794
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.920289855072465e-06,
+      "loss": 1.427,
+      "step": 795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.918840579710145e-06,
+      "loss": 1.3455,
+      "step": 796
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.917391304347827e-06,
+      "loss": 1.6258,
+      "step": 797
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.915942028985508e-06,
+      "loss": 1.4019,
+      "step": 798
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.91449275362319e-06,
+      "loss": 1.8572,
+      "step": 799
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.91304347826087e-06,
+      "loss": 1.3809,
+      "step": 800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.911594202898551e-06,
+      "loss": 1.4102,
+      "step": 801
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.910144927536232e-06,
+      "loss": 1.6432,
+      "step": 802
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.908695652173913e-06,
+      "loss": 1.5589,
+      "step": 803
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.907246376811594e-06,
+      "loss": 1.4182,
+      "step": 804
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.905797101449275e-06,
+      "loss": 1.1432,
+      "step": 805
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.904347826086958e-06,
+      "loss": 1.4933,
+      "step": 806
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.902898550724637e-06,
+      "loss": 2.0056,
+      "step": 807
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.90144927536232e-06,
+      "loss": 1.7011,
+      "step": 808
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 1.3532,
+      "step": 809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.898550724637682e-06,
+      "loss": 1.67,
+      "step": 810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.897101449275363e-06,
+      "loss": 1.4631,
+      "step": 811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.895652173913044e-06,
+      "loss": 1.3802,
+      "step": 812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.894202898550725e-06,
+      "loss": 1.3856,
+      "step": 813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.892753623188408e-06,
+      "loss": 1.7515,
+      "step": 814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.891304347826087e-06,
+      "loss": 1.5418,
+      "step": 815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.889855072463768e-06,
+      "loss": 1.581,
+      "step": 816
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.888405797101451e-06,
+      "loss": 1.5294,
+      "step": 817
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.88695652173913e-06,
+      "loss": 1.3623,
+      "step": 818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.885507246376813e-06,
+      "loss": 1.5296,
+      "step": 819
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.884057971014494e-06,
+      "loss": 1.0802,
+      "step": 820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.882608695652175e-06,
+      "loss": 1.6858,
+      "step": 821
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.881159420289856e-06,
+      "loss": 1.6703,
+      "step": 822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.879710144927537e-06,
+      "loss": 1.3031,
+      "step": 823
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.878260869565218e-06,
+      "loss": 1.631,
+      "step": 824
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.876811594202899e-06,
+      "loss": 1.6039,
+      "step": 825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.87536231884058e-06,
+      "loss": 1.6679,
+      "step": 826
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.873913043478261e-06,
+      "loss": 1.3576,
+      "step": 827
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.872463768115944e-06,
+      "loss": 1.6226,
+      "step": 828
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.871014492753623e-06,
+      "loss": 1.0331,
+      "step": 829
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.869565217391306e-06,
+      "loss": 1.3601,
+      "step": 830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.868115942028987e-06,
+      "loss": 1.5058,
+      "step": 831
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.866666666666668e-06,
+      "loss": 1.6157,
+      "step": 832
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.865217391304349e-06,
+      "loss": 1.2857,
+      "step": 833
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.86376811594203e-06,
+      "loss": 1.6806,
+      "step": 834
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.86231884057971e-06,
+      "loss": 1.0666,
+      "step": 835
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.860869565217392e-06,
+      "loss": 1.4058,
+      "step": 836
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.859420289855073e-06,
+      "loss": 1.489,
+      "step": 837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.857971014492754e-06,
+      "loss": 1.5739,
+      "step": 838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.856521739130436e-06,
+      "loss": 1.6802,
+      "step": 839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.855072463768116e-06,
+      "loss": 1.7244,
+      "step": 840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.853623188405798e-06,
+      "loss": 1.6474,
+      "step": 841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.85217391304348e-06,
+      "loss": 1.1037,
+      "step": 842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.85072463768116e-06,
+      "loss": 1.7256,
+      "step": 843
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.849275362318841e-06,
+      "loss": 2.0597,
+      "step": 844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.847826086956522e-06,
+      "loss": 1.6283,
+      "step": 845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.846376811594203e-06,
+      "loss": 1.393,
+      "step": 846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.844927536231884e-06,
+      "loss": 2.1503,
+      "step": 847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.843478260869565e-06,
+      "loss": 1.6364,
+      "step": 848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.842028985507248e-06,
+      "loss": 1.6713,
+      "step": 849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.840579710144929e-06,
+      "loss": 1.4895,
+      "step": 850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.839130434782608e-06,
+      "loss": 1.8004,
+      "step": 851
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.837681159420291e-06,
+      "loss": 2.0732,
+      "step": 852
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.836231884057972e-06,
+      "loss": 1.3199,
+      "step": 853
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.834782608695653e-06,
+      "loss": 1.3149,
+      "step": 854
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 1.8689,
+      "step": 855
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.831884057971015e-06,
+      "loss": 1.631,
+      "step": 856
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.830434782608696e-06,
+      "loss": 1.3619,
+      "step": 857
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.828985507246377e-06,
+      "loss": 1.63,
+      "step": 858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.827536231884058e-06,
+      "loss": 1.7442,
+      "step": 859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.82608695652174e-06,
+      "loss": 1.3376,
+      "step": 860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.824637681159422e-06,
+      "loss": 1.6102,
+      "step": 861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.823188405797101e-06,
+      "loss": 1.0392,
+      "step": 862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.821739130434784e-06,
+      "loss": 1.3544,
+      "step": 863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.820289855072465e-06,
+      "loss": 1.7174,
+      "step": 864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.818840579710146e-06,
+      "loss": 1.514,
+      "step": 865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.817391304347827e-06,
+      "loss": 1.5826,
+      "step": 866
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.815942028985508e-06,
+      "loss": 1.0349,
+      "step": 867
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.814492753623189e-06,
+      "loss": 1.4234,
+      "step": 868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.81304347826087e-06,
+      "loss": 1.9743,
+      "step": 869
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.81159420289855e-06,
+      "loss": 1.6062,
+      "step": 870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.810144927536233e-06,
+      "loss": 1.298,
+      "step": 871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.808695652173914e-06,
+      "loss": 1.6511,
+      "step": 872
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.807246376811594e-06,
+      "loss": 1.6493,
+      "step": 873
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.805797101449276e-06,
+      "loss": 1.636,
+      "step": 874
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.804347826086957e-06,
+      "loss": 2.1221,
+      "step": 875
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.802898550724638e-06,
+      "loss": 1.3905,
+      "step": 876
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.80144927536232e-06,
+      "loss": 1.6418,
+      "step": 877
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.8e-06,
+      "loss": 1.5917,
+      "step": 878
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.798550724637681e-06,
+      "loss": 1.6737,
+      "step": 879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.797101449275362e-06,
+      "loss": 1.6475,
+      "step": 880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.795652173913043e-06,
+      "loss": 1.2844,
+      "step": 881
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.794202898550726e-06,
+      "loss": 1.3881,
+      "step": 882
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.792753623188407e-06,
+      "loss": 2.0465,
+      "step": 883
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.791304347826088e-06,
+      "loss": 1.6243,
+      "step": 884
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.789855072463769e-06,
+      "loss": 1.1605,
+      "step": 885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.78840579710145e-06,
+      "loss": 1.6039,
+      "step": 886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.786956521739131e-06,
+      "loss": 1.463,
+      "step": 887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.785507246376812e-06,
+      "loss": 2.0862,
+      "step": 888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.784057971014493e-06,
+      "loss": 1.4761,
+      "step": 889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.782608695652174e-06,
+      "loss": 1.8063,
+      "step": 890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.781159420289855e-06,
+      "loss": 1.338,
+      "step": 891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.779710144927536e-06,
+      "loss": 1.6039,
+      "step": 892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.778260869565219e-06,
+      "loss": 1.5846,
+      "step": 893
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.7768115942029e-06,
+      "loss": 1.6199,
+      "step": 894
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.775362318840581e-06,
+      "loss": 1.6212,
+      "step": 895
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.773913043478262e-06,
+      "loss": 1.8586,
+      "step": 896
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.772463768115943e-06,
+      "loss": 1.6133,
+      "step": 897
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.771014492753624e-06,
+      "loss": 1.6282,
+      "step": 898
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.769565217391305e-06,
+      "loss": 1.5694,
+      "step": 899
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.768115942028986e-06,
+      "loss": 1.3378,
+      "step": 900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.766666666666669e-06,
+      "loss": 2.0807,
+      "step": 901
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.765217391304348e-06,
+      "loss": 1.0157,
+      "step": 902
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.763768115942029e-06,
+      "loss": 1.614,
+      "step": 903
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.762318840579712e-06,
+      "loss": 1.6484,
+      "step": 904
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.760869565217393e-06,
+      "loss": 1.4103,
+      "step": 905
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.759420289855074e-06,
+      "loss": 1.5811,
+      "step": 906
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.757971014492755e-06,
+      "loss": 1.5991,
+      "step": 907
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.756521739130436e-06,
+      "loss": 1.6452,
+      "step": 908
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.755072463768117e-06,
+      "loss": 1.2972,
+      "step": 909
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.753623188405798e-06,
+      "loss": 1.9945,
+      "step": 910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.752173913043479e-06,
+      "loss": 1.3154,
+      "step": 911
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.750724637681161e-06,
+      "loss": 1.6209,
+      "step": 912
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.74927536231884e-06,
+      "loss": 1.0825,
+      "step": 913
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.747826086956522e-06,
+      "loss": 1.2306,
+      "step": 914
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.746376811594204e-06,
+      "loss": 1.3046,
+      "step": 915
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.744927536231885e-06,
+      "loss": 1.6236,
+      "step": 916
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.743478260869566e-06,
+      "loss": 1.3973,
+      "step": 917
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.742028985507247e-06,
+      "loss": 1.6536,
+      "step": 918
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.740579710144928e-06,
+      "loss": 1.4497,
+      "step": 919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.73913043478261e-06,
+      "loss": 1.6378,
+      "step": 920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.73768115942029e-06,
+      "loss": 1.6431,
+      "step": 921
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.736231884057971e-06,
+      "loss": 1.3656,
+      "step": 922
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.734782608695654e-06,
+      "loss": 1.6428,
+      "step": 923
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 1.5969,
+      "step": 924
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.731884057971014e-06,
+      "loss": 1.7146,
+      "step": 925
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.730434782608697e-06,
+      "loss": 1.3212,
+      "step": 926
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.728985507246378e-06,
+      "loss": 1.6779,
+      "step": 927
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.727536231884059e-06,
+      "loss": 1.552,
+      "step": 928
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.72608695652174e-06,
+      "loss": 0.9555,
+      "step": 929
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.724637681159421e-06,
+      "loss": 1.5145,
+      "step": 930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.723188405797102e-06,
+      "loss": 1.8755,
+      "step": 931
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.721739130434783e-06,
+      "loss": 1.6382,
+      "step": 932
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.720289855072464e-06,
+      "loss": 1.4731,
+      "step": 933
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.718840579710147e-06,
+      "loss": 1.2701,
+      "step": 934
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.717391304347826e-06,
+      "loss": 1.9602,
+      "step": 935
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.715942028985509e-06,
+      "loss": 1.2025,
+      "step": 936
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.71449275362319e-06,
+      "loss": 1.7536,
+      "step": 937
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.71304347826087e-06,
+      "loss": 1.6947,
+      "step": 938
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.711594202898552e-06,
+      "loss": 1.6108,
+      "step": 939
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.710144927536233e-06,
+      "loss": 1.6532,
+      "step": 940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.708695652173914e-06,
+      "loss": 1.2633,
+      "step": 941
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.707246376811595e-06,
+      "loss": 1.5519,
+      "step": 942
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.705797101449276e-06,
+      "loss": 1.6296,
+      "step": 943
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.704347826086957e-06,
+      "loss": 1.6413,
+      "step": 944
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.70289855072464e-06,
+      "loss": 1.4248,
+      "step": 945
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.701449275362319e-06,
+      "loss": 1.5846,
+      "step": 946
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.9216,
+      "step": 947
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.698550724637682e-06,
+      "loss": 1.873,
+      "step": 948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.697101449275362e-06,
+      "loss": 1.4761,
+      "step": 949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 1.4822,
+      "step": 950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.694202898550725e-06,
+      "loss": 1.1829,
+      "step": 951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.692753623188406e-06,
+      "loss": 1.6215,
+      "step": 952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.691304347826087e-06,
+      "loss": 1.3494,
+      "step": 953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.689855072463768e-06,
+      "loss": 1.3998,
+      "step": 954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.68840579710145e-06,
+      "loss": 1.2456,
+      "step": 955
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.686956521739132e-06,
+      "loss": 0.9332,
+      "step": 956
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.685507246376811e-06,
+      "loss": 1.5734,
+      "step": 957
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.684057971014494e-06,
+      "loss": 1.6028,
+      "step": 958
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.682608695652175e-06,
+      "loss": 2.0519,
+      "step": 959
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.681159420289854e-06,
+      "loss": 1.4303,
+      "step": 960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.679710144927537e-06,
+      "loss": 1.6758,
+      "step": 961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.678260869565218e-06,
+      "loss": 1.6497,
+      "step": 962
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.676811594202899e-06,
+      "loss": 1.6136,
+      "step": 963
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.67536231884058e-06,
+      "loss": 1.4699,
+      "step": 964
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.673913043478261e-06,
+      "loss": 1.7557,
+      "step": 965
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.672463768115942e-06,
+      "loss": 1.7439,
+      "step": 966
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.671014492753625e-06,
+      "loss": 1.3189,
+      "step": 967
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.669565217391304e-06,
+      "loss": 0.9965,
+      "step": 968
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.668115942028987e-06,
+      "loss": 1.6572,
+      "step": 969
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.3122,
+      "step": 970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.665217391304349e-06,
+      "loss": 1.9254,
+      "step": 971
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.66376811594203e-06,
+      "loss": 2.1671,
+      "step": 972
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.66231884057971e-06,
+      "loss": 1.5877,
+      "step": 973
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.660869565217392e-06,
+      "loss": 1.231,
+      "step": 974
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.659420289855073e-06,
+      "loss": 1.7459,
+      "step": 975
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.657971014492754e-06,
+      "loss": 1.7577,
+      "step": 976
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.656521739130437e-06,
+      "loss": 1.2372,
+      "step": 977
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.655072463768118e-06,
+      "loss": 1.8361,
+      "step": 978
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.653623188405797e-06,
+      "loss": 1.2816,
+      "step": 979
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.65217391304348e-06,
+      "loss": 1.5858,
+      "step": 980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.65072463768116e-06,
+      "loss": 1.301,
+      "step": 981
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.649275362318842e-06,
+      "loss": 1.726,
+      "step": 982
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.647826086956523e-06,
+      "loss": 0.6892,
+      "step": 983
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.646376811594204e-06,
+      "loss": 1.7132,
+      "step": 984
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.644927536231885e-06,
+      "loss": 1.6219,
+      "step": 985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.643478260869566e-06,
+      "loss": 1.6076,
+      "step": 986
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.642028985507247e-06,
+      "loss": 2.1782,
+      "step": 987
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.64057971014493e-06,
+      "loss": 1.8341,
+      "step": 988
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.63913043478261e-06,
+      "loss": 1.5936,
+      "step": 989
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.63768115942029e-06,
+      "loss": 1.7613,
+      "step": 990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.636231884057972e-06,
+      "loss": 1.6068,
+      "step": 991
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.634782608695653e-06,
+      "loss": 1.2279,
+      "step": 992
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.633333333333334e-06,
+      "loss": 1.6399,
+      "step": 993
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.631884057971015e-06,
+      "loss": 1.2478,
+      "step": 994
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.630434782608696e-06,
+      "loss": 2.0538,
+      "step": 995
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.628985507246377e-06,
+      "loss": 1.4712,
+      "step": 996
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.627536231884058e-06,
+      "loss": 1.2299,
+      "step": 997
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.62608695652174e-06,
+      "loss": 1.6196,
+      "step": 998
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.624637681159422e-06,
+      "loss": 1.6362,
+      "step": 999
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.623188405797103e-06,
+      "loss": 1.5785,
+      "step": 1000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.621739130434782e-06,
+      "loss": 1.3924,
+      "step": 1001
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.620289855072465e-06,
+      "loss": 1.6185,
+      "step": 1002
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.618840579710146e-06,
+      "loss": 2.1901,
+      "step": 1003
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.617391304347827e-06,
+      "loss": 2.0595,
+      "step": 1004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.615942028985508e-06,
+      "loss": 1.6766,
+      "step": 1005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.614492753623189e-06,
+      "loss": 1.6389,
+      "step": 1006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.61304347826087e-06,
+      "loss": 1.7314,
+      "step": 1007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.611594202898551e-06,
+      "loss": 1.6674,
+      "step": 1008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.610144927536232e-06,
+      "loss": 1.3401,
+      "step": 1009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.608695652173915e-06,
+      "loss": 1.1663,
+      "step": 1010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.607246376811594e-06,
+      "loss": 1.6514,
+      "step": 1011
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.605797101449277e-06,
+      "loss": 1.5315,
+      "step": 1012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.604347826086958e-06,
+      "loss": 1.5383,
+      "step": 1013
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.602898550724639e-06,
+      "loss": 1.3553,
+      "step": 1014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.60144927536232e-06,
+      "loss": 1.5457,
+      "step": 1015
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.6e-06,
+      "loss": 1.3184,
+      "step": 1016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.598550724637682e-06,
+      "loss": 1.2739,
+      "step": 1017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.597101449275363e-06,
+      "loss": 1.4994,
+      "step": 1018
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.595652173913044e-06,
+      "loss": 2.101,
+      "step": 1019
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.594202898550725e-06,
+      "loss": 0.9226,
+      "step": 1020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.592753623188407e-06,
+      "loss": 2.0916,
+      "step": 1021
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.591304347826087e-06,
+      "loss": 1.239,
+      "step": 1022
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.58985507246377e-06,
+      "loss": 1.3842,
+      "step": 1023
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.58840579710145e-06,
+      "loss": 2.0445,
+      "step": 1024
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.586956521739131e-06,
+      "loss": 1.5843,
+      "step": 1025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.585507246376812e-06,
+      "loss": 1.6085,
+      "step": 1026
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.584057971014493e-06,
+      "loss": 1.5612,
+      "step": 1027
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.582608695652174e-06,
+      "loss": 1.7083,
+      "step": 1028
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.581159420289857e-06,
+      "loss": 1.7694,
+      "step": 1029
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.579710144927536e-06,
+      "loss": 1.5726,
+      "step": 1030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.578260869565217e-06,
+      "loss": 1.2603,
+      "step": 1031
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.5768115942029e-06,
+      "loss": 1.577,
+      "step": 1032
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.57536231884058e-06,
+      "loss": 1.7439,
+      "step": 1033
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.573913043478262e-06,
+      "loss": 1.8478,
+      "step": 1034
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.572463768115943e-06,
+      "loss": 1.9454,
+      "step": 1035
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.571014492753624e-06,
+      "loss": 2.0301,
+      "step": 1036
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.569565217391305e-06,
+      "loss": 1.3032,
+      "step": 1037
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.568115942028986e-06,
+      "loss": 1.5631,
+      "step": 1038
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.566666666666667e-06,
+      "loss": 1.4493,
+      "step": 1039
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.56521739130435e-06,
+      "loss": 1.5062,
+      "step": 1040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.563768115942029e-06,
+      "loss": 1.1835,
+      "step": 1041
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.56231884057971e-06,
+      "loss": 1.5594,
+      "step": 1042
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.560869565217393e-06,
+      "loss": 1.574,
+      "step": 1043
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.559420289855072e-06,
+      "loss": 1.417,
+      "step": 1044
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.557971014492755e-06,
+      "loss": 1.6158,
+      "step": 1045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.556521739130436e-06,
+      "loss": 1.6089,
+      "step": 1046
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.555072463768117e-06,
+      "loss": 1.5959,
+      "step": 1047
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.553623188405798e-06,
+      "loss": 1.1365,
+      "step": 1048
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.552173913043479e-06,
+      "loss": 1.5452,
+      "step": 1049
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.55072463768116e-06,
+      "loss": 1.6091,
+      "step": 1050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.549275362318842e-06,
+      "loss": 1.5758,
+      "step": 1051
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.547826086956522e-06,
+      "loss": 1.9756,
+      "step": 1052
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.546376811594203e-06,
+      "loss": 1.2248,
+      "step": 1053
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.544927536231886e-06,
+      "loss": 1.6829,
+      "step": 1054
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.543478260869565e-06,
+      "loss": 1.5365,
+      "step": 1055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.542028985507248e-06,
+      "loss": 1.2465,
+      "step": 1056
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.540579710144929e-06,
+      "loss": 1.1443,
+      "step": 1057
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.53913043478261e-06,
+      "loss": 1.5799,
+      "step": 1058
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.53768115942029e-06,
+      "loss": 1.4459,
+      "step": 1059
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.536231884057972e-06,
+      "loss": 1.6145,
+      "step": 1060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.534782608695653e-06,
+      "loss": 1.2652,
+      "step": 1061
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.5374,
+      "step": 1062
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.531884057971015e-06,
+      "loss": 1.2796,
+      "step": 1063
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.530434782608697e-06,
+      "loss": 1.4143,
+      "step": 1064
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.528985507246378e-06,
+      "loss": 1.8644,
+      "step": 1065
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.527536231884058e-06,
+      "loss": 1.5324,
+      "step": 1066
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.52608695652174e-06,
+      "loss": 1.6757,
+      "step": 1067
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.524637681159421e-06,
+      "loss": 1.7388,
+      "step": 1068
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.523188405797102e-06,
+      "loss": 1.7442,
+      "step": 1069
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.521739130434783e-06,
+      "loss": 1.6181,
+      "step": 1070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.520289855072464e-06,
+      "loss": 1.8289,
+      "step": 1071
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.518840579710145e-06,
+      "loss": 0.8582,
+      "step": 1072
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.517391304347826e-06,
+      "loss": 1.9393,
+      "step": 1073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.515942028985507e-06,
+      "loss": 0.9032,
+      "step": 1074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.51449275362319e-06,
+      "loss": 1.2197,
+      "step": 1075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.513043478260871e-06,
+      "loss": 1.1744,
+      "step": 1076
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.51159420289855e-06,
+      "loss": 2.1565,
+      "step": 1077
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.510144927536233e-06,
+      "loss": 1.5228,
+      "step": 1078
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.508695652173914e-06,
+      "loss": 1.3355,
+      "step": 1079
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.507246376811595e-06,
+      "loss": 1.4649,
+      "step": 1080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.505797101449276e-06,
+      "loss": 1.7083,
+      "step": 1081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.504347826086957e-06,
+      "loss": 1.11,
+      "step": 1082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.502898550724638e-06,
+      "loss": 1.6182,
+      "step": 1083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.501449275362319e-06,
+      "loss": 1.0143,
+      "step": 1084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.5e-06,
+      "loss": 1.5621,
+      "step": 1085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.498550724637683e-06,
+      "loss": 1.7645,
+      "step": 1086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.497101449275364e-06,
+      "loss": 1.5755,
+      "step": 1087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.495652173913043e-06,
+      "loss": 1.2481,
+      "step": 1088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.494202898550726e-06,
+      "loss": 1.3187,
+      "step": 1089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.492753623188407e-06,
+      "loss": 1.4351,
+      "step": 1090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.491304347826088e-06,
+      "loss": 1.7358,
+      "step": 1091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.489855072463769e-06,
+      "loss": 1.5221,
+      "step": 1092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.48840579710145e-06,
+      "loss": 1.1428,
+      "step": 1093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.48695652173913e-06,
+      "loss": 1.5474,
+      "step": 1094
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.485507246376812e-06,
+      "loss": 1.2951,
+      "step": 1095
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.484057971014493e-06,
+      "loss": 1.4674,
+      "step": 1096
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.482608695652175e-06,
+      "loss": 1.576,
+      "step": 1097
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.481159420289856e-06,
+      "loss": 1.4011,
+      "step": 1098
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.479710144927537e-06,
+      "loss": 1.9923,
+      "step": 1099
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.478260869565218e-06,
+      "loss": 1.6475,
+      "step": 1100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.4768115942029e-06,
+      "loss": 1.6367,
+      "step": 1101
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.47536231884058e-06,
+      "loss": 1.5326,
+      "step": 1102
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.473913043478261e-06,
+      "loss": 1.3325,
+      "step": 1103
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.472463768115942e-06,
+      "loss": 1.2671,
+      "step": 1104
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.471014492753623e-06,
+      "loss": 1.4676,
+      "step": 1105
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.469565217391304e-06,
+      "loss": 2.0244,
+      "step": 1106
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.468115942028985e-06,
+      "loss": 1.5288,
+      "step": 1107
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.466666666666668e-06,
+      "loss": 1.3624,
+      "step": 1108
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.465217391304349e-06,
+      "loss": 1.842,
+      "step": 1109
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.46376811594203e-06,
+      "loss": 1.6999,
+      "step": 1110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.462318840579711e-06,
+      "loss": 1.5333,
+      "step": 1111
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.460869565217392e-06,
+      "loss": 1.0082,
+      "step": 1112
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.459420289855073e-06,
+      "loss": 1.8825,
+      "step": 1113
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.457971014492754e-06,
+      "loss": 1.1323,
+      "step": 1114
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.456521739130435e-06,
+      "loss": 1.7599,
+      "step": 1115
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.455072463768118e-06,
+      "loss": 1.7637,
+      "step": 1116
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.453623188405797e-06,
+      "loss": 1.8934,
+      "step": 1117
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.452173913043478e-06,
+      "loss": 1.5562,
+      "step": 1118
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.45072463768116e-06,
+      "loss": 1.8159,
+      "step": 1119
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.449275362318842e-06,
+      "loss": 1.6183,
+      "step": 1120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.447826086956523e-06,
+      "loss": 1.2808,
+      "step": 1121
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.446376811594204e-06,
+      "loss": 1.4768,
+      "step": 1122
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.444927536231885e-06,
+      "loss": 1.2477,
+      "step": 1123
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.443478260869566e-06,
+      "loss": 0.9047,
+      "step": 1124
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.442028985507247e-06,
+      "loss": 1.0423,
+      "step": 1125
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.440579710144928e-06,
+      "loss": 1.7144,
+      "step": 1126
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.43913043478261e-06,
+      "loss": 1.5293,
+      "step": 1127
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.43768115942029e-06,
+      "loss": 1.0835,
+      "step": 1128
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.43623188405797e-06,
+      "loss": 1.947,
+      "step": 1129
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.434782608695653e-06,
+      "loss": 1.6353,
+      "step": 1130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.433333333333334e-06,
+      "loss": 1.5546,
+      "step": 1131
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.431884057971015e-06,
+      "loss": 1.1484,
+      "step": 1132
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.430434782608696e-06,
+      "loss": 1.3999,
+      "step": 1133
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.428985507246377e-06,
+      "loss": 1.8009,
+      "step": 1134
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.427536231884058e-06,
+      "loss": 1.7983,
+      "step": 1135
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.42608695652174e-06,
+      "loss": 1.5496,
+      "step": 1136
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.42463768115942e-06,
+      "loss": 1.7295,
+      "step": 1137
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.423188405797103e-06,
+      "loss": 1.2769,
+      "step": 1138
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.421739130434782e-06,
+      "loss": 1.1207,
+      "step": 1139
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.420289855072463e-06,
+      "loss": 1.9511,
+      "step": 1140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.418840579710146e-06,
+      "loss": 1.2796,
+      "step": 1141
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.417391304347827e-06,
+      "loss": 1.7075,
+      "step": 1142
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.415942028985508e-06,
+      "loss": 1.5392,
+      "step": 1143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.41449275362319e-06,
+      "loss": 1.208,
+      "step": 1144
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.41304347826087e-06,
+      "loss": 1.2763,
+      "step": 1145
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.411594202898551e-06,
+      "loss": 1.1577,
+      "step": 1146
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.410144927536232e-06,
+      "loss": 1.5211,
+      "step": 1147
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.408695652173913e-06,
+      "loss": 1.7107,
+      "step": 1148
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.407246376811596e-06,
+      "loss": 1.6298,
+      "step": 1149
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.405797101449275e-06,
+      "loss": 1.4763,
+      "step": 1150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.404347826086958e-06,
+      "loss": 1.9542,
+      "step": 1151
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.402898550724639e-06,
+      "loss": 1.3417,
+      "step": 1152
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.40144927536232e-06,
+      "loss": 1.4074,
+      "step": 1153
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.1268,
+      "step": 1154
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.398550724637682e-06,
+      "loss": 1.5713,
+      "step": 1155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.397101449275363e-06,
+      "loss": 1.2304,
+      "step": 1156
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.395652173913044e-06,
+      "loss": 1.2647,
+      "step": 1157
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.394202898550725e-06,
+      "loss": 1.1173,
+      "step": 1158
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.392753623188406e-06,
+      "loss": 1.566,
+      "step": 1159
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.391304347826089e-06,
+      "loss": 1.5604,
+      "step": 1160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.389855072463768e-06,
+      "loss": 1.2965,
+      "step": 1161
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.38840579710145e-06,
+      "loss": 1.6188,
+      "step": 1162
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.386956521739132e-06,
+      "loss": 0.9827,
+      "step": 1163
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.385507246376813e-06,
+      "loss": 1.6553,
+      "step": 1164
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.384057971014494e-06,
+      "loss": 1.1789,
+      "step": 1165
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.382608695652175e-06,
+      "loss": 1.5063,
+      "step": 1166
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.381159420289856e-06,
+      "loss": 0.8784,
+      "step": 1167
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.379710144927537e-06,
+      "loss": 1.9971,
+      "step": 1168
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.378260869565218e-06,
+      "loss": 1.0741,
+      "step": 1169
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.376811594202899e-06,
+      "loss": 1.8786,
+      "step": 1170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.375362318840581e-06,
+      "loss": 1.539,
+      "step": 1171
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.37391304347826e-06,
+      "loss": 1.6553,
+      "step": 1172
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.372463768115943e-06,
+      "loss": 1.4476,
+      "step": 1173
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.371014492753624e-06,
+      "loss": 1.9176,
+      "step": 1174
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.369565217391305e-06,
+      "loss": 1.6476,
+      "step": 1175
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.368115942028986e-06,
+      "loss": 1.5643,
+      "step": 1176
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.366666666666667e-06,
+      "loss": 1.5905,
+      "step": 1177
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.365217391304348e-06,
+      "loss": 0.8998,
+      "step": 1178
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36376811594203e-06,
+      "loss": 1.283,
+      "step": 1179
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36231884057971e-06,
+      "loss": 1.3829,
+      "step": 1180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.360869565217391e-06,
+      "loss": 1.2278,
+      "step": 1181
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359420289855074e-06,
+      "loss": 1.2628,
+      "step": 1182
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.357971014492753e-06,
+      "loss": 1.5743,
+      "step": 1183
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356521739130436e-06,
+      "loss": 1.4387,
+      "step": 1184
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.355072463768117e-06,
+      "loss": 1.6435,
+      "step": 1185
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.353623188405798e-06,
+      "loss": 1.7863,
+      "step": 1186
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.352173913043479e-06,
+      "loss": 1.4813,
+      "step": 1187
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.35072463768116e-06,
+      "loss": 2.004,
+      "step": 1188
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.349275362318841e-06,
+      "loss": 1.3384,
+      "step": 1189
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.347826086956522e-06,
+      "loss": 1.2469,
+      "step": 1190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.346376811594203e-06,
+      "loss": 1.3411,
+      "step": 1191
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.344927536231886e-06,
+      "loss": 0.9357,
+      "step": 1192
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.343478260869567e-06,
+      "loss": 1.323,
+      "step": 1193
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.342028985507246e-06,
+      "loss": 1.2898,
+      "step": 1194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.340579710144929e-06,
+      "loss": 1.3359,
+      "step": 1195
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.33913043478261e-06,
+      "loss": 1.6105,
+      "step": 1196
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.33768115942029e-06,
+      "loss": 1.3856,
+      "step": 1197
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.336231884057972e-06,
+      "loss": 1.677,
+      "step": 1198
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.334782608695653e-06,
+      "loss": 1.739,
+      "step": 1199
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.2557,
+      "step": 1200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.331884057971015e-06,
+      "loss": 1.6529,
+      "step": 1201
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.330434782608696e-06,
+      "loss": 1.9315,
+      "step": 1202
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.328985507246378e-06,
+      "loss": 1.2352,
+      "step": 1203
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.32753623188406e-06,
+      "loss": 1.5359,
+      "step": 1204
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.326086956521739e-06,
+      "loss": 1.6702,
+      "step": 1205
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.324637681159421e-06,
+      "loss": 1.4797,
+      "step": 1206
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.323188405797102e-06,
+      "loss": 1.3935,
+      "step": 1207
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.321739130434783e-06,
+      "loss": 1.6322,
+      "step": 1208
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.320289855072464e-06,
+      "loss": 0.6089,
+      "step": 1209
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.318840579710145e-06,
+      "loss": 1.6063,
+      "step": 1210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.317391304347826e-06,
+      "loss": 0.8827,
+      "step": 1211
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.315942028985507e-06,
+      "loss": 1.2968,
+      "step": 1212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.314492753623188e-06,
+      "loss": 1.2662,
+      "step": 1213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.313043478260871e-06,
+      "loss": 1.1847,
+      "step": 1214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.311594202898552e-06,
+      "loss": 1.2261,
+      "step": 1215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.310144927536231e-06,
+      "loss": 1.6993,
+      "step": 1216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.308695652173914e-06,
+      "loss": 1.4307,
+      "step": 1217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.307246376811595e-06,
+      "loss": 1.2382,
+      "step": 1218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.305797101449276e-06,
+      "loss": 1.1909,
+      "step": 1219
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.304347826086957e-06,
+      "loss": 1.2404,
