diff --git "a/Chinese_tiny_llm-zh_105k/trainer_state.json" "b/Chinese_tiny_llm-zh_105k/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Chinese_tiny_llm-zh_105k/trainer_state.json"
@@ -0,0 +1,14799 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9990867579908675,
+  "eval_steps": 500,
+  "global_step": 2463,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6129032258064518e-07,
+      "loss": 2.5207,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 2.5145,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838709677419355e-07,
+      "loss": 2.5788,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 2.5129,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 2.5563,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 2.6393,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1290322580645162e-06,
+      "loss": 3.243,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 2.577,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4516129032258066e-06,
+      "loss": 2.433,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 2.3083,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.774193548387097e-06,
+      "loss": 2.6382,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 2.9139,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.096774193548387e-06,
+      "loss": 2.7043,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 2.4413,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 2.7663,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 2.6213,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7419354838709676e-06,
+      "loss": 2.3298,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 2.5933,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0645161290322584e-06,
+      "loss": 2.5047,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 2.2711,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3870967741935484e-06,
+      "loss": 2.3215,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 2.2406,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7096774193548392e-06,
+      "loss": 2.3551,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 2.282,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 2.3598,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 2.2977,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.35483870967742e-06,
+      "loss": 2.213,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 2.1279,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.67741935483871e-06,
+      "loss": 2.2377,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 2.1427,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 2.1985,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 2.3242,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.322580645161291e-06,
+      "loss": 2.0709,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 2.1687,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 1.9598,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 2.429,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.967741935483872e-06,
+      "loss": 2.1136,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 1.9272,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.290322580645162e-06,
+      "loss": 2.0787,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 1.953,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.612903225806452e-06,
+      "loss": 2.0808,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 2.2803,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.935483870967743e-06,
+      "loss": 2.1082,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 2.0066,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 2.1115,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 2.3001,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.580645161290323e-06,
+      "loss": 2.199,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 2.1237,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.903225806451613e-06,
+      "loss": 1.9017,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 2.0559,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.225806451612904e-06,
+      "loss": 2.0735,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 2.1449,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.548387096774194e-06,
+      "loss": 1.8838,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 2.0179,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 2.0281,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 1.9958,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.193548387096775e-06,
+      "loss": 2.0776,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 2.1152,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.516129032258065e-06,
+      "loss": 2.1996,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 2.1103,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.838709677419356e-06,
+      "loss": 2.24,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 2.1092,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0161290322580646e-05,
+      "loss": 2.0969,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 1.6788,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 2.2611,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 2.0715,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0806451612903225e-05,
+      "loss": 1.903,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 1.922,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1129032258064516e-05,
+      "loss": 2.0652,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 2.0523,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1451612903225808e-05,
+      "loss": 2.4514,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 1.9532,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1774193548387098e-05,
+      "loss": 2.1664,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 1.9944,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 2.085,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 2.1888,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2419354838709679e-05,
+      "loss": 2.0487,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 2.1206,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.274193548387097e-05,
+      "loss": 2.1393,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 2.0151,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3064516129032258e-05,
+      "loss": 1.9447,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 1.8504,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3387096774193548e-05,
+      "loss": 1.9581,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 2.2026,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 2.2578,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 1.9956,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4032258064516131e-05,
+      "loss": 1.8559,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 2.1944,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4354838709677421e-05,
+      "loss": 2.0336,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 1.8468,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4677419354838712e-05,
+      "loss": 1.8434,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 2.2704,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.853,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 2.0141,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 2.2305,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 1.9325,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5645161290322583e-05,
+      "loss": 2.0163,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 2.2868,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.596774193548387e-05,
+      "loss": 2.1631,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 1.997,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6290322580645164e-05,
+      "loss": 2.0891,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 2.173,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6612903225806454e-05,
+      "loss": 2.0087,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 2.0722,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6935483870967744e-05,
+      "loss": 1.8766,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 2.1471,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7258064516129035e-05,
+      "loss": 1.8202,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 1.9646,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7580645161290325e-05,
+      "loss": 2.0087,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 2.089,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7903225806451612e-05,
+      "loss": 1.9932,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 2.0131,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8225806451612903e-05,
+      "loss": 1.9577,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 2.2636,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 2.0801,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 1.659,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8870967741935487e-05,
+      "loss": 1.9992,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 2.0917,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9193548387096777e-05,
+      "loss": 2.1657,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 2.0809,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9516129032258068e-05,
+      "loss": 2.1582,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 2.0755,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9838709677419358e-05,
+      "loss": 2.0114,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 1.7352,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999990979943856e-05,
+      "loss": 2.1899,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999963919791692e-05,
+      "loss": 1.7923,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999918819592326e-05,
+      "loss": 1.86,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999985567942712e-05,
+      "loss": 2.0418,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999774499409982e-05,
+      "loss": 2.1118,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999675279687354e-05,
+      "loss": 2.0651,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999955802043824e-05,
+      "loss": 2.2576,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999422721874167e-05,
+      "loss": 2.194,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999269384239222e-05,
+      "loss": 2.1086,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999098007810024e-05,
+      "loss": 1.7276,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999890859289574e-05,
+      "loss": 2.0556,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998701139838074e-05,
+      "loss": 1.9761,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998475649011278e-05,
+      "loss": 2.0811,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998232120822133e-05,
+      "loss": 1.8236,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997970555709977e-05,
+      "loss": 2.0507,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999769095414667e-05,
+      "loss": 2.1234,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997393316636616e-05,
+      "loss": 2.1476,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997077643716755e-05,
+      "loss": 2.1369,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999674393595657e-05,
+      "loss": 1.9781,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999639219395807e-05,
+      "loss": 1.8806,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99960224183558e-05,
+      "loss": 2.1054,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995634609816842e-05,
+      "loss": 2.0344,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995228769040808e-05,
+      "loss": 1.9186,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994804896759834e-05,
+      "loss": 1.7998,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994362993738595e-05,
+      "loss": 1.9016,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999390306077429e-05,
+      "loss": 1.9628,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999342509869664e-05,
+      "loss": 1.8174,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992929108367897e-05,
+      "loss": 2.2689,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999241509068283e-05,
+      "loss": 2.0546,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999188304656873e-05,
+      "loss": 2.027,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991332976985417e-05,
+      "loss": 2.0501,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990764882925223e-05,
+      "loss": 2.1867,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999017876541299e-05,
+      "loss": 1.8642,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989574625506084e-05,
+      "loss": 1.8054,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998895246429438e-05,
+      "loss": 1.8356,
+      "step": 159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9988312282900266e-05,
+      "loss": 2.0384,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987654082478633e-05,
+      "loss": 2.0395,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986977864216885e-05,
+      "loss": 2.2216,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986283629334918e-05,
+      "loss": 2.2781,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9985571379085155e-05,
+      "loss": 2.0535,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9984841114752494e-05,
+      "loss": 2.0179,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998409283765434e-05,
+      "loss": 2.0948,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9983326549140598e-05,
+      "loss": 2.0524,
+      "step": 167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982542250593653e-05,
+      "loss": 2.1748,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9981739943428392e-05,
+      "loss": 1.9585,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9980919629092192e-05,
+      "loss": 1.8113,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9980081309064904e-05,
+      "loss": 1.6517,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9979224984858868e-05,
+      "loss": 1.9969,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9978350658018902e-05,
+      "loss": 1.9404,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9977458330122297e-05,
+      "loss": 1.9997,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997654800277883e-05,
+      "loss": 2.2508,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997561967763074e-05,
+      "loss": 1.9998,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9974673356352734e-05,
+      "loss": 2.0957,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9973709040651984e-05,
+      "loss": 1.8238,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9972726732268132e-05,
+      "loss": 1.9144,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997172643297327e-05,
+      "loss": 2.1058,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997070814457195e-05,
+      "loss": 1.917,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969671868901173e-05,
+      "loss": 1.5747,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99686176078304e-05,
+      "loss": 1.8785,
+      "step": 183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996754536326152e-05,
+      "loss": 2.0738,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9966455137128877e-05,
+      "loss": 1.9648,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965346931399258e-05,
+      "loss": 1.9158,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964220748071868e-05,
+      "loss": 2.1141,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963076589178357e-05,
+      "loss": 2.0899,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961914456782813e-05,
+      "loss": 1.7556,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996073435298172e-05,
+      "loss": 2.1444,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959536279904005e-05,
+      "loss": 1.9723,
+      "step": 191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958320239711e-05,
+      "loss": 1.7663,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995708623459647e-05,
+      "loss": 1.8706,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955834266786555e-05,
+      "loss": 2.095,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995456433853983e-05,
+      "loss": 1.8566,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995327645214726e-05,
+      "loss": 2.0479,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9951970609932202e-05,
+      "loss": 1.9399,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950646814250415e-05,
+      "loss": 1.8787,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.994930506749004e-05,
+      "loss": 1.9375,
+      "step": 199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9947945372071595e-05,
+      "loss": 2.1012,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9946567730448e-05,
+      "loss": 2.1439,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9945172145104523e-05,
+      "loss": 1.9453,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943758618558822e-05,
+      "loss": 1.9632,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9942327153360917e-05,
+      "loss": 2.0806,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9940877752093183e-05,
+      "loss": 1.8044,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9939410417370357e-05,
+      "loss": 1.8723,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993792515183953e-05,
+      "loss": 1.9196,
+      "step": 207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993642195818013e-05,
+      "loss": 2.0066,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9934900839103944e-05,
+      "loss": 2.0548,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9933361797355086e-05,
+      "loss": 1.9798,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9931804835710003e-05,
+      "loss": 2.0358,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993022995697747e-05,
+      "loss": 2.0076,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9928637163998584e-05,
+      "loss": 1.9636,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992702645964677e-05,
+      "loss": 2.0984,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992539784682775e-05,
+      "loss": 1.8356,
+      "step": 215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9923751328479563e-05,
+      "loss": 1.9758,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9922086907572545e-05,
+      "loss": 2.1682,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9920404587109325e-05,
+      "loss": 2.0138,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9918704370124837e-05,
+      "loss": 2.0854,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9916986259686286e-05,
+      "loss": 2.197,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9915250258893163e-05,
+      "loss": 2.1957,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9913496370877236e-05,
+      "loss": 1.7028,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991172459880254e-05,
+      "loss": 1.9272,
+      "step": 223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909934945865365e-05,
