diff --git "a/checkpoint-1390/trainer_state.json" "b/checkpoint-1390/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1390/trainer_state.json"
@@ -0,0 +1,8376 @@
+{
+  "best_metric": 1.2769397497177124,
+  "best_model_checkpoint": "multiple_accent_classification/checkpoint-1390",
+  "epoch": 2.0,
+  "global_step": 1390,
+  "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
+    }
+  ],
+  "max_steps": 6950,
+  "num_train_epochs": 10,
+  "total_flos": 4.03676709768192e+17,
+  "trial_name": null,
+  "trial_params": null
+}