+      "step": 1220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.302898550724638e-06,
+      "loss": 2.0244,
+      "step": 1221
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.301449275362319e-06,
+      "loss": 1.1718,
+      "step": 1222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.3e-06,
+      "loss": 1.5409,
+      "step": 1223
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.298550724637681e-06,
+      "loss": 1.202,
+      "step": 1224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.297101449275364e-06,
+      "loss": 1.5235,
+      "step": 1225
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.295652173913045e-06,
+      "loss": 1.4762,
+      "step": 1226
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.294202898550726e-06,
+      "loss": 0.8406,
+      "step": 1227
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.292753623188407e-06,
+      "loss": 1.2052,
+      "step": 1228
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.291304347826088e-06,
+      "loss": 1.4428,
+      "step": 1229
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.289855072463769e-06,
+      "loss": 1.5495,
+      "step": 1230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.28840579710145e-06,
+      "loss": 1.3398,
+      "step": 1231
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.286956521739131e-06,
+      "loss": 1.3374,
+      "step": 1232
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.285507246376812e-06,
+      "loss": 1.4867,
+      "step": 1233
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.284057971014493e-06,
+      "loss": 0.9613,
+      "step": 1234
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.282608695652174e-06,
+      "loss": 1.1793,
+      "step": 1235
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.281159420289857e-06,
+      "loss": 1.3626,
+      "step": 1236
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.279710144927538e-06,
+      "loss": 1.4948,
+      "step": 1237
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.278260869565219e-06,
+      "loss": 1.6337,
+      "step": 1238
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.2768115942029e-06,
+      "loss": 1.596,
+      "step": 1239
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.27536231884058e-06,
+      "loss": 1.4803,
+      "step": 1240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.273913043478262e-06,
+      "loss": 1.2177,
+      "step": 1241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.272463768115943e-06,
+      "loss": 1.1992,
+      "step": 1242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.271014492753624e-06,
+      "loss": 1.2049,
+      "step": 1243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.269565217391306e-06,
+      "loss": 1.8858,
+      "step": 1244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.268115942028986e-06,
+      "loss": 1.1579,
+      "step": 1245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 2.0264,
+      "step": 1246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.26521739130435e-06,
+      "loss": 1.5869,
+      "step": 1247
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.26376811594203e-06,
+      "loss": 2.0683,
+      "step": 1248
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.262318840579711e-06,
+      "loss": 1.3512,
+      "step": 1249
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 2.113,
+      "step": 1250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.259420289855073e-06,
+      "loss": 2.0507,
+      "step": 1251
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.257971014492754e-06,
+      "loss": 1.6424,
+      "step": 1252
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.256521739130435e-06,
+      "loss": 1.5512,
+      "step": 1253
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.255072463768116e-06,
+      "loss": 1.4054,
+      "step": 1254
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.253623188405799e-06,
+      "loss": 1.0905,
+      "step": 1255
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.252173913043478e-06,
+      "loss": 1.171,
+      "step": 1256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.25072463768116e-06,
+      "loss": 1.6301,
+      "step": 1257
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.249275362318842e-06,
+      "loss": 1.4664,
+      "step": 1258
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.247826086956521e-06,
+      "loss": 1.5414,
+      "step": 1259
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.246376811594204e-06,
+      "loss": 1.5328,
+      "step": 1260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.244927536231885e-06,
+      "loss": 1.4568,
+      "step": 1261
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.243478260869566e-06,
+      "loss": 0.6284,
+      "step": 1262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.242028985507247e-06,
+      "loss": 1.4729,
+      "step": 1263
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.240579710144928e-06,
+      "loss": 1.4182,
+      "step": 1264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.239130434782609e-06,
+      "loss": 1.553,
+      "step": 1265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.237681159420292e-06,
+      "loss": 1.5542,
+      "step": 1266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.236231884057971e-06,
+      "loss": 1.2875,
+      "step": 1267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.234782608695652e-06,
+      "loss": 1.6609,
+      "step": 1268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.233333333333335e-06,
+      "loss": 1.6561,
+      "step": 1269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.231884057971014e-06,
+      "loss": 1.0855,
+      "step": 1270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.230434782608697e-06,
+      "loss": 1.2429,
+      "step": 1271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.228985507246378e-06,
+      "loss": 0.993,
+      "step": 1272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.227536231884059e-06,
+      "loss": 1.189,
+      "step": 1273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.22608695652174e-06,
+      "loss": 1.5407,
+      "step": 1274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.22463768115942e-06,
+      "loss": 1.5433,
+      "step": 1275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.223188405797102e-06,
+      "loss": 1.8614,
+      "step": 1276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.221739130434784e-06,
+      "loss": 1.4324,
+      "step": 1277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.220289855072464e-06,
+      "loss": 1.5163,
+      "step": 1278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.218840579710146e-06,
+      "loss": 1.2308,
+      "step": 1279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.217391304347827e-06,
+      "loss": 1.1901,
+      "step": 1280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.215942028985507e-06,
+      "loss": 1.95,
+      "step": 1281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.21449275362319e-06,
+      "loss": 1.7723,
+      "step": 1282
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.21304347826087e-06,
+      "loss": 2.0437,
+      "step": 1283
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.211594202898551e-06,
+      "loss": 1.8459,
+      "step": 1284
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.210144927536232e-06,
+      "loss": 0.9564,
+      "step": 1285
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.208695652173913e-06,
+      "loss": 1.5332,
+      "step": 1286
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.207246376811594e-06,
+      "loss": 1.0713,
+      "step": 1287
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.205797101449277e-06,
+      "loss": 1.7146,
+      "step": 1288
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.204347826086956e-06,
+      "loss": 1.2178,
+      "step": 1289
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.202898550724639e-06,
+      "loss": 1.5325,
+      "step": 1290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.20144927536232e-06,
+      "loss": 1.1564,
+      "step": 1291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.2e-06,
+      "loss": 1.6885,
+      "step": 1292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.198550724637682e-06,
+      "loss": 1.576,
+      "step": 1293
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.197101449275363e-06,
+      "loss": 1.5888,
+      "step": 1294
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.195652173913044e-06,
+      "loss": 1.4872,
+      "step": 1295
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.194202898550725e-06,
+      "loss": 1.5687,
+      "step": 1296
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.192753623188406e-06,
+      "loss": 1.1956,
+      "step": 1297
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.191304347826087e-06,
+      "loss": 1.2463,
+      "step": 1298
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.18985507246377e-06,
+      "loss": 1.7814,
+      "step": 1299
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.188405797101449e-06,
+      "loss": 1.3773,
+      "step": 1300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.186956521739132e-06,
+      "loss": 1.3483,
+      "step": 1301
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.185507246376813e-06,
+      "loss": 1.4455,
+      "step": 1302
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.184057971014492e-06,
+      "loss": 1.4734,
+      "step": 1303
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.182608695652175e-06,
+      "loss": 1.1343,
+      "step": 1304
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.181159420289856e-06,
+      "loss": 0.5646,
+      "step": 1305
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.179710144927537e-06,
+      "loss": 1.1056,
+      "step": 1306
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.178260869565218e-06,
+      "loss": 1.7115,
+      "step": 1307
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.176811594202899e-06,
+      "loss": 1.2889,
+      "step": 1308
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.17536231884058e-06,
+      "loss": 1.5364,
+      "step": 1309
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.173913043478263e-06,
+      "loss": 1.1741,
+      "step": 1310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.172463768115942e-06,
+      "loss": 1.8626,
+      "step": 1311
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.171014492753625e-06,
+      "loss": 1.1428,
+      "step": 1312
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.169565217391306e-06,
+      "loss": 1.0564,
+      "step": 1313
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.168115942028987e-06,
+      "loss": 1.1726,
+      "step": 1314
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 1.3215,
+      "step": 1315
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.165217391304349e-06,
+      "loss": 1.7501,
+      "step": 1316
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.16376811594203e-06,
+      "loss": 1.8432,
+      "step": 1317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.16231884057971e-06,
+      "loss": 0.6948,
+      "step": 1318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.160869565217392e-06,
+      "loss": 1.2027,
+      "step": 1319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.159420289855074e-06,
+      "loss": 0.8697,
+      "step": 1320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.157971014492754e-06,
+      "loss": 1.0386,
+      "step": 1321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.156521739130435e-06,
+      "loss": 1.1984,
+      "step": 1322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.155072463768117e-06,
+      "loss": 1.2252,
+      "step": 1323
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.153623188405798e-06,
+      "loss": 1.0652,
+      "step": 1324
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.15217391304348e-06,
+      "loss": 1.5158,
+      "step": 1325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.15072463768116e-06,
+      "loss": 1.2029,
+      "step": 1326
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.149275362318841e-06,
+      "loss": 1.3112,
+      "step": 1327
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.147826086956522e-06,
+      "loss": 1.5208,
+      "step": 1328
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.146376811594203e-06,
+      "loss": 1.0698,
+      "step": 1329
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.144927536231884e-06,
+      "loss": 1.4982,
+      "step": 1330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.143478260869567e-06,
+      "loss": 0.7541,
+      "step": 1331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.142028985507246e-06,
+      "loss": 1.3204,
+      "step": 1332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.140579710144927e-06,
+      "loss": 1.8861,
+      "step": 1333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.13913043478261e-06,
+      "loss": 1.2033,
+      "step": 1334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.137681159420291e-06,
+      "loss": 1.6387,
+      "step": 1335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.136231884057972e-06,
+      "loss": 1.647,
+      "step": 1336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.134782608695653e-06,
+      "loss": 1.3311,
+      "step": 1337
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 1.8995,
+      "step": 1338
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.131884057971015e-06,
+      "loss": 1.509,
+      "step": 1339
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.130434782608696e-06,
+      "loss": 0.919,
+      "step": 1340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.128985507246377e-06,
+      "loss": 1.0092,
+      "step": 1341
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.12753623188406e-06,
+      "loss": 1.5937,
+      "step": 1342
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.126086956521739e-06,
+      "loss": 1.1662,
+      "step": 1343
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.12463768115942e-06,
+      "loss": 1.381,
+      "step": 1344
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.123188405797103e-06,
+      "loss": 1.9683,
+      "step": 1345
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.121739130434784e-06,
+      "loss": 0.7581,
+      "step": 1346
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.120289855072465e-06,
+      "loss": 1.586,
+      "step": 1347
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.118840579710146e-06,
+      "loss": 1.9837,
+      "step": 1348
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.117391304347827e-06,
+      "loss": 0.979,
+      "step": 1349
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.115942028985508e-06,
+      "loss": 1.3359,
+      "step": 1350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.114492753623189e-06,
+      "loss": 1.3487,
+      "step": 1351
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.11304347826087e-06,
+      "loss": 1.678,
+      "step": 1352
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.111594202898552e-06,
+      "loss": 1.2371,
+      "step": 1353
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.110144927536232e-06,
+      "loss": 1.5089,
+      "step": 1354
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.108695652173914e-06,
+      "loss": 1.4335,
+      "step": 1355
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.107246376811595e-06,
+      "loss": 1.2756,
+      "step": 1356
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.105797101449276e-06,
+      "loss": 1.5723,
+      "step": 1357
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.104347826086957e-06,
+      "loss": 1.7552,
+      "step": 1358
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.102898550724638e-06,
+      "loss": 1.7005,
+      "step": 1359
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.10144927536232e-06,
+      "loss": 1.0595,
+      "step": 1360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.1e-06,
+      "loss": 1.1714,
+      "step": 1361
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.098550724637681e-06,
+      "loss": 1.1455,
+      "step": 1362
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.097101449275362e-06,
+      "loss": 1.1986,
+      "step": 1363
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.095652173913045e-06,
+      "loss": 1.5938,
+      "step": 1364
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.094202898550724e-06,
+      "loss": 2.2176,
+      "step": 1365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.092753623188407e-06,
+      "loss": 1.4507,
+      "step": 1366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.091304347826088e-06,
+      "loss": 1.5786,
+      "step": 1367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.089855072463769e-06,
+      "loss": 0.476,
+      "step": 1368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.08840579710145e-06,
+      "loss": 1.4674,
+      "step": 1369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.086956521739131e-06,
+      "loss": 1.42,
+      "step": 1370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.085507246376812e-06,
+      "loss": 1.7337,
+      "step": 1371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.084057971014495e-06,
+      "loss": 1.2889,
+      "step": 1372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.082608695652174e-06,
+      "loss": 1.5921,
+      "step": 1373
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.081159420289855e-06,
+      "loss": 1.9598,
+      "step": 1374
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.079710144927538e-06,
+      "loss": 1.5455,
+      "step": 1375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.078260869565217e-06,
+      "loss": 1.1318,
+      "step": 1376
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.0768115942029e-06,
+      "loss": 1.5546,
+      "step": 1377
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.07536231884058e-06,
+      "loss": 1.291,
+      "step": 1378
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.073913043478262e-06,
+      "loss": 1.3006,
+      "step": 1379
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.072463768115943e-06,
+      "loss": 0.9576,
+      "step": 1380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.071014492753624e-06,
+      "loss": 1.3393,
+      "step": 1381
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.069565217391305e-06,
+      "loss": 1.3858,
+      "step": 1382
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.068115942028986e-06,
+      "loss": 1.7063,
+      "step": 1383
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.066666666666667e-06,
+      "loss": 0.9991,
+      "step": 1384
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.065217391304348e-06,
+      "loss": 2.0504,
+      "step": 1385
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.06376811594203e-06,
+      "loss": 1.3623,
+      "step": 1386
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.06231884057971e-06,
+      "loss": 1.2719,
+      "step": 1387
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.060869565217392e-06,
+      "loss": 1.5965,
+      "step": 1388
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.059420289855073e-06,
+      "loss": 1.6982,
+      "step": 1389
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.057971014492754e-06,
+      "loss": 1.2451,
+      "step": 1390
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.4676258992805755,
+      "eval_loss": 1.2769397497177124,
+      "eval_roc_auc": 0.8359559009185776,
+      "eval_runtime": 85.4787,
+      "eval_samples_per_second": 8.131,
+      "eval_steps_per_second": 2.036,
+      "step": 1390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.056521739130435e-06,
+      "loss": 1.5127,
+      "step": 1391
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.055072463768117e-06,
+      "loss": 1.1711,
+      "step": 1392
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.053623188405798e-06,
+      "loss": 0.9796,
+      "step": 1393
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.052173913043479e-06,
+      "loss": 1.4316,
+      "step": 1394
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.05072463768116e-06,
+      "loss": 1.6129,
+      "step": 1395
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.04927536231884e-06,
+      "loss": 1.1991,
+      "step": 1396
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.047826086956523e-06,
+      "loss": 1.0694,
+      "step": 1397
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.046376811594203e-06,
+      "loss": 1.2941,
+      "step": 1398
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.044927536231885e-06,
+      "loss": 1.235,
+      "step": 1399
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.043478260869566e-06,
+      "loss": 0.6247,
+      "step": 1400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.042028985507247e-06,
+      "loss": 1.4204,
+      "step": 1401
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.040579710144928e-06,
+      "loss": 1.1813,
+      "step": 1402
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.03913043478261e-06,
+      "loss": 0.9655,
+      "step": 1403
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.03768115942029e-06,
+      "loss": 1.7045,
+      "step": 1404
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.036231884057971e-06,
+      "loss": 1.6167,
+      "step": 1405
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.034782608695652e-06,
+      "loss": 1.5009,
+      "step": 1406
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.033333333333335e-06,
+      "loss": 1.7278,
+      "step": 1407
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.031884057971016e-06,
+      "loss": 1.3106,
+      "step": 1408
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.030434782608695e-06,
+      "loss": 1.1879,
+      "step": 1409
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.028985507246378e-06,
+      "loss": 1.5061,
+      "step": 1410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.027536231884059e-06,
+      "loss": 1.2376,
+      "step": 1411
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.02608695652174e-06,
+      "loss": 0.8893,
+      "step": 1412
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.024637681159421e-06,
+      "loss": 1.1518,
+      "step": 1413
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.023188405797102e-06,
+      "loss": 1.3193,
+      "step": 1414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.021739130434783e-06,
+      "loss": 1.2978,
+      "step": 1415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.020289855072464e-06,
+      "loss": 1.1788,
+      "step": 1416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.018840579710145e-06,
+      "loss": 1.126,
+      "step": 1417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.017391304347828e-06,
+      "loss": 1.3164,
+      "step": 1418
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.015942028985509e-06,
+      "loss": 1.1014,
+      "step": 1419
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.014492753623188e-06,
+      "loss": 1.4145,
+      "step": 1420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.01304347826087e-06,
+      "loss": 0.823,
+      "step": 1421
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.011594202898552e-06,
+      "loss": 1.2001,
+      "step": 1422
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.010144927536233e-06,
+      "loss": 1.2046,
+      "step": 1423
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.008695652173914e-06,
+      "loss": 1.341,
+      "step": 1424
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.007246376811595e-06,
+      "loss": 1.163,
+      "step": 1425
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.005797101449276e-06,
+      "loss": 1.2218,
+      "step": 1426
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.004347826086957e-06,
+      "loss": 1.2342,
+      "step": 1427
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.002898550724638e-06,
+      "loss": 1.5054,
+      "step": 1428
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.00144927536232e-06,
+      "loss": 1.3444,
+      "step": 1429
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8582,
+      "step": 1430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.99855072463768e-06,
+      "loss": 1.1593,
+      "step": 1431
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.997101449275363e-06,
+      "loss": 0.9636,
+      "step": 1432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.995652173913044e-06,
+      "loss": 1.5227,
+      "step": 1433
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.994202898550725e-06,
+      "loss": 1.7717,
+      "step": 1434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.992753623188406e-06,
+      "loss": 1.3049,
+      "step": 1435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.991304347826087e-06,
+      "loss": 1.4721,
+      "step": 1436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.989855072463768e-06,
+      "loss": 1.2361,
+      "step": 1437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.98840579710145e-06,
+      "loss": 1.4296,
+      "step": 1438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.98695652173913e-06,
+      "loss": 1.348,
+      "step": 1439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.985507246376813e-06,
+      "loss": 1.1924,
+      "step": 1440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.984057971014494e-06,
+      "loss": 1.3608,
+      "step": 1441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.982608695652175e-06,
+      "loss": 1.6098,
+      "step": 1442
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.981159420289856e-06,
+      "loss": 1.4396,
+      "step": 1443
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.979710144927537e-06,
+      "loss": 1.4739,
+      "step": 1444
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.978260869565218e-06,
+      "loss": 1.4797,
+      "step": 1445
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.976811594202899e-06,
+      "loss": 1.4102,
+      "step": 1446
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.97536231884058e-06,
+      "loss": 1.4741,
+      "step": 1447
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.973913043478261e-06,
+      "loss": 1.4848,
+      "step": 1448
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.972463768115942e-06,
+      "loss": 1.3846,
+      "step": 1449
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.971014492753623e-06,
+      "loss": 1.1628,
+      "step": 1450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.969565217391306e-06,
+      "loss": 1.4921,
+      "step": 1451
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.968115942028987e-06,
+      "loss": 1.2695,
+      "step": 1452
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.966666666666668e-06,
+      "loss": 1.1064,
+      "step": 1453
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.965217391304349e-06,
+      "loss": 0.8997,
+      "step": 1454
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.96376811594203e-06,
+      "loss": 1.309,
+      "step": 1455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.96231884057971e-06,
+      "loss": 1.8353,
+      "step": 1456
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.960869565217392e-06,
+      "loss": 0.9684,
+      "step": 1457
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.959420289855073e-06,
+      "loss": 1.548,
+      "step": 1458
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.957971014492755e-06,
+      "loss": 1.6796,
+      "step": 1459
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.956521739130435e-06,
+      "loss": 1.7265,
+      "step": 1460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.955072463768116e-06,
+      "loss": 0.9362,
+      "step": 1461
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.953623188405798e-06,
+      "loss": 1.902,
+      "step": 1462
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.95217391304348e-06,
+      "loss": 0.8132,
+      "step": 1463
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.95072463768116e-06,
+      "loss": 0.6427,
+      "step": 1464
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.949275362318841e-06,
+      "loss": 1.3377,
+      "step": 1465
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.947826086956522e-06,
+      "loss": 0.9609,
+      "step": 1466
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.946376811594203e-06,
+      "loss": 1.4052,
+      "step": 1467
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.944927536231884e-06,
+      "loss": 2.0955,
+      "step": 1468
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.943478260869565e-06,
+      "loss": 1.5651,
+      "step": 1469
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.942028985507248e-06,
+      "loss": 1.2632,
+      "step": 1470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.940579710144927e-06,
+      "loss": 1.4993,
+      "step": 1471
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.939130434782608e-06,
+      "loss": 1.094,
+      "step": 1472
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.937681159420291e-06,
+      "loss": 0.887,
+      "step": 1473
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.936231884057972e-06,
+      "loss": 1.5292,
+      "step": 1474
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.934782608695653e-06,
+      "loss": 1.6132,
+      "step": 1475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.933333333333334e-06,
+      "loss": 2.005,
+      "step": 1476
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.931884057971015e-06,
+      "loss": 1.2737,
+      "step": 1477
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.930434782608696e-06,
+      "loss": 1.2534,
+      "step": 1478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.928985507246377e-06,
+      "loss": 1.0552,
+      "step": 1479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.927536231884058e-06,
+      "loss": 0.5767,
+      "step": 1480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.926086956521741e-06,
+      "loss": 1.0828,
+      "step": 1481
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.92463768115942e-06,
+      "loss": 1.0905,
+      "step": 1482
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.923188405797101e-06,
+      "loss": 0.8934,
+      "step": 1483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.921739130434784e-06,
+      "loss": 1.5714,
+      "step": 1484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.920289855072465e-06,
+      "loss": 1.2566,
+      "step": 1485
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.918840579710146e-06,
+      "loss": 1.0516,
+      "step": 1486
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.917391304347827e-06,
+      "loss": 1.2498,
+      "step": 1487
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.915942028985508e-06,
+      "loss": 1.4605,
+      "step": 1488
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.914492753623189e-06,
+      "loss": 1.1475,
+      "step": 1489
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.91304347826087e-06,
+      "loss": 1.4595,
+      "step": 1490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.911594202898551e-06,
+      "loss": 0.9023,
+      "step": 1491
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.910144927536234e-06,
+      "loss": 1.1079,
+      "step": 1492
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.908695652173913e-06,
+      "loss": 1.5325,
+      "step": 1493
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.907246376811596e-06,
+      "loss": 1.0282,
+      "step": 1494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.905797101449277e-06,
+      "loss": 1.2044,
+      "step": 1495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.904347826086958e-06,
+      "loss": 1.415,
+      "step": 1496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.902898550724639e-06,
+      "loss": 0.9598,
+      "step": 1497
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.90144927536232e-06,
+      "loss": 0.7817,
+      "step": 1498
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.9e-06,
+      "loss": 1.8142,
+      "step": 1499
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.898550724637682e-06,
+      "loss": 1.3765,
+      "step": 1500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.897101449275363e-06,
+      "loss": 1.4103,
+      "step": 1501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.895652173913044e-06,
+      "loss": 1.5495,
+      "step": 1502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.894202898550726e-06,
+      "loss": 1.7344,
+      "step": 1503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.892753623188406e-06,
+      "loss": 1.1979,
+      "step": 1504
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.891304347826088e-06,
+      "loss": 1.4298,
+      "step": 1505
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.88985507246377e-06,
+      "loss": 1.5998,
+      "step": 1506
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.888405797101449e-06,
+      "loss": 1.1131,
+      "step": 1507
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.886956521739131e-06,
+      "loss": 1.4125,
+      "step": 1508
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.885507246376812e-06,
+      "loss": 1.4481,
+      "step": 1509
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.884057971014493e-06,
+      "loss": 1.396,
+      "step": 1510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.882608695652174e-06,
+      "loss": 1.1031,
+      "step": 1511
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.881159420289855e-06,
+      "loss": 1.6642,
+      "step": 1512
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.879710144927536e-06,
+      "loss": 0.7832,
+      "step": 1513
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.878260869565219e-06,
+      "loss": 1.4425,
+      "step": 1514
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.876811594202898e-06,
+      "loss": 1.4056,
+      "step": 1515
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.875362318840581e-06,
+      "loss": 0.4806,
+      "step": 1516
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.873913043478262e-06,
+      "loss": 0.6235,
+      "step": 1517
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.872463768115941e-06,
+      "loss": 0.935,
+      "step": 1518
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.871014492753624e-06,
+      "loss": 0.6797,
+      "step": 1519
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.869565217391305e-06,
+      "loss": 0.5923,
+      "step": 1520
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.868115942028986e-06,
+      "loss": 1.1729,
+      "step": 1521
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.2409,
+      "step": 1522
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.865217391304348e-06,
+      "loss": 1.5392,
+      "step": 1523
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.863768115942029e-06,
+      "loss": 2.0757,
+      "step": 1524
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.862318840579712e-06,
+      "loss": 1.5004,
+      "step": 1525
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.860869565217391e-06,
+      "loss": 0.9814,
+      "step": 1526
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.859420289855074e-06,
+      "loss": 1.4758,
+      "step": 1527
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.857971014492755e-06,
+      "loss": 1.511,
+      "step": 1528
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.856521739130436e-06,
+      "loss": 1.6219,
+      "step": 1529
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.855072463768117e-06,
+      "loss": 1.2788,
+      "step": 1530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.853623188405798e-06,
+      "loss": 1.4254,
+      "step": 1531
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.852173913043479e-06,
+      "loss": 1.5492,
+      "step": 1532
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.85072463768116e-06,
+      "loss": 1.1438,
+      "step": 1533
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.84927536231884e-06,
+      "loss": 1.0311,
+      "step": 1534
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.847826086956523e-06,
+      "loss": 1.4931,
+      "step": 1535
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.846376811594204e-06,
+      "loss": 0.8152,
+      "step": 1536
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.844927536231884e-06,
+      "loss": 1.0495,
+      "step": 1537
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.843478260869566e-06,
+      "loss": 0.7774,
+      "step": 1538
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.842028985507247e-06,
+      "loss": 1.2007,
+      "step": 1539
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.840579710144928e-06,
+      "loss": 1.5492,
+      "step": 1540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.83913043478261e-06,
+      "loss": 2.0268,
+      "step": 1541
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.83768115942029e-06,
+      "loss": 1.102,
+      "step": 1542
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.836231884057971e-06,
+      "loss": 1.584,
+      "step": 1543
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.834782608695652e-06,
+      "loss": 1.1719,
+      "step": 1544
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 0.9323,
+      "step": 1545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.831884057971016e-06,
+      "loss": 2.0697,
+      "step": 1546
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.830434782608697e-06,
+      "loss": 1.9412,
+      "step": 1547
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.828985507246376e-06,
+      "loss": 1.6861,
+      "step": 1548
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.827536231884059e-06,
+      "loss": 1.1366,
+      "step": 1549
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 1.1518,
+      "step": 1550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.824637681159421e-06,
+      "loss": 1.8727,
+      "step": 1551
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.823188405797102e-06,
+      "loss": 1.3963,
+      "step": 1552
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.821739130434783e-06,
+      "loss": 1.2129,
+      "step": 1553
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.820289855072464e-06,
+      "loss": 1.1197,
+      "step": 1554
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.818840579710145e-06,
+      "loss": 1.2091,
+      "step": 1555
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.817391304347826e-06,
+      "loss": 1.4713,
+      "step": 1556
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.815942028985509e-06,
+      "loss": 0.8559,
+      "step": 1557
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.81449275362319e-06,
+      "loss": 0.9911,
+      "step": 1558
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.813043478260869e-06,
+      "loss": 1.3791,
+      "step": 1559
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.811594202898552e-06,
+      "loss": 0.8889,
+      "step": 1560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.810144927536233e-06,
+      "loss": 1.0605,
+      "step": 1561
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.808695652173914e-06,
+      "loss": 1.8643,
+      "step": 1562
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.807246376811595e-06,
+      "loss": 1.3198,
+      "step": 1563
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.805797101449276e-06,
+      "loss": 1.4576,
+      "step": 1564
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.804347826086957e-06,
+      "loss": 1.2589,
+      "step": 1565
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.802898550724638e-06,
+      "loss": 1.9371,
+      "step": 1566
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.801449275362319e-06,
+      "loss": 0.9285,
+      "step": 1567
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 1.0488,
+      "step": 1568
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.798550724637681e-06,
+      "loss": 0.7949,
+      "step": 1569
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.797101449275364e-06,
+      "loss": 0.617,
+      "step": 1570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.795652173913045e-06,
+      "loss": 1.7671,
+      "step": 1571
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.794202898550726e-06,
+      "loss": 1.5012,
+      "step": 1572
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.792753623188407e-06,
+      "loss": 1.1855,
+      "step": 1573
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.791304347826088e-06,
+      "loss": 1.6556,
+      "step": 1574
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.789855072463769e-06,
+      "loss": 1.5599,
+      "step": 1575
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.78840579710145e-06,
+      "loss": 1.1782,
+      "step": 1576
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.78695652173913e-06,
+      "loss": 0.9275,
+      "step": 1577
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.785507246376812e-06,
+      "loss": 1.5205,
+      "step": 1578
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.784057971014494e-06,
+      "loss": 1.2501,
+      "step": 1579
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.782608695652174e-06,
+      "loss": 1.1432,
+      "step": 1580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.781159420289856e-06,
+      "loss": 1.4431,
+      "step": 1581
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.779710144927537e-06,
+      "loss": 0.8884,
+      "step": 1582
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.778260869565218e-06,
+      "loss": 0.8783,
+      "step": 1583
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.7768115942029e-06,
+      "loss": 0.713,
+      "step": 1584
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.77536231884058e-06,
+      "loss": 0.9825,
+      "step": 1585
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.773913043478261e-06,
+      "loss": 1.2478,
+      "step": 1586
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.772463768115944e-06,
+      "loss": 1.0341,
+      "step": 1587
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.771014492753623e-06,
+      "loss": 1.5983,
+      "step": 1588
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.769565217391304e-06,
+      "loss": 1.0553,
+      "step": 1589
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.768115942028987e-06,
+      "loss": 0.857,
+      "step": 1590
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.766666666666666e-06,
+      "loss": 1.1763,
+      "step": 1591
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.765217391304349e-06,
+      "loss": 1.4758,
+      "step": 1592
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.76376811594203e-06,
+      "loss": 1.6646,
+      "step": 1593
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.762318840579711e-06,
+      "loss": 1.3264,
+      "step": 1594
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.760869565217392e-06,
+      "loss": 0.7588,
+      "step": 1595
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.759420289855073e-06,
+      "loss": 1.1513,
+      "step": 1596
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.757971014492754e-06,
+      "loss": 1.8423,
+      "step": 1597
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.756521739130437e-06,
+      "loss": 1.4726,
+      "step": 1598
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.755072463768116e-06,
+      "loss": 1.3054,
+      "step": 1599
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.753623188405797e-06,
+      "loss": 1.2112,
+      "step": 1600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.75217391304348e-06,
+      "loss": 1.0785,
+      "step": 1601
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.750724637681159e-06,
+      "loss": 0.945,
+      "step": 1602
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.749275362318842e-06,
+      "loss": 0.6925,
+      "step": 1603
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.747826086956523e-06,
+      "loss": 1.427,
+      "step": 1604
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.746376811594204e-06,
+      "loss": 1.2463,
+      "step": 1605
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.744927536231885e-06,
+      "loss": 1.2485,
+      "step": 1606
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.743478260869566e-06,
+      "loss": 1.3317,
+      "step": 1607
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.742028985507247e-06,
+      "loss": 1.0225,
+      "step": 1608
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.74057971014493e-06,
+      "loss": 1.1226,
+      "step": 1609
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.739130434782609e-06,
+      "loss": 1.2722,
+      "step": 1610
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.73768115942029e-06,
+      "loss": 1.5021,
+      "step": 1611
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.736231884057972e-06,
+      "loss": 1.1766,
+      "step": 1612
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.734782608695652e-06,
+      "loss": 1.2907,
+      "step": 1613
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.9463,
+      "step": 1614
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.731884057971015e-06,
+      "loss": 1.5496,
+      "step": 1615
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.730434782608696e-06,
+      "loss": 0.8126,
+      "step": 1616
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.728985507246377e-06,
+      "loss": 0.5231,
+      "step": 1617
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.727536231884058e-06,
+      "loss": 1.178,
+      "step": 1618
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.72608695652174e-06,
+      "loss": 1.4704,
+      "step": 1619
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.724637681159422e-06,
+      "loss": 1.6318,
+      "step": 1620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.723188405797101e-06,
+      "loss": 1.2274,
+      "step": 1621
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.721739130434784e-06,
+      "loss": 1.0833,
+      "step": 1622
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.720289855072465e-06,
+      "loss": 0.949,
+      "step": 1623
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.718840579710144e-06,
+      "loss": 1.0719,
+      "step": 1624
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.717391304347827e-06,
+      "loss": 1.2627,
+      "step": 1625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.715942028985508e-06,
+      "loss": 1.6239,
+      "step": 1626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.714492753623189e-06,
+      "loss": 0.5771,
+      "step": 1627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.71304347826087e-06,
+      "loss": 1.3467,
+      "step": 1628
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.711594202898551e-06,
+      "loss": 1.552,
+      "step": 1629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.710144927536232e-06,
+      "loss": 1.0003,
+      "step": 1630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.708695652173913e-06,
+      "loss": 1.325,
+      "step": 1631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.707246376811594e-06,
+      "loss": 1.3925,
+      "step": 1632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.705797101449277e-06,
+      "loss": 1.1213,
+      "step": 1633
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.704347826086958e-06,
+      "loss": 0.8087,
+      "step": 1634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.702898550724637e-06,
+      "loss": 1.9915,
+      "step": 1635
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.70144927536232e-06,
+      "loss": 1.8736,
+      "step": 1636
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.7e-06,
+      "loss": 1.1576,
+      "step": 1637
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.698550724637682e-06,
+      "loss": 1.959,
+      "step": 1638
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.697101449275363e-06,
+      "loss": 1.5379,
+      "step": 1639
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.695652173913044e-06,
+      "loss": 0.7664,
+      "step": 1640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.694202898550725e-06,
+      "loss": 0.9015,
+      "step": 1641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.692753623188406e-06,
+      "loss": 0.6585,
+      "step": 1642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.691304347826087e-06,