+      "loss": 1.7391,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908127415294264e-05,
+      "loss": 1.7921,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906302010350053e-05,
+      "loss": 2.0703,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9904458734325775e-05,
+      "loss": 2.2371,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990259759054672e-05,
+      "loss": 2.2859,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9900718582370415e-05,
+      "loss": 2.1366,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9898821713186614e-05,
+      "loss": 2.1681,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9896906986417286e-05,
+      "loss": 1.7152,
+      "step": 231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9894974405516624e-05,
+      "loss": 1.9233,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989302397397102e-05,
+      "loss": 2.0261,
+      "step": 233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9891055695299077e-05,
+      "loss": 1.7388,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9889069573051597e-05,
+      "loss": 1.8859,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988706561081156e-05,
+      "loss": 1.8946,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9885043812194134e-05,
+      "loss": 1.7558,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9883004180846678e-05,
+      "loss": 2.1105,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.98809467204487e-05,
+      "loss": 2.0252,
+      "step": 239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9878871434711882e-05,
+      "loss": 2.1284,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987677832738007e-05,
+      "loss": 2.2478,
+      "step": 241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9874667402229242e-05,
+      "loss": 2.1557,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9872538663067546e-05,
+      "loss": 2.1735,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9870392113735234e-05,
+      "loss": 1.9844,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9868227758104718e-05,
+      "loss": 1.8116,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9866045600080514e-05,
+      "loss": 2.1697,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9863845643599263e-05,
+      "loss": 1.941,
+      "step": 247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9861627892629703e-05,
+      "loss": 2.0004,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9859392351172694e-05,
+      "loss": 2.2738,
+      "step": 249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9857139023261168e-05,
+      "loss": 1.5824,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.985486791296016e-05,
+      "loss": 2.1265,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9852579024366776e-05,
+      "loss": 1.6058,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9850272361610195e-05,
+      "loss": 2.1647,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9847947928851667e-05,
+      "loss": 2.0199,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984560573028449e-05,
+      "loss": 1.8089,
+      "step": 255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9843245770134017e-05,
+      "loss": 1.952,
+      "step": 256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984086805265765e-05,
+      "loss": 2.1597,
+      "step": 257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9838472582144806e-05,
+      "loss": 2.0161,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9836059362916947e-05,
+      "loss": 2.2478,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983362839932755e-05,
+      "loss": 1.8431,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9831179695762102e-05,
+      "loss": 2.0511,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9828713256638082e-05,
+      "loss": 1.7073,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9826229086404983e-05,
+      "loss": 2.1016,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982372718954427e-05,
+      "loss": 2.1155,
+      "step": 264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9821207570569398e-05,
+      "loss": 1.9292,
+      "step": 265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9818670234025783e-05,
+      "loss": 1.5905,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9816115184490812e-05,
+      "loss": 1.8842,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9813542426573823e-05,
+      "loss": 2.052,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.98109519649161e-05,
+      "loss": 2.2254,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9808343804190867e-05,
+      "loss": 1.9298,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980571794910327e-05,
+      "loss": 1.8935,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980307440439038e-05,
+      "loss": 1.9568,
+      "step": 272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980041317482119e-05,
+      "loss": 2.3348,
+      "step": 273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.979773426519658e-05,
+      "loss": 1.8759,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.979503768034934e-05,
+      "loss": 2.0232,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9792323425144135e-05,
+      "loss": 1.7306,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978959150447751e-05,
+      "loss": 1.7832,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9786841923277883e-05,
+      "loss": 2.058,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978407468650553e-05,
+      "loss": 1.8205,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978128979915258e-05,
+      "loss": 1.8996,
+      "step": 280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9778487266242992e-05,
+      "loss": 2.0165,
+      "step": 281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9775667092832576e-05,
+      "loss": 1.8078,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977282928400895e-05,
+      "loss": 1.811,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9769973844891558e-05,
+      "loss": 1.7681,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9767100780631638e-05,
+      "loss": 1.9626,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9764210096412238e-05,
+      "loss": 2.0063,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9761301797448183e-05,
+      "loss": 2.1175,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975837588898607e-05,
+      "loss": 1.9754,
+      "step": 288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9755432376304278e-05,
+      "loss": 2.2613,
+      "step": 289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975247126471294e-05,
+      "loss": 1.8823,
+      "step": 290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.974949255955392e-05,
+      "loss": 2.0303,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9746496266200853e-05,
+      "loss": 1.9506,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.974348239005907e-05,
+      "loss": 1.9468,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9740450936565653e-05,
+      "loss": 1.9065,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9737401911189365e-05,
+      "loss": 2.1423,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9734335319430687e-05,
+      "loss": 1.9627,
+      "step": 296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.973125116682179e-05,
+      "loss": 2.1064,
+      "step": 297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9728149458926512e-05,
+      "loss": 2.0344,
+      "step": 298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9725030201340372e-05,
+      "loss": 1.8448,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9721893399690547e-05,
+      "loss": 2.0863,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9718739059635858e-05,
+      "loss": 1.701,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9715567186866776e-05,
+      "loss": 1.9583,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.971237778710539e-05,
+      "loss": 2.0764,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9709170866105418e-05,
+      "loss": 2.1598,
+      "step": 304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9705946429652177e-05,
+      "loss": 1.906,
+      "step": 305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9702704483562586e-05,
+      "loss": 1.8985,
+      "step": 306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9699445033685158e-05,
+      "loss": 1.8673,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.969616808589997e-05,
+      "loss": 2.095,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.969287364611868e-05,
+      "loss": 2.1897,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9689561720284487e-05,
+      "loss": 1.7424,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.968623231437215e-05,
+      "loss": 2.1236,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9682885434387945e-05,
+      "loss": 2.1478,
+      "step": 312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967952108636969e-05,
+      "loss": 2.1148,
+      "step": 313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.96761392763867e-05,
+      "loss": 2.0232,
+      "step": 314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9672740010539805e-05,
+      "loss": 2.1999,
+      "step": 315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9669323294961315e-05,
+      "loss": 1.9045,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9665889135815025e-05,
+      "loss": 2.0132,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9662437539296192e-05,
+      "loss": 1.7199,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9658968511631538e-05,
+      "loss": 1.82,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9655482059079226e-05,
+      "loss": 1.7305,
+      "step": 320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.965197818792886e-05,
+      "loss": 2.03,
+      "step": 321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964845690450146e-05,
+      "loss": 2.032,
+      "step": 322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9644918215149463e-05,
+      "loss": 1.9535,
+      "step": 323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.96413621262567e-05,
+      "loss": 2.1946,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9637788644238397e-05,
+      "loss": 1.8656,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9634197775541155e-05,
+      "loss": 2.1057,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9630589526642946e-05,
+      "loss": 1.897,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962696390405309e-05,
+      "loss": 1.7691,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9623320914312246e-05,
+      "loss": 2.0512,
+      "step": 329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9619660563992414e-05,
+      "loss": 1.7593,
+      "step": 330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.96159828596969e-05,
+      "loss": 1.8791,
+      "step": 331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9612287808060334e-05,
+      "loss": 2.1388,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960857541574862e-05,
+      "loss": 1.8134,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9604845689458963e-05,
+      "loss": 1.7492,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960109863591983e-05,
+      "loss": 1.9242,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9597334261890948e-05,
+      "loss": 2.0666,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9593552574163286e-05,
+      "loss": 2.1282,
+      "step": 337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9589753579559053e-05,
+      "loss": 2.0797,
+      "step": 338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.958593728493168e-05,
+      "loss": 1.9703,
+      "step": 339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9582103697165804e-05,
+      "loss": 1.7957,
+      "step": 340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.957825282317726e-05,
+      "loss": 2.0525,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9574384669913065e-05,
+      "loss": 1.6631,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9570499244351416e-05,
+      "loss": 2.1923,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9566596553501664e-05,
+      "loss": 1.9094,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9562676604404306e-05,
+      "loss": 2.1605,
+      "step": 345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.955873940413097e-05,
+      "loss": 2.0452,
+      "step": 346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9554784959784414e-05,
+      "loss": 2.0895,
+      "step": 347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9550813278498502e-05,
+      "loss": 1.9968,
+      "step": 348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9546824367438186e-05,
+      "loss": 2.0513,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9542818233799513e-05,
+      "loss": 1.8328,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9538794884809587e-05,
+      "loss": 1.9633,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9534754327726574e-05,
+      "loss": 1.79,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9530696569839692e-05,
+      "loss": 1.7187,
+      "step": 353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9526621618469175e-05,
+      "loss": 2.0143,
+      "step": 354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9522529480966284e-05,
+      "loss": 2.0626,
+      "step": 355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.951842016471328e-05,
+      "loss": 2.1984,
+      "step": 356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9514293677123415e-05,
+      "loss": 2.0069,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9510150025640922e-05,
+      "loss": 2.0029,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950598921774099e-05,
+      "loss": 1.8949,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9501811260929767e-05,
+      "loss": 2.036,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.949761616274433e-05,
+      "loss": 1.8641,
+      "step": 361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.949340393075269e-05,
+      "loss": 1.961,
+      "step": 362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9489174572553756e-05,
+      "loss": 2.1743,
+      "step": 363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948492809577734e-05,
+      "loss": 2.2121,
+      "step": 364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9480664508084127e-05,
+      "loss": 1.9521,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9476383817165687e-05,
+      "loss": 1.8577,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9472086030744427e-05,
+      "loss": 1.8453,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9467771156573607e-05,
+      "loss": 1.656,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9463439202437304e-05,
+      "loss": 2.192,
+      "step": 369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9459090176150416e-05,
+      "loss": 2.1148,
+      "step": 370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9454724085558633e-05,
+      "loss": 2.1243,
+      "step": 371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9450340938538433e-05,
+      "loss": 1.7917,
+      "step": 372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.944594074299706e-05,
+      "loss": 1.7953,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9441523506872518e-05,
+      "loss": 1.8143,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.943708923813355e-05,
+      "loss": 1.7996,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9432637944779627e-05,
+      "loss": 2.1183,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.942816963484093e-05,
+      "loss": 1.9369,
+      "step": 377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9423684316378347e-05,
+      "loss": 1.7545,
+      "step": 378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9419181997483434e-05,
+      "loss": 1.8895,
+      "step": 379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.941466268627843e-05,
+      "loss": 2.048,
+      "step": 380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9410126390916222e-05,
+      "loss": 2.0853,
+      "step": 381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.940557311958034e-05,
+      "loss": 2.1418,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.940100288048493e-05,
+      "loss": 2.0857,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939641568187476e-05,
+      "loss": 2.1174,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939181153202519e-05,
+      "loss": 2.0177,
+      "step": 385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.938719043924215e-05,
+      "loss": 2.1718,
+      "step": 386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9382552411862156e-05,
+      "loss": 2.0568,
+      "step": 387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9377897458252246e-05,
+      "loss": 2.1037,
+      "step": 388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9373225586810017e-05,
+      "loss": 2.0926,
+      "step": 389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.936853680596358e-05,
+      "loss": 1.9285,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9363831124171544e-05,
+      "loss": 2.0107,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9359108549923013e-05,
+      "loss": 2.0444,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9354369091737565e-05,
+      "loss": 2.2889,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9349612758165233e-05,
+      "loss": 1.7483,
+      "step": 394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9344839557786498e-05,
+      "loss": 1.918,
+      "step": 395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9340049499212266e-05,
+      "loss": 2.0738,
+      "step": 396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.933524259108386e-05,
+      "loss": 1.8562,
+      "step": 397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9330418842072993e-05,
+      "loss": 1.8352,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932557826088176e-05,
+      "loss": 2.0174,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932072085624263e-05,
+      "loss": 1.9453,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9315846636918413e-05,
+      "loss": 2.1166,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.931095561170225e-05,
+      "loss": 1.801,
+      "step": 402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9306047789417613e-05,
+      "loss": 1.8258,
+      "step": 403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9301123178918265e-05,
+      "loss": 2.0901,
+      "step": 404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929618178908826e-05,
+      "loss": 2.0536,
+      "step": 405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929122362884192e-05,
+      "loss": 1.9614,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9286248707123826e-05,
+      "loss": 2.0506,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9281257032908782e-05,
+      "loss": 2.1316,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927624861520183e-05,
+      "loss": 1.9514,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927122346303822e-05,
+      "loss": 2.415,
+      "step": 410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.926618158548337e-05,
+      "loss": 1.748,
+      "step": 411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9261122991632887e-05,
+      "loss": 1.7835,
+      "step": 412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9256047690612537e-05,
+      "loss": 2.0501,
+      "step": 413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9250955691578216e-05,
+      "loss": 1.9744,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924584700371595e-05,
+      "loss": 1.9035,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9240721636241864e-05,
+      "loss": 1.8804,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9235579598402182e-05,
+      "loss": 1.9426,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9230420899473197e-05,
+      "loss": 2.0108,
+      "step": 418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9225245548761262e-05,
+      "loss": 1.9597,
+      "step": 419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9220053555602765e-05,
+      "loss": 1.7114,
+      "step": 420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9214844929364124e-05,
+      "loss": 1.7995,
+      "step": 421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9209619679441748e-05,
+      "loss": 1.9698,
+      "step": 422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9204377815262062e-05,
+      "loss": 2.1224,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9199119346281437e-05,
+      "loss": 1.8972,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919384428198622e-05,
+      "loss": 1.9163,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9188552631892674e-05,
+      "loss": 2.0286,
+      "step": 426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9183244405547005e-05,
+      "loss": 1.8741,
+      "step": 427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9177919612525303e-05,
+      "loss": 2.0267,
+      "step": 428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917257826243357e-05,
+      "loss": 1.9333,
+      "step": 429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9167220364907645e-05,
+      "loss": 1.7863,
+      "step": 430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9161845929613242e-05,
+      "loss": 1.8852,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9156454966245905e-05,
+      "loss": 1.9128,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915104748453099e-05,
+      "loss": 1.8365,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9145623494223655e-05,
+      "loss": 2.1575,
+      "step": 434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9140183005108842e-05,
+      "loss": 2.0408,
+      "step": 435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9134726027001248e-05,
+      "loss": 2.2291,
+      "step": 436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.912925256974533e-05,
+      "loss": 1.8621,
+      "step": 437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9123762643215262e-05,
+      "loss": 2.0789,
+      "step": 438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9118256257314934e-05,
+      "loss": 1.9532,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.911273342197793e-05,
+      "loss": 1.836,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.91071941471675e-05,
+      "loss": 1.9574,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.910163844287657e-05,
+      "loss": 2.1402,
+      "step": 442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.909606631912768e-05,
+      "loss": 2.0509,
+      "step": 443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9090477785973015e-05,
+      "loss": 1.8478,
+      "step": 444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9084872853494346e-05,
+      "loss": 2.1025,
+      "step": 445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9079251531803034e-05,
+      "loss": 1.801,
+      "step": 446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9073613831040005e-05,
+      "loss": 1.9521,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.906795976137574e-05,
+      "loss": 1.9663,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9062289333010236e-05,
+      "loss": 2.0949,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905660255617302e-05,
+      "loss": 2.0318,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905089944112309e-05,
+      "loss": 1.9098,
+      "step": 451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.904517999814894e-05,
+      "loss": 1.9667,
+      "step": 452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.90394442375685e-05,
+      "loss": 2.0213,
+      "step": 453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.903369216972915e-05,
+      "loss": 1.6855,
+      "step": 454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.902792380500769e-05,
+      "loss": 2.1313,
+      "step": 455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.902213915381031e-05,
+      "loss": 1.9896,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9016338226572582e-05,
+      "loss": 1.7646,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9010521033759455e-05,
+      "loss": 1.8734,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.90046875858652e-05,
+      "loss": 1.8666,
+      "step": 459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8998837893413425e-05,
+      "loss": 1.7822,
+      "step": 460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8992971966957044e-05,
+      "loss": 2.0583,
+      "step": 461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.898708981707825e-05,
+      "loss": 1.9916,
+      "step": 462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8981191454388506e-05,
+      "loss": 1.9042,
+      "step": 463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8975276889528535e-05,
+      "loss": 2.023,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8969346133168268e-05,
+      "loss": 2.0079,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.896339919600686e-05,
+      "loss": 1.797,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895743608877265e-05,
+      "loss": 1.8778,
+      "step": 467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895145682222315e-05,
+      "loss": 1.9398,
+      "step": 468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894546140714503e-05,
+      "loss": 2.1481,
+      "step": 469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8939449854354084e-05,
+      "loss": 2.032,
+      "step": 470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8933422174695217e-05,
+      "loss": 2.1306,
+      "step": 471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.892737837904243e-05,
+      "loss": 2.1779,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8921318478298804e-05,
+      "loss": 1.9308,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8915242483396468e-05,
+      "loss": 1.9035,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8909150405296576e-05,
+      "loss": 2.0592,
+      "step": 475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8903042254989314e-05,
+      "loss": 1.9142,
+      "step": 476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889691804349385e-05,
+      "loss": 1.7983,
+      "step": 477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8890777781858335e-05,
+      "loss": 1.83,
+      "step": 478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8884621481159864e-05,
+      "loss": 1.9545,
+      "step": 479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8878449152504473e-05,
+      "loss": 1.9859,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.887226080702712e-05,
+      "loss": 2.2106,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.886605645589164e-05,
+      "loss": 1.9682,
+      "step": 482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8859836110290754e-05,
+      "loss": 1.8038,
+      "step": 483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.885359978144604e-05,
+      "loss": 1.7911,
+      "step": 484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8847347480607904e-05,
+      "loss": 1.9743,
+      "step": 485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.884107921905556e-05,
+      "loss": 2.0165,
+      "step": 486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.883479500809703e-05,
+      "loss": 2.0757,
+      "step": 487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.88284948590691e-05,
+      "loss": 2.2154,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.882217878333731e-05,
+      "loss": 1.8193,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.881584679229593e-05,
+      "loss": 2.2537,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.880949889736794e-05,
+      "loss": 1.9783,
+      "step": 491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8803135110005018e-05,
+      "loss": 2.2349,
+      "step": 492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.879675544168751e-05,
+      "loss": 2.2371,
+      "step": 493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.87903599039244e-05,
+      "loss": 1.6828,
+      "step": 494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8783948508253317e-05,
+      "loss": 2.0191,
+      "step": 495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8777521266240492e-05,
+      "loss": 1.9493,
+      "step": 496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.877107818948074e-05,
+      "loss": 1.9168,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.876461928959744e-05,
+      "loss": 1.901,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8758144578242525e-05,
+      "loss": 1.7627,
+      "step": 499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8751654067096444e-05,
+      "loss": 1.6406,
+      "step": 500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8745147767868153e-05,
+      "loss": 1.8524,
+      "step": 501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.873862569229509e-05,
+      "loss": 1.9375,
+      "step": 502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.873208785214315e-05,
+      "loss": 2.1537,
+      "step": 503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8725534259206673e-05,
+      "loss": 1.8391,
+      "step": 504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.871896492530841e-05,
+      "loss": 2.0802,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.871237986229952e-05,
+      "loss": 2.096,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8705779082059524e-05,
+      "loss": 1.6819,
+      "step": 507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8699162596496304e-05,
+      "loss": 1.7636,
+      "step": 508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.869253041754608e-05,
+      "loss": 1.7488,
+      "step": 509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.868588255717337e-05,
+      "loss": 1.9398,
+      "step": 510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8679219027371e-05,
+      "loss": 1.9298,
+      "step": 511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.867253984016004e-05,
+      "loss": 1.9779,
+      "step": 512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.866584500758982e-05,
+      "loss": 2.2037,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.86591345417379e-05,
+      "loss": 1.8737,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8652408454710037e-05,
+      "loss": 1.8001,
+      "step": 515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.864566675864016e-05,
+      "loss": 1.8971,
+      "step": 516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8638909465690366e-05,
+      "loss": 1.9949,
+      "step": 517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.863213658805089e-05,
+      "loss": 1.8185,
+      "step": 518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8625348137940082e-05,
+      "loss": 1.712,
+      "step": 519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8618544127604377e-05,
+      "loss": 1.8451,
+      "step": 520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8611724569318283e-05,
+      "loss": 2.1305,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.860488947538437e-05,
+      "loss": 2.1112,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8598038858133213e-05,
+      "loss": 1.9001,
+      "step": 523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.859117272992341e-05,
+      "loss": 2.1771,
+      "step": 524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.858429110314153e-05,
+      "loss": 2.2063,
+      "step": 525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8577393990202112e-05,
+      "loss": 1.9402,
+      "step": 526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8570481403547617e-05,
+      "loss": 2.0979,
+      "step": 527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.856355335564843e-05,
+      "loss": 1.7555,
+      "step": 528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8556609859002826e-05,
+      "loss": 2.0374,
+      "step": 529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.854965092613696e-05,
+      "loss": 2.0082,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8542676569604814e-05,
+      "loss": 2.0658,
+      "step": 531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.853568680198821e-05,
+      "loss": 2.1145,
+      "step": 532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852868163589677e-05,
+      "loss": 1.974,
+      "step": 533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852166108396789e-05,
+      "loss": 1.8787,
+      "step": 534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.851462515886672e-05,
+      "loss": 2.0187,
+      "step": 535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.850757387328616e-05,
+      "loss": 2.0691,
+      "step": 536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.850050723994679e-05,
+      "loss": 2.1734,
+      "step": 537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8493425271596913e-05,
+      "loss": 2.0136,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8486327981012472e-05,
+      "loss": 1.945,
+      "step": 539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8479215380997058e-05,
+      "loss": 2.041,
+      "step": 540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8472087484381883e-05,
+      "loss": 1.7386,
+      "step": 541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8464944304025756e-05,
+      "loss": 1.9296,
+      "step": 542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.845778585281505e-05,
+      "loss": 2.1317,
+      "step": 543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.845061214366369e-05,
+      "loss": 2.1729,
+      "step": 544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8443423189513132e-05,
+      "loss": 1.8606,
+      "step": 545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8436219003332326e-05,
+      "loss": 2.0245,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.842899959811771e-05,
+      "loss": 1.9842,
+      "step": 547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.842176498689317e-05,
+      "loss": 2.3579,
+      "step": 548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.841451518271002e-05,
+      "loss": 1.4143,
+      "step": 549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8407250198647e-05,
+      "loss": 1.9396,
+      "step": 550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.839997004781021e-05,
+      "loss": 1.9443,
+      "step": 551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8392674743333133e-05,
+      "loss": 2.1404,
+      "step": 552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8385364298376575e-05,
+      "loss": 1.9791,
+      "step": 553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8378038726128664e-05,
+      "loss": 1.5049,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8370698039804814e-05,
+      "loss": 2.1548,
+      "step": 555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8363342252647704e-05,
+      "loss": 2.0946,
+      "step": 556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.835597137792726e-05,
+      "loss": 2.0252,
+      "step": 557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.834858542894062e-05,
+      "loss": 1.9359,
+      "step": 558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8341184419012118e-05,
+      "loss": 2.0362,
+      "step": 559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8333768361493258e-05,
+      "loss": 2.0648,
+      "step": 560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.83263372697627e-05,
+      "loss": 2.0586,
+      "step": 561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8318891157226205e-05,
+      "loss": 2.0873,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8311430037316647e-05,
+      "loss": 2.0371,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.830395392349398e-05,
+      "loss": 1.8845,
+      "step": 564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8296462829245184e-05,
+      "loss": 1.9745,
+      "step": 565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8288956768084283e-05,
+      "loss": 2.0463,
+      "step": 566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8281435753552294e-05,
+      "loss": 2.0185,
+      "step": 567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8273899799217214e-05,
+      "loss": 1.7125,
+      "step": 568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.826634891867399e-05,
+      "loss": 1.8645,
+      "step": 569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.825878312554449e-05,
+      "loss": 2.0158,
+      "step": 570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8251202433477494e-05,
+      "loss": 1.7707,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8243606856148657e-05,
+      "loss": 2.1736,
+      "step": 572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.823599640726048e-05,
+      "loss": 1.8956,
+      "step": 573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8228371100542307e-05,
+      "loss": 2.2077,
+      "step": 574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8220730949750266e-05,
+      "loss": 1.8022,
+      "step": 575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8213075968667283e-05,
+      "loss": 1.8447,
+      "step": 576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8205406171103027e-05,
+      "loss": 1.962,
+      "step": 577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.81977215708939e-05,
+      "loss": 1.889,
+      "step": 578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8190022181903006e-05,
+      "loss": 2.0223,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8182308018020127e-05,
+      "loss": 1.9409,
+      "step": 580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8174579093161708e-05,
+      "loss": 1.9427,
+      "step": 581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.816683542127081e-05,
+      "loss": 2.001,
+      "step": 582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8159077016317105e-05,
+      "loss": 1.9679,
+      "step": 583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8151303892296845e-05,
+      "loss": 1.8686,
+      "step": 584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8143516063232833e-05,