+      "loss": 1.2508,
+      "step": 1643
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.68985507246377e-06,
+      "loss": 0.9974,
+      "step": 1644
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.68840579710145e-06,
+      "loss": 1.1534,
+      "step": 1645
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.68695652173913e-06,
+      "loss": 0.994,
+      "step": 1646
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.685507246376813e-06,
+      "loss": 0.6765,
+      "step": 1647
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.684057971014494e-06,
+      "loss": 1.482,
+      "step": 1648
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.682608695652175e-06,
+      "loss": 1.0951,
+      "step": 1649
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.681159420289856e-06,
+      "loss": 0.8017,
+      "step": 1650
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.679710144927537e-06,
+      "loss": 1.1623,
+      "step": 1651
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.678260869565218e-06,
+      "loss": 2.0656,
+      "step": 1652
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.676811594202899e-06,
+      "loss": 1.1294,
+      "step": 1653
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.67536231884058e-06,
+      "loss": 0.933,
+      "step": 1654
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.673913043478262e-06,
+      "loss": 0.8141,
+      "step": 1655
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.672463768115943e-06,
+      "loss": 0.9573,
+      "step": 1656
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.671014492753624e-06,
+      "loss": 0.789,
+      "step": 1657
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.669565217391305e-06,
+      "loss": 1.4806,
+      "step": 1658
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.668115942028986e-06,
+      "loss": 0.9206,
+      "step": 1659
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 1.374,
+      "step": 1660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.665217391304348e-06,
+      "loss": 1.1696,
+      "step": 1661
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.66376811594203e-06,
+      "loss": 0.9187,
+      "step": 1662
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.662318840579712e-06,
+      "loss": 1.0952,
+      "step": 1663
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.660869565217391e-06,
+      "loss": 1.6913,
+      "step": 1664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.659420289855072e-06,
+      "loss": 1.401,
+      "step": 1665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.657971014492755e-06,
+      "loss": 1.2225,
+      "step": 1666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.656521739130436e-06,
+      "loss": 0.6846,
+      "step": 1667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.655072463768117e-06,
+      "loss": 1.1961,
+      "step": 1668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.653623188405798e-06,
+      "loss": 0.7051,
+      "step": 1669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.652173913043479e-06,
+      "loss": 1.6178,
+      "step": 1670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.65072463768116e-06,
+      "loss": 1.6362,
+      "step": 1671
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.649275362318841e-06,
+      "loss": 1.1819,
+      "step": 1672
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.647826086956522e-06,
+      "loss": 1.8543,
+      "step": 1673
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.646376811594205e-06,
+      "loss": 2.1984,
+      "step": 1674
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.644927536231884e-06,
+      "loss": 1.5383,
+      "step": 1675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.643478260869565e-06,
+      "loss": 1.1337,
+      "step": 1676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.642028985507248e-06,
+      "loss": 1.4695,
+      "step": 1677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.640579710144929e-06,
+      "loss": 0.7925,
+      "step": 1678
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.63913043478261e-06,
+      "loss": 1.3694,
+      "step": 1679
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.63768115942029e-06,
+      "loss": 1.4396,
+      "step": 1680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.636231884057972e-06,
+      "loss": 1.298,
+      "step": 1681
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.634782608695653e-06,
+      "loss": 1.8412,
+      "step": 1682
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.633333333333334e-06,
+      "loss": 1.4036,
+      "step": 1683
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.631884057971015e-06,
+      "loss": 0.8159,
+      "step": 1684
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.630434782608697e-06,
+      "loss": 1.4856,
+      "step": 1685
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.628985507246377e-06,
+      "loss": 1.3938,
+      "step": 1686
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.6275362318840585e-06,
+      "loss": 1.0227,
+      "step": 1687
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.6260869565217395e-06,
+      "loss": 0.8692,
+      "step": 1688
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.624637681159421e-06,
+      "loss": 0.7677,
+      "step": 1689
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.6231884057971015e-06,
+      "loss": 1.1956,
+      "step": 1690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.621739130434783e-06,
+      "loss": 0.7832,
+      "step": 1691
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.620289855072464e-06,
+      "loss": 1.0222,
+      "step": 1692
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.618840579710145e-06,
+      "loss": 1.8238,
+      "step": 1693
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.617391304347826e-06,
+      "loss": 1.0749,
+      "step": 1694
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.615942028985508e-06,
+      "loss": 0.7191,
+      "step": 1695
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.614492753623189e-06,
+      "loss": 1.0856,
+      "step": 1696
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.613043478260869e-06,
+      "loss": 1.4753,
+      "step": 1697
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.611594202898551e-06,
+      "loss": 2.0209,
+      "step": 1698
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.610144927536233e-06,
+      "loss": 1.1155,
+      "step": 1699
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.608695652173914e-06,
+      "loss": 1.1686,
+      "step": 1700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.607246376811594e-06,
+      "loss": 1.1026,
+      "step": 1701
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.605797101449276e-06,
+      "loss": 1.2574,
+      "step": 1702
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.604347826086957e-06,
+      "loss": 1.4368,
+      "step": 1703
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.602898550724638e-06,
+      "loss": 1.3174,
+      "step": 1704
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.601449275362319e-06,
+      "loss": 1.0716,
+      "step": 1705
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.629,
+      "step": 1706
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.598550724637682e-06,
+      "loss": 0.8522,
+      "step": 1707
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.597101449275363e-06,
+      "loss": 1.5118,
+      "step": 1708
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.595652173913044e-06,
+      "loss": 1.0325,
+      "step": 1709
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.594202898550726e-06,
+      "loss": 1.1489,
+      "step": 1710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.592753623188407e-06,
+      "loss": 1.0384,
+      "step": 1711
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.591304347826087e-06,
+      "loss": 1.2985,
+      "step": 1712
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.589855072463769e-06,
+      "loss": 0.8631,
+      "step": 1713
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.58840579710145e-06,
+      "loss": 0.7083,
+      "step": 1714
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.586956521739131e-06,
+      "loss": 1.5897,
+      "step": 1715
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.585507246376812e-06,
+      "loss": 1.2666,
+      "step": 1716
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.584057971014494e-06,
+      "loss": 2.0971,
+      "step": 1717
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.582608695652175e-06,
+      "loss": 1.1013,
+      "step": 1718
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.581159420289856e-06,
+      "loss": 1.932,
+      "step": 1719
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.579710144927537e-06,
+      "loss": 1.7338,
+      "step": 1720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5782608695652185e-06,
+      "loss": 1.3716,
+      "step": 1721
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5768115942028995e-06,
+      "loss": 0.5462,
+      "step": 1722
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.57536231884058e-06,
+      "loss": 0.9168,
+      "step": 1723
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5739130434782615e-06,
+      "loss": 1.8779,
+      "step": 1724
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.572463768115943e-06,
+      "loss": 1.0136,
+      "step": 1725
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5710144927536235e-06,
+      "loss": 0.4658,
+      "step": 1726
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5695652173913045e-06,
+      "loss": 1.1266,
+      "step": 1727
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.568115942028986e-06,
+      "loss": 1.1104,
+      "step": 1728
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.566666666666667e-06,
+      "loss": 1.1654,
+      "step": 1729
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.565217391304348e-06,
+      "loss": 1.0244,
+      "step": 1730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.563768115942029e-06,
+      "loss": 1.6626,
+      "step": 1731
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.562318840579711e-06,
+      "loss": 1.9483,
+      "step": 1732
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.560869565217392e-06,
+      "loss": 0.7457,
+      "step": 1733
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.559420289855073e-06,
+      "loss": 0.5819,
+      "step": 1734
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.557971014492754e-06,
+      "loss": 1.448,
+      "step": 1735
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.556521739130436e-06,
+      "loss": 1.1425,
+      "step": 1736
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.555072463768116e-06,
+      "loss": 1.0007,
+      "step": 1737
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.553623188405797e-06,
+      "loss": 1.1417,
+      "step": 1738
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.552173913043479e-06,
+      "loss": 1.8732,
+      "step": 1739
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.550724637681161e-06,
+      "loss": 1.4463,
+      "step": 1740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.549275362318841e-06,
+      "loss": 1.375,
+      "step": 1741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.547826086956522e-06,
+      "loss": 1.1419,
+      "step": 1742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.546376811594204e-06,
+      "loss": 0.775,
+      "step": 1743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.544927536231885e-06,
+      "loss": 1.1819,
+      "step": 1744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.543478260869566e-06,
+      "loss": 0.9708,
+      "step": 1745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.542028985507247e-06,
+      "loss": 1.0451,
+      "step": 1746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.540579710144929e-06,
+      "loss": 1.203,
+      "step": 1747
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.539130434782609e-06,
+      "loss": 1.1399,
+      "step": 1748
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.53768115942029e-06,
+      "loss": 1.4766,
+      "step": 1749
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.536231884057972e-06,
+      "loss": 1.1635,
+      "step": 1750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.534782608695654e-06,
+      "loss": 1.1641,
+      "step": 1751
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.533333333333334e-06,
+      "loss": 1.4592,
+      "step": 1752
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.531884057971015e-06,
+      "loss": 0.884,
+      "step": 1753
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.530434782608697e-06,
+      "loss": 1.3327,
+      "step": 1754
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.528985507246377e-06,
+      "loss": 1.6411,
+      "step": 1755
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.527536231884059e-06,
+      "loss": 1.8822,
+      "step": 1756
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.52608695652174e-06,
+      "loss": 1.5773,
+      "step": 1757
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.5246376811594215e-06,
+      "loss": 1.4882,
+      "step": 1758
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.523188405797102e-06,
+      "loss": 0.992,
+      "step": 1759
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.5217391304347835e-06,
+      "loss": 0.8799,
+      "step": 1760
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.5202898550724645e-06,
+      "loss": 0.799,
+      "step": 1761
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.518840579710146e-06,
+      "loss": 0.5583,
+      "step": 1762
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.5173913043478265e-06,
+      "loss": 1.4356,
+      "step": 1763
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.5159420289855075e-06,
+      "loss": 1.4021,
+      "step": 1764
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.514492753623189e-06,
+      "loss": 1.3816,
+      "step": 1765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.5130434782608695e-06,
+      "loss": 1.0287,
+      "step": 1766
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.511594202898551e-06,
+      "loss": 1.1781,
+      "step": 1767
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.510144927536232e-06,
+      "loss": 1.3,
+      "step": 1768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.508695652173914e-06,
+      "loss": 1.4985,
+      "step": 1769
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.507246376811594e-06,
+      "loss": 1.0786,
+      "step": 1770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.505797101449276e-06,
+      "loss": 1.0521,
+      "step": 1771
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.504347826086957e-06,
+      "loss": 1.2067,
+      "step": 1772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.502898550724639e-06,
+      "loss": 1.064,
+      "step": 1773
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.501449275362319e-06,
+      "loss": 1.2208,
+      "step": 1774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0581,
+      "step": 1775
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.498550724637682e-06,
+      "loss": 1.4607,
+      "step": 1776
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.497101449275362e-06,
+      "loss": 1.3887,
+      "step": 1777
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.495652173913044e-06,
+      "loss": 1.3934,
+      "step": 1778
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.494202898550725e-06,
+      "loss": 1.393,
+      "step": 1779
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.492753623188407e-06,
+      "loss": 0.8851,
+      "step": 1780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.491304347826087e-06,
+      "loss": 1.4414,
+      "step": 1781
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.489855072463769e-06,
+      "loss": 1.485,
+      "step": 1782
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.48840579710145e-06,
+      "loss": 1.0167,
+      "step": 1783
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.486956521739132e-06,
+      "loss": 1.1255,
+      "step": 1784
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.485507246376812e-06,
+      "loss": 1.51,
+      "step": 1785
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.484057971014494e-06,
+      "loss": 1.3184,
+      "step": 1786
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.482608695652175e-06,
+      "loss": 1.4151,
+      "step": 1787
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.481159420289855e-06,
+      "loss": 1.5831,
+      "step": 1788
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.479710144927537e-06,
+      "loss": 1.1826,
+      "step": 1789
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.478260869565218e-06,
+      "loss": 1.0626,
+      "step": 1790
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.4768115942029e-06,
+      "loss": 1.0941,
+      "step": 1791
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.47536231884058e-06,
+      "loss": 1.2211,
+      "step": 1792
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.473913043478262e-06,
+      "loss": 1.1451,
+      "step": 1793
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.472463768115943e-06,
+      "loss": 1.6536,
+      "step": 1794
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.4710144927536245e-06,
+      "loss": 1.4907,
+      "step": 1795
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.469565217391305e-06,
+      "loss": 0.9592,
+      "step": 1796
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4681159420289865e-06,
+      "loss": 0.874,
+      "step": 1797
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.9908,
+      "step": 1798
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.465217391304348e-06,
+      "loss": 1.4588,
+      "step": 1799
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4637681159420295e-06,
+      "loss": 0.9686,
+      "step": 1800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.462318840579711e-06,
+      "loss": 1.2349,
+      "step": 1801
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.460869565217392e-06,
+      "loss": 1.2203,
+      "step": 1802
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4594202898550725e-06,
+      "loss": 1.1895,
+      "step": 1803
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.457971014492754e-06,
+      "loss": 1.2064,
+      "step": 1804
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.456521739130435e-06,
+      "loss": 1.2658,
+      "step": 1805
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.455072463768116e-06,
+      "loss": 2.0856,
+      "step": 1806
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.453623188405797e-06,
+      "loss": 1.4437,
+      "step": 1807
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.452173913043479e-06,
+      "loss": 0.3598,
+      "step": 1808
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.45072463768116e-06,
+      "loss": 1.3309,
+      "step": 1809
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.44927536231884e-06,
+      "loss": 1.2636,
+      "step": 1810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.447826086956522e-06,
+      "loss": 0.969,
+      "step": 1811
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.446376811594204e-06,
+      "loss": 1.193,
+      "step": 1812
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.444927536231885e-06,
+      "loss": 1.9162,
+      "step": 1813
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.443478260869565e-06,
+      "loss": 0.7652,
+      "step": 1814
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.442028985507247e-06,
+      "loss": 0.6324,
+      "step": 1815
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.440579710144928e-06,
+      "loss": 1.0985,
+      "step": 1816
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.439130434782609e-06,
+      "loss": 1.0048,
+      "step": 1817
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.43768115942029e-06,
+      "loss": 1.3825,
+      "step": 1818
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.436231884057972e-06,
+      "loss": 1.3322,
+      "step": 1819
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.434782608695653e-06,
+      "loss": 1.1325,
+      "step": 1820
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.433333333333334e-06,
+      "loss": 0.8759,
+      "step": 1821
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.431884057971015e-06,
+      "loss": 0.3761,
+      "step": 1822
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.430434782608697e-06,
+      "loss": 1.0249,
+      "step": 1823
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.428985507246378e-06,
+      "loss": 1.1223,
+      "step": 1824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.427536231884058e-06,
+      "loss": 1.0838,
+      "step": 1825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.42608695652174e-06,
+      "loss": 0.9034,
+      "step": 1826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.424637681159422e-06,
+      "loss": 0.9741,
+      "step": 1827
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.423188405797102e-06,
+      "loss": 1.759,
+      "step": 1828
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.421739130434783e-06,
+      "loss": 1.0673,
+      "step": 1829
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.420289855072465e-06,
+      "loss": 1.5009,
+      "step": 1830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.418840579710146e-06,
+      "loss": 0.9156,
+      "step": 1831
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.417391304347827e-06,
+      "loss": 1.3263,
+      "step": 1832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.415942028985508e-06,
+      "loss": 1.1102,
+      "step": 1833
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4144927536231894e-06,
+      "loss": 1.1779,
+      "step": 1834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4130434782608704e-06,
+      "loss": 1.392,
+      "step": 1835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4115942028985514e-06,
+      "loss": 1.4654,
+      "step": 1836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4101449275362324e-06,
+      "loss": 1.4454,
+      "step": 1837
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.408695652173914e-06,
+      "loss": 0.498,
+      "step": 1838
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.4072463768115944e-06,
+      "loss": 1.2573,
+      "step": 1839
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.4057971014492754e-06,
+      "loss": 1.0935,
+      "step": 1840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.404347826086957e-06,
+      "loss": 1.3044,
+      "step": 1841
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.402898550724638e-06,
+      "loss": 1.0334,
+      "step": 1842
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.401449275362319e-06,
+      "loss": 1.4182,
+      "step": 1843
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.4e-06,
+      "loss": 1.1644,
+      "step": 1844
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.398550724637682e-06,
+      "loss": 1.4915,
+      "step": 1845
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.397101449275363e-06,
+      "loss": 0.9576,
+      "step": 1846
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.395652173913044e-06,
+      "loss": 1.279,
+      "step": 1847
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.394202898550725e-06,
+      "loss": 1.4922,
+      "step": 1848
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.392753623188407e-06,
+      "loss": 0.86,
+      "step": 1849
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 1.4811,
+      "step": 1850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.389855072463768e-06,
+      "loss": 1.3353,
+      "step": 1851
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.38840579710145e-06,
+      "loss": 1.2279,
+      "step": 1852
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.386956521739132e-06,
+      "loss": 1.0055,
+      "step": 1853
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.385507246376812e-06,
+      "loss": 1.4575,
+      "step": 1854
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.384057971014493e-06,
+      "loss": 1.7541,
+      "step": 1855
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.382608695652175e-06,
+      "loss": 0.8942,
+      "step": 1856
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.381159420289856e-06,
+      "loss": 0.9465,
+      "step": 1857
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.379710144927537e-06,
+      "loss": 1.4035,
+      "step": 1858
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.378260869565218e-06,
+      "loss": 1.5848,
+      "step": 1859
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.3768115942029e-06,
+      "loss": 2.2714,
+      "step": 1860
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.37536231884058e-06,
+      "loss": 1.8595,
+      "step": 1861
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.373913043478262e-06,
+      "loss": 1.4251,
+      "step": 1862
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.372463768115943e-06,
+      "loss": 0.7934,
+      "step": 1863
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.3710144927536246e-06,
+      "loss": 1.8867,
+      "step": 1864
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.369565217391305e-06,
+      "loss": 1.2563,
+      "step": 1865
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.368115942028986e-06,
+      "loss": 1.2839,
+      "step": 1866
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3666666666666676e-06,
+      "loss": 1.8592,
+      "step": 1867
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.365217391304348e-06,
+      "loss": 1.0321,
+      "step": 1868
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3637681159420296e-06,
+      "loss": 1.1834,
+      "step": 1869
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3623188405797106e-06,
+      "loss": 1.0738,
+      "step": 1870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.360869565217392e-06,
+      "loss": 1.137,
+      "step": 1871
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3594202898550726e-06,
+      "loss": 1.3086,
+      "step": 1872
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3579710144927544e-06,
+      "loss": 1.2908,
+      "step": 1873
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.3565217391304354e-06,
+      "loss": 1.4272,
+      "step": 1874
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.355072463768117e-06,
+      "loss": 1.2771,
+      "step": 1875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.3536231884057974e-06,
+      "loss": 0.8257,
+      "step": 1876
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.3521739130434784e-06,
+      "loss": 0.4461,
+      "step": 1877
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.35072463768116e-06,
+      "loss": 0.6959,
+      "step": 1878
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.3492753623188404e-06,
+      "loss": 0.5324,
+      "step": 1879
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.347826086956522e-06,
+      "loss": 0.6942,
+      "step": 1880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.346376811594203e-06,
+      "loss": 0.9751,
+      "step": 1881
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.344927536231885e-06,
+      "loss": 1.0466,
+      "step": 1882
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.343478260869565e-06,
+      "loss": 0.5599,
+      "step": 1883
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.342028985507247e-06,
+      "loss": 1.4444,
+      "step": 1884
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.340579710144928e-06,
+      "loss": 1.2081,
+      "step": 1885
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.33913043478261e-06,
+      "loss": 0.759,
+      "step": 1886
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.33768115942029e-06,
+      "loss": 0.4754,
+      "step": 1887
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.336231884057972e-06,
+      "loss": 1.0188,
+      "step": 1888
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.334782608695653e-06,
+      "loss": 1.1535,
+      "step": 1889
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.9688,
+      "step": 1890
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.331884057971015e-06,
+      "loss": 1.5196,
+      "step": 1891
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.330434782608696e-06,
+      "loss": 1.1128,
+      "step": 1892
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.328985507246378e-06,
+      "loss": 1.2009,
+      "step": 1893
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.327536231884058e-06,
+      "loss": 1.3729,
+      "step": 1894
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.32608695652174e-06,
+      "loss": 0.6076,
+      "step": 1895
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.324637681159421e-06,
+      "loss": 2.0188,
+      "step": 1896
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.323188405797103e-06,
+      "loss": 1.4842,
+      "step": 1897
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.321739130434783e-06,
+      "loss": 0.9059,
+      "step": 1898
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.320289855072465e-06,
+      "loss": 1.5743,
+      "step": 1899
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.318840579710146e-06,
+      "loss": 2.0078,
+      "step": 1900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.317391304347826e-06,
+      "loss": 0.8285,
+      "step": 1901
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.315942028985508e-06,
+      "loss": 0.8775,
+      "step": 1902
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.314492753623189e-06,
+      "loss": 1.2195,
+      "step": 1903
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.3130434782608706e-06,
+      "loss": 1.4139,
+      "step": 1904
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.311594202898551e-06,
+      "loss": 1.2899,
+      "step": 1905
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.3101449275362326e-06,
+      "loss": 1.5005,
+      "step": 1906
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.3086956521739136e-06,
+      "loss": 0.9438,
+      "step": 1907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.307246376811595e-06,
+      "loss": 1.3899,
+      "step": 1908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3057971014492756e-06,
+      "loss": 1.2404,
+      "step": 1909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.304347826086957e-06,
+      "loss": 0.5808,
+      "step": 1910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.302898550724638e-06,
+      "loss": 0.5168,
+      "step": 1911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3014492753623186e-06,
+      "loss": 1.1355,
+      "step": 1912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3e-06,
+      "loss": 1.0358,
+      "step": 1913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.298550724637682e-06,
+      "loss": 1.3806,
+      "step": 1914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.297101449275363e-06,
+      "loss": 1.1614,
+      "step": 1915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.295652173913043e-06,
+      "loss": 1.1723,
+      "step": 1916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.294202898550725e-06,
+      "loss": 0.7258,
+      "step": 1917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.292753623188406e-06,
+      "loss": 1.6474,
+      "step": 1918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.291304347826088e-06,
+      "loss": 1.0113,
+      "step": 1919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.289855072463768e-06,
+      "loss": 1.1925,
+      "step": 1920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.28840579710145e-06,
+      "loss": 1.0008,
+      "step": 1921
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.286956521739131e-06,
+      "loss": 0.9638,
+      "step": 1922
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.285507246376812e-06,
+      "loss": 1.6418,
+      "step": 1923
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.284057971014493e-06,
+      "loss": 1.0296,
+      "step": 1924
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.282608695652175e-06,
+      "loss": 1.0475,
+      "step": 1925
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.281159420289856e-06,
+      "loss": 0.9555,
+      "step": 1926
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.279710144927536e-06,
+      "loss": 0.7951,
+      "step": 1927
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.278260869565218e-06,
+      "loss": 1.1826,
+      "step": 1928
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.276811594202899e-06,
+      "loss": 1.4789,
+      "step": 1929
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.27536231884058e-06,
+      "loss": 0.8723,
+      "step": 1930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.273913043478261e-06,
+      "loss": 1.2296,
+      "step": 1931
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.272463768115943e-06,
+      "loss": 0.7256,
+      "step": 1932
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.271014492753624e-06,
+      "loss": 1.1706,
+      "step": 1933
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.269565217391305e-06,
+      "loss": 1.1143,
+      "step": 1934
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.268115942028986e-06,
+      "loss": 1.392,
+      "step": 1935
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.266666666666668e-06,
+      "loss": 1.2859,
+      "step": 1936
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.265217391304349e-06,
+      "loss": 1.1925,
+      "step": 1937
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.263768115942029e-06,
+      "loss": 1.5836,
+      "step": 1938
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.262318840579711e-06,
+      "loss": 0.7888,
+      "step": 1939
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.2608695652173925e-06,
+      "loss": 0.539,
+      "step": 1940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.259420289855073e-06,
+      "loss": 1.2472,
+      "step": 1941
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.257971014492754e-06,
+      "loss": 0.9496,
+      "step": 1942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.2565217391304355e-06,
+      "loss": 0.2918,
+      "step": 1943
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.2550724637681165e-06,
+      "loss": 0.8006,
+      "step": 1944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.2536231884057975e-06,
+      "loss": 1.2876,
+      "step": 1945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.2521739130434785e-06,
+      "loss": 0.878,
+      "step": 1946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.25072463768116e-06,
+      "loss": 1.1044,
+      "step": 1947
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.249275362318841e-06,
+      "loss": 1.2342,
+      "step": 1948
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.247826086956522e-06,
+      "loss": 0.5587,
+      "step": 1949
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.246376811594203e-06,
+      "loss": 1.4875,
+      "step": 1950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.244927536231885e-06,
+      "loss": 1.5861,
+      "step": 1951
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.243478260869565e-06,
+      "loss": 1.6355,
+      "step": 1952
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.242028985507246e-06,
+      "loss": 0.8374,
+      "step": 1953
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.240579710144928e-06,
+      "loss": 1.0224,
+      "step": 1954
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.23913043478261e-06,
+      "loss": 0.7408,
+      "step": 1955
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.23768115942029e-06,
+      "loss": 0.9898,
+      "step": 1956
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.236231884057971e-06,
+      "loss": 1.4496,
+      "step": 1957
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.234782608695653e-06,
+      "loss": 1.6761,
+      "step": 1958
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.233333333333334e-06,
+      "loss": 1.0484,
+      "step": 1959
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.231884057971015e-06,
+      "loss": 0.4762,
+      "step": 1960
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.230434782608696e-06,
+      "loss": 1.0284,
+      "step": 1961
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.228985507246378e-06,
+      "loss": 1.2631,
+      "step": 1962
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.227536231884058e-06,
+      "loss": 1.4107,
+      "step": 1963
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.226086956521739e-06,
+      "loss": 0.9384,
+      "step": 1964
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.224637681159421e-06,
+      "loss": 0.7195,
+      "step": 1965
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.223188405797103e-06,
+      "loss": 1.0565,
+      "step": 1966
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.221739130434783e-06,
+      "loss": 0.7292,
+      "step": 1967
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.220289855072464e-06,
+      "loss": 1.2762,
+      "step": 1968
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.218840579710146e-06,
+      "loss": 0.7026,
+      "step": 1969
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.217391304347827e-06,
+      "loss": 1.2111,
+      "step": 1970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.215942028985508e-06,
+      "loss": 1.1313,
+      "step": 1971
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.214492753623189e-06,
+      "loss": 1.3335,
+      "step": 1972
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.213043478260871e-06,
+      "loss": 1.7858,
+      "step": 1973
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.211594202898551e-06,
+      "loss": 1.2352,
+      "step": 1974
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.210144927536233e-06,
+      "loss": 1.245,
+      "step": 1975
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.208695652173914e-06,
+      "loss": 1.1708,
+      "step": 1976
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.2072463768115955e-06,
+      "loss": 0.7833,
+      "step": 1977
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.205797101449276e-06,
+      "loss": 1.3545,
+      "step": 1978
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.204347826086957e-06,
+      "loss": 1.0363,
+      "step": 1979
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.2028985507246385e-06,
+      "loss": 0.4,
+      "step": 1980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.20144927536232e-06,
+      "loss": 0.3065,
+      "step": 1981
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.9724,
+      "step": 1982
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.1985507246376815e-06,
+      "loss": 1.091,
+      "step": 1983
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.197101449275363e-06,
+      "loss": 1.1912,
+      "step": 1984
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.1956521739130435e-06,
+      "loss": 1.6574,
+      "step": 1985
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.194202898550725e-06,
+      "loss": 0.8303,
+      "step": 1986
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.192753623188406e-06,
+      "loss": 0.799,
+      "step": 1987
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.191304347826088e-06,
+      "loss": 1.1932,
+      "step": 1988
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.189855072463768e-06,
+      "loss": 1.0871,
+      "step": 1989
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.18840579710145e-06,
+      "loss": 1.3647,
+      "step": 1990
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.186956521739131e-06,
+      "loss": 0.958,
+      "step": 1991
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.185507246376811e-06,
+      "loss": 0.5315,
+      "step": 1992
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.184057971014493e-06,
+      "loss": 0.9342,
+      "step": 1993
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.182608695652174e-06,
+      "loss": 0.7741,
+      "step": 1994
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.181159420289856e-06,
+      "loss": 1.446,
+      "step": 1995
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.179710144927536e-06,
+      "loss": 1.7873,
+      "step": 1996
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.178260869565218e-06,
+      "loss": 1.1226,
+      "step": 1997
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.176811594202899e-06,
+      "loss": 0.6292,
+      "step": 1998
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.175362318840581e-06,
+      "loss": 0.9599,
+      "step": 1999
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.173913043478261e-06,
+      "loss": 1.018,
+      "step": 2000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.172463768115943e-06,
+      "loss": 0.7975,
+      "step": 2001
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.171014492753624e-06,
+      "loss": 1.6427,
+      "step": 2002
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.169565217391304e-06,
+      "loss": 1.4549,
+      "step": 2003
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.168115942028986e-06,
+      "loss": 0.9134,
+      "step": 2004
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 1.2702,
+      "step": 2005
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.165217391304349e-06,
+      "loss": 0.8226,
+      "step": 2006
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.163768115942029e-06,
+      "loss": 1.2573,
+      "step": 2007
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.162318840579711e-06,
+      "loss": 0.9461,
+      "step": 2008
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.160869565217392e-06,
+      "loss": 1.2894,
+      "step": 2009
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.159420289855074e-06,
+      "loss": 0.6197,
+      "step": 2010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.157971014492754e-06,
+      "loss": 1.1463,
+      "step": 2011
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.156521739130436e-06,
+      "loss": 1.9825,
+      "step": 2012
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.155072463768117e-06,
+      "loss": 0.6588,
+      "step": 2013
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.153623188405797e-06,
+      "loss": 1.4279,
+      "step": 2014
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.152173913043479e-06,
+      "loss": 0.9431,
+      "step": 2015
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.1507246376811605e-06,
+      "loss": 1.067,
+      "step": 2016
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.1492753623188415e-06,
+      "loss": 1.1163,
+      "step": 2017
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.147826086956522e-06,
+      "loss": 1.3025,
+      "step": 2018
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.1463768115942035e-06,
+      "loss": 1.7732,
+      "step": 2019
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.1449275362318845e-06,
+      "loss": 1.249,
+      "step": 2020
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.143478260869566e-06,
+      "loss": 0.7366,
+      "step": 2021
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.1420289855072465e-06,
+      "loss": 0.8942,
+      "step": 2022
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.140579710144928e-06,
+      "loss": 0.6498,
+      "step": 2023
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.139130434782609e-06,
+      "loss": 1.3376,
+      "step": 2024
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.13768115942029e-06,
+      "loss": 0.769,
+      "step": 2025
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.136231884057971e-06,
+      "loss": 0.7943,
+      "step": 2026
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.134782608695653e-06,
+      "loss": 0.6845,
+      "step": 2027
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.133333333333334e-06,
+      "loss": 0.9168,
+      "step": 2028
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.131884057971014e-06,
+      "loss": 0.7639,
+      "step": 2029
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.130434782608696e-06,
+      "loss": 1.4677,
+      "step": 2030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.128985507246377e-06,
+      "loss": 0.8012,
+      "step": 2031
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.127536231884059e-06,
+      "loss": 1.756,
+      "step": 2032
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.126086956521739e-06,
+      "loss": 2.0854,
+      "step": 2033
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.124637681159421e-06,
+      "loss": 2.12,
+      "step": 2034
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.123188405797102e-06,
+      "loss": 1.2958,
+      "step": 2035
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.121739130434783e-06,
+      "loss": 1.8992,
+      "step": 2036
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.120289855072464e-06,
+      "loss": 1.5806,
+      "step": 2037
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.118840579710146e-06,
+      "loss": 1.5876,
+      "step": 2038
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.117391304347827e-06,
+      "loss": 1.3037,
+      "step": 2039
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.115942028985507e-06,
+      "loss": 0.8497,
+      "step": 2040
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.114492753623189e-06,
+      "loss": 1.4771,
+      "step": 2041
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.113043478260871e-06,
+      "loss": 0.9194,
+      "step": 2042
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.111594202898552e-06,
+      "loss": 0.7044,
+      "step": 2043
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.110144927536232e-06,
+      "loss": 0.9249,
+      "step": 2044
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.108695652173914e-06,
+      "loss": 1.1737,
+      "step": 2045
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.107246376811595e-06,
+      "loss": 0.468,
+      "step": 2046
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.105797101449276e-06,
+      "loss": 1.0607,
+      "step": 2047
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.104347826086957e-06,
+      "loss": 0.7155,
+      "step": 2048
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.102898550724639e-06,
+      "loss": 1.2606,
+      "step": 2049
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.10144927536232e-06,
+      "loss": 1.3982,
+      "step": 2050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.9631,
+      "step": 2051
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.098550724637682e-06,
+      "loss": 0.381,
+      "step": 2052
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.0971014492753635e-06,
+      "loss": 1.2072,
+      "step": 2053
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.095652173913044e-06,
+      "loss": 0.4985,
+      "step": 2054
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.094202898550725e-06,
+      "loss": 0.732,
+      "step": 2055
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.0927536231884065e-06,
+      "loss": 0.7517,
+      "step": 2056
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.0913043478260875e-06,
+      "loss": 0.8114,
+      "step": 2057
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.0898550724637685e-06,
+      "loss": 0.7438,
+      "step": 2058
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.0884057971014495e-06,
+      "loss": 0.7678,
+      "step": 2059
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.086956521739131e-06,
+      "loss": 1.3007,
+      "step": 2060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.085507246376812e-06,
+      "loss": 1.6539,
+      "step": 2061
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.084057971014493e-06,
+      "loss": 0.7623,
+      "step": 2062
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.082608695652174e-06,
+      "loss": 1.2284,
+      "step": 2063
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.081159420289856e-06,
+      "loss": 2.0534,
+      "step": 2064
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.079710144927536e-06,
+      "loss": 0.59,
+      "step": 2065
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.078260869565217e-06,