+      "loss": 1.9129,
+      "step": 585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8135713543174394e-05,
+      "loss": 2.0488,
+      "step": 586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.812789634619737e-05,
+      "loss": 2.2694,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.812006448640407e-05,
+      "loss": 1.8026,
+      "step": 588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8112217977923254e-05,
+      "loss": 1.798,
+      "step": 589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8104356834910112e-05,
+      "loss": 2.1176,
+      "step": 590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8096481071546237e-05,
+      "loss": 2.0498,
+      "step": 591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8088590702039593e-05,
+      "loss": 1.9803,
+      "step": 592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8080685740624498e-05,
+      "loss": 2.0453,
+      "step": 593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8072766201561584e-05,
+      "loss": 1.8586,
+      "step": 594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8064832099137794e-05,
+      "loss": 1.8423,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8056883447666335e-05,
+      "loss": 1.8153,
+      "step": 596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8048920261486668e-05,
+      "loss": 2.0496,
+      "step": 597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8040942554964464e-05,
+      "loss": 2.187,
+      "step": 598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8032950342491604e-05,
+      "loss": 2.4184,
+      "step": 599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8024943638486122e-05,
+      "loss": 1.6387,
+      "step": 600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8016922457392198e-05,
+      "loss": 1.7536,
+      "step": 601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8008886813680145e-05,
+      "loss": 1.8816,
+      "step": 602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8000836721846344e-05,
+      "loss": 1.7164,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7992772196413257e-05,
+      "loss": 2.117,
+      "step": 604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7984693251929376e-05,
+      "loss": 2.3539,
+      "step": 605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7976599902969205e-05,
+      "loss": 1.9322,
+      "step": 606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.796849216413324e-05,
+      "loss": 2.2824,
+      "step": 607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7960370050047934e-05,
+      "loss": 2.0018,
+      "step": 608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.795223357536567e-05,
+      "loss": 1.8168,
+      "step": 609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.794408275476474e-05,
+      "loss": 2.1572,
+      "step": 610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7935917602949316e-05,
+      "loss": 1.9455,
+      "step": 611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.792773813464942e-05,
+      "loss": 2.1374,
+      "step": 612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.791954436462091e-05,
+      "loss": 1.9496,
+      "step": 613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.791133630764544e-05,
+      "loss": 1.7077,
+      "step": 614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7903113978530432e-05,
+      "loss": 2.03,
+      "step": 615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7894877392109062e-05,
+      "loss": 1.8651,
+      "step": 616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7886626563240224e-05,
+      "loss": 1.912,
+      "step": 617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7878361506808505e-05,
+      "loss": 1.8377,
+      "step": 618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7870082237724165e-05,
+      "loss": 1.9901,
+      "step": 619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7861788770923093e-05,
+      "loss": 1.8039,
+      "step": 620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7853481121366796e-05,
+      "loss": 1.9472,
+      "step": 621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.784515930404237e-05,
+      "loss": 2.0067,
+      "step": 622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7836823333962466e-05,
+      "loss": 1.7341,
+      "step": 623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7828473226165264e-05,
+      "loss": 2.0152,
+      "step": 624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.782010899571446e-05,
+      "loss": 2.0719,
+      "step": 625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7811730657699213e-05,
+      "loss": 1.6528,
+      "step": 626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7803338227234142e-05,
+      "loss": 1.9729,
+      "step": 627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.779493171945928e-05,
+      "loss": 1.8491,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7786511149540072e-05,
+      "loss": 2.0517,
+      "step": 629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7778076532667317e-05,
+      "loss": 2.0365,
+      "step": 630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.776962788405715e-05,
+      "loss": 1.906,
+      "step": 631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.776116521895104e-05,
+      "loss": 1.8054,
+      "step": 632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.775268855261572e-05,
+      "loss": 1.9939,
+      "step": 633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.77441979003432e-05,
+      "loss": 1.9813,
+      "step": 634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7735693277450705e-05,
+      "loss": 2.0502,
+      "step": 635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.772717469928067e-05,
+      "loss": 2.2315,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7718642181200715e-05,
+      "loss": 2.226,
+      "step": 637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7710095738603587e-05,
+      "loss": 1.7076,
+      "step": 638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7701535386907173e-05,
+      "loss": 1.8973,
+      "step": 639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7692961141554433e-05,
+      "loss": 1.9998,
+      "step": 640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7684373018013415e-05,
+      "loss": 1.9482,
+      "step": 641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.767577103177718e-05,
+      "loss": 1.8322,
+      "step": 642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7667155198363816e-05,
+      "loss": 1.9925,
+      "step": 643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.765852553331638e-05,
+      "loss": 1.837,
+      "step": 644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7649882052202884e-05,
+      "loss": 1.9001,
+      "step": 645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.764122477061626e-05,
+      "loss": 2.2415,
+      "step": 646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7632553704174345e-05,
+      "loss": 2.1909,
+      "step": 647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7623868868519847e-05,
+      "loss": 1.7208,
+      "step": 648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7615170279320297e-05,
+      "loss": 2.1053,
+      "step": 649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7606457952268057e-05,
+      "loss": 1.925,
+      "step": 650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7597731903080254e-05,
+      "loss": 2.0768,
+      "step": 651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.758899214749878e-05,
+      "loss": 1.9063,
+      "step": 652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7580238701290258e-05,
+      "loss": 1.8607,
+      "step": 653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7571471580246e-05,
+      "loss": 1.8006,
+      "step": 654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.756269080018199e-05,
+      "loss": 2.0071,
+      "step": 655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7553896376938854e-05,
+      "loss": 2.2349,
+      "step": 656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7545088326381828e-05,
+      "loss": 1.8469,
+      "step": 657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7536266664400742e-05,
+      "loss": 1.8878,
+      "step": 658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7527431406909968e-05,
+      "loss": 1.7606,
+      "step": 659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7518582569848404e-05,
+      "loss": 2.0525,
+      "step": 660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7509720169179463e-05,
+      "loss": 2.2482,
+      "step": 661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7500844220891008e-05,
+      "loss": 1.969,
+      "step": 662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749195474099535e-05,
+      "loss": 2.3349,
+      "step": 663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7483051745529213e-05,
+      "loss": 1.9302,
+      "step": 664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74741352505537e-05,
+      "loss": 1.7586,
+      "step": 665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7465205272154265e-05,
+      "loss": 1.9079,
+      "step": 666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7456261826440696e-05,
+      "loss": 1.7754,
+      "step": 667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744730492954706e-05,
+      "loss": 1.8524,
+      "step": 668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7438334597631707e-05,
+      "loss": 1.7416,
+      "step": 669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7429350846877215e-05,
+      "loss": 2.1352,
+      "step": 670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.742035369349037e-05,
+      "loss": 1.7858,
+      "step": 671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741134315370214e-05,
+      "loss": 1.7515,
+      "step": 672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7402319243767632e-05,
+      "loss": 2.1605,
+      "step": 673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7393281979966094e-05,
+      "loss": 1.8692,
+      "step": 674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738423137860084e-05,
+      "loss": 1.9194,
+      "step": 675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7375167455999265e-05,
+      "loss": 1.722,
+      "step": 676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736609022851278e-05,
+      "loss": 1.5621,
+      "step": 677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7356999712516808e-05,
+      "loss": 1.8927,
+      "step": 678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7347895924410745e-05,
+      "loss": 1.9952,
+      "step": 679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7338778880617923e-05,
+      "loss": 2.0765,
+      "step": 680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7329648597585592e-05,
+      "loss": 2.2819,
+      "step": 681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7320505091784888e-05,
+      "loss": 1.9088,
+      "step": 682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7311348379710795e-05,
+      "loss": 1.9086,
+      "step": 683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7302178477882123e-05,
+      "loss": 1.7906,
+      "step": 684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7292995402841485e-05,
+      "loss": 1.8393,
+      "step": 685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7283799171155244e-05,
+      "loss": 1.9277,
+      "step": 686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.727458979941351e-05,
+      "loss": 1.8474,
+      "step": 687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7265367304230086e-05,
+      "loss": 2.0351,
+      "step": 688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7256131702242464e-05,
+      "loss": 2.0788,
+      "step": 689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7246883010111774e-05,
+      "loss": 1.8413,
+      "step": 690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7237621244522755e-05,
+      "loss": 2.132,
+      "step": 691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.722834642218374e-05,
+      "loss": 2.0788,
+      "step": 692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7219058559826614e-05,
+      "loss": 1.9369,
+      "step": 693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.720975767420678e-05,
+      "loss": 1.9287,
+      "step": 694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7200443782103145e-05,
+      "loss": 1.8068,
+      "step": 695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.719111690031807e-05,
+      "loss": 2.4016,
+      "step": 696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.718177704567736e-05,
+      "loss": 1.9975,
+      "step": 697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7172424235030214e-05,
+      "loss": 1.7385,
+      "step": 698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7163058485249207e-05,
+      "loss": 1.9118,
+      "step": 699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.715367981323026e-05,
+      "loss": 1.7521,
+      "step": 700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.71442882358926e-05,
+      "loss": 2.3891,
+      "step": 701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.713488377017874e-05,
+      "loss": 2.0714,
+      "step": 702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7125466433054438e-05,
+      "loss": 2.0739,
+      "step": 703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.711603624150868e-05,
+      "loss": 1.7363,
+      "step": 704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7106593212553635e-05,
+      "loss": 1.7815,
+      "step": 705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7097137363224635e-05,
+      "loss": 1.7189,
+      "step": 706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7087668710580136e-05,
+      "loss": 1.9974,
+      "step": 707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7078187271701698e-05,
+      "loss": 1.7478,
+      "step": 708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.706869306369394e-05,
+      "loss": 1.865,
+      "step": 709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7059186103684516e-05,
+      "loss": 2.0376,
+      "step": 710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.70496664088241e-05,
+      "loss": 1.9609,
+      "step": 711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7040133996286316e-05,
+      "loss": 1.7922,
+      "step": 712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.703058888326775e-05,
+      "loss": 1.9263,
+      "step": 713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7021031086987888e-05,
+      "loss": 1.9915,
+      "step": 714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.701146062468911e-05,
+      "loss": 1.7153,
+      "step": 715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7001877513636633e-05,
+      "loss": 1.8835,
+      "step": 716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6992281771118495e-05,
+      "loss": 1.8981,
+      "step": 717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.698267341444553e-05,
+      "loss": 2.204,
+      "step": 718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6973052460951305e-05,
+      "loss": 1.8763,
+      "step": 719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.696341892799215e-05,
+      "loss": 1.9203,
+      "step": 720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6953772832947054e-05,
+      "loss": 1.814,
+      "step": 721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6944114193217683e-05,
+      "loss": 1.7761,
+      "step": 722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6934443026228327e-05,
+      "loss": 1.9316,
+      "step": 723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6924759349425887e-05,
+      "loss": 1.7715,
+      "step": 724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.691506318027982e-05,
+      "loss": 2.2877,
+      "step": 725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6905354536282132e-05,
+      "loss": 2.1064,
+      "step": 726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6895633434947317e-05,
+      "loss": 1.737,
+      "step": 727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6885899893812355e-05,
+      "loss": 2.186,
+      "step": 728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6876153930436657e-05,
+      "loss": 1.8405,
+      "step": 729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6866395562402062e-05,
+      "loss": 1.9053,
+      "step": 730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6856624807312763e-05,
+      "loss": 1.6767,
+      "step": 731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6846841682795324e-05,
+      "loss": 2.0335,
+      "step": 732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6837046206498607e-05,
+      "loss": 1.8965,
+      "step": 733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6827238396093753e-05,
+      "loss": 1.8379,
+      "step": 734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.681741826927417e-05,
+      "loss": 2.1349,
+      "step": 735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.680758584375548e-05,
+      "loss": 2.2008,
+      "step": 736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6797741137275485e-05,
+      "loss": 2.0419,
+      "step": 737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.678788416759414e-05,
+      "loss": 1.9333,
+      "step": 738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6778014952493543e-05,
+      "loss": 1.8504,
+      "step": 739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6768133509777855e-05,
+      "loss": 1.956,
+      "step": 740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6758239857273316e-05,
+      "loss": 1.8159,
+      "step": 741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.674833401282819e-05,
+      "loss": 2.0801,
+      "step": 742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6738415994312725e-05,
+      "loss": 1.9113,
+      "step": 743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.672848581961914e-05,
+      "loss": 1.8125,
+      "step": 744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.671854350666158e-05,
+      "loss": 2.0734,
+      "step": 745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.670858907337609e-05,
+      "loss": 2.0964,
+      "step": 746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6698622537720584e-05,
+      "loss": 1.8896,
+      "step": 747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.66886439176748e-05,
+      "loss": 1.9396,
+      "step": 748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.667865323124028e-05,
+      "loss": 2.1868,
+      "step": 749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.666865049644034e-05,
+      "loss": 1.7182,
+      "step": 750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6658635731320016e-05,
+      "loss": 1.8557,
+      "step": 751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6648608953946066e-05,
+      "loss": 1.9393,
+      "step": 752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.66385701824069e-05,
+      "loss": 1.9173,
+      "step": 753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662851943481259e-05,
+      "loss": 2.0333,
+      "step": 754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661845672929478e-05,
+      "loss": 2.067,
+      "step": 755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.660838208400671e-05,
+      "loss": 2.0918,
+      "step": 756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6598295517123158e-05,
+      "loss": 2.0176,
+      "step": 757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.65881970468404e-05,
+      "loss": 2.0665,
+      "step": 758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.657808669137619e-05,
+      "loss": 1.9928,
+      "step": 759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6567964468969725e-05,
+      "loss": 2.0065,
+      "step": 760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6557830397881602e-05,
+      "loss": 1.9093,
+      "step": 761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6547684496393803e-05,
+      "loss": 2.2087,
+      "step": 762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6537526782809652e-05,
+      "loss": 1.9131,
+      "step": 763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.652735727545377e-05,
+      "loss": 1.9874,
+      "step": 764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.651717599267207e-05,
+      "loss": 2.3694,
+      "step": 765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.65069829528317e-05,
+      "loss": 1.8433,
+      "step": 766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6496778174321008e-05,
+      "loss": 1.9645,
+      "step": 767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.648656167554954e-05,
+      "loss": 1.9558,
+      "step": 768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6476333474947976e-05,
+      "loss": 1.7886,
+      "step": 769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6466093590968096e-05,
+      "loss": 1.7925,
+      "step": 770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.645584204208277e-05,
+      "loss": 1.7644,
+      "step": 771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6445578846785905e-05,
+      "loss": 1.8215,
+      "step": 772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6435304023592425e-05,
+      "loss": 1.7299,
+      "step": 773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6425017591038224e-05,
+      "loss": 2.0512,
+      "step": 774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.641471956768014e-05,
+      "loss": 1.8984,
+      "step": 775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.640440997209593e-05,
+      "loss": 1.9813,
+      "step": 776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6394088822884207e-05,
+      "loss": 1.6076,
+      "step": 777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.638375613866445e-05,
+      "loss": 2.2301,
+      "step": 778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.637341193807694e-05,
+      "loss": 2.0948,
+      "step": 779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.636305623978272e-05,
+      "loss": 1.8837,
+      "step": 780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6352689062463594e-05,
+      "loss": 1.9885,
+      "step": 781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6342310424822065e-05,
+      "loss": 1.9429,
+      "step": 782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6331920345581314e-05,
+      "loss": 1.9026,
+      "step": 783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6321518843485155e-05,
+      "loss": 2.0673,
+      "step": 784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.631110593729802e-05,
+      "loss": 2.1918,
+      "step": 785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6300681645804905e-05,
+      "loss": 2.1262,
+      "step": 786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.629024598781135e-05,
+      "loss": 1.9325,
+      "step": 787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6279798982143397e-05,
+      "loss": 1.9277,
+      "step": 788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6269340647647568e-05,
+      "loss": 1.7921,
+      "step": 789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.625887100319081e-05,
+      "loss": 1.9095,
+      "step": 790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.624839006766048e-05,
+      "loss": 1.8493,
+      "step": 791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6237897859964305e-05,
+      "loss": 1.8622,
+      "step": 792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6227394399030347e-05,
+      "loss": 1.8892,
+      "step": 793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6216879703806963e-05,
+      "loss": 1.7851,
+      "step": 794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6206353793262787e-05,
+      "loss": 1.965,
+      "step": 795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.619581668638667e-05,
+      "loss": 2.2624,
+      "step": 796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.618526840218768e-05,
+      "loss": 1.9518,
+      "step": 797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.617470895969504e-05,
+      "loss": 2.0194,
+      "step": 798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.61641383779581e-05,
+      "loss": 1.9845,
+      "step": 799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6153556676046306e-05,
+      "loss": 1.7868,
+      "step": 800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6142963873049174e-05,
+      "loss": 2.1953,
+      "step": 801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613235998807623e-05,
+      "loss": 1.7898,
+      "step": 802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612174504025701e-05,
+      "loss": 2.0114,
+      "step": 803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6111119048740994e-05,
+      "loss": 2.0065,
+      "step": 804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6100482032697596e-05,
+      "loss": 1.9887,
+      "step": 805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6089834011316105e-05,
+      "loss": 1.7774,
+      "step": 806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6079175003805673e-05,
+      "loss": 2.091,
+      "step": 807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6068505029395276e-05,
+      "loss": 1.7725,
+      "step": 808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6057824107333657e-05,
+      "loss": 1.9495,
+      "step": 809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604713225688933e-05,
+      "loss": 2.1128,
+      "step": 810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60364294973505e-05,
+      "loss": 1.7646,
+      "step": 811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602571584802508e-05,
+      "loss": 1.8575,
+      "step": 812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6014991328240605e-05,
+      "loss": 1.5529,
+      "step": 813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.600425595734423e-05,
+      "loss": 1.976,
+      "step": 814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5993509754702687e-05,
+      "loss": 2.0203,
+      "step": 815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5982752739702246e-05,
+      "loss": 1.495,
+      "step": 816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5971984931748678e-05,
+      "loss": 1.8719,
+      "step": 817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5961206350267235e-05,
+      "loss": 2.0461,
+      "step": 818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.59504170147026e-05,
+      "loss": 2.0654,
+      "step": 819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5939616944518853e-05,
+      "loss": 1.7793,
+      "step": 820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5928806159199438e-05,
+      "loss": 1.7812,
+      "step": 821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5917984678247137e-05,
+      "loss": 1.7138,
+      "step": 822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5907152521184025e-05,
+      "loss": 1.5119,
+      "step": 823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5896309707551432e-05,
+      "loss": 2.0122,
+      "step": 824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.588545625690992e-05,
+      "loss": 1.8118,
+      "step": 825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5874592188839226e-05,
+      "loss": 1.8077,
+      "step": 826
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5863717522938262e-05,
+      "loss": 1.7638,
+      "step": 827
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5852832278825038e-05,
+      "loss": 1.7655,
+      "step": 828
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5841936476136668e-05,
+      "loss": 1.7924,
+      "step": 829
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.583103013452929e-05,
+      "loss": 1.8736,
+      "step": 830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5820113273678073e-05,
+      "loss": 2.0161,
+      "step": 831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5809185913277158e-05,
+      "loss": 1.7046,
+      "step": 832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5798248073039623e-05,
+      "loss": 1.8043,
+      "step": 833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5787299772697452e-05,
+      "loss": 1.7387,
+      "step": 834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.577634103200151e-05,
+      "loss": 1.6105,
+      "step": 835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.576537187072148e-05,
+      "loss": 1.574,
+      "step": 836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5754392308645857e-05,
+      "loss": 1.7703,
+      "step": 837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.574340236558189e-05,
+      "loss": 1.7539,
+      "step": 838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5732402061355564e-05,
+      "loss": 1.6659,
+      "step": 839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5721391415811554e-05,
+      "loss": 1.4657,
+      "step": 840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.571037044881318e-05,
+      "loss": 1.9124,
+      "step": 841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5699339180242395e-05,
+      "loss": 1.9223,
+      "step": 842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.568829762999973e-05,
+      "loss": 2.001,
+      "step": 843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5677245818004267e-05,
+      "loss": 1.8484,
+      "step": 844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5666183764193595e-05,
+      "loss": 1.7617,
+      "step": 845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.565511148852379e-05,
+      "loss": 1.642,
+      "step": 846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.564402901096936e-05,
+      "loss": 1.8697,
+      "step": 847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5632936351523212e-05,
+      "loss": 1.738,
+      "step": 848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5621833530196635e-05,
+      "loss": 1.8316,
+      "step": 849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.561072056701924e-05,
+      "loss": 1.8437,
+      "step": 850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5599597482038943e-05,
+      "loss": 1.7922,
+      "step": 851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.558846429532191e-05,
+      "loss": 1.8237,
+      "step": 852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.557732102695253e-05,
+      "loss": 1.9518,
+      "step": 853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5566167697033394e-05,
+      "loss": 1.5705,
+      "step": 854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5555004325685226e-05,
+      "loss": 1.7483,
+      "step": 855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.554383093304688e-05,
+      "loss": 1.9097,
+      "step": 856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5532647539275275e-05,
+      "loss": 1.6185,
+      "step": 857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5521454164545387e-05,
+      "loss": 1.7133,
+      "step": 858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5510250829050183e-05,
+      "loss": 1.9302,
+      "step": 859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5499037553000608e-05,
+      "loss": 1.6367,
+      "step": 860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5487814356625542e-05,
+      "loss": 1.7239,
+      "step": 861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.547658126017175e-05,
+      "loss": 1.7912,
+      "step": 862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5465338283903868e-05,
+      "loss": 1.6617,
+      "step": 863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5454085448104353e-05,
+      "loss": 1.722,
+      "step": 864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5442822773073443e-05,
+      "loss": 1.8698,
+      "step": 865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5431550279129133e-05,
+      "loss": 1.6635,
+      "step": 866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5420267986607133e-05,
+      "loss": 1.8094,
+      "step": 867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5408975915860814e-05,
+      "loss": 2.0887,
+      "step": 868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5397674087261204e-05,
+      "loss": 1.6381,
+      "step": 869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5386362521196935e-05,
+      "loss": 1.6735,
+      "step": 870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.537504123807419e-05,
+      "loss": 1.8098,
+      "step": 871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.53637102583167e-05,
+      "loss": 1.81,
+      "step": 872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5352369602365672e-05,
+      "loss": 1.7992,
+      "step": 873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5341019290679776e-05,
+      "loss": 1.9782,
+      "step": 874
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.532965934373511e-05,
+      "loss": 1.9091,
+      "step": 875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.531828978202514e-05,
+      "loss": 1.7481,
+      "step": 876
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.530691062606068e-05,
+      "loss": 1.5683,
+      "step": 877
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.529552189636986e-05,
+      "loss": 1.7813,
+      "step": 878
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5284123613498077e-05,
+      "loss": 1.8491,
+      "step": 879
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.527271579800796e-05,
+      "loss": 1.9376,
+      "step": 880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5261298470479335e-05,
+      "loss": 1.8666,
+      "step": 881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5249871651509186e-05,
+      "loss": 1.531,
+      "step": 882
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5238435361711629e-05,
+      "loss": 1.8579,
+      "step": 883
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.522698962171786e-05,
+      "loss": 1.9701,
+      "step": 884
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5215534452176118e-05,
+      "loss": 1.7714,
+      "step": 885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5204069873751654e-05,
+      "loss": 1.5738,
+      "step": 886
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5192595907126703e-05,
+      "loss": 2.1066,
+      "step": 887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5181112573000429e-05,
+      "loss": 1.7597,
+      "step": 888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5169619892088896e-05,
+      "loss": 1.7255,
+      "step": 889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5158117885125025e-05,
+      "loss": 1.5984,
+      "step": 890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5146606572858568e-05,
+      "loss": 1.6127,
+      "step": 891
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5135085976056063e-05,
+      "loss": 1.775,
+      "step": 892
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5123556115500794e-05,
+      "loss": 1.7851,
+      "step": 893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5112017011992764e-05,
+      "loss": 1.7638,
+      "step": 894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5100468686348637e-05,
+      "loss": 1.984,
+      "step": 895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5088911159401727e-05,
+      "loss": 1.7119,
+      "step": 896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5077344452001943e-05,
+      "loss": 1.8547,
+      "step": 897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5065768585015753e-05,
+      "loss": 1.8017,
+      "step": 898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5054183579326149e-05,
+      "loss": 1.8689,
+      "step": 899
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5042589455832617e-05,
+      "loss": 1.6593,
+      "step": 900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5030986235451084e-05,
+      "loss": 1.911,
+      "step": 901
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5019373939113889e-05,
+      "loss": 1.7512,
+      "step": 902
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5007752587769741e-05,
+      "loss": 2.0166,
+      "step": 903
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4996122202383693e-05,
+      "loss": 1.6801,
+      "step": 904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4984482803937094e-05,
+      "loss": 1.8145,
+      "step": 905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4972834413427544e-05,
+      "loss": 1.5703,
+      "step": 906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4961177051868871e-05,
+      "loss": 1.5529,
+      "step": 907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4949510740291087e-05,
+      "loss": 1.5896,
+      "step": 908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4937835499740349e-05,
+      "loss": 1.7022,
+      "step": 909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4926151351278924e-05,
+      "loss": 1.6291,
+      "step": 910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4914458315985142e-05,
+      "loss": 1.7863,
+      "step": 911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4902756414953376e-05,
+      "loss": 1.931,
+      "step": 912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4891045669293985e-05,
+      "loss": 1.9136,
+      "step": 913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4879326100133282e-05,
+      "loss": 1.6361,
+      "step": 914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4867597728613503e-05,
+      "loss": 1.8148,
+      "step": 915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4855860575892766e-05,
+      "loss": 2.0616,
+      "step": 916
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4844114663145024e-05,
+      "loss": 1.491,
+      "step": 917
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4832360011560032e-05,
+      "loss": 1.6758,
+      "step": 918
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4820596642343316e-05,
+      "loss": 1.6791,
+      "step": 919
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4808824576716128e-05,