+      "loss": 1.1005,
+      "step": 2066
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.076811594202899e-06,
+      "loss": 0.8519,
+      "step": 2067
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.075362318840581e-06,
+      "loss": 1.1009,
+      "step": 2068
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.073913043478261e-06,
+      "loss": 0.8196,
+      "step": 2069
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.072463768115942e-06,
+      "loss": 0.6633,
+      "step": 2070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.071014492753624e-06,
+      "loss": 0.9723,
+      "step": 2071
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.069565217391305e-06,
+      "loss": 0.8762,
+      "step": 2072
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.068115942028986e-06,
+      "loss": 1.4685,
+      "step": 2073
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 0.5551,
+      "step": 2074
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.065217391304349e-06,
+      "loss": 0.5059,
+      "step": 2075
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.063768115942029e-06,
+      "loss": 0.8459,
+      "step": 2076
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.062318840579711e-06,
+      "loss": 1.1154,
+      "step": 2077
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.060869565217392e-06,
+      "loss": 0.6767,
+      "step": 2078
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.059420289855074e-06,
+      "loss": 0.7618,
+      "step": 2079
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.057971014492754e-06,
+      "loss": 0.8528,
+      "step": 2080
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.056521739130435e-06,
+      "loss": 0.9008,
+      "step": 2081
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.055072463768117e-06,
+      "loss": 1.0999,
+      "step": 2082
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.053623188405798e-06,
+      "loss": 1.4976,
+      "step": 2083
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.052173913043479e-06,
+      "loss": 0.8744,
+      "step": 2084
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.05072463768116e-06,
+      "loss": 0.8863,
+      "step": 2085
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.6892086330935252,
+      "eval_loss": 0.9276160597801208,
+      "eval_roc_auc": 0.9116289978324251,
+      "eval_runtime": 86.4124,
+      "eval_samples_per_second": 8.043,
+      "eval_steps_per_second": 2.014,
+      "step": 2085
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.049275362318842e-06,
+      "loss": 0.9234,
+      "step": 2086
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.047826086956522e-06,
+      "loss": 0.7962,
+      "step": 2087
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.046376811594204e-06,
+      "loss": 0.9956,
+      "step": 2088
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.044927536231885e-06,
+      "loss": 1.3502,
+      "step": 2089
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.0434782608695665e-06,
+      "loss": 1.0596,
+      "step": 2090
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.042028985507247e-06,
+      "loss": 0.9903,
+      "step": 2091
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.040579710144928e-06,
+      "loss": 1.0098,
+      "step": 2092
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.0391304347826095e-06,
+      "loss": 1.1917,
+      "step": 2093
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.037681159420291e-06,
+      "loss": 0.5551,
+      "step": 2094
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.0362318840579715e-06,
+      "loss": 1.0433,
+      "step": 2095
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.0347826086956525e-06,
+      "loss": 0.9903,
+      "step": 2096
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.033333333333334e-06,
+      "loss": 0.672,
+      "step": 2097
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.0318840579710145e-06,
+      "loss": 0.7802,
+      "step": 2098
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.030434782608696e-06,
+      "loss": 1.1596,
+      "step": 2099
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.028985507246377e-06,
+      "loss": 1.2375,
+      "step": 2100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.027536231884059e-06,
+      "loss": 1.0492,
+      "step": 2101
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.026086956521739e-06,
+      "loss": 0.464,
+      "step": 2102
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.024637681159421e-06,
+      "loss": 0.9496,
+      "step": 2103
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.023188405797102e-06,
+      "loss": 0.8253,
+      "step": 2104
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.021739130434784e-06,
+      "loss": 0.9479,
+      "step": 2105
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.020289855072464e-06,
+      "loss": 0.6441,
+      "step": 2106
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.018840579710145e-06,
+      "loss": 0.8197,
+      "step": 2107
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.017391304347827e-06,
+      "loss": 0.7777,
+      "step": 2108
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.015942028985507e-06,
+      "loss": 1.2476,
+      "step": 2109
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014492753623189e-06,
+      "loss": 1.2027,
+      "step": 2110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.01304347826087e-06,
+      "loss": 1.5652,
+      "step": 2111
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.011594202898552e-06,
+      "loss": 0.6775,
+      "step": 2112
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010144927536232e-06,
+      "loss": 1.0945,
+      "step": 2113
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008695652173914e-06,
+      "loss": 0.8473,
+      "step": 2114
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.007246376811595e-06,
+      "loss": 0.9054,
+      "step": 2115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.005797101449275e-06,
+      "loss": 0.8363,
+      "step": 2116
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.004347826086957e-06,
+      "loss": 0.7712,
+      "step": 2117
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.002898550724638e-06,
+      "loss": 1.1664,
+      "step": 2118
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.00144927536232e-06,
+      "loss": 0.6808,
+      "step": 2119
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7e-06,
+      "loss": 1.3135,
+      "step": 2120
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.998550724637682e-06,
+      "loss": 0.3413,
+      "step": 2121
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.997101449275363e-06,
+      "loss": 0.9244,
+      "step": 2122
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.995652173913045e-06,
+      "loss": 1.23,
+      "step": 2123
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.994202898550725e-06,
+      "loss": 0.8723,
+      "step": 2124
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.992753623188407e-06,
+      "loss": 1.5082,
+      "step": 2125
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.991304347826088e-06,
+      "loss": 0.9006,
+      "step": 2126
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.989855072463768e-06,
+      "loss": 0.6288,
+      "step": 2127
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.98840579710145e-06,
+      "loss": 1.2563,
+      "step": 2128
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9869565217391314e-06,
+      "loss": 1.0383,
+      "step": 2129
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9855072463768125e-06,
+      "loss": 0.8628,
+      "step": 2130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.984057971014493e-06,
+      "loss": 0.4789,
+      "step": 2131
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9826086956521745e-06,
+      "loss": 0.4988,
+      "step": 2132
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9811594202898555e-06,
+      "loss": 0.6352,
+      "step": 2133
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.979710144927537e-06,
+      "loss": 1.0686,
+      "step": 2134
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9782608695652175e-06,
+      "loss": 1.1937,
+      "step": 2135
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.976811594202899e-06,
+      "loss": 1.1976,
+      "step": 2136
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.97536231884058e-06,
+      "loss": 1.2407,
+      "step": 2137
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.973913043478261e-06,
+      "loss": 0.6994,
+      "step": 2138
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.972463768115942e-06,
+      "loss": 0.897,
+      "step": 2139
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.971014492753624e-06,
+      "loss": 1.1726,
+      "step": 2140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.969565217391305e-06,
+      "loss": 0.9912,
+      "step": 2141
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.968115942028985e-06,
+      "loss": 0.7912,
+      "step": 2142
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.966666666666667e-06,
+      "loss": 0.9025,
+      "step": 2143
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.965217391304349e-06,
+      "loss": 1.148,
+      "step": 2144
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.96376811594203e-06,
+      "loss": 1.8687,
+      "step": 2145
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.96231884057971e-06,
+      "loss": 0.8218,
+      "step": 2146
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.960869565217392e-06,
+      "loss": 1.2991,
+      "step": 2147
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.959420289855073e-06,
+      "loss": 0.5385,
+      "step": 2148
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.957971014492754e-06,
+      "loss": 0.7609,
+      "step": 2149
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 1.1232,
+      "step": 2150
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.955072463768117e-06,
+      "loss": 1.6891,
+      "step": 2151
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.953623188405798e-06,
+      "loss": 0.9106,
+      "step": 2152
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.952173913043478e-06,
+      "loss": 0.2035,
+      "step": 2153
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.95072463768116e-06,
+      "loss": 1.4197,
+      "step": 2154
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.949275362318842e-06,
+      "loss": 1.9393,
+      "step": 2155
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.947826086956523e-06,
+      "loss": 1.2118,
+      "step": 2156
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.946376811594203e-06,
+      "loss": 1.0655,
+      "step": 2157
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.944927536231885e-06,
+      "loss": 0.9278,
+      "step": 2158
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.943478260869566e-06,
+      "loss": 0.6112,
+      "step": 2159
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.942028985507247e-06,
+      "loss": 0.7541,
+      "step": 2160
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.940579710144928e-06,
+      "loss": 1.6925,
+      "step": 2161
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.93913043478261e-06,
+      "loss": 1.0305,
+      "step": 2162
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.937681159420291e-06,
+      "loss": 0.9425,
+      "step": 2163
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.936231884057972e-06,
+      "loss": 0.7233,
+      "step": 2164
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.934782608695653e-06,
+      "loss": 0.8757,
+      "step": 2165
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.5834,
+      "step": 2166
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.9318840579710154e-06,
+      "loss": 0.7642,
+      "step": 2167
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.930434782608696e-06,
+      "loss": 1.7773,
+      "step": 2168
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.9289855072463774e-06,
+      "loss": 0.4228,
+      "step": 2169
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.927536231884059e-06,
+      "loss": 1.3626,
+      "step": 2170
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.9260869565217394e-06,
+      "loss": 0.701,
+      "step": 2171
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.9246376811594204e-06,
+      "loss": 0.9872,
+      "step": 2172
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.923188405797102e-06,
+      "loss": 0.6636,
+      "step": 2173
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.921739130434783e-06,
+      "loss": 1.3163,
+      "step": 2174
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.920289855072464e-06,
+      "loss": 0.8492,
+      "step": 2175
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.918840579710145e-06,
+      "loss": 0.7609,
+      "step": 2176
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.917391304347827e-06,
+      "loss": 1.3146,
+      "step": 2177
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.915942028985507e-06,
+      "loss": 1.0857,
+      "step": 2178
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.914492753623189e-06,
+      "loss": 0.4182,
+      "step": 2179
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.91304347826087e-06,
+      "loss": 1.0792,
+      "step": 2180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.911594202898552e-06,
+      "loss": 1.3339,
+      "step": 2181
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.910144927536232e-06,
+      "loss": 1.3849,
+      "step": 2182
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.908695652173913e-06,
+      "loss": 0.645,
+      "step": 2183
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.907246376811595e-06,
+      "loss": 0.9565,
+      "step": 2184
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.905797101449276e-06,
+      "loss": 0.6397,
+      "step": 2185
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.904347826086957e-06,
+      "loss": 0.8698,
+      "step": 2186
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.902898550724638e-06,
+      "loss": 1.4796,
+      "step": 2187
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.90144927536232e-06,
+      "loss": 0.6674,
+      "step": 2188
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.9e-06,
+      "loss": 0.2939,
+      "step": 2189
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.898550724637682e-06,
+      "loss": 0.6171,
+      "step": 2190
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.897101449275363e-06,
+      "loss": 0.9375,
+      "step": 2191
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.895652173913045e-06,
+      "loss": 1.4676,
+      "step": 2192
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.894202898550725e-06,
+      "loss": 1.3358,
+      "step": 2193
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.892753623188406e-06,
+      "loss": 1.574,
+      "step": 2194
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.891304347826088e-06,
+      "loss": 1.7477,
+      "step": 2195
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.8898550724637696e-06,
+      "loss": 0.8825,
+      "step": 2196
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.88840579710145e-06,
+      "loss": 0.7116,
+      "step": 2197
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.886956521739131e-06,
+      "loss": 0.4596,
+      "step": 2198
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.8855072463768126e-06,
+      "loss": 2.1692,
+      "step": 2199
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.884057971014493e-06,
+      "loss": 0.3857,
+      "step": 2200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.8826086956521746e-06,
+      "loss": 1.1752,
+      "step": 2201
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.8811594202898556e-06,
+      "loss": 0.7583,
+      "step": 2202
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.879710144927537e-06,
+      "loss": 1.5623,
+      "step": 2203
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.8782608695652176e-06,
+      "loss": 1.1754,
+      "step": 2204
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.876811594202899e-06,
+      "loss": 1.8067,
+      "step": 2205
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.87536231884058e-06,
+      "loss": 0.9644,
+      "step": 2206
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.873913043478262e-06,
+      "loss": 0.8778,
+      "step": 2207
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.872463768115942e-06,
+      "loss": 1.0633,
+      "step": 2208
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.871014492753623e-06,
+      "loss": 1.2682,
+      "step": 2209
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.869565217391305e-06,
+      "loss": 1.4231,
+      "step": 2210
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.8681159420289854e-06,
+      "loss": 0.9678,
+      "step": 2211
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.866666666666667e-06,
+      "loss": 0.5412,
+      "step": 2212
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.865217391304348e-06,
+      "loss": 1.3336,
+      "step": 2213
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.86376811594203e-06,
+      "loss": 0.7463,
+      "step": 2214
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.86231884057971e-06,
+      "loss": 0.4054,
+      "step": 2215
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.860869565217392e-06,
+      "loss": 1.137,
+      "step": 2216
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.859420289855073e-06,
+      "loss": 2.6262,
+      "step": 2217
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.857971014492755e-06,
+      "loss": 1.0543,
+      "step": 2218
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.856521739130435e-06,
+      "loss": 1.0597,
+      "step": 2219
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.855072463768116e-06,
+      "loss": 0.7617,
+      "step": 2220
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.853623188405798e-06,
+      "loss": 0.9696,
+      "step": 2221
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.852173913043478e-06,
+      "loss": 1.3352,
+      "step": 2222
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.85072463768116e-06,
+      "loss": 0.5984,
+      "step": 2223
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.849275362318841e-06,
+      "loss": 0.8452,
+      "step": 2224
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.847826086956523e-06,
+      "loss": 1.2387,
+      "step": 2225
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.846376811594203e-06,
+      "loss": 0.2946,
+      "step": 2226
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.844927536231885e-06,
+      "loss": 0.6396,
+      "step": 2227
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.843478260869566e-06,
+      "loss": 0.3868,
+      "step": 2228
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.842028985507248e-06,
+      "loss": 0.6516,
+      "step": 2229
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.840579710144928e-06,
+      "loss": 1.2048,
+      "step": 2230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.83913043478261e-06,
+      "loss": 0.7766,
+      "step": 2231
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.837681159420291e-06,
+      "loss": 0.9984,
+      "step": 2232
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.836231884057971e-06,
+      "loss": 0.7542,
+      "step": 2233
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.834782608695653e-06,
+      "loss": 0.9731,
+      "step": 2234
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 1.1327,
+      "step": 2235
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8318840579710155e-06,
+      "loss": 0.1963,
+      "step": 2236
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.830434782608696e-06,
+      "loss": 1.1402,
+      "step": 2237
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8289855072463775e-06,
+      "loss": 0.7085,
+      "step": 2238
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8275362318840585e-06,
+      "loss": 0.8674,
+      "step": 2239
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8260869565217395e-06,
+      "loss": 1.9362,
+      "step": 2240
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8246376811594206e-06,
+      "loss": 1.1304,
+      "step": 2241
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.823188405797102e-06,
+      "loss": 0.7903,
+      "step": 2242
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.821739130434783e-06,
+      "loss": 0.7765,
+      "step": 2243
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.8202898550724636e-06,
+      "loss": 1.5134,
+      "step": 2244
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.818840579710145e-06,
+      "loss": 0.9165,
+      "step": 2245
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.817391304347826e-06,
+      "loss": 1.8925,
+      "step": 2246
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.815942028985508e-06,
+      "loss": 1.4386,
+      "step": 2247
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.814492753623188e-06,
+      "loss": 1.2145,
+      "step": 2248
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.81304347826087e-06,
+      "loss": 0.9405,
+      "step": 2249
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.811594202898551e-06,
+      "loss": 0.8739,
+      "step": 2250
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.810144927536232e-06,
+      "loss": 1.3505,
+      "step": 2251
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.808695652173913e-06,
+      "loss": 0.972,
+      "step": 2252
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.807246376811595e-06,
+      "loss": 1.046,
+      "step": 2253
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.805797101449276e-06,
+      "loss": 1.4585,
+      "step": 2254
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.804347826086956e-06,
+      "loss": 1.321,
+      "step": 2255
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.802898550724638e-06,
+      "loss": 1.1543,
+      "step": 2256
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.80144927536232e-06,
+      "loss": 1.051,
+      "step": 2257
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.6234,
+      "step": 2258
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.798550724637681e-06,
+      "loss": 1.1234,
+      "step": 2259
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.797101449275363e-06,
+      "loss": 0.468,
+      "step": 2260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.795652173913044e-06,
+      "loss": 1.3154,
+      "step": 2261
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.794202898550725e-06,
+      "loss": 0.8285,
+      "step": 2262
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.792753623188406e-06,
+      "loss": 0.6284,
+      "step": 2263
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.791304347826088e-06,
+      "loss": 0.5884,
+      "step": 2264
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.789855072463769e-06,
+      "loss": 1.0337,
+      "step": 2265
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.78840579710145e-06,
+      "loss": 1.4119,
+      "step": 2266
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.786956521739131e-06,
+      "loss": 1.1878,
+      "step": 2267
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.785507246376813e-06,
+      "loss": 0.7715,
+      "step": 2268
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.784057971014494e-06,
+      "loss": 1.1349,
+      "step": 2269
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.782608695652174e-06,
+      "loss": 0.6337,
+      "step": 2270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.781159420289856e-06,
+      "loss": 0.8005,
+      "step": 2271
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.779710144927537e-06,
+      "loss": 0.7676,
+      "step": 2272
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.778260869565218e-06,
+      "loss": 0.7048,
+      "step": 2273
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.776811594202899e-06,
+      "loss": 0.7608,
+      "step": 2274
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.7753623188405805e-06,
+      "loss": 0.9003,
+      "step": 2275
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.7739130434782615e-06,
+      "loss": 1.1944,
+      "step": 2276
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.7724637681159425e-06,
+      "loss": 1.0018,
+      "step": 2277
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.7710144927536235e-06,
+      "loss": 0.9131,
+      "step": 2278
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.769565217391305e-06,
+      "loss": 1.0769,
+      "step": 2279
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.768115942028986e-06,
+      "loss": 0.5078,
+      "step": 2280
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.7666666666666665e-06,
+      "loss": 0.5583,
+      "step": 2281
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.765217391304348e-06,
+      "loss": 0.9959,
+      "step": 2282
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.76376811594203e-06,
+      "loss": 1.036,
+      "step": 2283
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.76231884057971e-06,
+      "loss": 0.8959,
+      "step": 2284
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.760869565217391e-06,
+      "loss": 0.6594,
+      "step": 2285
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.759420289855073e-06,
+      "loss": 0.8633,
+      "step": 2286
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.757971014492754e-06,
+      "loss": 0.7975,
+      "step": 2287
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.756521739130435e-06,
+      "loss": 1.1728,
+      "step": 2288
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.755072463768116e-06,
+      "loss": 1.6183,
+      "step": 2289
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.753623188405798e-06,
+      "loss": 0.6997,
+      "step": 2290
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.752173913043479e-06,
+      "loss": 0.7778,
+      "step": 2291
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.75072463768116e-06,
+      "loss": 0.6756,
+      "step": 2292
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.749275362318841e-06,
+      "loss": 0.4743,
+      "step": 2293
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.747826086956523e-06,
+      "loss": 0.998,
+      "step": 2294
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.746376811594203e-06,
+      "loss": 1.1861,
+      "step": 2295
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.744927536231884e-06,
+      "loss": 0.6119,
+      "step": 2296
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.743478260869566e-06,
+      "loss": 1.1952,
+      "step": 2297
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.742028985507248e-06,
+      "loss": 0.7529,
+      "step": 2298
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.740579710144928e-06,
+      "loss": 0.8972,
+      "step": 2299
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.739130434782609e-06,
+      "loss": 1.0797,
+      "step": 2300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.737681159420291e-06,
+      "loss": 0.6787,
+      "step": 2301
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.736231884057971e-06,
+      "loss": 0.6807,
+      "step": 2302
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.734782608695653e-06,
+      "loss": 0.6036,
+      "step": 2303
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.733333333333334e-06,
+      "loss": 1.034,
+      "step": 2304
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.731884057971016e-06,
+      "loss": 0.6353,
+      "step": 2305
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.730434782608696e-06,
+      "loss": 1.2854,
+      "step": 2306
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.728985507246377e-06,
+      "loss": 0.6645,
+      "step": 2307
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.727536231884059e-06,
+      "loss": 1.3467,
+      "step": 2308
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7260869565217405e-06,
+      "loss": 0.5259,
+      "step": 2309
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.724637681159421e-06,
+      "loss": 0.5805,
+      "step": 2310
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.723188405797102e-06,
+      "loss": 1.0564,
+      "step": 2311
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.7217391304347835e-06,
+      "loss": 0.6784,
+      "step": 2312
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.720289855072464e-06,
+      "loss": 1.5788,
+      "step": 2313
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.7188405797101455e-06,
+      "loss": 0.6734,
+      "step": 2314
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.7173913043478265e-06,
+      "loss": 0.4004,
+      "step": 2315
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.715942028985508e-06,
+      "loss": 0.8242,
+      "step": 2316
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.7144927536231885e-06,
+      "loss": 1.4618,
+      "step": 2317
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.71304347826087e-06,
+      "loss": 0.7793,
+      "step": 2318
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.711594202898551e-06,
+      "loss": 0.5742,
+      "step": 2319
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.710144927536233e-06,
+      "loss": 0.9979,
+      "step": 2320
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.708695652173913e-06,
+      "loss": 1.2919,
+      "step": 2321
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.707246376811594e-06,
+      "loss": 0.3437,
+      "step": 2322
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.705797101449276e-06,
+      "loss": 0.5633,
+      "step": 2323
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.704347826086956e-06,
+      "loss": 1.3065,
+      "step": 2324
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.702898550724638e-06,
+      "loss": 0.214,
+      "step": 2325
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.701449275362319e-06,
+      "loss": 1.0148,
+      "step": 2326
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 1.1586,
+      "step": 2327
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.698550724637681e-06,
+      "loss": 0.4381,
+      "step": 2328
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.697101449275363e-06,
+      "loss": 0.6304,
+      "step": 2329
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.695652173913044e-06,
+      "loss": 0.6067,
+      "step": 2330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.694202898550726e-06,
+      "loss": 0.9432,
+      "step": 2331
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.692753623188406e-06,
+      "loss": 0.905,
+      "step": 2332
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.691304347826088e-06,
+      "loss": 0.9093,
+      "step": 2333
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.689855072463769e-06,
+      "loss": 1.4282,
+      "step": 2334
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.688405797101449e-06,
+      "loss": 0.5966,
+      "step": 2335
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.686956521739131e-06,
+      "loss": 1.1302,
+      "step": 2336
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.685507246376812e-06,
+      "loss": 1.0634,
+      "step": 2337
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.684057971014494e-06,
+      "loss": 0.8136,
+      "step": 2338
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.682608695652174e-06,
+      "loss": 0.341,
+      "step": 2339
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.681159420289856e-06,
+      "loss": 0.7822,
+      "step": 2340
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.679710144927537e-06,
+      "loss": 0.9666,
+      "step": 2341
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.678260869565219e-06,
+      "loss": 1.2171,
+      "step": 2342
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.676811594202899e-06,
+      "loss": 0.5045,
+      "step": 2343
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.675362318840581e-06,
+      "loss": 0.6893,
+      "step": 2344
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.673913043478262e-06,
+      "loss": 0.8662,
+      "step": 2345
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.672463768115942e-06,
+      "loss": 0.8196,
+      "step": 2346
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.671014492753624e-06,
+      "loss": 0.8574,
+      "step": 2347
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.669565217391305e-06,
+      "loss": 0.8285,
+      "step": 2348
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.6681159420289865e-06,
+      "loss": 1.0988,
+      "step": 2349
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.544,
+      "step": 2350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.6652173913043485e-06,
+      "loss": 1.3921,
+      "step": 2351
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.6637681159420295e-06,
+      "loss": 0.881,
+      "step": 2352
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.662318840579711e-06,
+      "loss": 1.5826,
+      "step": 2353
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.6608695652173915e-06,
+      "loss": 0.3996,
+      "step": 2354
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.659420289855073e-06,
+      "loss": 0.5637,
+      "step": 2355
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657971014492754e-06,
+      "loss": 0.8045,
+      "step": 2356
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.6565217391304345e-06,
+      "loss": 1.2335,
+      "step": 2357
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.655072463768116e-06,
+      "loss": 1.7507,
+      "step": 2358
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.653623188405798e-06,
+      "loss": 0.1786,
+      "step": 2359
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.652173913043479e-06,
+      "loss": 0.9094,
+      "step": 2360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.650724637681159e-06,
+      "loss": 0.8127,
+      "step": 2361
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.649275362318841e-06,
+      "loss": 0.6909,
+      "step": 2362
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.647826086956522e-06,
+      "loss": 0.6747,
+      "step": 2363
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.646376811594203e-06,
+      "loss": 1.3303,
+      "step": 2364
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.644927536231884e-06,
+      "loss": 0.4453,
+      "step": 2365
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.643478260869566e-06,
+      "loss": 0.7157,
+      "step": 2366
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.642028985507247e-06,
+      "loss": 0.9548,
+      "step": 2367
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.640579710144928e-06,
+      "loss": 0.5904,
+      "step": 2368
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.639130434782609e-06,
+      "loss": 0.7355,
+      "step": 2369
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.637681159420291e-06,
+      "loss": 0.5531,
+      "step": 2370
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.636231884057972e-06,
+      "loss": 0.7728,
+      "step": 2371
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.634782608695652e-06,
+      "loss": 0.6888,
+      "step": 2372
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.633333333333334e-06,
+      "loss": 0.7219,
+      "step": 2373
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.631884057971015e-06,
+      "loss": 1.1249,
+      "step": 2374
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.630434782608696e-06,
+      "loss": 0.6412,
+      "step": 2375
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.628985507246377e-06,
+      "loss": 0.8247,
+      "step": 2376
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.627536231884059e-06,
+      "loss": 1.6719,
+      "step": 2377
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.62608695652174e-06,
+      "loss": 0.8318,
+      "step": 2378
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.624637681159421e-06,
+      "loss": 1.4003,
+      "step": 2379
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.623188405797102e-06,
+      "loss": 0.8085,
+      "step": 2380
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.621739130434784e-06,
+      "loss": 1.1814,
+      "step": 2381
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.620289855072465e-06,
+      "loss": 0.9597,
+      "step": 2382
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.618840579710145e-06,
+      "loss": 1.2313,
+      "step": 2383
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.617391304347827e-06,
+      "loss": 1.0975,
+      "step": 2384
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.6159420289855085e-06,
+      "loss": 0.8266,
+      "step": 2385
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.614492753623189e-06,
+      "loss": 0.902,
+      "step": 2386
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.61304347826087e-06,
+      "loss": 1.0258,
+      "step": 2387
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.6115942028985515e-06,
+      "loss": 0.9905,
+      "step": 2388
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.6101449275362325e-06,
+      "loss": 0.7603,
+      "step": 2389
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.6086956521739135e-06,
+      "loss": 1.663,
+      "step": 2390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.6072463768115945e-06,
+      "loss": 0.6563,
+      "step": 2391
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.605797101449276e-06,
+      "loss": 1.3225,
+      "step": 2392
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.604347826086957e-06,
+      "loss": 0.9979,
+      "step": 2393
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.602898550724638e-06,
+      "loss": 0.8795,
+      "step": 2394
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.601449275362319e-06,
+      "loss": 0.9345,
+      "step": 2395
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 1.0243,
+      "step": 2396
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.598550724637681e-06,
+      "loss": 0.7143,
+      "step": 2397
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.597101449275362e-06,
+      "loss": 0.579,
+      "step": 2398
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.595652173913044e-06,
+      "loss": 0.9438,
+      "step": 2399
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.594202898550725e-06,
+      "loss": 0.7392,
+      "step": 2400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.592753623188406e-06,
+      "loss": 1.362,
+      "step": 2401
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.591304347826087e-06,
+      "loss": 0.5814,
+      "step": 2402
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.589855072463769e-06,
+      "loss": 0.5149,
+      "step": 2403
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.58840579710145e-06,
+      "loss": 0.5485,
+      "step": 2404
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.586956521739131e-06,
+      "loss": 0.7098,
+      "step": 2405
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.585507246376812e-06,
+      "loss": 1.4333,
+      "step": 2406
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.584057971014494e-06,
+      "loss": 0.9351,
+      "step": 2407
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.582608695652174e-06,
+      "loss": 0.839,
+      "step": 2408
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.581159420289855e-06,
+      "loss": 0.6026,
+      "step": 2409
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.579710144927537e-06,
+      "loss": 1.3569,
+      "step": 2410
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.578260869565219e-06,
+      "loss": 0.9892,
+      "step": 2411
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.576811594202899e-06,
+      "loss": 0.769,
+      "step": 2412
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.57536231884058e-06,
+      "loss": 1.5723,
+      "step": 2413
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.573913043478262e-06,
+      "loss": 0.9177,
+      "step": 2414
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.572463768115943e-06,
+      "loss": 0.7807,
+      "step": 2415
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.571014492753624e-06,
+      "loss": 1.5001,
+      "step": 2416
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.569565217391305e-06,
+      "loss": 0.5723,
+      "step": 2417
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.568115942028987e-06,
+      "loss": 1.2502,
+      "step": 2418
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.566666666666667e-06,
+      "loss": 0.8859,
+      "step": 2419
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.565217391304349e-06,
+      "loss": 1.4504,
+      "step": 2420
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.56376811594203e-06,
+      "loss": 1.2069,
+      "step": 2421
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.5623188405797115e-06,
+      "loss": 1.2082,
+      "step": 2422
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.560869565217392e-06,
+      "loss": 0.8906,
+      "step": 2423
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.559420289855073e-06,
+      "loss": 0.981,
+      "step": 2424
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.5579710144927545e-06,
+      "loss": 0.5661,
+      "step": 2425
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.556521739130435e-06,
+      "loss": 0.8553,
+      "step": 2426
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.5550724637681165e-06,
+      "loss": 1.1127,
+      "step": 2427
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.5536231884057975e-06,
+      "loss": 0.9405,
+      "step": 2428
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.552173913043479e-06,
+      "loss": 1.1978,
+      "step": 2429
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.5507246376811595e-06,
+      "loss": 0.6162,
+      "step": 2430
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.549275362318841e-06,
+      "loss": 1.0996,
+      "step": 2431
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.547826086956522e-06,
+      "loss": 1.0173,
+      "step": 2432
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.546376811594204e-06,
+      "loss": 0.5302,
+      "step": 2433
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.544927536231884e-06,
+      "loss": 1.5877,
+      "step": 2434
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.543478260869565e-06,
+      "loss": 1.0153,
+      "step": 2435
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.542028985507247e-06,
+      "loss": 0.3251,
+      "step": 2436
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.540579710144927e-06,
+      "loss": 0.4795,
+      "step": 2437
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.539130434782609e-06,
+      "loss": 1.3365,
+      "step": 2438
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.53768115942029e-06,
+      "loss": 1.4143,
+      "step": 2439
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.536231884057972e-06,
+      "loss": 0.4466,
+      "step": 2440
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.534782608695652e-06,
+      "loss": 1.2455,
+      "step": 2441
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 1.8604,
+      "step": 2442
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.531884057971015e-06,
+      "loss": 0.7781,
+      "step": 2443
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.530434782608697e-06,
+      "loss": 1.4257,
+      "step": 2444
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.528985507246377e-06,
+      "loss": 0.4584,
+      "step": 2445
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.527536231884059e-06,
+      "loss": 0.8364,
+      "step": 2446
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.52608695652174e-06,
+      "loss": 0.6802,
+      "step": 2447
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.52463768115942e-06,
+      "loss": 0.6894,
+      "step": 2448
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.523188405797102e-06,
+      "loss": 2.0829,
+      "step": 2449
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 1.073,
+      "step": 2450
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.520289855072465e-06,
+      "loss": 1.4186,
+      "step": 2451
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.518840579710145e-06,
+      "loss": 0.3644,
+      "step": 2452
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.517391304347827e-06,
+      "loss": 1.1495,
+      "step": 2453
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.515942028985508e-06,
+      "loss": 0.6136,
+      "step": 2454
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.51449275362319e-06,
+      "loss": 1.5541,
+      "step": 2455
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.51304347826087e-06,
+      "loss": 0.6025,
+      "step": 2456
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.511594202898552e-06,
+      "loss": 0.5372,
+      "step": 2457
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.510144927536233e-06,
+      "loss": 1.0113,
+      "step": 2458
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.508695652173913e-06,
+      "loss": 0.7771,
+      "step": 2459
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.507246376811595e-06,
+      "loss": 1.1078,
+      "step": 2460
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.505797101449276e-06,
+      "loss": 0.4086,
+      "step": 2461
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.5043478260869574e-06,
+      "loss": 0.9668,
+      "step": 2462
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.502898550724638e-06,
+      "loss": 1.087,
+      "step": 2463
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.5014492753623194e-06,
+      "loss": 1.2085,
+      "step": 2464
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.9227,
+      "step": 2465
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.498550724637682e-06,
+      "loss": 0.695,
+      "step": 2466
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.4971014492753624e-06,
+      "loss": 0.8538,
+      "step": 2467
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.495652173913044e-06,
+      "loss": 1.452,
+      "step": 2468
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.494202898550725e-06,
+      "loss": 1.0919,
+      "step": 2469
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.4927536231884055e-06,
+      "loss": 0.4163,
+      "step": 2470
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.491304347826087e-06,
+      "loss": 1.0322,
+      "step": 2471
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.489855072463769e-06,
+      "loss": 0.7719,
+      "step": 2472
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.48840579710145e-06,
+      "loss": 0.5952,
+      "step": 2473
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.48695652173913e-06,
+      "loss": 0.8861,
+      "step": 2474
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.485507246376812e-06,
+      "loss": 1.0597,
+      "step": 2475
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.484057971014493e-06,
+      "loss": 0.7672,
+      "step": 2476
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.482608695652175e-06,
+      "loss": 0.4667,
+      "step": 2477
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.481159420289855e-06,
+      "loss": 0.9616,
+      "step": 2478
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.479710144927537e-06,
+      "loss": 0.9268,
+      "step": 2479
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.478260869565218e-06,
+      "loss": 0.5055,
+      "step": 2480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.476811594202899e-06,
+      "loss": 1.929,
+      "step": 2481
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.47536231884058e-06,
+      "loss": 0.1055,
+      "step": 2482
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.473913043478262e-06,
+      "loss": 0.6656,
+      "step": 2483
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.472463768115943e-06,
+      "loss": 1.2847,
+      "step": 2484
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.471014492753623e-06,
+      "loss": 1.045,
+      "step": 2485
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.469565217391305e-06,
+      "loss": 0.8888,
+      "step": 2486
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.468115942028987e-06,
+      "loss": 1.9586,
+      "step": 2487
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.466666666666667e-06,
+      "loss": 0.3266,
+      "step": 2488
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.465217391304348e-06,
+      "loss": 1.9615,
+      "step": 2489
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.46376811594203e-06,
+      "loss": 0.7907,
+      "step": 2490
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.462318840579711e-06,
+      "loss": 0.9478,
+      "step": 2491
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.460869565217392e-06,
+      "loss": 0.8784,
+      "step": 2492
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.459420289855073e-06,
+      "loss": 0.4785,
+      "step": 2493
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.4579710144927546e-06,
+      "loss": 1.0034,
+      "step": 2494
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.4565217391304356e-06,
+      "loss": 0.5232,
+      "step": 2495
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.455072463768116e-06,
+      "loss": 0.8012,
+      "step": 2496
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.4536231884057976e-06,
+      "loss": 0.2209,
+      "step": 2497
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.4521739130434794e-06,
+      "loss": 0.7375,
+      "step": 2498
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.45072463768116e-06,
+      "loss": 0.18,
+      "step": 2499
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.449275362318841e-06,
+      "loss": 0.9189,
+      "step": 2500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.4478260869565224e-06,
+      "loss": 1.7202,
+      "step": 2501
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.4463768115942034e-06,
+      "loss": 0.1852,
+      "step": 2502
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.4449275362318844e-06,
+      "loss": 1.091,
+      "step": 2503
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.4434782608695654e-06,
+      "loss": 0.7008,
+      "step": 2504
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.442028985507247e-06,
+      "loss": 0.7596,
+      "step": 2505
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.440579710144928e-06,
+      "loss": 0.4781,
+      "step": 2506
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.439130434782609e-06,
+      "loss": 0.8013,
+      "step": 2507
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.43768115942029e-06,
+      "loss": 0.6078,
+      "step": 2508
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.436231884057972e-06,
+      "loss": 0.855,
+      "step": 2509
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.434782608695652e-06,
+      "loss": 0.3054,
+      "step": 2510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.433333333333333e-06,
+      "loss": 0.6309,
+      "step": 2511
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.431884057971015e-06,
+      "loss": 1.2653,
+      "step": 2512
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.430434782608697e-06,
+      "loss": 0.5413,
+      "step": 2513
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.428985507246377e-06,
+      "loss": 1.3946,
+      "step": 2514
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.427536231884058e-06,
+      "loss": 0.7746,
+      "step": 2515
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.42608695652174e-06,
+      "loss": 0.3998,
+      "step": 2516
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.424637681159421e-06,
+      "loss": 0.7896,
+      "step": 2517
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.423188405797102e-06,
+      "loss": 0.3753,
+      "step": 2518
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.421739130434783e-06,
+      "loss": 0.7483,
+      "step": 2519
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.420289855072465e-06,
+      "loss": 0.9919,
+      "step": 2520
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.418840579710145e-06,
+      "loss": 0.9296,
+      "step": 2521
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.417391304347827e-06,
+      "loss": 0.8731,
+      "step": 2522
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.415942028985508e-06,
+      "loss": 0.834,
+      "step": 2523
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.41449275362319e-06,
+      "loss": 0.6301,
+      "step": 2524
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.41304347826087e-06,
+      "loss": 0.6084,
+      "step": 2525
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.411594202898551e-06,
+      "loss": 0.1715,
+      "step": 2526
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.410144927536233e-06,
+      "loss": 0.5768,
+      "step": 2527
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.408695652173914e-06,
+      "loss": 0.6789,
+      "step": 2528
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.407246376811595e-06,
+      "loss": 0.9382,
+      "step": 2529
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.405797101449276e-06,
+      "loss": 0.5895,
+      "step": 2530
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.4043478260869576e-06,
+      "loss": 1.6974,
+      "step": 2531
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.402898550724638e-06,
+      "loss": 1.3714,
+      "step": 2532
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.4014492753623196e-06,
+      "loss": 1.0277,
+      "step": 2533
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.4077,
+      "step": 2534
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.398550724637682e-06,
+      "loss": 0.4013,
+      "step": 2535
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.3971014492753626e-06,
+      "loss": 0.5884,
+      "step": 2536
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.3956521739130436e-06,
+      "loss": 1.4061,
+      "step": 2537
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.394202898550725e-06,
+      "loss": 0.9214,
+      "step": 2538
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.392753623188407e-06,
+      "loss": 0.9533,
+      "step": 2539
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.391304347826087e-06,
+      "loss": 0.9821,
+      "step": 2540
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.389855072463768e-06,
+      "loss": 0.4952,
+      "step": 2541
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.38840579710145e-06,
+      "loss": 3.2807,
+      "step": 2542
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.38695652173913e-06,
+      "loss": 1.1112,
+      "step": 2543
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.385507246376812e-06,
+      "loss": 0.745,
+      "step": 2544
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.384057971014493e-06,
+      "loss": 0.8384,
+      "step": 2545
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.382608695652175e-06,
+      "loss": 0.6519,
+      "step": 2546
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.381159420289855e-06,
+      "loss": 0.5824,
+      "step": 2547
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.379710144927537e-06,
+      "loss": 1.867,
+      "step": 2548
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.378260869565218e-06,
+      "loss": 0.4994,
+      "step": 2549
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.376811594202898e-06,
+      "loss": 1.5887,
+      "step": 2550
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.37536231884058e-06,
+      "loss": 0.6361,
+      "step": 2551
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.373913043478261e-06,
+      "loss": 0.9683,
+      "step": 2552
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.372463768115943e-06,
+      "loss": 0.9818,
+      "step": 2553
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.371014492753623e-06,
+      "loss": 1.2393,
+      "step": 2554
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.369565217391305e-06,
+      "loss": 1.7686,
+      "step": 2555
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.368115942028986e-06,
+      "loss": 0.9042,
+      "step": 2556
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.366666666666668e-06,
+      "loss": 0.5771,
+      "step": 2557
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.365217391304348e-06,
+      "loss": 0.6324,
+      "step": 2558
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.36376811594203e-06,
+      "loss": 0.6909,
+      "step": 2559
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.362318840579711e-06,
+      "loss": 0.6181,
+      "step": 2560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.360869565217391e-06,
+      "loss": 0.1086,
+      "step": 2561
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.359420289855073e-06,
+      "loss": 0.3525,
+      "step": 2562
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.357971014492754e-06,
+      "loss": 1.4366,
+      "step": 2563
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.356521739130436e-06,
+      "loss": 1.464,
+      "step": 2564
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.355072463768116e-06,
+      "loss": 0.9447,
+      "step": 2565
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.353623188405798e-06,
+      "loss": 0.2981,
+      "step": 2566
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.352173913043479e-06,
+      "loss": 0.9193,
+      "step": 2567
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.3507246376811605e-06,
+      "loss": 0.8514,
+      "step": 2568
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.349275362318841e-06,
+      "loss": 0.812,
+      "step": 2569
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.3478260869565225e-06,
+      "loss": 0.5205,
+      "step": 2570
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.3463768115942035e-06,
+      "loss": 0.3244,
+      "step": 2571
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.344927536231884e-06,
+      "loss": 0.628,
+      "step": 2572
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.3434782608695655e-06,
+      "loss": 0.8033,
+      "step": 2573
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.342028985507247e-06,
+      "loss": 0.2947,
+      "step": 2574
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.340579710144928e-06,
+      "loss": 0.7197,
+      "step": 2575
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.3391304347826085e-06,
+      "loss": 0.8801,
+      "step": 2576
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.33768115942029e-06,
+      "loss": 0.7767,
+      "step": 2577
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.336231884057971e-06,
+      "loss": 1.0974,
+      "step": 2578
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.334782608695653e-06,
+      "loss": 0.4589,
+      "step": 2579
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 0.8899,
+      "step": 2580
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.331884057971015e-06,
+      "loss": 1.24,
+      "step": 2581
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.330434782608696e-06,
+      "loss": 0.4863,
+      "step": 2582
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.328985507246377e-06,
+      "loss": 0.3577,
+      "step": 2583
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.327536231884058e-06,
+      "loss": 1.0548,
+      "step": 2584
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.32608695652174e-06,
+      "loss": 0.8714,
+      "step": 2585
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.324637681159421e-06,
+      "loss": 1.5395,
+      "step": 2586
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.323188405797101e-06,
+      "loss": 0.4342,
+      "step": 2587
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.321739130434783e-06,
+      "loss": 0.4355,
+      "step": 2588
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.320289855072464e-06,
+      "loss": 0.3316,
+      "step": 2589
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.318840579710146e-06,
+      "loss": 0.675,
+      "step": 2590
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.317391304347826e-06,
+      "loss": 1.2265,
+      "step": 2591
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.315942028985508e-06,
+      "loss": 2.0779,
+      "step": 2592
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.314492753623189e-06,
+      "loss": 0.6463,
+      "step": 2593
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.31304347826087e-06,
+      "loss": 0.5241,
+      "step": 2594
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.311594202898551e-06,
+      "loss": 1.0443,
+      "step": 2595
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.310144927536233e-06,
+      "loss": 0.412,
+      "step": 2596
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.308695652173914e-06,
+      "loss": 0.5677,
+      "step": 2597
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.307246376811594e-06,
+      "loss": 0.604,
+      "step": 2598
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.305797101449276e-06,
+      "loss": 0.8842,
+      "step": 2599
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.304347826086958e-06,
+      "loss": 2.3409,
+      "step": 2600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.302898550724639e-06,
+      "loss": 1.0532,
+      "step": 2601
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.301449275362319e-06,
+      "loss": 0.6758,
+      "step": 2602
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.6271,
+      "step": 2603
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.298550724637682e-06,
+      "loss": 1.2646,
+      "step": 2604
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.297101449275363e-06,
+      "loss": 0.4973,
+      "step": 2605
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.295652173913044e-06,
+      "loss": 0.7166,
+      "step": 2606
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.2942028985507255e-06,
+      "loss": 1.6303,
+      "step": 2607
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.2927536231884065e-06,
+      "loss": 0.9962,
+      "step": 2608
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.2913043478260875e-06,
+      "loss": 1.358,
+      "step": 2609
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.2898550724637685e-06,
+      "loss": 0.9498,
+      "step": 2610
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.28840579710145e-06,
+      "loss": 1.15,
+      "step": 2611
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.2869565217391305e-06,
+      "loss": 0.4363,
+      "step": 2612
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.2855072463768115e-06,
+      "loss": 1.1387,
+      "step": 2613
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.284057971014493e-06,
+      "loss": 1.2992,
+      "step": 2614
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.282608695652174e-06,
+      "loss": 0.1494,
+      "step": 2615
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.281159420289855e-06,
+      "loss": 0.5802,
+      "step": 2616
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.279710144927536e-06,
+      "loss": 0.3427,
+      "step": 2617
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.278260869565218e-06,
+      "loss": 0.6827,
+      "step": 2618
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.276811594202899e-06,
+      "loss": 1.401,
+      "step": 2619
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.27536231884058e-06,
+      "loss": 1.0346,
+      "step": 2620
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.273913043478261e-06,
+      "loss": 0.6231,
+      "step": 2621
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.272463768115943e-06,
+      "loss": 0.6351,
+      "step": 2622
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.271014492753623e-06,
+      "loss": 0.8087,
+      "step": 2623
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.269565217391304e-06,
+      "loss": 0.699,
+      "step": 2624
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.268115942028986e-06,
+      "loss": 0.963,
+      "step": 2625
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 1.4286,
+      "step": 2626
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.265217391304348e-06,
+      "loss": 1.0951,
+      "step": 2627
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.263768115942029e-06,
+      "loss": 2.0945,
+      "step": 2628
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.262318840579711e-06,
+      "loss": 0.7744,
+      "step": 2629
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.260869565217392e-06,
+      "loss": 0.7637,
+      "step": 2630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.259420289855073e-06,
+      "loss": 0.5106,
+      "step": 2631
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.257971014492754e-06,
+      "loss": 0.6293,
+      "step": 2632
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.256521739130436e-06,
+      "loss": 0.2026,
+      "step": 2633
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.255072463768116e-06,
+      "loss": 0.5567,
+      "step": 2634
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.253623188405798e-06,
+      "loss": 0.439,
+      "step": 2635
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.252173913043479e-06,
+      "loss": 0.6995,
+      "step": 2636
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.250724637681161e-06,
+      "loss": 1.0805,
+      "step": 2637
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.249275362318841e-06,
+      "loss": 1.0686,
+      "step": 2638
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.247826086956522e-06,
+      "loss": 1.0693,
+      "step": 2639
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.246376811594204e-06,
+      "loss": 0.3977,
+      "step": 2640
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2449275362318855e-06,
+      "loss": 0.9133,
+      "step": 2641
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.243478260869566e-06,
+      "loss": 0.9795,
+      "step": 2642
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.242028985507247e-06,
+      "loss": 0.714,
+      "step": 2643
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2405797101449285e-06,
+      "loss": 0.3617,
+      "step": 2644
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.239130434782609e-06,
+      "loss": 1.1814,
+      "step": 2645
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.2376811594202905e-06,
+      "loss": 0.5065,
+      "step": 2646
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.2362318840579715e-06,
+      "loss": 0.6972,
+      "step": 2647
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.234782608695653e-06,
+      "loss": 1.4924,
+      "step": 2648
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.2333333333333335e-06,
+      "loss": 0.3178,
+      "step": 2649
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.2318840579710145e-06,
+      "loss": 0.7106,
+      "step": 2650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.230434782608696e-06,
+      "loss": 0.1585,
+      "step": 2651
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.228985507246378e-06,
+      "loss": 0.8154,
+      "step": 2652
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.227536231884058e-06,
+      "loss": 0.2786,
+      "step": 2653
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.226086956521739e-06,
+      "loss": 1.3843,
+      "step": 2654
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.224637681159421e-06,
+      "loss": 0.7287,
+      "step": 2655
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.223188405797101e-06,
+      "loss": 1.8239,
+      "step": 2656
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.221739130434783e-06,
+      "loss": 1.0767,
+      "step": 2657
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.220289855072464e-06,
+      "loss": 1.3345,
+      "step": 2658
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.218840579710146e-06,
+      "loss": 0.5081,
+      "step": 2659
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.217391304347826e-06,
+      "loss": 0.6222,
+      "step": 2660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.215942028985508e-06,
+      "loss": 0.885,
+      "step": 2661
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.214492753623189e-06,
+      "loss": 1.1507,
+      "step": 2662
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.213043478260871e-06,
+      "loss": 0.5298,
+      "step": 2663
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.211594202898551e-06,
+      "loss": 1.6794,
+      "step": 2664
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.210144927536232e-06,
+      "loss": 0.8313,
+      "step": 2665
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.208695652173914e-06,
+      "loss": 0.6257,
+      "step": 2666
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.207246376811594e-06,
+      "loss": 1.4252,
+      "step": 2667
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.205797101449276e-06,
+      "loss": 0.6922,
+      "step": 2668
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.204347826086957e-06,
+      "loss": 0.5642,
+      "step": 2669
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.202898550724639e-06,
+      "loss": 1.0682,
+      "step": 2670
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.201449275362319e-06,
+      "loss": 0.451,
+      "step": 2671
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.6381,
+      "step": 2672
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.198550724637682e-06,
+      "loss": 1.616,
+      "step": 2673
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.197101449275362e-06,
+      "loss": 0.6542,
+      "step": 2674
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.195652173913044e-06,
+      "loss": 0.449,
+      "step": 2675
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.194202898550725e-06,
+      "loss": 1.7029,
+      "step": 2676
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.192753623188407e-06,
+      "loss": 1.2467,
+      "step": 2677
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.191304347826087e-06,
+      "loss": 1.4235,
+      "step": 2678
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.189855072463769e-06,
+      "loss": 1.52,
+      "step": 2679
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.18840579710145e-06,
+      "loss": 0.3964,
+      "step": 2680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.1869565217391315e-06,
+      "loss": 0.5047,
+      "step": 2681
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.185507246376812e-06,
+      "loss": 0.4869,
+      "step": 2682
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.1840579710144935e-06,
+      "loss": 1.0384,
+      "step": 2683
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.1826086956521745e-06,
+      "loss": 0.881,
+      "step": 2684
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.181159420289855e-06,
+      "loss": 0.8739,
+      "step": 2685
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.1797101449275365e-06,
+      "loss": 0.0882,
+      "step": 2686
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.178260869565218e-06,
+      "loss": 0.534,
+      "step": 2687
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.176811594202899e-06,
+      "loss": 0.5507,
+      "step": 2688
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.1753623188405795e-06,
+      "loss": 1.1771,
+      "step": 2689
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.173913043478261e-06,
+      "loss": 0.3209,
+      "step": 2690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.172463768115942e-06,
+      "loss": 0.6169,
+      "step": 2691
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.171014492753624e-06,
+      "loss": 1.476,
+      "step": 2692
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.169565217391304e-06,
+      "loss": 1.233,
+      "step": 2693
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.168115942028986e-06,
+      "loss": 0.311,
+      "step": 2694
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 1.1653,
+      "step": 2695
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.165217391304348e-06,
+      "loss": 0.407,
+      "step": 2696
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.163768115942029e-06,
+      "loss": 1.4325,
+      "step": 2697
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.162318840579711e-06,
+      "loss": 0.3885,
+      "step": 2698
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.160869565217392e-06,
+      "loss": 0.9306,
+      "step": 2699
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.159420289855072e-06,
+      "loss": 0.9126,
+      "step": 2700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.157971014492754e-06,
+      "loss": 1.1295,
+      "step": 2701
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.156521739130436e-06,
+      "loss": 0.6717,
+      "step": 2702
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.155072463768117e-06,
+      "loss": 0.1937,
+      "step": 2703
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.153623188405797e-06,
+      "loss": 0.3897,
+      "step": 2704
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.152173913043479e-06,
+      "loss": 1.4487,
+      "step": 2705
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.15072463768116e-06,
+      "loss": 0.7351,
+      "step": 2706
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.149275362318841e-06,
+      "loss": 0.7172,
+      "step": 2707
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.147826086956522e-06,
+      "loss": 0.7791,
+      "step": 2708
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.146376811594204e-06,
+      "loss": 2.0596,
+      "step": 2709
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.144927536231885e-06,
+      "loss": 0.7875,
+      "step": 2710
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.143478260869565e-06,
+      "loss": 0.6989,
+      "step": 2711
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.142028985507247e-06,
+      "loss": 0.7731,
+      "step": 2712
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.140579710144929e-06,
+      "loss": 1.3144,
+      "step": 2713
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.13913043478261e-06,
+      "loss": 0.5277,
+      "step": 2714
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.13768115942029e-06,
+      "loss": 0.8121,
+      "step": 2715
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.136231884057972e-06,
+      "loss": 0.471,
+      "step": 2716
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.134782608695653e-06,
+      "loss": 1.1206,
+      "step": 2717
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.6364,
+      "step": 2718
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.131884057971015e-06,
+      "loss": 0.586,
+      "step": 2719
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.1304347826086965e-06,
+      "loss": 3.0846,
+      "step": 2720
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1289855072463775e-06,
+      "loss": 0.5698,
+      "step": 2721
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1275362318840585e-06,
+      "loss": 1.511,
+      "step": 2722
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1260869565217395e-06,
+      "loss": 0.6784,
+      "step": 2723
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.124637681159421e-06,
+      "loss": 1.0426,
+      "step": 2724
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.123188405797102e-06,
+      "loss": 0.7012,
+      "step": 2725
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1217391304347825e-06,
+      "loss": 0.7497,
+      "step": 2726
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.120289855072464e-06,
+      "loss": 0.4425,
+      "step": 2727
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.118840579710146e-06,
+      "loss": 0.6323,
+      "step": 2728
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.117391304347826e-06,
+      "loss": 0.2292,
+      "step": 2729
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.115942028985507e-06,
+      "loss": 1.0067,
+      "step": 2730
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.114492753623189e-06,
+      "loss": 0.9498,
+      "step": 2731
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.11304347826087e-06,
+      "loss": 0.7603,
+      "step": 2732
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.111594202898551e-06,
+      "loss": 0.6877,
+      "step": 2733
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.110144927536232e-06,
+      "loss": 0.7033,
+      "step": 2734
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.108695652173914e-06,
+      "loss": 0.3737,
+      "step": 2735
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.107246376811594e-06,
+      "loss": 1.8098,
+      "step": 2736
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.105797101449276e-06,
+      "loss": 0.4277,
+      "step": 2737
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.104347826086957e-06,
+      "loss": 0.6465,
+      "step": 2738
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.102898550724639e-06,
+      "loss": 0.4711,
+      "step": 2739
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.101449275362319e-06,
+      "loss": 1.712,
+      "step": 2740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.1e-06,
+      "loss": 0.5994,
+      "step": 2741
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.098550724637682e-06,
+      "loss": 0.6801,
+      "step": 2742
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.097101449275363e-06,
+      "loss": 0.9849,
+      "step": 2743
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.095652173913044e-06,
+      "loss": 1.6298,
+      "step": 2744
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.094202898550725e-06,
+      "loss": 1.6144,
+      "step": 2745
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.092753623188407e-06,
+      "loss": 0.8141,
+      "step": 2746
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.091304347826087e-06,
+      "loss": 0.5685,
+      "step": 2747
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.089855072463769e-06,
+      "loss": 0.6665,
+      "step": 2748
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.08840579710145e-06,
+      "loss": 0.6953,
+      "step": 2749
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.5637,
+      "step": 2750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.085507246376812e-06,
+      "loss": 0.5528,
+      "step": 2751
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.084057971014493e-06,
+      "loss": 1.0772,
+      "step": 2752
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.082608695652175e-06,
+      "loss": 0.9031,
+      "step": 2753
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.0811594202898564e-06,
+      "loss": 0.3594,
+      "step": 2754
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.079710144927537e-06,
+      "loss": 0.3972,
+      "step": 2755
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.078260869565218e-06,
+      "loss": 0.4921,
+      "step": 2756
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.0768115942028995e-06,
+      "loss": 0.7632,
+      "step": 2757
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.07536231884058e-06,
+      "loss": 0.658,
+      "step": 2758
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.0739130434782615e-06,
+      "loss": 1.1073,
+      "step": 2759
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.0724637681159425e-06,
+      "loss": 0.3441,
+      "step": 2760
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.071014492753624e-06,
+      "loss": 0.9864,
+      "step": 2761
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.0695652173913045e-06,
+      "loss": 0.8863,
+      "step": 2762
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.068115942028986e-06,
+      "loss": 0.6358,
+      "step": 2763
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.066666666666667e-06,
+      "loss": 1.152,
+      "step": 2764
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.065217391304349e-06,
+      "loss": 0.8965,
+      "step": 2765
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.063768115942029e-06,
+      "loss": 0.7359,
+      "step": 2766
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.06231884057971e-06,
+      "loss": 0.4293,
+      "step": 2767
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.060869565217392e-06,
+      "loss": 1.6614,
+      "step": 2768
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.059420289855072e-06,
+      "loss": 0.1556,
+      "step": 2769
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.057971014492754e-06,
+      "loss": 1.5181,
+      "step": 2770
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.056521739130435e-06,
+      "loss": 0.881,
+      "step": 2771
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.055072463768117e-06,
+      "loss": 0.5605,
+      "step": 2772
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.053623188405797e-06,
+      "loss": 0.6606,
+      "step": 2773
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.052173913043479e-06,
+      "loss": 0.462,
+      "step": 2774
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.05072463768116e-06,
+      "loss": 0.5226,
+      "step": 2775
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.049275362318842e-06,
+      "loss": 0.7095,
+      "step": 2776
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.047826086956522e-06,
+      "loss": 1.0415,
+      "step": 2777
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.046376811594203e-06,
+      "loss": 1.2299,
+      "step": 2778
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.044927536231885e-06,
+      "loss": 0.9617,
+      "step": 2779
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.043478260869565e-06,
+      "loss": 0.234,
+      "step": 2780
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.8345323741007195,
+      "eval_loss": 0.5955384373664856,
+      "eval_roc_auc": 0.9672990913802821,
+      "eval_runtime": 88.4541,
+      "eval_samples_per_second": 7.857,
+      "eval_steps_per_second": 1.967,
+      "step": 2780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.042028985507247e-06,
+      "loss": 1.9964,
+      "step": 2781
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.040579710144928e-06,
+      "loss": 0.8363,
+      "step": 2782
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.03913043478261e-06,
+      "loss": 1.0024,
+      "step": 2783
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.03768115942029e-06,
+      "loss": 0.8231,
+      "step": 2784
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.036231884057972e-06,
+      "loss": 0.5501,
+      "step": 2785
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.034782608695653e-06,
+      "loss": 0.5998,
+      "step": 2786
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.033333333333335e-06,
+      "loss": 0.5293,
+      "step": 2787
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.031884057971015e-06,
+      "loss": 0.613,
+      "step": 2788
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.030434782608697e-06,
+      "loss": 0.3846,
+      "step": 2789
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.028985507246378e-06,
+      "loss": 0.5377,
+      "step": 2790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.027536231884058e-06,
+      "loss": 1.7082,
+      "step": 2791
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.02608695652174e-06,
+      "loss": 0.863,
+      "step": 2792
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.024637681159421e-06,
+      "loss": 1.5062,
+      "step": 2793
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.0231884057971024e-06,
+      "loss": 0.5786,
+      "step": 2794
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.021739130434783e-06,
+      "loss": 0.9788,
+      "step": 2795
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.0202898550724644e-06,
+      "loss": 0.5422,
+      "step": 2796
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.0188405797101454e-06,
+      "loss": 0.4249,
+      "step": 2797
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.0173913043478264e-06,
+      "loss": 0.9293,
+      "step": 2798
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.0159420289855074e-06,
+      "loss": 0.4894,
+      "step": 2799
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.014492753623189e-06,
+      "loss": 0.4647,
+      "step": 2800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.01304347826087e-06,
+      "loss": 0.3523,
+      "step": 2801
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.0115942028985504e-06,
+      "loss": 1.166,
+      "step": 2802
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.010144927536232e-06,
+      "loss": 1.7054,
+      "step": 2803
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.008695652173913e-06,
+      "loss": 0.3769,
+      "step": 2804
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.007246376811595e-06,
+      "loss": 0.7295,
+      "step": 2805
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.005797101449275e-06,
+      "loss": 0.605,
+      "step": 2806
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.004347826086957e-06,
+      "loss": 1.7912,
+      "step": 2807
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.002898550724638e-06,
+      "loss": 0.6318,
+      "step": 2808
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.001449275362319e-06,
+      "loss": 0.7427,
+      "step": 2809
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6e-06,
+      "loss": 0.8665,
+      "step": 2810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.998550724637682e-06,
+      "loss": 0.6109,
+      "step": 2811
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.997101449275363e-06,
+      "loss": 0.4635,
+      "step": 2812
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.995652173913043e-06,
+      "loss": 0.8608,
+      "step": 2813
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.994202898550725e-06,
+      "loss": 0.9308,
+      "step": 2814
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.992753623188407e-06,
+      "loss": 0.9681,
+      "step": 2815
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.991304347826088e-06,
+      "loss": 0.5316,
+      "step": 2816
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.989855072463768e-06,
+      "loss": 0.6112,
+      "step": 2817
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.98840579710145e-06,
+      "loss": 0.5792,
+      "step": 2818
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.986956521739131e-06,
+      "loss": 1.4065,
+      "step": 2819
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.985507246376812e-06,
+      "loss": 1.164,
+      "step": 2820
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.984057971014493e-06,
+      "loss": 0.6948,
+      "step": 2821
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.982608695652175e-06,
+      "loss": 0.4467,
+      "step": 2822
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.981159420289856e-06,
+      "loss": 0.4376,
+      "step": 2823
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.979710144927537e-06,
+      "loss": 0.5349,
+      "step": 2824
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.978260869565218e-06,
+      "loss": 0.368,
+      "step": 2825
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9768115942028996e-06,
+      "loss": 1.4963,
+      "step": 2826
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9753623188405806e-06,
+      "loss": 1.7742,
+      "step": 2827
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.973913043478261e-06,
+      "loss": 2.0123,
+      "step": 2828
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9724637681159426e-06,
+      "loss": 0.8611,
+      "step": 2829
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9710144927536236e-06,
+      "loss": 0.581,
+      "step": 2830
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9695652173913046e-06,
+      "loss": 0.2993,
+      "step": 2831
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9681159420289856e-06,
+      "loss": 0.4602,
+      "step": 2832
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.966666666666667e-06,
+      "loss": 0.6504,
+      "step": 2833
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.965217391304348e-06,
+      "loss": 0.6902,
+      "step": 2834
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.963768115942029e-06,
+      "loss": 0.6643,
+      "step": 2835
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.9623188405797104e-06,
+      "loss": 0.6715,
+      "step": 2836
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.960869565217392e-06,
+      "loss": 0.3606,
+      "step": 2837
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.959420289855073e-06,
+      "loss": 1.266,
+      "step": 2838
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.9579710144927534e-06,
+      "loss": 0.1747,
+      "step": 2839
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.956521739130435e-06,
+      "loss": 0.7273,
+      "step": 2840
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.955072463768117e-06,
+      "loss": 0.9406,
+      "step": 2841
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.953623188405797e-06,
+      "loss": 0.9015,
+      "step": 2842
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.952173913043478e-06,
+      "loss": 0.9683,
+      "step": 2843
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.95072463768116e-06,
+      "loss": 0.5716,
+      "step": 2844
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.949275362318841e-06,
+      "loss": 1.0105,
+      "step": 2845
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.947826086956522e-06,
+      "loss": 0.8913,
+      "step": 2846
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.946376811594203e-06,
+      "loss": 1.109,
+      "step": 2847
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.944927536231885e-06,
+      "loss": 1.0728,
+      "step": 2848
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.943478260869566e-06,
+      "loss": 0.6249,
+      "step": 2849
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.942028985507247e-06,
+      "loss": 0.2967,
+      "step": 2850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.940579710144928e-06,
+      "loss": 0.3121,
+      "step": 2851
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.93913043478261e-06,
+      "loss": 0.6788,
+      "step": 2852
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.93768115942029e-06,
+      "loss": 0.8754,
+      "step": 2853
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.936231884057971e-06,
+      "loss": 0.6351,
+      "step": 2854
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.934782608695653e-06,
+      "loss": 0.5049,
+      "step": 2855
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.933333333333335e-06,
+      "loss": 0.8322,
+      "step": 2856
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.931884057971015e-06,
+      "loss": 1.4463,
+      "step": 2857
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.930434782608696e-06,
+      "loss": 1.0276,
+      "step": 2858
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.928985507246378e-06,
+      "loss": 0.4974,
+      "step": 2859
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.927536231884058e-06,
+      "loss": 0.4735,
+      "step": 2860
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.92608695652174e-06,
+      "loss": 0.4375,
+      "step": 2861
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.924637681159421e-06,
+      "loss": 2.0241,
+      "step": 2862
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.9231884057971025e-06,
+      "loss": 0.5578,
+      "step": 2863
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.921739130434783e-06,
+      "loss": 1.1856,
+      "step": 2864
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.920289855072464e-06,
+      "loss": 0.7348,
+      "step": 2865
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.9188405797101455e-06,
+      "loss": 0.5347,
+      "step": 2866
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.917391304347827e-06,
+      "loss": 0.185,
+      "step": 2867
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.9159420289855076e-06,
+      "loss": 0.7002,
+      "step": 2868
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.9144927536231886e-06,
+      "loss": 0.7863,
+      "step": 2869
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.91304347826087e-06,
+      "loss": 0.7584,
+      "step": 2870
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.9115942028985506e-06,
+      "loss": 0.6449,
+      "step": 2871
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.910144927536232e-06,
+      "loss": 0.424,
+      "step": 2872
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.908695652173913e-06,
+      "loss": 0.499,
+      "step": 2873
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.907246376811595e-06,
+      "loss": 0.3078,
+      "step": 2874
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.905797101449275e-06,
+      "loss": 0.6111,
+      "step": 2875
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.904347826086957e-06,
+      "loss": 0.3014,
+      "step": 2876
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.902898550724638e-06,
+      "loss": 2.5268,
+      "step": 2877
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.90144927536232e-06,
+      "loss": 0.8471,
+      "step": 2878
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.9e-06,
+      "loss": 0.6852,
+      "step": 2879
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.898550724637681e-06,
+      "loss": 0.656,
+      "step": 2880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.897101449275363e-06,
+      "loss": 0.8636,
+      "step": 2881
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.895652173913043e-06,
+      "loss": 0.8477,
+      "step": 2882
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.894202898550725e-06,
+      "loss": 1.1484,
+      "step": 2883
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.892753623188406e-06,
+      "loss": 0.4017,
+      "step": 2884
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.891304347826088e-06,
+      "loss": 0.3782,
+      "step": 2885
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.889855072463768e-06,
+      "loss": 1.2894,
+      "step": 2886
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.88840579710145e-06,
+      "loss": 0.4131,
+      "step": 2887
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.886956521739131e-06,
+      "loss": 0.6051,
+      "step": 2888
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.885507246376813e-06,
+      "loss": 0.4059,
+      "step": 2889
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.884057971014493e-06,
+      "loss": 0.8455,
+      "step": 2890
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.882608695652175e-06,
+      "loss": 0.518,
+      "step": 2891
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.881159420289856e-06,
+      "loss": 0.5883,
+      "step": 2892
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.879710144927536e-06,
+      "loss": 0.5803,
+      "step": 2893
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.878260869565218e-06,
+      "loss": 0.6739,
+      "step": 2894
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.876811594202899e-06,
+      "loss": 1.0422,
+      "step": 2895
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.875362318840581e-06,
+      "loss": 0.6637,
+      "step": 2896
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.873913043478261e-06,
+      "loss": 0.6317,
+      "step": 2897
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.872463768115943e-06,
+      "loss": 0.5593,
+      "step": 2898
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.871014492753624e-06,
+      "loss": 1.2294,
+      "step": 2899
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.8695652173913055e-06,
+      "loss": 0.6467,
+      "step": 2900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.868115942028986e-06,
+      "loss": 0.4249,
+      "step": 2901
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 0.6229,
+      "step": 2902
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.8652173913043485e-06,
+      "loss": 0.9553,
+      "step": 2903
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.863768115942029e-06,
+      "loss": 0.5551,
+      "step": 2904
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.8623188405797105e-06,
+      "loss": 0.3011,
+      "step": 2905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.8608695652173915e-06,
+      "loss": 0.6673,
+      "step": 2906
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.859420289855073e-06,