+      "loss": 1.5006,
+      "step": 920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4797043835915406e-05,
+      "loss": 1.6676,
+      "step": 921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4785254441193738e-05,
+      "loss": 1.7311,
+      "step": 922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.477345641381932e-05,
+      "loss": 1.5569,
+      "step": 923
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4761649775075933e-05,
+      "loss": 1.9139,
+      "step": 924
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4749834546262883e-05,
+      "loss": 1.4766,
+      "step": 925
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4738010748694973e-05,
+      "loss": 1.8229,
+      "step": 926
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4726178403702472e-05,
+      "loss": 1.9456,
+      "step": 927
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4714337532631058e-05,
+      "loss": 1.5488,
+      "step": 928
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4702488156841797e-05,
+      "loss": 1.5405,
+      "step": 929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4690630297711097e-05,
+      "loss": 1.7235,
+      "step": 930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.467876397663067e-05,
+      "loss": 1.7998,
+      "step": 931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.466688921500749e-05,
+      "loss": 1.7048,
+      "step": 932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4655006034263761e-05,
+      "loss": 1.8505,
+      "step": 933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4643114455836878e-05,
+      "loss": 1.8884,
+      "step": 934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4631214501179374e-05,
+      "loss": 1.5529,
+      "step": 935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4619306191758909e-05,
+      "loss": 2.0871,
+      "step": 936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4607389549058202e-05,
+      "loss": 1.6605,
+      "step": 937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.459546459457501e-05,
+      "loss": 1.7873,
+      "step": 938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4583531349822092e-05,
+      "loss": 1.5173,
+      "step": 939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4571589836327145e-05,
+      "loss": 2.0764,
+      "step": 940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.45596400756328e-05,
+      "loss": 1.7494,
+      "step": 941
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4547682089296558e-05,
+      "loss": 1.7465,
+      "step": 942
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4535715898890764e-05,
+      "loss": 2.0958,
+      "step": 943
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4523741526002551e-05,
+      "loss": 1.9423,
+      "step": 944
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4511758992233831e-05,
+      "loss": 1.8285,
+      "step": 945
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4499768319201226e-05,
+      "loss": 1.9261,
+      "step": 946
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4487769528536042e-05,
+      "loss": 1.5187,
+      "step": 947
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.447576264188424e-05,
+      "loss": 1.7609,
+      "step": 948
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.446374768090637e-05,
+      "loss": 1.8391,
+      "step": 949
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.445172466727756e-05,
+      "loss": 1.7483,
+      "step": 950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4439693622687463e-05,
+      "loss": 1.7084,
+      "step": 951
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4427654568840215e-05,
+      "loss": 1.6872,
+      "step": 952
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4415607527454406e-05,
+      "loss": 1.7083,
+      "step": 953
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.440355252026304e-05,
+      "loss": 1.6865,
+      "step": 954
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4391489569013475e-05,
+      "loss": 1.8598,
+      "step": 955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4379418695467415e-05,
+      "loss": 1.7496,
+      "step": 956
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4367339921400851e-05,
+      "loss": 1.6848,
+      "step": 957
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.435525326860403e-05,
+      "loss": 1.8336,
+      "step": 958
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4343158758881409e-05,
+      "loss": 1.6063,
+      "step": 959
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4331056414051614e-05,
+      "loss": 1.7836,
+      "step": 960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4318946255947412e-05,
+      "loss": 2.0158,
+      "step": 961
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.430682830641567e-05,
+      "loss": 1.5559,
+      "step": 962
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4294702587317302e-05,
+      "loss": 1.7787,
+      "step": 963
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4282569120527238e-05,
+      "loss": 1.7592,
+      "step": 964
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4270427927934396e-05,
+      "loss": 1.863,
+      "step": 965
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4258279031441615e-05,
+      "loss": 1.9742,
+      "step": 966
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4246122452965645e-05,
+      "loss": 1.7259,
+      "step": 967
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4233958214437088e-05,
+      "loss": 1.8017,
+      "step": 968
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4221786337800375e-05,
+      "loss": 1.9566,
+      "step": 969
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4209606845013699e-05,
+      "loss": 1.6463,
+      "step": 970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4197419758049003e-05,
+      "loss": 1.9166,
+      "step": 971
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4185225098891931e-05,
+      "loss": 1.8433,
+      "step": 972
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4173022889541785e-05,
+      "loss": 1.7534,
+      "step": 973
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4160813152011486e-05,
+      "loss": 1.6399,
+      "step": 974
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4148595908327543e-05,
+      "loss": 1.7516,
+      "step": 975
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.413637118052999e-05,
+      "loss": 1.8831,
+      "step": 976
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4124138990672384e-05,
+      "loss": 1.7359,
+      "step": 977
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4111899360821726e-05,
+      "loss": 1.7846,
+      "step": 978
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.409965231305845e-05,
+      "loss": 1.5996,
+      "step": 979
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4087397869476364e-05,
+      "loss": 1.6926,
+      "step": 980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.407513605218263e-05,
+      "loss": 1.7231,
+      "step": 981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4062866883297693e-05,
+      "loss": 1.6368,
+      "step": 982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4050590384955273e-05,
+      "loss": 1.803,
+      "step": 983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4038306579302318e-05,
+      "loss": 1.7042,
+      "step": 984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4026015488498948e-05,
+      "loss": 1.8718,
+      "step": 985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.401371713471843e-05,
+      "loss": 1.5429,
+      "step": 986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4001411540147128e-05,
+      "loss": 1.6121,
+      "step": 987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3989098726984476e-05,
+      "loss": 1.8772,
+      "step": 988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3976778717442925e-05,
+      "loss": 1.6582,
+      "step": 989
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3964451533747914e-05,
+      "loss": 1.963,
+      "step": 990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.395211719813782e-05,
+      "loss": 1.8725,
+      "step": 991
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.393977573286392e-05,
+      "loss": 1.7127,
+      "step": 992
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3927427160190358e-05,
+      "loss": 1.4399,
+      "step": 993
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3915071502394097e-05,
+      "loss": 1.6566,
+      "step": 994
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3902708781764885e-05,
+      "loss": 1.8646,
+      "step": 995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3890339020605206e-05,
+      "loss": 1.5859,
+      "step": 996
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.387796224123025e-05,
+      "loss": 1.6833,
+      "step": 997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3865578465967864e-05,
+      "loss": 1.9577,
+      "step": 998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3853187717158518e-05,
+      "loss": 1.7017,
+      "step": 999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3840790017155264e-05,
+      "loss": 1.5799,
+      "step": 1000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3828385388323688e-05,
+      "loss": 1.6971,
+      "step": 1001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3815973853041886e-05,
+      "loss": 1.7544,
+      "step": 1002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3803555433700401e-05,
+      "loss": 1.8026,
+      "step": 1003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3791130152702203e-05,
+      "loss": 1.8926,
+      "step": 1004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3778698032462638e-05,
+      "loss": 1.8523,
+      "step": 1005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3766259095409393e-05,
+      "loss": 1.5979,
+      "step": 1006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3753813363982447e-05,
+      "loss": 1.5312,
+      "step": 1007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3741360860634042e-05,
+      "loss": 1.8463,
+      "step": 1008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3728901607828632e-05,
+      "loss": 1.7193,
+      "step": 1009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3716435628042849e-05,
+      "loss": 1.5545,
+      "step": 1010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.370396294376546e-05,
+      "loss": 1.7971,
+      "step": 1011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3691483577497329e-05,
+      "loss": 1.6151,
+      "step": 1012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3678997551751374e-05,
+      "loss": 1.8212,
+      "step": 1013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3666504889052523e-05,
+      "loss": 1.9736,
+      "step": 1014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.365400561193768e-05,
+      "loss": 1.4222,
+      "step": 1015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3641499742955682e-05,
+      "loss": 1.4836,
+      "step": 1016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3628987304667255e-05,
+      "loss": 1.8388,
+      "step": 1017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3616468319644983e-05,
+      "loss": 1.7007,
+      "step": 1018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3603942810473254e-05,
+      "loss": 1.8641,
+      "step": 1019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3591410799748223e-05,
+      "loss": 1.5868,
+      "step": 1020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3578872310077782e-05,
+      "loss": 1.8256,
+      "step": 1021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3566327364081506e-05,
+      "loss": 1.5545,
+      "step": 1022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3553775984390618e-05,
+      "loss": 1.6633,
+      "step": 1023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3541218193647953e-05,
+      "loss": 1.7258,
+      "step": 1024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3528654014507898e-05,
+      "loss": 1.5288,
+      "step": 1025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3516083469636377e-05,
+      "loss": 1.549,
+      "step": 1026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3503506581710797e-05,
+      "loss": 1.6946,
+      "step": 1027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.349092337342e-05,
+      "loss": 1.6817,
+      "step": 1028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3478333867464239e-05,
+      "loss": 1.6187,
+      "step": 1029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3465738086555122e-05,
+      "loss": 1.4273,
+      "step": 1030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3453136053415578e-05,
+      "loss": 1.8014,
+      "step": 1031
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3440527790779822e-05,
+      "loss": 1.9505,
+      "step": 1032
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3427913321393295e-05,
+      "loss": 1.664,
+      "step": 1033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3415292668012642e-05,
+      "loss": 1.7148,
+      "step": 1034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3402665853405671e-05,
+      "loss": 1.4967,
+      "step": 1035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.339003290035129e-05,
+      "loss": 1.7897,
+      "step": 1036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3377393831639486e-05,
+      "loss": 1.9791,
+      "step": 1037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3364748670071289e-05,
+      "loss": 1.9908,
+      "step": 1038
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3352097438458705e-05,
+      "loss": 1.7829,
+      "step": 1039
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3339440159624703e-05,
+      "loss": 1.634,
+      "step": 1040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3326776856403155e-05,
+      "loss": 1.5434,
+      "step": 1041
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.33141075516388e-05,
+      "loss": 1.8311,
+      "step": 1042
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3301432268187206e-05,
+      "loss": 1.5993,
+      "step": 1043
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.328875102891473e-05,
+      "loss": 1.6892,
+      "step": 1044
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3276063856698466e-05,
+      "loss": 1.6102,
+      "step": 1045
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3263370774426224e-05,
+      "loss": 1.6739,
+      "step": 1046
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3250671804996453e-05,
+      "loss": 1.8062,
+      "step": 1047
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3237966971318247e-05,
+      "loss": 1.9935,
+      "step": 1048
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3225256296311264e-05,
+      "loss": 1.6113,
+      "step": 1049
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3212539802905706e-05,
+      "loss": 1.7978,
+      "step": 1050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3199817514042273e-05,
+      "loss": 1.81,
+      "step": 1051
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3187089452672111e-05,
+      "loss": 1.9428,
+      "step": 1052
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3174355641756788e-05,
+      "loss": 1.8519,
+      "step": 1053
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3161616104268243e-05,
+      "loss": 2.023,
+      "step": 1054
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3148870863188742e-05,
+      "loss": 1.9122,
+      "step": 1055
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.313611994151085e-05,
+      "loss": 1.7681,
+      "step": 1056
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3123363362237363e-05,
+      "loss": 1.5049,
+      "step": 1057
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.31106011483813e-05,
+      "loss": 1.4662,
+      "step": 1058
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3097833322965833e-05,
+      "loss": 1.7361,
+      "step": 1059
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3085059909024269e-05,
+      "loss": 1.7074,
+      "step": 1060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3072280929599985e-05,
+      "loss": 1.7558,
+      "step": 1061
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3059496407746407e-05,
+      "loss": 1.5883,
+      "step": 1062
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3046706366526953e-05,
+      "loss": 1.9173,
+      "step": 1063
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3033910829015004e-05,
+      "loss": 1.6502,
+      "step": 1064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3021109818293849e-05,
+      "loss": 1.5945,
+      "step": 1065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3008303357456657e-05,
+      "loss": 1.8538,
+      "step": 1066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2995491469606431e-05,
+      "loss": 1.6556,
+      "step": 1067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2982674177855954e-05,
+      "loss": 2.0285,
+      "step": 1068
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2969851505327768e-05,
+      "loss": 1.7749,
+      "step": 1069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.295702347515412e-05,
+      "loss": 1.6655,
+      "step": 1070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2944190110476915e-05,
+      "loss": 1.647,
+      "step": 1071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2931351434447692e-05,
+      "loss": 1.722,
+      "step": 1072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2918507470227567e-05,
+      "loss": 1.6751,
+      "step": 1073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2905658240987192e-05,
+      "loss": 1.7321,
+      "step": 1074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.289280376990672e-05,
+      "loss": 1.6686,
+      "step": 1075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2879944080175765e-05,
+      "loss": 1.8305,
+      "step": 1076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.286707919499335e-05,
+      "loss": 1.9242,
+      "step": 1077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2854209137567874e-05,
+      "loss": 1.7498,
+      "step": 1078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2841333931117063e-05,
+      "loss": 1.7217,
+      "step": 1079
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2828453598867934e-05,
+      "loss": 1.8248,
+      "step": 1080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2815568164056754e-05,
+      "loss": 1.9462,
+      "step": 1081
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2802677649928988e-05,
+      "loss": 1.9232,
+      "step": 1082
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.278978207973927e-05,
+      "loss": 1.6523,
+      "step": 1083
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2776881476751355e-05,
+      "loss": 1.5722,
+      "step": 1084
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2763975864238072e-05,
+      "loss": 1.7854,
+      "step": 1085
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2751065265481293e-05,