+      "loss": 1.1261,
+      "step": 2907
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.8579710144927535e-06,
+      "loss": 0.5053,
+      "step": 2908
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.856521739130435e-06,
+      "loss": 0.2684,
+      "step": 2909
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.855072463768116e-06,
+      "loss": 0.2071,
+      "step": 2910
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.853623188405798e-06,
+      "loss": 0.5592,
+      "step": 2911
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.852173913043478e-06,
+      "loss": 1.5819,
+      "step": 2912
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.85072463768116e-06,
+      "loss": 0.792,
+      "step": 2913
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.849275362318841e-06,
+      "loss": 0.9497,
+      "step": 2914
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.847826086956521e-06,
+      "loss": 0.3311,
+      "step": 2915
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.846376811594203e-06,
+      "loss": 0.1631,
+      "step": 2916
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.844927536231885e-06,
+      "loss": 0.2983,
+      "step": 2917
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.843478260869566e-06,
+      "loss": 0.8611,
+      "step": 2918
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.842028985507246e-06,
+      "loss": 0.3105,
+      "step": 2919
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.840579710144928e-06,
+      "loss": 0.7163,
+      "step": 2920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.839130434782609e-06,
+      "loss": 1.2301,
+      "step": 2921
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.83768115942029e-06,
+      "loss": 1.3942,
+      "step": 2922
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.836231884057971e-06,
+      "loss": 0.5467,
+      "step": 2923
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.834782608695653e-06,
+      "loss": 0.8895,
+      "step": 2924
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.557,
+      "step": 2925
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.831884057971015e-06,
+      "loss": 0.9951,
+      "step": 2926
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.830434782608696e-06,
+      "loss": 1.1833,
+      "step": 2927
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.828985507246378e-06,
+      "loss": 0.8592,
+      "step": 2928
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.827536231884059e-06,
+      "loss": 0.4715,
+      "step": 2929
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.826086956521739e-06,
+      "loss": 1.0611,
+      "step": 2930
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.824637681159421e-06,
+      "loss": 1.5649,
+      "step": 2931
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.823188405797102e-06,
+      "loss": 0.7529,
+      "step": 2932
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.821739130434783e-06,
+      "loss": 0.3839,
+      "step": 2933
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.820289855072464e-06,
+      "loss": 0.4574,
+      "step": 2934
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.818840579710146e-06,
+      "loss": 0.7514,
+      "step": 2935
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.817391304347827e-06,
+      "loss": 0.2845,
+      "step": 2936
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.815942028985508e-06,
+      "loss": 0.431,
+      "step": 2937
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.814492753623189e-06,
+      "loss": 0.2081,
+      "step": 2938
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8130434782608705e-06,
+      "loss": 0.7127,
+      "step": 2939
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8115942028985515e-06,
+      "loss": 0.6374,
+      "step": 2940
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.810144927536232e-06,
+      "loss": 0.398,
+      "step": 2941
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8086956521739135e-06,
+      "loss": 0.801,
+      "step": 2942
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.807246376811595e-06,
+      "loss": 0.3066,
+      "step": 2943
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8057971014492755e-06,
+      "loss": 1.6148,
+      "step": 2944
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8043478260869565e-06,
+      "loss": 1.4797,
+      "step": 2945
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.802898550724638e-06,
+      "loss": 0.7859,
+      "step": 2946
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.801449275362319e-06,
+      "loss": 0.2505,
+      "step": 2947
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8e-06,
+      "loss": 1.2281,
+      "step": 2948
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.798550724637681e-06,
+      "loss": 1.1464,
+      "step": 2949
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.797101449275363e-06,
+      "loss": 1.1362,
+      "step": 2950
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.795652173913044e-06,
+      "loss": 1.1989,
+      "step": 2951
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.794202898550725e-06,
+      "loss": 1.4487,
+      "step": 2952
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.792753623188406e-06,
+      "loss": 0.2446,
+      "step": 2953
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.791304347826088e-06,
+      "loss": 0.8114,
+      "step": 2954
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.789855072463768e-06,
+      "loss": 0.5103,
+      "step": 2955
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.788405797101449e-06,
+      "loss": 1.0151,
+      "step": 2956
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.786956521739131e-06,
+      "loss": 0.6697,
+      "step": 2957
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.785507246376812e-06,
+      "loss": 0.6599,
+      "step": 2958
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.784057971014493e-06,
+      "loss": 0.8742,
+      "step": 2959
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.782608695652174e-06,
+      "loss": 0.5925,
+      "step": 2960
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.781159420289856e-06,
+      "loss": 0.4813,
+      "step": 2961
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.779710144927537e-06,
+      "loss": 0.6337,
+      "step": 2962
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.778260869565218e-06,
+      "loss": 0.4228,
+      "step": 2963
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.776811594202899e-06,
+      "loss": 1.149,
+      "step": 2964
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.775362318840581e-06,
+      "loss": 0.4372,
+      "step": 2965
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.773913043478261e-06,
+      "loss": 0.7484,
+      "step": 2966
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.772463768115942e-06,
+      "loss": 0.8435,
+      "step": 2967
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.771014492753624e-06,
+      "loss": 0.753,
+      "step": 2968
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.769565217391306e-06,
+      "loss": 1.3434,
+      "step": 2969
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.768115942028986e-06,
+      "loss": 0.434,
+      "step": 2970
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.766666666666667e-06,
+      "loss": 1.2898,
+      "step": 2971
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.765217391304349e-06,
+      "loss": 1.034,
+      "step": 2972
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.76376811594203e-06,
+      "loss": 0.2644,
+      "step": 2973
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.762318840579711e-06,
+      "loss": 0.2696,
+      "step": 2974
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.760869565217392e-06,
+      "loss": 1.0691,
+      "step": 2975
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.7594202898550735e-06,
+      "loss": 0.91,
+      "step": 2976
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.757971014492754e-06,
+      "loss": 0.6959,
+      "step": 2977
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.7565217391304355e-06,
+      "loss": 1.1346,
+      "step": 2978
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.7550724637681165e-06,
+      "loss": 0.6573,
+      "step": 2979
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.753623188405798e-06,
+      "loss": 0.5783,
+      "step": 2980
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.7521739130434785e-06,
+      "loss": 0.5252,
+      "step": 2981
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.7507246376811595e-06,
+      "loss": 0.3078,
+      "step": 2982
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.749275362318841e-06,
+      "loss": 0.5719,
+      "step": 2983
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.7478260869565215e-06,
+      "loss": 1.8119,
+      "step": 2984
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.746376811594203e-06,
+      "loss": 0.4805,
+      "step": 2985
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.744927536231884e-06,
+      "loss": 0.991,
+      "step": 2986
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.743478260869566e-06,
+      "loss": 0.8537,
+      "step": 2987
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.742028985507246e-06,
+      "loss": 0.8468,
+      "step": 2988
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.740579710144928e-06,
+      "loss": 0.2011,
+      "step": 2989
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.739130434782609e-06,
+      "loss": 1.2012,
+      "step": 2990
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.737681159420291e-06,
+      "loss": 0.4553,
+      "step": 2991
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.736231884057971e-06,
+      "loss": 1.1401,
+      "step": 2992
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.734782608695652e-06,
+      "loss": 1.5433,
+      "step": 2993
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 0.6233,
+      "step": 2994
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.731884057971014e-06,
+      "loss": 0.4444,
+      "step": 2995
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.730434782608696e-06,
+      "loss": 0.64,
+      "step": 2996
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.728985507246377e-06,
+      "loss": 1.0934,
+      "step": 2997
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.727536231884059e-06,
+      "loss": 0.5123,
+      "step": 2998
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.726086956521739e-06,
+      "loss": 0.4674,
+      "step": 2999
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.724637681159421e-06,
+      "loss": 0.8897,
+      "step": 3000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.723188405797102e-06,
+      "loss": 0.395,
+      "step": 3001
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.721739130434784e-06,
+      "loss": 0.4521,
+      "step": 3002
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.720289855072464e-06,
+      "loss": 0.6284,
+      "step": 3003
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.718840579710146e-06,
+      "loss": 0.7767,
+      "step": 3004
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.717391304347827e-06,
+      "loss": 0.6763,
+      "step": 3005
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.715942028985507e-06,
+      "loss": 0.5164,
+      "step": 3006
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.714492753623189e-06,
+      "loss": 0.4289,
+      "step": 3007
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.71304347826087e-06,
+      "loss": 1.2683,
+      "step": 3008
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.711594202898552e-06,
+      "loss": 0.9035,
+      "step": 3009
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.710144927536232e-06,
+      "loss": 0.6301,
+      "step": 3010
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.708695652173914e-06,
+      "loss": 1.5773,
+      "step": 3011
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.707246376811595e-06,
+      "loss": 1.1549,
+      "step": 3012
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.7057971014492765e-06,
+      "loss": 0.6753,
+      "step": 3013
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.704347826086957e-06,
+      "loss": 0.4742,
+      "step": 3014
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.7028985507246385e-06,
+      "loss": 0.5963,
+      "step": 3015
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.7014492753623195e-06,
+      "loss": 0.8185,
+      "step": 3016
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.7e-06,
+      "loss": 0.5766,
+      "step": 3017
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.6985507246376815e-06,
+      "loss": 0.6028,
+      "step": 3018
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.6971014492753625e-06,
+      "loss": 0.2232,
+      "step": 3019
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.695652173913044e-06,
+      "loss": 0.1356,
+      "step": 3020
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.6942028985507245e-06,
+      "loss": 0.3484,
+      "step": 3021
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.692753623188406e-06,
+      "loss": 0.4187,
+      "step": 3022
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.691304347826087e-06,
+      "loss": 1.238,
+      "step": 3023
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.689855072463769e-06,
+      "loss": 0.4806,
+      "step": 3024
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.688405797101449e-06,
+      "loss": 1.0191,
+      "step": 3025
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.686956521739131e-06,
+      "loss": 1.5565,
+      "step": 3026
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.685507246376812e-06,
+      "loss": 0.2592,
+      "step": 3027
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.684057971014492e-06,
+      "loss": 0.6731,
+      "step": 3028
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.682608695652174e-06,
+      "loss": 0.6906,
+      "step": 3029
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.681159420289856e-06,
+      "loss": 0.3894,
+      "step": 3030
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.679710144927537e-06,
+      "loss": 1.5348,
+      "step": 3031
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.678260869565217e-06,
+      "loss": 0.6173,
+      "step": 3032
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.676811594202899e-06,
+      "loss": 0.5762,
+      "step": 3033
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.67536231884058e-06,
+      "loss": 0.552,
+      "step": 3034
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.673913043478262e-06,
+      "loss": 0.244,
+      "step": 3035
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.672463768115942e-06,
+      "loss": 1.2036,
+      "step": 3036
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.671014492753624e-06,
+      "loss": 0.3716,
+      "step": 3037
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.669565217391305e-06,
+      "loss": 0.7356,
+      "step": 3038
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.668115942028986e-06,
+      "loss": 1.5765,
+      "step": 3039
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 1.0223,
+      "step": 3040
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.665217391304349e-06,
+      "loss": 1.5789,
+      "step": 3041
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.66376811594203e-06,
+      "loss": 0.7084,
+      "step": 3042
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.66231884057971e-06,
+      "loss": 0.8372,
+      "step": 3043
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.660869565217392e-06,
+      "loss": 0.7234,
+      "step": 3044
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.659420289855074e-06,
+      "loss": 1.7128,
+      "step": 3045
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.657971014492754e-06,
+      "loss": 0.7941,
+      "step": 3046
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.656521739130435e-06,
+      "loss": 0.7772,
+      "step": 3047
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.655072463768117e-06,
+      "loss": 0.3774,
+      "step": 3048
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.653623188405798e-06,
+      "loss": 0.4611,
+      "step": 3049
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 0.5068,
+      "step": 3050
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.65072463768116e-06,
+      "loss": 0.3404,
+      "step": 3051
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.6492753623188415e-06,
+      "loss": 0.77,
+      "step": 3052
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.6478260869565225e-06,
+      "loss": 0.0836,
+      "step": 3053
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.646376811594203e-06,
+      "loss": 0.4288,
+      "step": 3054
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.6449275362318845e-06,
+      "loss": 0.7402,
+      "step": 3055
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.643478260869566e-06,
+      "loss": 0.5434,
+      "step": 3056
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.6420289855072465e-06,
+      "loss": 1.132,
+      "step": 3057
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.6405797101449275e-06,
+      "loss": 0.2705,
+      "step": 3058
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.639130434782609e-06,
+      "loss": 0.8962,
+      "step": 3059
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.63768115942029e-06,
+      "loss": 1.1154,
+      "step": 3060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.636231884057971e-06,
+      "loss": 0.4702,
+      "step": 3061
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.634782608695652e-06,
+      "loss": 1.5294,
+      "step": 3062
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.633333333333334e-06,
+      "loss": 0.5048,
+      "step": 3063
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.631884057971015e-06,
+      "loss": 0.1374,
+      "step": 3064
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.630434782608696e-06,
+      "loss": 0.4114,
+      "step": 3065
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.628985507246377e-06,
+      "loss": 1.5224,
+      "step": 3066
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.627536231884059e-06,
+      "loss": 1.6664,
+      "step": 3067
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.626086956521739e-06,
+      "loss": 0.5298,
+      "step": 3068
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.62463768115942e-06,
+      "loss": 0.3563,
+      "step": 3069
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.623188405797102e-06,
+      "loss": 0.3272,
+      "step": 3070
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.621739130434784e-06,
+      "loss": 1.0799,
+      "step": 3071
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.620289855072464e-06,
+      "loss": 1.1386,
+      "step": 3072
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.618840579710145e-06,
+      "loss": 0.4074,
+      "step": 3073
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.617391304347827e-06,
+      "loss": 1.624,
+      "step": 3074
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.615942028985508e-06,
+      "loss": 0.225,
+      "step": 3075
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.614492753623189e-06,
+      "loss": 0.7434,
+      "step": 3076
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.61304347826087e-06,
+      "loss": 0.3941,
+      "step": 3077
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.611594202898552e-06,
+      "loss": 0.4993,
+      "step": 3078
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.610144927536232e-06,
+      "loss": 0.1985,
+      "step": 3079
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.608695652173914e-06,
+      "loss": 0.4884,
+      "step": 3080
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.607246376811595e-06,
+      "loss": 0.6487,
+      "step": 3081
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.605797101449277e-06,
+      "loss": 1.4234,
+      "step": 3082
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.604347826086957e-06,
+      "loss": 0.6262,
+      "step": 3083
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.602898550724638e-06,
+      "loss": 0.9148,
+      "step": 3084
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.60144927536232e-06,
+      "loss": 0.2104,
+      "step": 3085
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.4441,
+      "step": 3086
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.598550724637682e-06,
+      "loss": 0.302,
+      "step": 3087
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.597101449275363e-06,
+      "loss": 0.7835,
+      "step": 3088
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.5956521739130444e-06,
+      "loss": 0.142,
+      "step": 3089
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.594202898550725e-06,
+      "loss": 0.3065,
+      "step": 3090
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.5927536231884064e-06,
+      "loss": 0.9859,
+      "step": 3091
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.5913043478260874e-06,
+      "loss": 0.9891,
+      "step": 3092
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.589855072463769e-06,
+      "loss": 1.1432,
+      "step": 3093
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.5884057971014495e-06,
+      "loss": 0.2975,
+      "step": 3094
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.5869565217391305e-06,
+      "loss": 0.4706,
+      "step": 3095
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.585507246376812e-06,
+      "loss": 0.3927,
+      "step": 3096
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.584057971014494e-06,
+      "loss": 0.4667,
+      "step": 3097
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.582608695652174e-06,
+      "loss": 0.3644,
+      "step": 3098
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.581159420289855e-06,
+      "loss": 0.4934,
+      "step": 3099
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.579710144927537e-06,
+      "loss": 0.6441,
+      "step": 3100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.578260869565217e-06,
+      "loss": 0.2186,
+      "step": 3101
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.576811594202899e-06,
+      "loss": 0.4441,
+      "step": 3102
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.57536231884058e-06,
+      "loss": 0.1859,
+      "step": 3103
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.573913043478262e-06,
+      "loss": 0.4242,
+      "step": 3104
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.572463768115942e-06,
+      "loss": 0.4705,
+      "step": 3105
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.571014492753624e-06,
+      "loss": 0.8124,
+      "step": 3106
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.569565217391305e-06,
+      "loss": 1.1978,
+      "step": 3107
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.568115942028985e-06,
+      "loss": 0.5201,
+      "step": 3108
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.566666666666667e-06,
+      "loss": 1.297,
+      "step": 3109
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.565217391304348e-06,
+      "loss": 0.3519,
+      "step": 3110
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.56376811594203e-06,
+      "loss": 0.2551,
+      "step": 3111
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.56231884057971e-06,
+      "loss": 0.2929,
+      "step": 3112
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.560869565217392e-06,
+      "loss": 1.0397,
+      "step": 3113
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.559420289855073e-06,
+      "loss": 0.2595,
+      "step": 3114
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.557971014492755e-06,
+      "loss": 0.2051,
+      "step": 3115
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.556521739130435e-06,
+      "loss": 0.6857,
+      "step": 3116
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.555072463768117e-06,
+      "loss": 1.0026,
+      "step": 3117
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.553623188405798e-06,
+      "loss": 0.5759,
+      "step": 3118
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.552173913043478e-06,
+      "loss": 0.4428,
+      "step": 3119
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.55072463768116e-06,
+      "loss": 0.9378,
+      "step": 3120
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.549275362318841e-06,
+      "loss": 0.9292,
+      "step": 3121
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.5478260869565226e-06,
+      "loss": 0.0995,
+      "step": 3122
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.546376811594203e-06,
+      "loss": 0.6559,
+      "step": 3123
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.544927536231885e-06,
+      "loss": 0.2804,
+      "step": 3124
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.543478260869566e-06,
+      "loss": 0.6593,
+      "step": 3125
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.5420289855072474e-06,
+      "loss": 0.3148,
+      "step": 3126
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.540579710144928e-06,
+      "loss": 0.2474,
+      "step": 3127
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.5391304347826094e-06,
+      "loss": 0.3086,
+      "step": 3128
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.5376811594202904e-06,
+      "loss": 0.4969,
+      "step": 3129
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.536231884057971e-06,
+      "loss": 0.6229,
+      "step": 3130
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.5347826086956524e-06,
+      "loss": 0.3681,
+      "step": 3131
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.533333333333334e-06,
+      "loss": 0.7927,
+      "step": 3132
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.531884057971015e-06,
+      "loss": 0.3564,
+      "step": 3133
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.5304347826086954e-06,
+      "loss": 0.5868,
+      "step": 3134
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.528985507246377e-06,
+      "loss": 0.4045,
+      "step": 3135
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.527536231884058e-06,
+      "loss": 0.6791,
+      "step": 3136
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.52608695652174e-06,
+      "loss": 0.672,
+      "step": 3137
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.52463768115942e-06,
+      "loss": 0.1996,
+      "step": 3138
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.523188405797102e-06,
+      "loss": 0.4883,
+      "step": 3139
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.521739130434783e-06,
+      "loss": 1.1941,
+      "step": 3140
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.520289855072464e-06,
+      "loss": 0.0994,
+      "step": 3141
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.518840579710145e-06,
+      "loss": 0.2595,
+      "step": 3142
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.517391304347827e-06,
+      "loss": 0.8514,
+      "step": 3143
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.515942028985508e-06,
+      "loss": 1.1645,
+      "step": 3144
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.514492753623188e-06,
+      "loss": 1.8704,
+      "step": 3145
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.51304347826087e-06,
+      "loss": 0.5711,
+      "step": 3146
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.511594202898551e-06,
+      "loss": 0.2366,
+      "step": 3147
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.510144927536233e-06,
+      "loss": 1.713,
+      "step": 3148
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.508695652173913e-06,
+      "loss": 0.6483,
+      "step": 3149
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.507246376811595e-06,
+      "loss": 0.9638,
+      "step": 3150
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.505797101449276e-06,
+      "loss": 0.2946,
+      "step": 3151
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.504347826086957e-06,
+      "loss": 0.334,
+      "step": 3152
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.502898550724638e-06,
+      "loss": 1.4452,
+      "step": 3153
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.50144927536232e-06,
+      "loss": 0.9171,
+      "step": 3154
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.3132,
+      "step": 3155
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.498550724637681e-06,
+      "loss": 0.4692,
+      "step": 3156
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.497101449275363e-06,
+      "loss": 0.815,
+      "step": 3157
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.4956521739130446e-06,
+      "loss": 1.0865,
+      "step": 3158
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4942028985507256e-06,
+      "loss": 0.301,
+      "step": 3159
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.492753623188406e-06,
+      "loss": 0.3538,
+      "step": 3160
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4913043478260876e-06,
+      "loss": 0.4305,
+      "step": 3161
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4898550724637686e-06,
+      "loss": 0.1968,
+      "step": 3162
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4884057971014496e-06,
+      "loss": 0.6328,
+      "step": 3163
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4869565217391306e-06,
+      "loss": 0.3781,
+      "step": 3164
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.485507246376812e-06,
+      "loss": 0.7975,
+      "step": 3165
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.484057971014493e-06,
+      "loss": 0.8038,
+      "step": 3166
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.482608695652174e-06,
+      "loss": 0.4808,
+      "step": 3167
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.481159420289855e-06,
+      "loss": 0.8767,
+      "step": 3168
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.479710144927537e-06,
+      "loss": 0.4256,
+      "step": 3169
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.478260869565217e-06,
+      "loss": 0.4048,
+      "step": 3170
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.476811594202898e-06,
+      "loss": 0.3817,
+      "step": 3171
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.47536231884058e-06,
+      "loss": 0.2003,
+      "step": 3172
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.473913043478261e-06,
+      "loss": 1.17,
+      "step": 3173
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.472463768115942e-06,
+      "loss": 0.093,
+      "step": 3174
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.471014492753623e-06,
+      "loss": 1.0099,
+      "step": 3175
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.469565217391305e-06,
+      "loss": 0.4879,
+      "step": 3176
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.468115942028986e-06,
+      "loss": 0.787,
+      "step": 3177
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 0.2833,
+      "step": 3178
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.465217391304348e-06,
+      "loss": 0.4792,
+      "step": 3179
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.46376811594203e-06,
+      "loss": 0.2985,
+      "step": 3180
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.46231884057971e-06,
+      "loss": 0.4983,
+      "step": 3181
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.460869565217391e-06,
+      "loss": 0.7881,
+      "step": 3182
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.459420289855073e-06,
+      "loss": 0.5872,
+      "step": 3183
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.457971014492755e-06,
+      "loss": 0.5566,
+      "step": 3184
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.456521739130435e-06,
+      "loss": 0.926,
+      "step": 3185
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.455072463768116e-06,
+      "loss": 0.2258,
+      "step": 3186
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.453623188405798e-06,
+      "loss": 1.0952,
+      "step": 3187
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.452173913043479e-06,
+      "loss": 0.8376,
+      "step": 3188
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.45072463768116e-06,
+      "loss": 0.3677,
+      "step": 3189
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.449275362318841e-06,
+      "loss": 1.8837,
+      "step": 3190
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.447826086956523e-06,
+      "loss": 0.4066,
+      "step": 3191
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.446376811594203e-06,
+      "loss": 0.2082,
+      "step": 3192
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.444927536231885e-06,
+      "loss": 0.4044,
+      "step": 3193
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.443478260869566e-06,
+      "loss": 0.829,
+      "step": 3194
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4420289855072475e-06,
+      "loss": 0.255,
+      "step": 3195
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.440579710144928e-06,
+      "loss": 0.5936,
+      "step": 3196
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.439130434782609e-06,
+      "loss": 0.359,
+      "step": 3197
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4376811594202905e-06,
+      "loss": 0.6834,
+      "step": 3198
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.436231884057972e-06,
+      "loss": 0.3496,
+      "step": 3199
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 0.7319,
+      "step": 3200
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.4333333333333335e-06,
+      "loss": 0.2716,
+      "step": 3201
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.431884057971015e-06,
+      "loss": 0.7405,
+      "step": 3202
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.4304347826086955e-06,
+      "loss": 0.3359,
+      "step": 3203
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.428985507246377e-06,
+      "loss": 0.3162,
+      "step": 3204
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.427536231884058e-06,
+      "loss": 0.3502,
+      "step": 3205
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.42608695652174e-06,
+      "loss": 1.2372,
+      "step": 3206
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.42463768115942e-06,
+      "loss": 0.7124,
+      "step": 3207
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.423188405797101e-06,
+      "loss": 0.7049,
+      "step": 3208
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.421739130434783e-06,
+      "loss": 1.026,
+      "step": 3209
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.420289855072465e-06,
+      "loss": 0.3258,
+      "step": 3210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.418840579710145e-06,
+      "loss": 0.4502,
+      "step": 3211
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.417391304347826e-06,
+      "loss": 0.2663,
+      "step": 3212
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.415942028985508e-06,
+      "loss": 1.1003,
+      "step": 3213
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.414492753623188e-06,
+      "loss": 0.8475,
+      "step": 3214
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.41304347826087e-06,
+      "loss": 0.6342,
+      "step": 3215
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.411594202898551e-06,
+      "loss": 0.5827,
+      "step": 3216
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.410144927536233e-06,
+      "loss": 1.1766,
+      "step": 3217
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.408695652173913e-06,
+      "loss": 0.6489,
+      "step": 3218
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.407246376811595e-06,
+      "loss": 1.1801,
+      "step": 3219
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.405797101449276e-06,
+      "loss": 0.381,
+      "step": 3220
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.404347826086958e-06,
+      "loss": 1.1066,
+      "step": 3221
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.402898550724638e-06,
+      "loss": 0.8818,
+      "step": 3222
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.401449275362319e-06,
+      "loss": 0.3729,
+      "step": 3223
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.1668,
+      "step": 3224
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.398550724637681e-06,
+      "loss": 0.6118,
+      "step": 3225
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.397101449275363e-06,
+      "loss": 0.5083,
+      "step": 3226
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.395652173913044e-06,
+      "loss": 0.093,
+      "step": 3227
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.394202898550726e-06,
+      "loss": 0.4465,
+      "step": 3228
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.392753623188406e-06,
+      "loss": 0.6983,
+      "step": 3229
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.391304347826088e-06,
+      "loss": 0.7914,
+      "step": 3230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.389855072463769e-06,
+      "loss": 0.528,
+      "step": 3231
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.388405797101449e-06,
+      "loss": 0.7946,
+      "step": 3232
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.386956521739131e-06,
+      "loss": 0.3747,
+      "step": 3233
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.3855072463768125e-06,
+      "loss": 1.2386,
+      "step": 3234
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.3840579710144935e-06,
+      "loss": 1.2374,
+      "step": 3235
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.382608695652174e-06,
+      "loss": 0.9709,
+      "step": 3236
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.3811594202898555e-06,
+      "loss": 0.6546,
+      "step": 3237
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.3797101449275365e-06,
+      "loss": 2.0529,
+      "step": 3238
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.378260869565218e-06,
+      "loss": 0.2375,
+      "step": 3239
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.3768115942028985e-06,
+      "loss": 0.9623,
+      "step": 3240
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.37536231884058e-06,
+      "loss": 0.4241,
+      "step": 3241
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.373913043478261e-06,
+      "loss": 1.0615,
+      "step": 3242
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.3724637681159415e-06,
+      "loss": 0.6929,
+      "step": 3243
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.371014492753623e-06,
+      "loss": 1.1428,
+      "step": 3244
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.369565217391305e-06,
+      "loss": 0.4566,
+      "step": 3245
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.368115942028986e-06,
+      "loss": 1.2938,
+      "step": 3246
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.366666666666666e-06,
+      "loss": 0.6299,
+      "step": 3247
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.365217391304348e-06,
+      "loss": 0.2176,
+      "step": 3248
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.363768115942029e-06,
+      "loss": 0.159,
+      "step": 3249
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.362318840579711e-06,
+      "loss": 0.4317,
+      "step": 3250
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.360869565217391e-06,
+      "loss": 0.2067,
+      "step": 3251
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.359420289855073e-06,
+      "loss": 0.2255,
+      "step": 3252
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.357971014492754e-06,
+      "loss": 0.2697,
+      "step": 3253
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.356521739130435e-06,
+      "loss": 0.2203,
+      "step": 3254
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.355072463768116e-06,
+      "loss": 0.1352,
+      "step": 3255
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.353623188405798e-06,
+      "loss": 0.3105,
+      "step": 3256
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.352173913043479e-06,
+      "loss": 1.4637,
+      "step": 3257
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.350724637681159e-06,
+      "loss": 0.7782,
+      "step": 3258
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.349275362318841e-06,
+      "loss": 0.176,
+      "step": 3259
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.347826086956523e-06,
+      "loss": 0.2021,
+      "step": 3260
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.346376811594204e-06,
+      "loss": 0.2126,
+      "step": 3261
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.344927536231884e-06,
+      "loss": 0.1917,
+      "step": 3262
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.343478260869566e-06,
+      "loss": 0.2227,
+      "step": 3263
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.342028985507247e-06,
+      "loss": 0.6633,
+      "step": 3264
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.340579710144928e-06,
+      "loss": 0.0903,
+      "step": 3265
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.339130434782609e-06,
+      "loss": 0.1869,
+      "step": 3266
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.337681159420291e-06,
+      "loss": 1.6773,
+      "step": 3267
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.336231884057972e-06,
+      "loss": 1.2338,
+      "step": 3268
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.334782608695653e-06,
+      "loss": 0.3707,
+      "step": 3269
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.4417,
+      "step": 3270
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.3318840579710155e-06,
+      "loss": 0.4012,
+      "step": 3271
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.3304347826086965e-06,
+      "loss": 0.8613,
+      "step": 3272
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.328985507246377e-06,
+      "loss": 0.4828,
+      "step": 3273
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.3275362318840585e-06,
+      "loss": 1.7011,
+      "step": 3274
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.3260869565217395e-06,
+      "loss": 0.7426,
+      "step": 3275
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.3246376811594205e-06,
+      "loss": 2.6618,
+      "step": 3276
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.3231884057971015e-06,
+      "loss": 0.7748,
+      "step": 3277
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.321739130434783e-06,
+      "loss": 1.1349,
+      "step": 3278
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.320289855072464e-06,
+      "loss": 0.5568,
+      "step": 3279
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.318840579710145e-06,
+      "loss": 0.307,
+      "step": 3280
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.317391304347826e-06,
+      "loss": 0.5853,
+      "step": 3281
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.315942028985508e-06,
+      "loss": 0.4704,
+      "step": 3282
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.314492753623189e-06,
+      "loss": 1.4949,
+      "step": 3283
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.313043478260869e-06,
+      "loss": 0.0785,
+      "step": 3284
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.311594202898551e-06,
+      "loss": 0.153,
+      "step": 3285
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.310144927536233e-06,
+      "loss": 0.5336,
+      "step": 3286
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.308695652173913e-06,
+      "loss": 0.4415,
+      "step": 3287
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.307246376811594e-06,
+      "loss": 0.5187,
+      "step": 3288
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.305797101449276e-06,
+      "loss": 0.6965,
+      "step": 3289
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.304347826086957e-06,
+      "loss": 1.157,
+      "step": 3290
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.302898550724638e-06,
+      "loss": 0.4995,
+      "step": 3291
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.301449275362319e-06,
+      "loss": 0.7681,
+      "step": 3292
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.4241,
+      "step": 3293
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.298550724637681e-06,
+      "loss": 0.7907,
+      "step": 3294
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.297101449275363e-06,
+      "loss": 0.3847,
+      "step": 3295
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.295652173913044e-06,
+      "loss": 0.5981,
+      "step": 3296
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.294202898550726e-06,
+      "loss": 0.6118,
+      "step": 3297
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.292753623188406e-06,
+      "loss": 0.2378,
+      "step": 3298
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.291304347826087e-06,
+      "loss": 0.2399,
+      "step": 3299
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.289855072463769e-06,
+      "loss": 1.705,
+      "step": 3300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.28840579710145e-06,
+      "loss": 0.2642,
+      "step": 3301
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.286956521739131e-06,
+      "loss": 0.6207,
+      "step": 3302
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.285507246376812e-06,
+      "loss": 0.9125,
+      "step": 3303
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.284057971014494e-06,
+      "loss": 1.5532,
+      "step": 3304
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.282608695652174e-06,
+      "loss": 0.5342,
+      "step": 3305
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.281159420289856e-06,
+      "loss": 0.7291,
+      "step": 3306
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.279710144927537e-06,
+      "loss": 1.0526,
+      "step": 3307
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.2782608695652185e-06,
+      "loss": 0.5911,
+      "step": 3308
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.276811594202899e-06,
+      "loss": 0.4425,
+      "step": 3309
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.27536231884058e-06,
+      "loss": 0.2305,
+      "step": 3310
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.2739130434782615e-06,
+      "loss": 0.5593,
+      "step": 3311
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.272463768115943e-06,
+      "loss": 0.4723,
+      "step": 3312
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.2710144927536235e-06,
+      "loss": 0.5273,
+      "step": 3313
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.2695652173913045e-06,
+      "loss": 1.2672,
+      "step": 3314
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.268115942028986e-06,
+      "loss": 0.1008,
+      "step": 3315
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.2666666666666665e-06,
+      "loss": 0.9693,
+      "step": 3316
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.265217391304348e-06,
+      "loss": 0.3035,
+      "step": 3317
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.263768115942029e-06,
+      "loss": 0.3664,
+      "step": 3318
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.262318840579711e-06,
+      "loss": 0.3534,
+      "step": 3319
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.260869565217391e-06,
+      "loss": 0.2563,
+      "step": 3320
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.259420289855073e-06,
+      "loss": 0.7765,
+      "step": 3321
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.257971014492754e-06,
+      "loss": 1.338,
+      "step": 3322
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.256521739130436e-06,
+      "loss": 0.2613,
+      "step": 3323
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.255072463768116e-06,
+      "loss": 0.4171,
+      "step": 3324
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.253623188405797e-06,
+      "loss": 0.8609,
+      "step": 3325
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.252173913043479e-06,
+      "loss": 1.4549,
+      "step": 3326
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.250724637681159e-06,
+      "loss": 0.7417,
+      "step": 3327
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.249275362318841e-06,
+      "loss": 1.2536,
+      "step": 3328
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.247826086956522e-06,
+      "loss": 0.5257,
+      "step": 3329
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.246376811594204e-06,