+      "loss": 1.8023,
+      "step": 1086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2738149703771885e-05,
+      "loss": 1.6042,
+      "step": 1087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2725229202409662e-05,
+      "loss": 2.093,
+      "step": 1088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2712303784703362e-05,
+      "loss": 1.6153,
+      "step": 1089
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.269937347397057e-05,
+      "loss": 1.2245,
+      "step": 1090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2686438293537725e-05,
+      "loss": 1.872,
+      "step": 1091
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2673498266740029e-05,
+      "loss": 1.7853,
+      "step": 1092
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.266055341692144e-05,
+      "loss": 1.9354,
+      "step": 1093
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2647603767434612e-05,
+      "loss": 1.7227,
+      "step": 1094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2634649341640856e-05,
+      "loss": 1.5924,
+      "step": 1095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2621690162910105e-05,
+      "loss": 1.9701,
+      "step": 1096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.260872625462086e-05,
+      "loss": 1.9044,
+      "step": 1097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2595757640160157e-05,
+      "loss": 1.4248,
+      "step": 1098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2582784342923524e-05,
+      "loss": 1.8282,
+      "step": 1099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2569806386314935e-05,
+      "loss": 1.7229,
+      "step": 1100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2556823793746768e-05,
+      "loss": 1.9624,
+      "step": 1101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2543836588639767e-05,
+      "loss": 1.7878,
+      "step": 1102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2530844794422993e-05,
+      "loss": 1.6919,
+      "step": 1103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2517848434533793e-05,
+      "loss": 1.8428,
+      "step": 1104
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2504847532417746e-05,
+      "loss": 1.8294,
+      "step": 1105
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2491842111528618e-05,
+      "loss": 1.5887,
+      "step": 1106
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2478832195328343e-05,
+      "loss": 1.7019,
+      "step": 1107
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2465817807286951e-05,
+      "loss": 1.453,
+      "step": 1108
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2452798970882547e-05,
+      "loss": 1.8649,
+      "step": 1109
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2439775709601258e-05,
+      "loss": 1.6922,
+      "step": 1110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2426748046937191e-05,
+      "loss": 1.6768,
+      "step": 1111
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2413716006392398e-05,
+      "loss": 1.8377,
+      "step": 1112
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2400679611476824e-05,
+      "loss": 1.4469,
+      "step": 1113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2387638885708274e-05,
+      "loss": 1.6142,
+      "step": 1114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2374593852612363e-05,
+      "loss": 1.981,
+      "step": 1115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2361544535722481e-05,
+      "loss": 1.5887,
+      "step": 1116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2348490958579736e-05,
+      "loss": 1.7549,
+      "step": 1117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2335433144732929e-05,
+      "loss": 1.9017,
+      "step": 1118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2322371117738504e-05,
+      "loss": 2.1026,
+      "step": 1119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2309304901160506e-05,
+      "loss": 1.9094,
+      "step": 1120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2296234518570534e-05,
+      "loss": 1.6542,
+      "step": 1121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2283159993547708e-05,
+      "loss": 1.8173,
+      "step": 1122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2270081349678615e-05,
+      "loss": 1.6728,
+      "step": 1123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2256998610557275e-05,
+      "loss": 1.7757,
+      "step": 1124
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.22439117997851e-05,
+      "loss": 1.9248,
+      "step": 1125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2230820940970839e-05,
+      "loss": 1.3587,
+      "step": 1126
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2217726057730553e-05,
+      "loss": 2.0036,
+      "step": 1127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2204627173687554e-05,
+      "loss": 1.4565,
+      "step": 1128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2191524312472376e-05,
+      "loss": 1.6243,
+      "step": 1129
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.217841749772273e-05,
+      "loss": 1.695,
+      "step": 1130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.216530675308346e-05,
+      "loss": 1.5623,
+      "step": 1131
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2152192102206489e-05,
+      "loss": 1.7931,
+      "step": 1132
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2139073568750801e-05,
+      "loss": 1.6992,
+      "step": 1133
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2125951176382373e-05,
+      "loss": 1.9345,
+      "step": 1134
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2112824948774153e-05,
+      "loss": 1.8393,
+      "step": 1135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2099694909606e-05,
+      "loss": 1.8088,
+      "step": 1136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2086561082564649e-05,
+      "loss": 1.542,
+      "step": 1137
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2073423491343679e-05,
+      "loss": 1.699,
+      "step": 1138
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2060282159643448e-05,
+      "loss": 1.7892,
+      "step": 1139
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2047137111171063e-05,
+      "loss": 1.8486,
+      "step": 1140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2033988369640344e-05,
+      "loss": 1.5031,
+      "step": 1141
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2020835958771766e-05,
+      "loss": 1.5387,
+      "step": 1142
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2007679902292424e-05,
+      "loss": 1.7015,
+      "step": 1143
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1994520223935997e-05,
+      "loss": 1.6494,
+      "step": 1144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.198135694744269e-05,
+      "loss": 1.8222,
+      "step": 1145
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1968190096559197e-05,
+      "loss": 1.6857,
+      "step": 1146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1955019695038672e-05,
+      "loss": 1.8216,
+      "step": 1147
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1941845766640663e-05,
+      "loss": 1.7607,
+      "step": 1148
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1928668335131087e-05,
+      "loss": 1.6955,
+      "step": 1149
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1915487424282175e-05,
+      "loss": 1.736,
+      "step": 1150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1902303057872445e-05,
+      "loss": 1.8572,
+      "step": 1151
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1889115259686634e-05,
+      "loss": 1.6275,
+      "step": 1152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1875924053515682e-05,
+      "loss": 1.7289,
+      "step": 1153
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.186272946315667e-05,
+      "loss": 1.737,
+      "step": 1154
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1849531512412794e-05,
+      "loss": 1.8558,
+      "step": 1155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1836330225093298e-05,
+      "loss": 1.5853,
+      "step": 1156
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1823125625013459e-05,
+      "loss": 1.8138,
+      "step": 1157
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1809917735994519e-05,
+      "loss": 1.6845,
+      "step": 1158
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1796706581863657e-05,
+      "loss": 1.5249,
+      "step": 1159
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1783492186453949e-05,
+      "loss": 1.4558,
+      "step": 1160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1770274573604309e-05,
+      "loss": 1.5507,
+      "step": 1161
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.175705376715946e-05,
+      "loss": 1.6599,
+      "step": 1162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1743829790969885e-05,
+      "loss": 1.8786,
+      "step": 1163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1730602668891784e-05,
+      "loss": 1.7782,
+      "step": 1164
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1717372424787037e-05,
+      "loss": 1.5701,
+      "step": 1165
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1704139082523149e-05,
+      "loss": 1.892,
+      "step": 1166
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1690902665973221e-05,
+      "loss": 1.884,
+      "step": 1167
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1677663199015895e-05,
+      "loss": 1.6552,
+      "step": 1168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1664420705535324e-05,
+      "loss": 1.8881,
+      "step": 1169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1651175209421108e-05,
+      "loss": 1.7133,
+      "step": 1170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1637926734568272e-05,
+      "loss": 1.5572,
+      "step": 1171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1624675304877216e-05,
+      "loss": 1.7387,
+      "step": 1172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1611420944253664e-05,
+      "loss": 1.8226,
+      "step": 1173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1598163676608637e-05,
+      "loss": 1.823,
+      "step": 1174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.158490352585839e-05,
+      "loss": 1.8104,
+      "step": 1175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1571640515924386e-05,
+      "loss": 1.5717,
+      "step": 1176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.155837467073324e-05,
+      "loss": 1.5925,
+      "step": 1177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1545106014216691e-05,
+      "loss": 1.7911,
+      "step": 1178
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1531834570311544e-05,
+      "loss": 1.4883,
+      "step": 1179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1518560362959627e-05,
+      "loss": 1.8797,
+      "step": 1180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.150528341610777e-05,
+      "loss": 1.6152,
+      "step": 1181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1492003753707724e-05,
+      "loss": 1.715,
+      "step": 1182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1478721399716149e-05,
+      "loss": 1.6458,
+      "step": 1183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1465436378094568e-05,
+      "loss": 2.0381,
+      "step": 1184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1452148712809304e-05,
+      "loss": 1.79,
+      "step": 1185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1438858427831456e-05,
+      "loss": 1.5757,
+      "step": 1186
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1425565547136848e-05,
+      "loss": 1.9503,
+      "step": 1187
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1412270094705983e-05,
+      "loss": 1.9518,
+      "step": 1188
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1398972094524012e-05,
+      "loss": 1.9339,
+      "step": 1189
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1385671570580673e-05,
+      "loss": 1.734,
+      "step": 1190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1372368546870258e-05,
+      "loss": 1.6447,
+      "step": 1191
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1359063047391576e-05,
+      "loss": 1.7506,
+      "step": 1192
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1345755096147895e-05,
+      "loss": 1.7517,
+      "step": 1193
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1332444717146908e-05,
+      "loss": 1.6569,
+      "step": 1194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1319131934400689e-05,
+      "loss": 1.8714,
+      "step": 1195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1305816771925651e-05,
+      "loss": 1.845,
+      "step": 1196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.129249925374249e-05,
+      "loss": 1.7908,
+      "step": 1197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1279179403876164e-05,
+      "loss": 1.9938,
+      "step": 1198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1265857246355827e-05,
+      "loss": 1.6952,
+      "step": 1199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1252532805214804e-05,
+      "loss": 1.6272,
+      "step": 1200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1239206104490537e-05,
+      "loss": 1.4211,
+      "step": 1201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1225877168224544e-05,
+      "loss": 1.9437,
+      "step": 1202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1212546020462368e-05,
+      "loss": 1.64,
+      "step": 1203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.119921268525356e-05,
+      "loss": 1.6666,
+      "step": 1204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1185877186651602e-05,
+      "loss": 1.4501,
+      "step": 1205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1172539548713881e-05,
+      "loss": 1.7632,
+      "step": 1206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1159199795501647e-05,
+      "loss": 1.6141,
+      "step": 1207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1145857951079967e-05,
+      "loss": 1.5467,
+      "step": 1208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1132514039517677e-05,
+      "loss": 1.9927,
+      "step": 1209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1119168084887339e-05,
+      "loss": 1.7743,
+      "step": 1210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.110582011126521e-05,
+      "loss": 1.3532,
+      "step": 1211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1092470142731185e-05,
+      "loss": 1.62,
+      "step": 1212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1079118203368755e-05,
+      "loss": 1.7714,
+      "step": 1213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1065764317264969e-05,
+      "loss": 1.5992,
+      "step": 1214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1052408508510383e-05,
+      "loss": 1.5276,
+      "step": 1215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1039050801199035e-05,
+      "loss": 1.7298,
+      "step": 1216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1025691219428371e-05,
+      "loss": 1.9522,
+      "step": 1217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1012329787299232e-05,
+      "loss": 1.956,
+      "step": 1218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0998966528915786e-05,
+      "loss": 1.5012,
+      "step": 1219
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0985601468385508e-05,
+      "loss": 1.4808,
+      "step": 1220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.097223462981911e-05,
+      "loss": 1.4052,
+      "step": 1221
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0958866037330524e-05,
+      "loss": 1.8629,
+      "step": 1222
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.094549571503684e-05,
+      "loss": 1.813,
+      "step": 1223
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0932123687058269e-05,
+      "loss": 1.4428,
+      "step": 1224
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.09187499775181e-05,
+      "loss": 1.7306,
+      "step": 1225
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0905374610542654e-05,
+      "loss": 1.6758,
+      "step": 1226
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0891997610261245e-05,
+      "loss": 1.8463,
+      "step": 1227
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0878619000806128e-05,
+      "loss": 1.8612,
+      "step": 1228
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0865238806312473e-05,
+      "loss": 1.7799,
+      "step": 1229
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.085185705091829e-05,
+      "loss": 1.8219,
+      "step": 1230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0838473758764422e-05,
+      "loss": 1.7008,
+      "step": 1231
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.082508895399448e-05,
+      "loss": 1.7243,
+      "step": 1232
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0811702660754802e-05,
+      "loss": 1.5484,
+      "step": 1233
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0798314903194404e-05,
+      "loss": 1.7378,
+      "step": 1234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0784925705464962e-05,
+      "loss": 1.8455,
+      "step": 1235
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0771535091720731e-05,
+      "loss": 1.8444,
+      "step": 1236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0758143086118528e-05,
+      "loss": 1.7598,
+      "step": 1237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0744749712817688e-05,
+      "loss": 1.5471,
+      "step": 1238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.073135499598e-05,
+      "loss": 1.6465,
+      "step": 1239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0717958959769692e-05,
+      "loss": 1.5503,
+      "step": 1240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0704561628353356e-05,
+      "loss": 1.7158,
+      "step": 1241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0691163025899929e-05,
+      "loss": 1.8676,
+      "step": 1242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0677763176580645e-05,
+      "loss": 1.7463,
+      "step": 1243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0664362104568977e-05,
+      "loss": 1.6699,
+      "step": 1244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0650959834040615e-05,
+      "loss": 1.6827,
+      "step": 1245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0637556389173402e-05,
+      "loss": 1.5257,
+      "step": 1246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0624151794147302e-05,
+      "loss": 1.7982,
+      "step": 1247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0610746073144359e-05,
+      "loss": 1.5078,
+      "step": 1248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0597339250348639e-05,
+      "loss": 1.6466,
+      "step": 1249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0583931349946206e-05,
+      "loss": 1.3655,
+      "step": 1250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.057052239612506e-05,