+      "loss": 1.0183,
+      "step": 3330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.244927536231884e-06,
+      "loss": 0.3547,
+      "step": 3331
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.243478260869566e-06,
+      "loss": 0.4941,
+      "step": 3332
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.242028985507247e-06,
+      "loss": 0.2326,
+      "step": 3333
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.240579710144929e-06,
+      "loss": 0.3306,
+      "step": 3334
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.239130434782609e-06,
+      "loss": 0.4498,
+      "step": 3335
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.23768115942029e-06,
+      "loss": 1.2001,
+      "step": 3336
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.236231884057972e-06,
+      "loss": 0.8674,
+      "step": 3337
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.234782608695652e-06,
+      "loss": 0.894,
+      "step": 3338
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.233333333333334e-06,
+      "loss": 0.84,
+      "step": 3339
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.231884057971015e-06,
+      "loss": 0.2299,
+      "step": 3340
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.230434782608697e-06,
+      "loss": 1.2979,
+      "step": 3341
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.228985507246377e-06,
+      "loss": 0.5987,
+      "step": 3342
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.227536231884059e-06,
+      "loss": 1.5267,
+      "step": 3343
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.22608695652174e-06,
+      "loss": 1.2736,
+      "step": 3344
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.2246376811594215e-06,
+      "loss": 0.7708,
+      "step": 3345
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.223188405797102e-06,
+      "loss": 1.1647,
+      "step": 3346
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2217391304347835e-06,
+      "loss": 0.4766,
+      "step": 3347
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2202898550724645e-06,
+      "loss": 1.1997,
+      "step": 3348
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.218840579710145e-06,
+      "loss": 0.3237,
+      "step": 3349
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 0.3863,
+      "step": 3350
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2159420289855075e-06,
+      "loss": 0.1785,
+      "step": 3351
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.214492753623189e-06,
+      "loss": 1.3768,
+      "step": 3352
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2130434782608695e-06,
+      "loss": 0.3645,
+      "step": 3353
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.211594202898551e-06,
+      "loss": 0.663,
+      "step": 3354
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.210144927536232e-06,
+      "loss": 0.2677,
+      "step": 3355
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.208695652173913e-06,
+      "loss": 0.6643,
+      "step": 3356
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.207246376811594e-06,
+      "loss": 0.6373,
+      "step": 3357
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.205797101449276e-06,
+      "loss": 0.4794,
+      "step": 3358
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.204347826086957e-06,
+      "loss": 0.4823,
+      "step": 3359
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.202898550724637e-06,
+      "loss": 0.2284,
+      "step": 3360
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.201449275362319e-06,
+      "loss": 0.278,
+      "step": 3361
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2e-06,
+      "loss": 0.2944,
+      "step": 3362
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.198550724637682e-06,
+      "loss": 0.943,
+      "step": 3363
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.197101449275362e-06,
+      "loss": 0.3087,
+      "step": 3364
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.195652173913044e-06,
+      "loss": 0.1758,
+      "step": 3365
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.194202898550725e-06,
+      "loss": 0.5488,
+      "step": 3366
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.192753623188406e-06,
+      "loss": 0.227,
+      "step": 3367
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.191304347826087e-06,
+      "loss": 0.203,
+      "step": 3368
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.189855072463769e-06,
+      "loss": 0.8805,
+      "step": 3369
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.18840579710145e-06,
+      "loss": 0.3896,
+      "step": 3370
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.18695652173913e-06,
+      "loss": 0.2035,
+      "step": 3371
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.185507246376812e-06,
+      "loss": 0.5259,
+      "step": 3372
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.184057971014494e-06,
+      "loss": 0.6384,
+      "step": 3373
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.182608695652175e-06,
+      "loss": 0.276,
+      "step": 3374
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.181159420289855e-06,
+      "loss": 0.9844,
+      "step": 3375
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.179710144927537e-06,
+      "loss": 0.5738,
+      "step": 3376
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.178260869565218e-06,
+      "loss": 0.2912,
+      "step": 3377
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.176811594202899e-06,
+      "loss": 0.1902,
+      "step": 3378
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.17536231884058e-06,
+      "loss": 0.5141,
+      "step": 3379
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.173913043478262e-06,
+      "loss": 1.1567,
+      "step": 3380
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.172463768115943e-06,
+      "loss": 0.2215,
+      "step": 3381
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.171014492753624e-06,
+      "loss": 0.2659,
+      "step": 3382
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.169565217391305e-06,
+      "loss": 0.9547,
+      "step": 3383
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.1681159420289865e-06,
+      "loss": 0.2507,
+      "step": 3384
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 0.4863,
+      "step": 3385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.165217391304348e-06,
+      "loss": 1.203,
+      "step": 3386
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.1637681159420295e-06,
+      "loss": 0.6429,
+      "step": 3387
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.162318840579711e-06,
+      "loss": 0.6633,
+      "step": 3388
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.1608695652173915e-06,
+      "loss": 0.5707,
+      "step": 3389
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.1594202898550725e-06,
+      "loss": 0.3065,
+      "step": 3390
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.157971014492754e-06,
+      "loss": 0.893,
+      "step": 3391
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.156521739130435e-06,
+      "loss": 0.3986,
+      "step": 3392
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.155072463768116e-06,
+      "loss": 0.4695,
+      "step": 3393
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.153623188405797e-06,
+      "loss": 0.2518,
+      "step": 3394
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.152173913043479e-06,
+      "loss": 0.9113,
+      "step": 3395
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.15072463768116e-06,
+      "loss": 0.7048,
+      "step": 3396
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.14927536231884e-06,
+      "loss": 0.2793,
+      "step": 3397
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.147826086956522e-06,
+      "loss": 0.7606,
+      "step": 3398
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.146376811594204e-06,
+      "loss": 0.2887,
+      "step": 3399
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.144927536231884e-06,
+      "loss": 0.4229,
+      "step": 3400
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.143478260869565e-06,
+      "loss": 1.2011,
+      "step": 3401
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.142028985507247e-06,
+      "loss": 0.1823,
+      "step": 3402
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.140579710144928e-06,
+      "loss": 0.6861,
+      "step": 3403
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.139130434782609e-06,
+      "loss": 0.3625,
+      "step": 3404
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.13768115942029e-06,
+      "loss": 1.062,
+      "step": 3405
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.136231884057972e-06,
+      "loss": 0.7167,
+      "step": 3406
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.134782608695653e-06,
+      "loss": 0.9267,
+      "step": 3407
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.133333333333334e-06,
+      "loss": 0.9321,
+      "step": 3408
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.131884057971015e-06,
+      "loss": 0.9583,
+      "step": 3409
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.130434782608697e-06,
+      "loss": 0.35,
+      "step": 3410
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.128985507246377e-06,
+      "loss": 0.3879,
+      "step": 3411
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.127536231884058e-06,
+      "loss": 0.3743,
+      "step": 3412
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.12608695652174e-06,
+      "loss": 0.6682,
+      "step": 3413
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.124637681159422e-06,
+      "loss": 0.1773,
+      "step": 3414
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.123188405797102e-06,
+      "loss": 0.95,
+      "step": 3415
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.121739130434783e-06,
+      "loss": 0.1531,
+      "step": 3416
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.120289855072465e-06,
+      "loss": 0.4276,
+      "step": 3417
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.118840579710145e-06,
+      "loss": 0.2389,
+      "step": 3418
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.117391304347827e-06,
+      "loss": 0.6423,
+      "step": 3419
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.115942028985508e-06,
+      "loss": 0.2073,
+      "step": 3420
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.1144927536231894e-06,
+      "loss": 0.3724,
+      "step": 3421
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.11304347826087e-06,
+      "loss": 0.298,
+      "step": 3422
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1115942028985514e-06,
+      "loss": 0.3304,
+      "step": 3423
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1101449275362324e-06,
+      "loss": 0.4044,
+      "step": 3424
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.108695652173914e-06,
+      "loss": 0.5952,
+      "step": 3425
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1072463768115944e-06,
+      "loss": 0.5149,
+      "step": 3426
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1057971014492754e-06,
+      "loss": 0.2534,
+      "step": 3427
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.104347826086957e-06,
+      "loss": 0.603,
+      "step": 3428
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1028985507246374e-06,
+      "loss": 0.205,
+      "step": 3429
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.101449275362319e-06,
+      "loss": 0.2639,
+      "step": 3430
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.1e-06,
+      "loss": 0.5028,
+      "step": 3431
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.098550724637682e-06,
+      "loss": 0.3838,
+      "step": 3432
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.097101449275362e-06,
+      "loss": 0.8056,
+      "step": 3433
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.095652173913044e-06,
+      "loss": 0.6127,
+      "step": 3434
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.094202898550725e-06,
+      "loss": 0.4144,
+      "step": 3435
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.092753623188407e-06,
+      "loss": 0.4904,
+      "step": 3436
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.091304347826087e-06,
+      "loss": 1.0261,
+      "step": 3437
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.089855072463768e-06,
+      "loss": 0.3759,
+      "step": 3438
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.08840579710145e-06,
+      "loss": 0.2718,
+      "step": 3439
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.08695652173913e-06,
+      "loss": 0.4091,
+      "step": 3440
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.085507246376812e-06,
+      "loss": 0.2516,
+      "step": 3441
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.084057971014493e-06,
+      "loss": 0.3776,
+      "step": 3442
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.082608695652175e-06,
+      "loss": 1.2104,
+      "step": 3443
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.081159420289855e-06,
+      "loss": 0.8751,
+      "step": 3444
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.079710144927537e-06,
+      "loss": 0.7136,
+      "step": 3445
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.078260869565218e-06,
+      "loss": 0.2362,
+      "step": 3446
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.0768115942029e-06,
+      "loss": 0.3554,
+      "step": 3447
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.07536231884058e-06,
+      "loss": 0.0757,
+      "step": 3448
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.073913043478262e-06,
+      "loss": 0.2498,
+      "step": 3449
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.072463768115943e-06,
+      "loss": 1.925,
+      "step": 3450
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.071014492753623e-06,
+      "loss": 0.7119,
+      "step": 3451
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.069565217391305e-06,
+      "loss": 0.3432,
+      "step": 3452
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.068115942028986e-06,
+      "loss": 0.7072,
+      "step": 3453
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 0.3922,
+      "step": 3454
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.065217391304348e-06,
+      "loss": 1.3804,
+      "step": 3455
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.0637681159420296e-06,
+      "loss": 0.2037,
+      "step": 3456
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.0623188405797106e-06,
+      "loss": 0.1507,
+      "step": 3457
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.060869565217392e-06,
+      "loss": 0.3991,
+      "step": 3458
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.0594202898550726e-06,
+      "loss": 0.4141,
+      "step": 3459
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.057971014492754e-06,
+      "loss": 0.7301,
+      "step": 3460
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.0565217391304354e-06,
+      "loss": 0.4129,
+      "step": 3461
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.055072463768116e-06,
+      "loss": 1.599,
+      "step": 3462
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.0536231884057974e-06,
+      "loss": 0.2026,
+      "step": 3463
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.0521739130434784e-06,
+      "loss": 1.2929,
+      "step": 3464
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.05072463768116e-06,
+      "loss": 1.1254,
+      "step": 3465
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.0492753623188404e-06,
+      "loss": 0.5758,
+      "step": 3466
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.047826086956522e-06,
+      "loss": 0.2535,
+      "step": 3467
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.046376811594203e-06,
+      "loss": 1.7333,
+      "step": 3468
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.044927536231885e-06,
+      "loss": 1.7225,
+      "step": 3469
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.043478260869565e-06,
+      "loss": 0.8146,
+      "step": 3470
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.042028985507247e-06,
+      "loss": 0.2937,
+      "step": 3471
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.040579710144928e-06,
+      "loss": 0.5057,
+      "step": 3472
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.039130434782608e-06,
+      "loss": 1.283,
+      "step": 3473
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.03768115942029e-06,
+      "loss": 0.0818,
+      "step": 3474
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.036231884057972e-06,
+      "loss": 0.4146,
+      "step": 3475
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9122302158273381,
+      "eval_loss": 0.43118616938591003,
+      "eval_roc_auc": 0.9759308554627323,
+      "eval_runtime": 86.7293,
+      "eval_samples_per_second": 8.013,
+      "eval_steps_per_second": 2.006,
+      "step": 3475
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.034782608695653e-06,
+      "loss": 0.427,
+      "step": 3476
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.033333333333333e-06,
+      "loss": 0.679,
+      "step": 3477
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.031884057971015e-06,
+      "loss": 0.2157,
+      "step": 3478
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.030434782608696e-06,
+      "loss": 0.7291,
+      "step": 3479
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.028985507246377e-06,
+      "loss": 0.987,
+      "step": 3480
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.027536231884058e-06,
+      "loss": 0.4652,
+      "step": 3481
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.02608695652174e-06,
+      "loss": 0.7263,
+      "step": 3482
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.024637681159421e-06,
+      "loss": 0.522,
+      "step": 3483
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.023188405797102e-06,
+      "loss": 0.6626,
+      "step": 3484
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.021739130434783e-06,
+      "loss": 0.7048,
+      "step": 3485
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.020289855072465e-06,
+      "loss": 1.4436,
+      "step": 3486
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.018840579710146e-06,
+      "loss": 1.504,
+      "step": 3487
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.017391304347826e-06,
+      "loss": 1.6383,
+      "step": 3488
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.015942028985508e-06,
+      "loss": 0.6624,
+      "step": 3489
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.014492753623189e-06,
+      "loss": 0.2483,
+      "step": 3490
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.01304347826087e-06,
+      "loss": 1.0865,
+      "step": 3491
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.011594202898551e-06,
+      "loss": 2.8032,
+      "step": 3492
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0101449275362326e-06,
+      "loss": 0.6008,
+      "step": 3493
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0086956521739136e-06,
+      "loss": 0.3407,
+      "step": 3494
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0072463768115946e-06,
+      "loss": 0.6241,
+      "step": 3495
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0057971014492756e-06,
+      "loss": 0.3955,
+      "step": 3496
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.004347826086957e-06,
+      "loss": 0.8743,
+      "step": 3497
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.002898550724638e-06,
+      "loss": 0.1038,
+      "step": 3498
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0014492753623186e-06,
+      "loss": 0.3793,
+      "step": 3499
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5e-06,
+      "loss": 1.2762,
+      "step": 3500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.998550724637681e-06,
+      "loss": 0.5014,
+      "step": 3501
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.997101449275362e-06,
+      "loss": 0.5768,
+      "step": 3502
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995652173913043e-06,
+      "loss": 0.7675,
+      "step": 3503
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.994202898550725e-06,
+      "loss": 0.1934,
+      "step": 3504
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.992753623188406e-06,
+      "loss": 0.2452,
+      "step": 3505
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.991304347826087e-06,
+      "loss": 1.6044,
+      "step": 3506
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.989855072463768e-06,
+      "loss": 0.2607,
+      "step": 3507
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.98840579710145e-06,
+      "loss": 0.3201,
+      "step": 3508
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.986956521739131e-06,
+      "loss": 0.8526,
+      "step": 3509
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.985507246376812e-06,
+      "loss": 0.2904,
+      "step": 3510
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.984057971014493e-06,
+      "loss": 0.3146,
+      "step": 3511
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.982608695652174e-06,
+      "loss": 0.7232,
+      "step": 3512
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.981159420289856e-06,
+      "loss": 0.3669,
+      "step": 3513
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.979710144927536e-06,
+      "loss": 0.1555,
+      "step": 3514
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.978260869565218e-06,
+      "loss": 0.2624,
+      "step": 3515
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.976811594202899e-06,
+      "loss": 1.9786,
+      "step": 3516
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.97536231884058e-06,
+      "loss": 1.7239,
+      "step": 3517
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.973913043478261e-06,
+      "loss": 0.1743,
+      "step": 3518
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.972463768115943e-06,
+      "loss": 0.68,
+      "step": 3519
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.971014492753624e-06,
+      "loss": 0.697,
+      "step": 3520
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.969565217391305e-06,
+      "loss": 0.4202,
+      "step": 3521
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.968115942028986e-06,
+      "loss": 0.324,
+      "step": 3522
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.966666666666667e-06,
+      "loss": 0.5349,
+      "step": 3523
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.965217391304349e-06,
+      "loss": 0.2679,
+      "step": 3524
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.963768115942029e-06,
+      "loss": 0.5341,
+      "step": 3525
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.962318840579711e-06,
+      "loss": 0.334,
+      "step": 3526
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.960869565217392e-06,
+      "loss": 0.3556,
+      "step": 3527
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.959420289855073e-06,
+      "loss": 0.7467,
+      "step": 3528
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.957971014492754e-06,
+      "loss": 0.2004,
+      "step": 3529
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9565217391304355e-06,
+      "loss": 0.401,
+      "step": 3530
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9550724637681165e-06,
+      "loss": 2.1918,
+      "step": 3531
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9536231884057975e-06,
+      "loss": 0.4968,
+      "step": 3532
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9521739130434785e-06,
+      "loss": 0.5075,
+      "step": 3533
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9507246376811595e-06,
+      "loss": 0.2242,
+      "step": 3534
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.949275362318841e-06,
+      "loss": 0.0762,
+      "step": 3535
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.947826086956522e-06,
+      "loss": 0.2116,
+      "step": 3536
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.946376811594203e-06,
+      "loss": 0.2736,
+      "step": 3537
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.944927536231884e-06,
+      "loss": 0.9324,
+      "step": 3538
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.943478260869566e-06,
+      "loss": 0.4071,
+      "step": 3539
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.942028985507246e-06,
+      "loss": 0.7763,
+      "step": 3540
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.940579710144928e-06,
+      "loss": 1.658,
+      "step": 3541
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.939130434782609e-06,
+      "loss": 1.4214,
+      "step": 3542
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.93768115942029e-06,
+      "loss": 0.296,
+      "step": 3543
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.936231884057971e-06,
+      "loss": 0.3358,
+      "step": 3544
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.934782608695652e-06,
+      "loss": 0.7107,
+      "step": 3545
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 0.4986,
+      "step": 3546
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.931884057971015e-06,
+      "loss": 0.2526,
+      "step": 3547
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.930434782608696e-06,
+      "loss": 1.194,
+      "step": 3548
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.928985507246377e-06,
+      "loss": 0.4611,
+      "step": 3549
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.927536231884059e-06,
+      "loss": 0.1628,
+      "step": 3550
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.926086956521739e-06,
+      "loss": 0.1923,
+      "step": 3551
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.92463768115942e-06,
+      "loss": 0.3649,
+      "step": 3552
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.923188405797102e-06,
+      "loss": 0.2429,
+      "step": 3553
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.921739130434783e-06,
+      "loss": 0.6276,
+      "step": 3554
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.920289855072464e-06,
+      "loss": 0.0784,
+      "step": 3555
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.918840579710145e-06,
+      "loss": 0.122,
+      "step": 3556
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.917391304347827e-06,
+      "loss": 0.0688,
+      "step": 3557
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.915942028985508e-06,
+      "loss": 1.3038,
+      "step": 3558
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.914492753623189e-06,
+      "loss": 0.2165,
+      "step": 3559
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.91304347826087e-06,
+      "loss": 0.09,
+      "step": 3560
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.911594202898552e-06,
+      "loss": 0.5588,
+      "step": 3561
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.910144927536233e-06,
+      "loss": 0.1491,
+      "step": 3562
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.908695652173913e-06,
+      "loss": 0.5871,
+      "step": 3563
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.907246376811595e-06,
+      "loss": 0.2045,
+      "step": 3564
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.905797101449276e-06,
+      "loss": 1.6891,
+      "step": 3565
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.904347826086957e-06,
+      "loss": 1.5033,
+      "step": 3566
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.902898550724638e-06,
+      "loss": 0.9235,
+      "step": 3567
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.9014492753623195e-06,
+      "loss": 0.2157,
+      "step": 3568
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.3284,
+      "step": 3569
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.8985507246376815e-06,
+      "loss": 0.46,
+      "step": 3570
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.8971014492753625e-06,
+      "loss": 1.0129,
+      "step": 3571
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.895652173913044e-06,
+      "loss": 1.3011,
+      "step": 3572
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.894202898550725e-06,
+      "loss": 0.2578,
+      "step": 3573
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.892753623188406e-06,
+      "loss": 0.4771,
+      "step": 3574
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.891304347826087e-06,
+      "loss": 0.2852,
+      "step": 3575
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.889855072463768e-06,
+      "loss": 0.2076,
+      "step": 3576
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.888405797101449e-06,
+      "loss": 0.2129,
+      "step": 3577
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.88695652173913e-06,
+      "loss": 0.1122,
+      "step": 3578
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.885507246376812e-06,
+      "loss": 1.0411,
+      "step": 3579
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.884057971014493e-06,
+      "loss": 0.7841,
+      "step": 3580
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.882608695652174e-06,
+      "loss": 0.5166,
+      "step": 3581
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.881159420289855e-06,
+      "loss": 0.3435,
+      "step": 3582
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.879710144927536e-06,
+      "loss": 0.5456,
+      "step": 3583
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.878260869565218e-06,
+      "loss": 0.6966,
+      "step": 3584
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.876811594202899e-06,
+      "loss": 0.3773,
+      "step": 3585
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.87536231884058e-06,
+      "loss": 0.257,
+      "step": 3586
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.873913043478261e-06,
+      "loss": 0.335,
+      "step": 3587
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.872463768115943e-06,
+      "loss": 0.1557,
+      "step": 3588
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.871014492753623e-06,
+      "loss": 0.4118,
+      "step": 3589
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.869565217391305e-06,
+      "loss": 0.8092,
+      "step": 3590
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.868115942028986e-06,
+      "loss": 0.9213,
+      "step": 3591
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.866666666666667e-06,
+      "loss": 0.4117,
+      "step": 3592
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.865217391304348e-06,
+      "loss": 0.0846,
+      "step": 3593
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.863768115942029e-06,
+      "loss": 1.4902,
+      "step": 3594
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.862318840579711e-06,
+      "loss": 0.6647,
+      "step": 3595
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.860869565217392e-06,
+      "loss": 0.2264,
+      "step": 3596
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.859420289855073e-06,
+      "loss": 0.2438,
+      "step": 3597
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.857971014492754e-06,
+      "loss": 1.3863,
+      "step": 3598
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.856521739130436e-06,
+      "loss": 0.342,
+      "step": 3599
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.855072463768117e-06,
+      "loss": 1.3015,
+      "step": 3600
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.853623188405798e-06,
+      "loss": 0.6361,
+      "step": 3601
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.852173913043479e-06,
+      "loss": 0.2554,
+      "step": 3602
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.8507246376811605e-06,
+      "loss": 0.2803,
+      "step": 3603
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.849275362318841e-06,
+      "loss": 0.4034,
+      "step": 3604
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.847826086956522e-06,
+      "loss": 0.9563,
+      "step": 3605
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.8463768115942035e-06,
+      "loss": 1.0396,
+      "step": 3606
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.8449275362318845e-06,
+      "loss": 0.5218,
+      "step": 3607
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.8434782608695655e-06,
+      "loss": 0.479,
+      "step": 3608
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.8420289855072465e-06,
+      "loss": 0.8006,
+      "step": 3609
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.840579710144928e-06,
+      "loss": 0.4582,
+      "step": 3610
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.839130434782609e-06,
+      "loss": 0.2189,
+      "step": 3611
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.83768115942029e-06,
+      "loss": 0.2789,
+      "step": 3612
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.836231884057971e-06,
+      "loss": 0.137,
+      "step": 3613
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.834782608695652e-06,
+      "loss": 0.3836,
+      "step": 3614
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 0.8283,
+      "step": 3615
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.831884057971014e-06,
+      "loss": 0.253,
+      "step": 3616
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.830434782608696e-06,
+      "loss": 0.4353,
+      "step": 3617
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.828985507246377e-06,
+      "loss": 0.8749,
+      "step": 3618
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.827536231884058e-06,
+      "loss": 0.3398,
+      "step": 3619
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.826086956521739e-06,
+      "loss": 0.3132,
+      "step": 3620
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.824637681159421e-06,
+      "loss": 1.1384,
+      "step": 3621
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.823188405797102e-06,
+      "loss": 0.7858,
+      "step": 3622
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.821739130434783e-06,
+      "loss": 0.1168,
+      "step": 3623
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.820289855072464e-06,
+      "loss": 0.3209,
+      "step": 3624
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.818840579710145e-06,
+      "loss": 0.2272,
+      "step": 3625
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.817391304347827e-06,
+      "loss": 0.2601,
+      "step": 3626
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.815942028985507e-06,
+      "loss": 1.4374,
+      "step": 3627
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.814492753623189e-06,
+      "loss": 0.8212,
+      "step": 3628
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.81304347826087e-06,
+      "loss": 0.3793,
+      "step": 3629
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.811594202898551e-06,
+      "loss": 0.546,
+      "step": 3630
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.810144927536232e-06,
+      "loss": 0.2117,
+      "step": 3631
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.808695652173914e-06,
+      "loss": 1.2978,
+      "step": 3632
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.807246376811595e-06,
+      "loss": 0.888,
+      "step": 3633
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.805797101449276e-06,
+      "loss": 0.2716,
+      "step": 3634
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.804347826086957e-06,
+      "loss": 0.2888,
+      "step": 3635
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.802898550724638e-06,
+      "loss": 0.4556,
+      "step": 3636
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.80144927536232e-06,
+      "loss": 0.2944,
+      "step": 3637
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.4708,
+      "step": 3638
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.798550724637682e-06,
+      "loss": 0.4108,
+      "step": 3639
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.797101449275363e-06,
+      "loss": 0.2244,
+      "step": 3640
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.795652173913044e-06,
+      "loss": 0.2245,
+      "step": 3641
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.794202898550725e-06,
+      "loss": 1.1004,
+      "step": 3642
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7927536231884065e-06,
+      "loss": 0.5,
+      "step": 3643
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7913043478260875e-06,
+      "loss": 0.49,
+      "step": 3644
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7898550724637685e-06,
+      "loss": 0.564,
+      "step": 3645
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7884057971014495e-06,
+      "loss": 0.8764,
+      "step": 3646
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7869565217391305e-06,
+      "loss": 0.2027,
+      "step": 3647
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.785507246376812e-06,
+      "loss": 0.4041,
+      "step": 3648
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.784057971014493e-06,
+      "loss": 1.2441,
+      "step": 3649
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 0.6228,
+      "step": 3650
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.781159420289855e-06,
+      "loss": 0.3665,
+      "step": 3651
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.779710144927537e-06,
+      "loss": 0.4806,
+      "step": 3652
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.778260869565217e-06,
+      "loss": 0.287,
+      "step": 3653
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.776811594202899e-06,
+      "loss": 1.1521,
+      "step": 3654
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.77536231884058e-06,
+      "loss": 0.2158,
+      "step": 3655
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.773913043478261e-06,
+      "loss": 1.378,
+      "step": 3656
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.772463768115942e-06,
+      "loss": 0.1165,
+      "step": 3657
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.771014492753623e-06,
+      "loss": 0.234,
+      "step": 3658
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.769565217391305e-06,
+      "loss": 0.0584,
+      "step": 3659
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.768115942028986e-06,
+      "loss": 0.3307,
+      "step": 3660
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.766666666666667e-06,
+      "loss": 0.0715,
+      "step": 3661
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.765217391304348e-06,
+      "loss": 0.3708,
+      "step": 3662
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.76376811594203e-06,
+      "loss": 0.1291,
+      "step": 3663
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.762318840579711e-06,
+      "loss": 0.2862,
+      "step": 3664
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.760869565217392e-06,
+      "loss": 0.3641,
+      "step": 3665
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.759420289855073e-06,
+      "loss": 0.2129,
+      "step": 3666
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.757971014492754e-06,
+      "loss": 0.4839,
+      "step": 3667
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.756521739130435e-06,
+      "loss": 0.3639,
+      "step": 3668
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.755072463768116e-06,
+      "loss": 0.2,
+      "step": 3669
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.753623188405798e-06,
+      "loss": 1.1159,
+      "step": 3670
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.752173913043479e-06,
+      "loss": 0.19,
+      "step": 3671
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.75072463768116e-06,
+      "loss": 0.4224,
+      "step": 3672
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.749275362318841e-06,
+      "loss": 0.2662,
+      "step": 3673
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.747826086956523e-06,
+      "loss": 0.4827,
+      "step": 3674
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.746376811594204e-06,
+      "loss": 0.3467,
+      "step": 3675
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.744927536231884e-06,
+      "loss": 0.6264,
+      "step": 3676
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.743478260869566e-06,
+      "loss": 1.0593,
+      "step": 3677
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.742028985507247e-06,
+      "loss": 0.7857,
+      "step": 3678
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.740579710144928e-06,
+      "loss": 0.2708,
+      "step": 3679
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.739130434782609e-06,
+      "loss": 0.1773,
+      "step": 3680
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7376811594202905e-06,
+      "loss": 0.1455,
+      "step": 3681
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7362318840579715e-06,
+      "loss": 0.2954,
+      "step": 3682
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7347826086956525e-06,
+      "loss": 1.0701,
+      "step": 3683
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7333333333333335e-06,
+      "loss": 0.2896,
+      "step": 3684
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.731884057971015e-06,
+      "loss": 0.4847,
+      "step": 3685
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.730434782608696e-06,
+      "loss": 0.1803,
+      "step": 3686
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.728985507246377e-06,
+      "loss": 0.3328,
+      "step": 3687
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.727536231884058e-06,
+      "loss": 0.2721,
+      "step": 3688
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.726086956521739e-06,
+      "loss": 0.7486,
+      "step": 3689
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.724637681159421e-06,
+      "loss": 1.1567,
+      "step": 3690
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.723188405797101e-06,
+      "loss": 0.4513,
+      "step": 3691
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.721739130434783e-06,
+      "loss": 0.5569,
+      "step": 3692
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.720289855072464e-06,
+      "loss": 0.3573,
+      "step": 3693
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.718840579710145e-06,
+      "loss": 0.3102,
+      "step": 3694
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.717391304347826e-06,
+      "loss": 1.4344,
+      "step": 3695
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.715942028985508e-06,
+      "loss": 0.7702,
+      "step": 3696
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.714492753623189e-06,
+      "loss": 0.4971,
+      "step": 3697
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.71304347826087e-06,
+      "loss": 0.8675,
+      "step": 3698
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.711594202898551e-06,
+      "loss": 0.2856,
+      "step": 3699
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.710144927536232e-06,
+      "loss": 0.2197,
+      "step": 3700
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.708695652173914e-06,
+      "loss": 1.6015,
+      "step": 3701
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.707246376811595e-06,
+      "loss": 0.1876,
+      "step": 3702
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.705797101449276e-06,
+      "loss": 0.6594,
+      "step": 3703
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.704347826086957e-06,
+      "loss": 1.0094,
+      "step": 3704
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.702898550724638e-06,
+      "loss": 1.081,
+      "step": 3705
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.701449275362319e-06,
+      "loss": 0.2872,
+      "step": 3706
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.7e-06,
+      "loss": 0.3788,
+      "step": 3707
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.698550724637682e-06,
+      "loss": 0.3398,
+      "step": 3708
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.697101449275363e-06,
+      "loss": 0.2698,
+      "step": 3709
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.695652173913044e-06,
+      "loss": 1.2336,
+      "step": 3710
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.694202898550725e-06,
+      "loss": 0.1977,
+      "step": 3711
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.692753623188407e-06,
+      "loss": 0.3945,
+      "step": 3712
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.691304347826088e-06,
+      "loss": 0.2963,
+      "step": 3713
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.689855072463769e-06,
+      "loss": 0.2436,
+      "step": 3714
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.68840579710145e-06,
+      "loss": 0.1452,
+      "step": 3715
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.6869565217391314e-06,
+      "loss": 0.959,
+      "step": 3716
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.685507246376812e-06,
+      "loss": 1.0293,
+      "step": 3717
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.684057971014493e-06,
+      "loss": 0.2189,
+      "step": 3718
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.6826086956521745e-06,
+      "loss": 0.4069,
+      "step": 3719
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.6811594202898555e-06,
+      "loss": 0.9629,
+      "step": 3720
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.6797101449275365e-06,
+      "loss": 0.3999,
+      "step": 3721
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.6782608695652175e-06,
+      "loss": 0.3179,
+      "step": 3722
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.676811594202899e-06,
+      "loss": 0.2478,
+      "step": 3723
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.67536231884058e-06,
+      "loss": 0.1797,
+      "step": 3724
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.673913043478261e-06,
+      "loss": 1.1612,
+      "step": 3725
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.672463768115942e-06,
+      "loss": 0.3256,
+      "step": 3726
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.671014492753624e-06,
+      "loss": 0.1381,
+      "step": 3727
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.669565217391305e-06,
+      "loss": 1.1141,
+      "step": 3728
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.668115942028985e-06,
+      "loss": 0.6304,
+      "step": 3729
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.0751,
+      "step": 3730
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.665217391304348e-06,
+      "loss": 0.8712,
+      "step": 3731
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.663768115942029e-06,
+      "loss": 0.4432,
+      "step": 3732
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.66231884057971e-06,
+      "loss": 0.2562,
+      "step": 3733
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.660869565217392e-06,
+      "loss": 0.5195,
+      "step": 3734
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.659420289855073e-06,
+      "loss": 0.9,
+      "step": 3735
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.657971014492754e-06,
+      "loss": 0.2375,
+      "step": 3736
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.656521739130435e-06,
+      "loss": 0.243,
+      "step": 3737
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.655072463768116e-06,
+      "loss": 0.2275,
+      "step": 3738
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.653623188405798e-06,
+      "loss": 0.28,
+      "step": 3739
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.652173913043478e-06,
+      "loss": 0.2488,
+      "step": 3740
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.65072463768116e-06,
+      "loss": 0.3461,
+      "step": 3741
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.649275362318841e-06,
+      "loss": 0.5903,
+      "step": 3742
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.647826086956522e-06,
+      "loss": 0.572,
+      "step": 3743
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.646376811594203e-06,
+      "loss": 0.2536,
+      "step": 3744
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.644927536231885e-06,
+      "loss": 0.1916,
+      "step": 3745
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.643478260869566e-06,
+      "loss": 0.6997,
+      "step": 3746
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.642028985507247e-06,
+      "loss": 0.2286,
+      "step": 3747
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.640579710144928e-06,
+      "loss": 0.1544,
+      "step": 3748
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.639130434782609e-06,
+      "loss": 0.6544,
+      "step": 3749
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.637681159420291e-06,
+      "loss": 0.1473,
+      "step": 3750
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.636231884057972e-06,
+      "loss": 1.2153,
+      "step": 3751
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.634782608695653e-06,
+      "loss": 1.0197,
+      "step": 3752
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.633333333333334e-06,
+      "loss": 0.2858,
+      "step": 3753
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.6318840579710154e-06,
+      "loss": 0.6838,
+      "step": 3754
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.630434782608696e-06,
+      "loss": 0.4793,
+      "step": 3755
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.6289855072463774e-06,
+      "loss": 1.687,
+      "step": 3756
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.6275362318840584e-06,
+      "loss": 0.7739,
+      "step": 3757
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.6260869565217394e-06,
+      "loss": 0.0591,
+      "step": 3758
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.6246376811594204e-06,
+      "loss": 0.6913,
+      "step": 3759
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.6231884057971014e-06,
+      "loss": 0.209,
+      "step": 3760
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.621739130434783e-06,
+      "loss": 0.2371,
+      "step": 3761
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.620289855072464e-06,
+      "loss": 1.3547,
+      "step": 3762
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.618840579710145e-06,
+      "loss": 0.2134,
+      "step": 3763
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.617391304347826e-06,
+      "loss": 0.2219,
+      "step": 3764
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.615942028985508e-06,
+      "loss": 0.2024,
+      "step": 3765
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.614492753623188e-06,
+      "loss": 0.1234,
+      "step": 3766
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.61304347826087e-06,
+      "loss": 0.273,
+      "step": 3767
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.611594202898551e-06,
+      "loss": 0.2714,
+      "step": 3768
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.610144927536232e-06,
+      "loss": 0.2196,
+      "step": 3769
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.608695652173913e-06,
+      "loss": 0.5666,
+      "step": 3770
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.607246376811594e-06,
+      "loss": 0.212,
+      "step": 3771
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.605797101449276e-06,
+      "loss": 0.1588,
+      "step": 3772
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.604347826086957e-06,
+      "loss": 0.3535,
+      "step": 3773
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.602898550724638e-06,
+      "loss": 0.2638,
+      "step": 3774
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.601449275362319e-06,
+      "loss": 0.1715,
+      "step": 3775
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.1541,
+      "step": 3776
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.598550724637682e-06,
+      "loss": 0.1365,
+      "step": 3777
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.597101449275363e-06,
+      "loss": 0.4522,
+      "step": 3778
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.595652173913044e-06,
+      "loss": 0.2453,
+      "step": 3779
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.594202898550725e-06,
+      "loss": 0.8969,
+      "step": 3780
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.592753623188406e-06,
+      "loss": 1.47,
+      "step": 3781
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.591304347826087e-06,
+      "loss": 0.1991,
+      "step": 3782
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.589855072463769e-06,
+      "loss": 0.2243,
+      "step": 3783
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.58840579710145e-06,
+      "loss": 0.8664,
+      "step": 3784
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.586956521739131e-06,
+      "loss": 0.3692,
+      "step": 3785
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.585507246376812e-06,
+      "loss": 0.2413,
+      "step": 3786
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.5840579710144936e-06,
+      "loss": 0.1864,
+      "step": 3787
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.5826086956521746e-06,
+      "loss": 0.764,
+      "step": 3788
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.5811594202898556e-06,
+      "loss": 0.2595,
+      "step": 3789
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.5797101449275366e-06,
+      "loss": 0.3062,
+      "step": 3790
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.5782608695652176e-06,
+      "loss": 0.1851,
+      "step": 3791
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.576811594202899e-06,
+      "loss": 0.1959,
+      "step": 3792
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.5753623188405796e-06,
+      "loss": 0.5691,
+      "step": 3793
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.573913043478261e-06,
+      "loss": 0.3895,
+      "step": 3794
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.572463768115942e-06,
+      "loss": 0.4116,
+      "step": 3795
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.571014492753623e-06,
+      "loss": 1.2362,
+      "step": 3796
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.569565217391304e-06,
+      "loss": 0.4492,
+      "step": 3797
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.568115942028986e-06,
+      "loss": 0.124,
+      "step": 3798
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.566666666666667e-06,
+      "loss": 0.551,
+      "step": 3799
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 0.7587,
+      "step": 3800
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.563768115942029e-06,
+      "loss": 1.2855,
+      "step": 3801
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.56231884057971e-06,
+      "loss": 0.7879,
+      "step": 3802
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.560869565217392e-06,
+      "loss": 0.3484,
+      "step": 3803
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.559420289855072e-06,
+      "loss": 0.2055,
+      "step": 3804
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.557971014492754e-06,
+      "loss": 0.2109,
+      "step": 3805
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.556521739130435e-06,
+      "loss": 0.8313,
+      "step": 3806
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.555072463768116e-06,
+      "loss": 0.1157,
+      "step": 3807
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.553623188405797e-06,
+      "loss": 0.2643,
+      "step": 3808
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.552173913043479e-06,
+      "loss": 0.295,
+      "step": 3809
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.55072463768116e-06,
+      "loss": 0.2718,
+      "step": 3810
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.549275362318841e-06,
+      "loss": 0.2346,
+      "step": 3811
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.547826086956522e-06,
+      "loss": 0.6378,
+      "step": 3812
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.546376811594203e-06,
+      "loss": 0.5072,
+      "step": 3813
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.544927536231885e-06,
+      "loss": 0.32,
+      "step": 3814
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.543478260869566e-06,
+      "loss": 0.1157,
+      "step": 3815
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.542028985507247e-06,
+      "loss": 0.8203,
+      "step": 3816
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.540579710144928e-06,
+      "loss": 0.2857,
+      "step": 3817
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.53913043478261e-06,
+      "loss": 0.0703,
+      "step": 3818
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.53768115942029e-06,
+      "loss": 0.2688,
+      "step": 3819
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.536231884057972e-06,
+      "loss": 0.9432,
+      "step": 3820
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.534782608695653e-06,
+      "loss": 0.22,
+      "step": 3821
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.2825,
+      "step": 3822
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.531884057971015e-06,
+      "loss": 0.7324,
+      "step": 3823
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.530434782608696e-06,
+      "loss": 0.8482,
+      "step": 3824
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.5289855072463775e-06,
+      "loss": 0.1832,
+      "step": 3825
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.5275362318840585e-06,
+      "loss": 0.4407,
+      "step": 3826
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.5260869565217395e-06,
+      "loss": 0.2729,
+      "step": 3827
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.5246376811594205e-06,
+      "loss": 0.1808,
+      "step": 3828
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.523188405797102e-06,
+      "loss": 0.642,
+      "step": 3829
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.5217391304347826e-06,
+      "loss": 0.1143,
+      "step": 3830
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.5202898550724636e-06,
+      "loss": 0.9596,
+      "step": 3831
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.518840579710145e-06,
+      "loss": 1.527,
+      "step": 3832
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.517391304347826e-06,
+      "loss": 0.9219,
+      "step": 3833
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.515942028985507e-06,
+      "loss": 0.359,
+      "step": 3834
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.514492753623188e-06,
+      "loss": 0.2436,
+      "step": 3835
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.51304347826087e-06,
+      "loss": 0.1124,
+      "step": 3836
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.511594202898551e-06,
+      "loss": 0.4781,
+      "step": 3837
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.510144927536232e-06,
+      "loss": 0.0712,
+      "step": 3838
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.508695652173913e-06,
+      "loss": 0.1976,
+      "step": 3839
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.507246376811595e-06,
+      "loss": 0.8822,
+      "step": 3840
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.505797101449276e-06,
+      "loss": 0.2635,
+      "step": 3841
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.504347826086956e-06,
+      "loss": 0.1364,
+      "step": 3842
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.502898550724638e-06,
+      "loss": 0.8555,
+      "step": 3843
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.501449275362319e-06,
+      "loss": 0.4842,
+      "step": 3844
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0541,
+      "step": 3845
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.498550724637681e-06,
+      "loss": 0.1016,
+      "step": 3846
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.497101449275363e-06,
+      "loss": 0.1512,
+      "step": 3847
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.495652173913044e-06,
+      "loss": 0.1577,
+      "step": 3848
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.494202898550725e-06,
+      "loss": 0.1799,
+      "step": 3849
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.492753623188406e-06,
+      "loss": 0.3332,
+      "step": 3850
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.491304347826088e-06,
+      "loss": 0.3894,
+      "step": 3851
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.489855072463769e-06,
+      "loss": 0.1596,
+      "step": 3852
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.48840579710145e-06,
+      "loss": 1.0164,
+      "step": 3853
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.486956521739131e-06,
+      "loss": 0.211,
+      "step": 3854
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.485507246376812e-06,
+      "loss": 0.2228,
+      "step": 3855
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.484057971014494e-06,
+      "loss": 0.2467,
+      "step": 3856
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.482608695652174e-06,
+      "loss": 0.0548,
+      "step": 3857
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.481159420289856e-06,
+      "loss": 0.6021,
+      "step": 3858
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.479710144927537e-06,
+      "loss": 0.169,
+      "step": 3859
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.478260869565218e-06,
+      "loss": 0.2874,
+      "step": 3860
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.476811594202899e-06,
+      "loss": 0.1087,
+      "step": 3861
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.47536231884058e-06,
+      "loss": 0.1569,
+      "step": 3862
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.4739130434782615e-06,
+      "loss": 0.2041,
+      "step": 3863
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.4724637681159425e-06,
+      "loss": 0.1383,
+      "step": 3864
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.4710144927536235e-06,
+      "loss": 0.209,
+      "step": 3865
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.4695652173913045e-06,
+      "loss": 0.3568,
+      "step": 3866
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.468115942028986e-06,
+      "loss": 0.601,
+      "step": 3867
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 0.1893,
+      "step": 3868
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.465217391304348e-06,
+      "loss": 0.502,
+      "step": 3869
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.463768115942029e-06,
+      "loss": 0.6136,
+      "step": 3870
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.46231884057971e-06,
+      "loss": 0.0989,
+      "step": 3871
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.460869565217391e-06,
+      "loss": 0.1416,
+      "step": 3872
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.459420289855072e-06,
+      "loss": 0.3158,
+      "step": 3873
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.457971014492754e-06,
+      "loss": 0.1187,
+      "step": 3874
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.456521739130435e-06,
+      "loss": 0.1172,
+      "step": 3875
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.455072463768116e-06,
+      "loss": 0.0711,
+      "step": 3876
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.453623188405797e-06,
+      "loss": 1.8144,
+      "step": 3877
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.452173913043479e-06,
+      "loss": 0.1375,
+      "step": 3878
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.45072463768116e-06,
+      "loss": 0.142,
+      "step": 3879
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.449275362318841e-06,
+      "loss": 0.4734,
+      "step": 3880
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.447826086956522e-06,
+      "loss": 0.6576,
+      "step": 3881
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.446376811594204e-06,
+      "loss": 1.6508,
+      "step": 3882
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.444927536231884e-06,
+      "loss": 0.1884,
+      "step": 3883
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.443478260869565e-06,
+      "loss": 0.2547,
+      "step": 3884
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.442028985507247e-06,
+      "loss": 0.1788,
+      "step": 3885
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.440579710144928e-06,
+      "loss": 0.9005,
+      "step": 3886
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.439130434782609e-06,
+      "loss": 0.0619,
+      "step": 3887
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.43768115942029e-06,
+      "loss": 0.113,
+      "step": 3888
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.436231884057972e-06,
+      "loss": 0.3791,
+      "step": 3889
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.434782608695653e-06,
+      "loss": 0.2691,
+      "step": 3890
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.433333333333334e-06,
+      "loss": 0.1242,
+      "step": 3891
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.431884057971015e-06,
+      "loss": 1.032,
+      "step": 3892
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.430434782608696e-06,
+      "loss": 0.3569,
+      "step": 3893
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.428985507246377e-06,
+      "loss": 0.0976,
+      "step": 3894
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.427536231884058e-06,
+      "loss": 1.2185,
+      "step": 3895
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.42608695652174e-06,
+      "loss": 0.0499,
+      "step": 3896
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.424637681159421e-06,
+      "loss": 0.1468,
+      "step": 3897
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.423188405797102e-06,
+      "loss": 0.2218,
+      "step": 3898
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.421739130434783e-06,
+      "loss": 1.8478,
+      "step": 3899
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.4202898550724645e-06,
+      "loss": 0.2545,
+      "step": 3900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.4188405797101455e-06,
+      "loss": 0.2187,
+      "step": 3901
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.4173913043478265e-06,
+      "loss": 0.1065,
+      "step": 3902
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.4159420289855075e-06,
+      "loss": 0.222,
+      "step": 3903
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.4144927536231885e-06,
+      "loss": 1.1975,
+      "step": 3904
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.41304347826087e-06,
+      "loss": 0.212,
+      "step": 3905
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.4115942028985505e-06,
+      "loss": 0.2768,
+      "step": 3906
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.410144927536232e-06,
+      "loss": 0.1831,
+      "step": 3907
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.408695652173913e-06,
+      "loss": 0.2858,
+      "step": 3908
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.407246376811594e-06,
+      "loss": 0.1434,
+      "step": 3909
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.405797101449275e-06,
+      "loss": 0.2808,
+      "step": 3910
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.404347826086957e-06,
+      "loss": 0.1567,
+      "step": 3911
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.402898550724638e-06,
+      "loss": 0.3135,
+      "step": 3912
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.401449275362319e-06,
+      "loss": 0.6598,
+      "step": 3913
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.4e-06,
+      "loss": 0.2965,
+      "step": 3914
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.398550724637681e-06,
+      "loss": 0.18,
+      "step": 3915
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.397101449275363e-06,
+      "loss": 1.4576,
+      "step": 3916
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.395652173913044e-06,
+      "loss": 0.1999,
+      "step": 3917
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.394202898550725e-06,
+      "loss": 0.1049,
+      "step": 3918
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.392753623188406e-06,
+      "loss": 0.2754,
+      "step": 3919
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.391304347826087e-06,
+      "loss": 0.4418,
+      "step": 3920
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.389855072463768e-06,
+      "loss": 0.3159,
+      "step": 3921
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.38840579710145e-06,
+      "loss": 0.334,
+      "step": 3922
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.386956521739131e-06,
+      "loss": 0.4129,
+      "step": 3923
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.385507246376812e-06,
+      "loss": 0.1754,
+      "step": 3924
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.384057971014493e-06,
+      "loss": 0.853,
+      "step": 3925
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.382608695652174e-06,
+      "loss": 0.3325,
+      "step": 3926
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.381159420289856e-06,
+      "loss": 0.2335,
+      "step": 3927
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.379710144927537e-06,
+      "loss": 1.4532,
+      "step": 3928
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.378260869565218e-06,
+      "loss": 0.8352,
+      "step": 3929
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.376811594202899e-06,
+      "loss": 0.1991,
+      "step": 3930
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.375362318840581e-06,
+      "loss": 0.54,
+      "step": 3931
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.373913043478261e-06,
+      "loss": 0.057,
+      "step": 3932
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.372463768115943e-06,
+      "loss": 0.214,
+      "step": 3933
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.371014492753624e-06,
+      "loss": 0.1192,
+      "step": 3934
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.369565217391305e-06,
+      "loss": 0.6681,
+      "step": 3935
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.368115942028986e-06,
+      "loss": 0.289,
+      "step": 3936
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.366666666666667e-06,
+      "loss": 0.115,
+      "step": 3937
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.3652173913043485e-06,
+      "loss": 1.0487,
+      "step": 3938
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.3637681159420295e-06,
+      "loss": 0.1617,
+      "step": 3939
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.3623188405797105e-06,
+      "loss": 0.1485,
+      "step": 3940
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.3608695652173915e-06,
+      "loss": 0.1668,
+      "step": 3941
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.359420289855073e-06,
+      "loss": 0.7899,
+      "step": 3942
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.357971014492754e-06,
+      "loss": 0.46,
+      "step": 3943
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.356521739130435e-06,
+      "loss": 0.6655,
+      "step": 3944
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.355072463768116e-06,
+      "loss": 0.2907,
+      "step": 3945
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.353623188405797e-06,
+      "loss": 0.1286,
+      "step": 3946
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.352173913043478e-06,
+      "loss": 0.1436,
+      "step": 3947
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.350724637681159e-06,
+      "loss": 0.1381,
+      "step": 3948
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.349275362318841e-06,
+      "loss": 1.0764,
+      "step": 3949
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.9898,
+      "step": 3950
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.346376811594203e-06,
+      "loss": 0.1907,
+      "step": 3951
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.344927536231884e-06,
+      "loss": 0.2222,
+      "step": 3952
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.343478260869566e-06,
+      "loss": 0.1682,
+      "step": 3953
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.342028985507247e-06,
+      "loss": 1.2423,
+      "step": 3954
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.340579710144927e-06,
+      "loss": 0.4933,
+      "step": 3955
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.339130434782609e-06,
+      "loss": 0.2623,
+      "step": 3956
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.33768115942029e-06,
+      "loss": 0.0378,
+      "step": 3957
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.336231884057971e-06,
+      "loss": 1.4138,
+      "step": 3958
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.334782608695652e-06,
+      "loss": 0.86,
+      "step": 3959
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 1.2682,
+      "step": 3960
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.331884057971015e-06,
+      "loss": 0.4251,
+      "step": 3961
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.330434782608696e-06,
+      "loss": 0.1352,
+      "step": 3962
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.328985507246377e-06,
+      "loss": 1.1118,
+      "step": 3963
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.327536231884059e-06,
+      "loss": 1.168,
+      "step": 3964
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.32608695652174e-06,
+      "loss": 0.2678,
+      "step": 3965
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.324637681159421e-06,
+      "loss": 1.1091,
+      "step": 3966
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.323188405797102e-06,
+      "loss": 0.1031,
+      "step": 3967
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.321739130434783e-06,
+      "loss": 0.463,
+      "step": 3968
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.320289855072465e-06,
+      "loss": 0.1847,
+      "step": 3969
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.318840579710145e-06,
+      "loss": 0.8644,
+      "step": 3970
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.317391304347827e-06,
+      "loss": 0.0936,
+      "step": 3971
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.315942028985508e-06,
+      "loss": 0.0916,
+      "step": 3972
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.314492753623189e-06,
+      "loss": 0.0765,
+      "step": 3973
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.31304347826087e-06,
+      "loss": 1.3824,
+      "step": 3974
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.3115942028985515e-06,
+      "loss": 0.2529,
+      "step": 3975
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.3101449275362325e-06,
+      "loss": 1.7186,
+      "step": 3976
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.3086956521739135e-06,
+      "loss": 1.3961,
+      "step": 3977
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.3072463768115945e-06,
+      "loss": 0.1865,
+      "step": 3978
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.3057971014492755e-06,
+      "loss": 0.9719,
+      "step": 3979
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.304347826086957e-06,
+      "loss": 0.1324,
+      "step": 3980
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.302898550724638e-06,
+      "loss": 0.953,
+      "step": 3981
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.301449275362319e-06,
+      "loss": 0.2388,
+      "step": 3982
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0851,
+      "step": 3983
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.298550724637681e-06,
+      "loss": 0.1339,
+      "step": 3984
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.297101449275362e-06,
+      "loss": 0.2847,
+      "step": 3985
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.295652173913043e-06,
+      "loss": 0.3809,
+      "step": 3986
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.294202898550725e-06,
+      "loss": 1.7042,
+      "step": 3987
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.292753623188406e-06,
+      "loss": 0.1713,
+      "step": 3988
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.291304347826087e-06,
+      "loss": 0.5222,
+      "step": 3989
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.289855072463768e-06,
+      "loss": 1.0706,
+      "step": 3990
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.28840579710145e-06,
+      "loss": 0.6784,
+      "step": 3991
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.286956521739131e-06,
+      "loss": 0.9764,
+      "step": 3992
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.285507246376812e-06,
+      "loss": 0.0978,
+      "step": 3993
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.284057971014493e-06,
+      "loss": 0.0474,
+      "step": 3994
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.282608695652175e-06,
+      "loss": 0.2418,
+      "step": 3995
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.281159420289855e-06,
+      "loss": 0.2045,
+      "step": 3996
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.279710144927536e-06,
+      "loss": 2.8682,
+      "step": 3997
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.278260869565218e-06,
+      "loss": 0.1913,
+      "step": 3998
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.276811594202899e-06,
+      "loss": 0.1533,
+      "step": 3999
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.27536231884058e-06,
+      "loss": 0.2027,
+      "step": 4000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.273913043478261e-06,
+      "loss": 0.0947,
+      "step": 4001
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.272463768115943e-06,
+      "loss": 0.0928,
+      "step": 4002
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.271014492753624e-06,
+      "loss": 0.1321,
+      "step": 4003
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.269565217391305e-06,
+      "loss": 0.3533,
+      "step": 4004
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.268115942028986e-06,
+      "loss": 0.1453,
+      "step": 4005
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.7836,
+      "step": 4006
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.265217391304349e-06,
+      "loss": 0.8818,
+      "step": 4007
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.263768115942029e-06,
+      "loss": 0.1565,
+      "step": 4008
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.262318840579711e-06,
+      "loss": 0.7528,
+      "step": 4009
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.260869565217392e-06,
+      "loss": 0.1719,
+      "step": 4010
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.259420289855073e-06,
+      "loss": 0.9927,
+      "step": 4011
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.257971014492754e-06,
+      "loss": 0.1433,
+      "step": 4012
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.2565217391304355e-06,
+      "loss": 0.7415,
+      "step": 4013
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2550724637681165e-06,
+      "loss": 0.1744,
+      "step": 4014
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2536231884057975e-06,
+      "loss": 0.055,
+      "step": 4015
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2521739130434785e-06,
+      "loss": 0.0912,
+      "step": 4016
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2507246376811595e-06,
+      "loss": 2.4495,
+      "step": 4017
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.249275362318841e-06,
+      "loss": 0.1304,
+      "step": 4018
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2478260869565215e-06,
+      "loss": 0.2779,
+      "step": 4019
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.246376811594203e-06,
+      "loss": 0.0565,
+      "step": 4020
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.244927536231884e-06,
+      "loss": 0.1474,
+      "step": 4021
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.243478260869565e-06,
+      "loss": 0.1678,
+      "step": 4022
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.242028985507246e-06,
+      "loss": 0.1788,
+      "step": 4023
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.240579710144928e-06,
+      "loss": 0.1087,
+      "step": 4024
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.239130434782609e-06,
+      "loss": 1.2389,
+      "step": 4025
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.23768115942029e-06,
+      "loss": 0.2254,
+      "step": 4026
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.236231884057971e-06,
+      "loss": 0.0814,
+      "step": 4027
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.234782608695652e-06,
+      "loss": 0.3499,
+      "step": 4028
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.233333333333334e-06,
+      "loss": 3.2256,
+      "step": 4029
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.231884057971015e-06,
+      "loss": 0.0465,
+      "step": 4030
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.230434782608696e-06,
+      "loss": 0.1153,
+      "step": 4031
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.228985507246377e-06,
+      "loss": 1.8082,
+      "step": 4032
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.227536231884059e-06,
+      "loss": 0.0531,
+      "step": 4033
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.226086956521739e-06,
+      "loss": 0.2322,
+      "step": 4034
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.224637681159421e-06,
+      "loss": 0.2532,
+      "step": 4035
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.223188405797102e-06,
+      "loss": 0.2969,
+      "step": 4036
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.221739130434783e-06,
+      "loss": 0.1219,
+      "step": 4037
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.220289855072464e-06,
+      "loss": 0.1266,
+      "step": 4038
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.218840579710145e-06,
+      "loss": 0.0912,
+      "step": 4039
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.217391304347827e-06,
+      "loss": 0.7878,
+      "step": 4040
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.215942028985508e-06,
+      "loss": 0.2761,
+      "step": 4041
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.214492753623189e-06,
+      "loss": 0.0965,
+      "step": 4042
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.21304347826087e-06,
+      "loss": 0.4293,
+      "step": 4043
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.211594202898552e-06,
+      "loss": 0.1446,
+      "step": 4044
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.210144927536232e-06,
+      "loss": 0.0518,
+      "step": 4045
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.208695652173914e-06,
+      "loss": 0.046,
+      "step": 4046
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.207246376811595e-06,
+      "loss": 0.8073,
+      "step": 4047
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.205797101449276e-06,
+      "loss": 1.2858,
+      "step": 4048
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.204347826086957e-06,
+      "loss": 1.3378,
+      "step": 4049
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.202898550724638e-06,
+      "loss": 0.1578,
+      "step": 4050
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.2014492753623194e-06,
+      "loss": 0.2814,
+      "step": 4051
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0779,
+      "step": 4052
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1985507246376814e-06,
+      "loss": 0.0837,
+      "step": 4053
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1971014492753624e-06,
+      "loss": 1.2062,
+      "step": 4054
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.195652173913044e-06,
+      "loss": 0.2897,
+      "step": 4055
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.194202898550725e-06,
+      "loss": 0.8654,
+      "step": 4056
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.192753623188406e-06,
+      "loss": 1.4031,
+      "step": 4057
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.191304347826087e-06,
+      "loss": 0.141,
+      "step": 4058
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.189855072463768e-06,
+      "loss": 0.0949,
+      "step": 4059
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.188405797101449e-06,
+      "loss": 0.11,
+      "step": 4060
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.18695652173913e-06,
+      "loss": 0.2613,
+      "step": 4061
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.185507246376812e-06,
+      "loss": 0.372,
+      "step": 4062
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.184057971014493e-06,
+      "loss": 0.6136,
+      "step": 4063
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.182608695652174e-06,
+      "loss": 1.1772,
+      "step": 4064
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.181159420289855e-06,
+      "loss": 0.6292,
+      "step": 4065
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.179710144927537e-06,
+      "loss": 0.0959,
+      "step": 4066
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.178260869565218e-06,
+      "loss": 0.2054,
+      "step": 4067
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.176811594202899e-06,
+      "loss": 0.359,
+      "step": 4068
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.17536231884058e-06,
+      "loss": 1.109,
+      "step": 4069
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.173913043478261e-06,
+      "loss": 0.1455,
+      "step": 4070
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.172463768115943e-06,
+      "loss": 0.1335,
+      "step": 4071
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.171014492753623e-06,
+      "loss": 0.1277,
+      "step": 4072
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.169565217391305e-06,
+      "loss": 0.3778,
+      "step": 4073
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.168115942028986e-06,
+      "loss": 2.3952,
+      "step": 4074
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.1534,
+      "step": 4075
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.165217391304348e-06,
+      "loss": 0.1729,
+      "step": 4076
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.16376811594203e-06,
+      "loss": 0.0778,
+      "step": 4077
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.162318840579711e-06,
+      "loss": 0.1441,
+      "step": 4078
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.160869565217392e-06,
+      "loss": 0.3546,
+      "step": 4079
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.159420289855073e-06,
+      "loss": 1.7534,
+      "step": 4080
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.157971014492754e-06,
+      "loss": 0.0969,
+      "step": 4081
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.1565217391304356e-06,
+      "loss": 0.1021,
+      "step": 4082
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.155072463768116e-06,
+      "loss": 0.1474,
+      "step": 4083
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1536231884057976e-06,
+      "loss": 0.1084,
+      "step": 4084
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1521739130434786e-06,
+      "loss": 0.118,
+      "step": 4085
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1507246376811596e-06,
+      "loss": 0.3069,
+      "step": 4086
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.149275362318841e-06,
+      "loss": 0.1757,
+      "step": 4087
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1478260869565224e-06,
+      "loss": 1.1796,
+      "step": 4088
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1463768115942034e-06,
+      "loss": 1.1259,
+      "step": 4089
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1449275362318844e-06,
+      "loss": 0.0773,
+      "step": 4090
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.1434782608695654e-06,
+      "loss": 0.1292,
+      "step": 4091
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.1420289855072464e-06,
+      "loss": 0.1744,
+      "step": 4092
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.140579710144928e-06,
+      "loss": 0.1282,
+      "step": 4093
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.139130434782609e-06,
+      "loss": 0.1932,
+      "step": 4094
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.13768115942029e-06,
+      "loss": 0.2569,
+      "step": 4095
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.136231884057971e-06,
+      "loss": 0.114,
+      "step": 4096
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.134782608695653e-06,
+      "loss": 0.1431,
+      "step": 4097
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 0.0466,
+      "step": 4098
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.131884057971015e-06,
+      "loss": 0.2426,
+      "step": 4099
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.130434782608696e-06,
+      "loss": 0.11,
+      "step": 4100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.128985507246377e-06,
+      "loss": 0.8204,
+      "step": 4101
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.127536231884058e-06,
+      "loss": 0.1134,
+      "step": 4102
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.126086956521739e-06,
+      "loss": 0.4387,
+      "step": 4103
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.124637681159421e-06,
+      "loss": 0.0653,
+      "step": 4104
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.123188405797102e-06,
+      "loss": 1.2444,
+      "step": 4105
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.121739130434783e-06,
+      "loss": 0.827,
+      "step": 4106
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.120289855072464e-06,
+      "loss": 0.4144,
+      "step": 4107
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.118840579710146e-06,
+      "loss": 0.1117,
+      "step": 4108
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.117391304347826e-06,
+      "loss": 0.4255,
+      "step": 4109
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.115942028985507e-06,
+      "loss": 0.0705,
+      "step": 4110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.114492753623189e-06,
+      "loss": 1.5372,
+      "step": 4111
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.11304347826087e-06,
+      "loss": 0.2426,
+      "step": 4112
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.111594202898551e-06,
+      "loss": 0.0972,
+      "step": 4113
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.110144927536232e-06,
+      "loss": 0.4538,
+      "step": 4114
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.108695652173914e-06,
+      "loss": 0.6555,
+      "step": 4115
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.107246376811595e-06,
+      "loss": 0.0868,
+      "step": 4116
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.105797101449276e-06,
+      "loss": 1.3329,
+      "step": 4117
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.104347826086957e-06,
+      "loss": 0.162,
+      "step": 4118
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.1028985507246386e-06,
+      "loss": 0.3548,
+      "step": 4119
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.1014492753623196e-06,
+      "loss": 0.1527,
+      "step": 4120
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.1e-06,
+      "loss": 0.1072,
+      "step": 4121
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0985507246376816e-06,
+      "loss": 0.1648,
+      "step": 4122
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0971014492753626e-06,
+      "loss": 0.8045,
+      "step": 4123
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0956521739130436e-06,
+      "loss": 1.3249,
+      "step": 4124
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0942028985507246e-06,
+      "loss": 0.2987,
+      "step": 4125
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.092753623188406e-06,
+      "loss": 0.8809,
+      "step": 4126
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.091304347826087e-06,
+      "loss": 0.1357,
+      "step": 4127
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.089855072463768e-06,
+      "loss": 0.4468,
+      "step": 4128
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.088405797101449e-06,
+      "loss": 0.2358,
+      "step": 4129
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.086956521739131e-06,
+      "loss": 0.4758,
+      "step": 4130
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.085507246376812e-06,
+      "loss": 0.1509,
+      "step": 4131
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.084057971014493e-06,
+      "loss": 0.114,
+      "step": 4132
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.082608695652174e-06,
+      "loss": 0.1449,
+      "step": 4133
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.081159420289855e-06,
+      "loss": 1.0676,
+      "step": 4134
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.079710144927537e-06,
+      "loss": 0.291,
+      "step": 4135
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.078260869565217e-06,
+      "loss": 0.1116,
+      "step": 4136
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.076811594202899e-06,
+      "loss": 0.1357,
+      "step": 4137
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.07536231884058e-06,
+      "loss": 1.4207,
+      "step": 4138
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.073913043478261e-06,
+      "loss": 0.1202,
+      "step": 4139
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.072463768115942e-06,
+      "loss": 0.0982,
+      "step": 4140
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.071014492753623e-06,
+      "loss": 0.0653,
+      "step": 4141
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.069565217391305e-06,
+      "loss": 1.6376,
+      "step": 4142
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.068115942028986e-06,
+      "loss": 0.0766,
+      "step": 4143
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.066666666666667e-06,
+      "loss": 0.1586,
+      "step": 4144
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.065217391304348e-06,
+      "loss": 0.3503,
+      "step": 4145
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.06376811594203e-06,
+      "loss": 1.103,
+      "step": 4146
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.06231884057971e-06,
+      "loss": 1.3051,
+      "step": 4147
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.060869565217392e-06,
+      "loss": 0.089,
+      "step": 4148
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.059420289855073e-06,
+      "loss": 0.0798,
+      "step": 4149
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.057971014492754e-06,
+      "loss": 0.8442,
+      "step": 4150
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.056521739130435e-06,
+      "loss": 0.6468,
+      "step": 4151
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.055072463768116e-06,
+      "loss": 1.7923,
+      "step": 4152
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.053623188405798e-06,
+      "loss": 0.9793,
+      "step": 4153
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.052173913043479e-06,
+      "loss": 0.1145,
+      "step": 4154
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.05072463768116e-06,
+      "loss": 0.0392,
+      "step": 4155
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.049275362318841e-06,
+      "loss": 0.1019,
+      "step": 4156
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0478260869565225e-06,
+      "loss": 0.8367,
+      "step": 4157
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0463768115942035e-06,
+      "loss": 0.0675,
+      "step": 4158
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0449275362318845e-06,
+      "loss": 0.9371,
+      "step": 4159
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0434782608695655e-06,
+      "loss": 0.0784,
+      "step": 4160
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.042028985507247e-06,
+      "loss": 0.2056,
+      "step": 4161
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0405797101449275e-06,
+      "loss": 0.0937,
+      "step": 4162
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0391304347826085e-06,
+      "loss": 0.4021,
+      "step": 4163
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.03768115942029e-06,
+      "loss": 0.2703,
+      "step": 4164
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.036231884057971e-06,
+      "loss": 1.1871,
+      "step": 4165
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.034782608695652e-06,
+      "loss": 0.3573,
+      "step": 4166
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.033333333333333e-06,
+      "loss": 0.1365,
+      "step": 4167
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.031884057971015e-06,
+      "loss": 0.181,
+      "step": 4168
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.030434782608696e-06,
+      "loss": 0.1776,
+      "step": 4169
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.028985507246377e-06,
+      "loss": 0.0347,
+      "step": 4170
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9568345323741008,
+      "eval_loss": 0.21844688057899475,
+      "eval_roc_auc": 0.9918676422650899,
+      "eval_runtime": 85.9187,
+      "eval_samples_per_second": 8.089,
+      "eval_steps_per_second": 2.025,
+      "step": 4170
+    }
+  ],
+  "max_steps": 6950,
+  "num_train_epochs": 10,
+  "total_flos": 1.211030129304576e+18,
+  "trial_name": null,
+  "trial_params": null
+}