+      "loss": 1.8426,
+      "step": 1251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0557112413075106e-05,
+      "loss": 1.6883,
+      "step": 1252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0543701424988102e-05,
+      "loss": 1.6952,
+      "step": 1253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.053028945605762e-05,
+      "loss": 1.6964,
+      "step": 1254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0516876530479008e-05,
+      "loss": 1.8324,
+      "step": 1255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0503462672449331e-05,
+      "loss": 1.837,
+      "step": 1256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0490047906167338e-05,
+      "loss": 1.8782,
+      "step": 1257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0476632255833422e-05,
+      "loss": 1.7143,
+      "step": 1258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0463215745649559e-05,
+      "loss": 1.7256,
+      "step": 1259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0449798399819293e-05,
+      "loss": 1.9712,
+      "step": 1260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0436380242547662e-05,
+      "loss": 1.6597,
+      "step": 1261
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0422961298041177e-05,
+      "loss": 1.8649,
+      "step": 1262
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0409541590507755e-05,
+      "loss": 1.7863,
+      "step": 1263
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0396121144156708e-05,
+      "loss": 1.4775,
+      "step": 1264
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0382699983198668e-05,
+      "loss": 1.5864,
+      "step": 1265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0369278131845561e-05,
+      "loss": 1.8892,
+      "step": 1266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0355855614310559e-05,
+      "loss": 1.4543,
+      "step": 1267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0342432454808031e-05,
+      "loss": 1.6199,
+      "step": 1268
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0329008677553509e-05,
+      "loss": 1.8391,
+      "step": 1269
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.031558430676364e-05,
+      "loss": 1.665,
+      "step": 1270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0302159366656137e-05,
+      "loss": 1.7104,
+      "step": 1271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0288733881449746e-05,
+      "loss": 1.9364,
+      "step": 1272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0275307875364187e-05,
+      "loss": 1.837,
+      "step": 1273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0261881372620131e-05,
+      "loss": 1.8914,
+      "step": 1274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0248454397439139e-05,
+      "loss": 1.9814,
+      "step": 1275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.023502697404362e-05,
+      "loss": 1.4755,
+      "step": 1276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0221599126656806e-05,
+      "loss": 1.7376,
+      "step": 1277
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0208170879502677e-05,
+      "loss": 1.7352,
+      "step": 1278
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0194742256805943e-05,
+      "loss": 1.7646,
+      "step": 1279
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0181313282791991e-05,
+      "loss": 1.4554,
+      "step": 1280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0167883981686842e-05,
+      "loss": 1.6564,
+      "step": 1281
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0154454377717102e-05,
+      "loss": 1.9236,
+      "step": 1282
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0141024495109935e-05,
+      "loss": 1.7105,
+      "step": 1283
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0127594358092994e-05,
+      "loss": 1.6853,
+      "step": 1284
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0114163990894396e-05,
+      "loss": 1.7031,
+      "step": 1285
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0100733417742677e-05,
+      "loss": 1.9128,
+      "step": 1286
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.008730266286674e-05,
+      "loss": 1.4167,
+      "step": 1287
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.007387175049582e-05,
+      "loss": 1.7772,
+      "step": 1288
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0060440704859435e-05,
+      "loss": 1.812,
+      "step": 1289
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0047009550187339e-05,
+      "loss": 1.5706,
+      "step": 1290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0033578310709484e-05,
+      "loss": 1.4725,
+      "step": 1291
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.002014701065598e-05,
+      "loss": 1.9494,
+      "step": 1292
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0006715674257043e-05,
+      "loss": 1.6827,
+      "step": 1293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.993284325742955e-06,
+      "loss": 1.6718,
+      "step": 1294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.979852989344021e-06,
+      "loss": 1.7175,
+      "step": 1295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.966421689290516e-06,
+      "loss": 1.9115,
+      "step": 1296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.952990449812664e-06,
+      "loss": 1.9041,
+      "step": 1297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.939559295140568e-06,
+      "loss": 1.6117,
+      "step": 1298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.92612824950418e-06,
+      "loss": 1.5911,
+      "step": 1299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.912697337133261e-06,
+      "loss": 1.6191,
+      "step": 1300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.899266582257325e-06,
+      "loss": 1.6391,
+      "step": 1301
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.885836009105607e-06,
+      "loss": 1.898,
+      "step": 1302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.872405641907013e-06,
+      "loss": 1.5127,
+      "step": 1303
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.858975504890067e-06,
+      "loss": 1.9172,
+      "step": 1304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.8455456222829e-06,
+      "loss": 1.4616,
+      "step": 1305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.83211601831316e-06,
+      "loss": 1.641,
+      "step": 1306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.81868671720801e-06,
+      "loss": 1.7476,
+      "step": 1307
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.805257743194062e-06,
+      "loss": 1.5965,
+      "step": 1308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.791829120497326e-06,
+      "loss": 1.3396,
+      "step": 1309
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.7784008733432e-06,
+      "loss": 1.7196,
+      "step": 1310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.76497302595638e-06,
+      "loss": 1.5882,
+      "step": 1311
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.751545602560866e-06,
+      "loss": 1.6935,
+      "step": 1312
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.738118627379874e-06,
+      "loss": 1.6973,
+      "step": 1313
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.724692124635814e-06,
+      "loss": 1.6877,
+      "step": 1314
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.71126611855026e-06,
+      "loss": 1.9789,
+      "step": 1315
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.697840633343863e-06,
+      "loss": 1.6287,
+      "step": 1316
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.684415693236363e-06,
+      "loss": 1.6557,
+      "step": 1317
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.670991322446495e-06,
+      "loss": 1.8147,
+      "step": 1318
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.65756754519197e-06,
+      "loss": 1.8534,
+      "step": 1319
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.644144385689446e-06,
+      "loss": 1.6912,
+      "step": 1320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.63072186815444e-06,
+      "loss": 1.7225,
+      "step": 1321
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.617300016801336e-06,
+      "loss": 1.6309,
+      "step": 1322
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.603878855843292e-06,
+      "loss": 1.9473,
+      "step": 1323
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.590458409492247e-06,
+      "loss": 1.8752,
+      "step": 1324
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.57703870195883e-06,
+      "loss": 1.6092,
+      "step": 1325
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.563619757452338e-06,
+      "loss": 1.934,
+      "step": 1326
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.550201600180709e-06,
+      "loss": 1.757,
+      "step": 1327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.536784254350441e-06,
+      "loss": 1.6582,
+      "step": 1328
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.523367744166583e-06,
+      "loss": 1.585,
+      "step": 1329
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.509952093832666e-06,
+      "loss": 1.6271,
+      "step": 1330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.49653732755067e-06,
+      "loss": 1.75,
+      "step": 1331
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.483123469520995e-06,
+      "loss": 1.8332,
+      "step": 1332
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.46971054394238e-06,
+      "loss": 1.6848,
+      "step": 1333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.4562985750119e-06,
+      "loss": 1.6287,
+      "step": 1334
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.4428875869249e-06,
+      "loss": 1.7761,
+      "step": 1335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.429477603874942e-06,
+      "loss": 1.6918,
+      "step": 1336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.416068650053798e-06,
+      "loss": 1.8352,
+      "step": 1337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.402660749651361e-06,
+      "loss": 1.6401,
+      "step": 1338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.389253926855645e-06,
+      "loss": 1.7546,
+      "step": 1339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.375848205852702e-06,
+      "loss": 1.8476,
+      "step": 1340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.362443610826602e-06,
+      "loss": 1.6021,
+      "step": 1341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.349040165959388e-06,
+      "loss": 1.3842,
+      "step": 1342
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.335637895431023e-06,
+      "loss": 1.5799,
+      "step": 1343
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.322236823419358e-06,
+      "loss": 1.8308,
+      "step": 1344
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.308836974100075e-06,
+      "loss": 1.7813,
+      "step": 1345
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.295438371646647e-06,
+      "loss": 1.5405,
+      "step": 1346
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.282041040230312e-06,
+      "loss": 1.8285,
+      "step": 1347
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.268645004019997e-06,
+      "loss": 2.0266,
+      "step": 1348
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.255250287182315e-06,
+      "loss": 1.8779,
+      "step": 1349
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.24185691388147e-06,
+      "loss": 1.5952,
+      "step": 1350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.228464908279272e-06,
+      "loss": 1.7218,
+      "step": 1351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.215074294535043e-06,
+      "loss": 1.7289,
+      "step": 1352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.201685096805596e-06,
+      "loss": 1.5467,
+      "step": 1353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.188297339245203e-06,
+      "loss": 1.8675,
+      "step": 1354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.17491104600552e-06,
+      "loss": 1.7553,
+      "step": 1355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.16152624123558e-06,
+      "loss": 1.4933,
+      "step": 1356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.148142949081715e-06,
+      "loss": 1.8433,
+      "step": 1357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.134761193687532e-06,
+      "loss": 1.8575,
+      "step": 1358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.121380999193874e-06,
+      "loss": 1.5974,
+      "step": 1359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.108002389738757e-06,
+      "loss": 1.6541,
+      "step": 1360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.09462538945735e-06,
+      "loss": 1.9351,
+      "step": 1361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.081250022481905e-06,
+      "loss": 1.6969,
+      "step": 1362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.067876312941733e-06,
+      "loss": 1.386,
+      "step": 1363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.054504284963163e-06,
+      "loss": 1.946,
+      "step": 1364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.041133962669478e-06,
+      "loss": 1.6746,
+      "step": 1365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.027765370180894e-06,
+      "loss": 1.5457,
+      "step": 1366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.014398531614497e-06,
+      "loss": 1.8753,
+      "step": 1367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.001033471084215e-06,
+      "loss": 1.8461,
+      "step": 1368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.987670212700773e-06,
+      "loss": 1.4871,
+      "step": 1369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.97430878057163e-06,
+      "loss": 1.6499,
+      "step": 1370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.960949198800969e-06,
+      "loss": 1.7061,
+      "step": 1371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.94759149148962e-06,
+      "loss": 1.5575,
+      "step": 1372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.934235682735034e-06,
+      "loss": 1.8041,
+      "step": 1373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.920881796631249e-06,
+      "loss": 1.7836,
+      "step": 1374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.907529857268815e-06,
+      "loss": 1.9916,
+      "step": 1375
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.894179888734791e-06,
+      "loss": 1.598,
+      "step": 1376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.88083191511266e-06,
+      "loss": 1.5448,
+      "step": 1377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.867485960482328e-06,
+      "loss": 1.8919,
+      "step": 1378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.854142048920038e-06,
+      "loss": 1.7798,
+      "step": 1379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.840800204498355e-06,
+      "loss": 1.8522,
+      "step": 1380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.827460451286122e-06,
+      "loss": 1.8883,
+      "step": 1381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.8141228133484e-06,
+      "loss": 1.7034,
+      "step": 1382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.800787314746442e-06,
+      "loss": 1.8917,
+      "step": 1383
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.787453979537635e-06,
+      "loss": 1.7761,
+      "step": 1384
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.77412283177546e-06,
+      "loss": 1.7843,
+      "step": 1385
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.760793895509466e-06,
+      "loss": 1.7409,
+      "step": 1386
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.747467194785196e-06,
+      "loss": 2.0236,
+      "step": 1387
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.734142753644175e-06,
+      "loss": 1.7622,
+      "step": 1388
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720820596123839e-06,
+      "loss": 1.6407,
+      "step": 1389
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.707500746257512e-06,
+      "loss": 1.7216,
+      "step": 1390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.694183228074354e-06,
+      "loss": 1.8548,
+      "step": 1391
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.680868065599311e-06,
+      "loss": 1.9963,
+      "step": 1392
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.667555282853095e-06,
+      "loss": 1.7686,
+      "step": 1393
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.654244903852108e-06,
+      "loss": 1.7075,
+      "step": 1394
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.640936952608427e-06,
+      "loss": 1.5105,
+      "step": 1395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.627631453129745e-06,
+      "loss": 1.4883,
+      "step": 1396
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.614328429419329e-06,
+      "loss": 1.9429,
+      "step": 1397
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.601027905475991e-06,
+      "loss": 1.4451,
+      "step": 1398
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.587729905294018e-06,
+      "loss": 1.7012,
+      "step": 1399
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.574434452863153e-06,
+      "loss": 1.6766,
+      "step": 1400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.561141572168547e-06,
+      "loss": 1.7036,
+      "step": 1401
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.547851287190698e-06,
+      "loss": 1.8051,
+      "step": 1402
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.534563621905436e-06,
+      "loss": 1.7219,
+      "step": 1403
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.521278600283851e-06,
+      "loss": 1.6784,
+      "step": 1404
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.507996246292281e-06,
+      "loss": 1.8754,
+      "step": 1405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.494716583892232e-06,
+      "loss": 1.6943,
+      "step": 1406
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.481439637040372e-06,
+      "loss": 1.872,
+      "step": 1407
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.46816542968846e-06,
+      "loss": 1.9107,
+      "step": 1408
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.454893985783307e-06,
+      "loss": 1.8026,
+      "step": 1409
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.44162532926676e-06,
+      "loss": 1.6449,
+      "step": 1410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.428359484075618e-06,
+      "loss": 1.9324,
+      "step": 1411
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.415096474141613e-06,
+      "loss": 2.0195,
+      "step": 1412
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.401836323391367e-06,
+      "loss": 1.7585,
+      "step": 1413
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.388579055746337e-06,
+      "loss": 1.7768,
+      "step": 1414
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.375324695122787e-06,
+      "loss": 1.7469,
+      "step": 1415
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.36207326543173e-06,
+      "loss": 1.8059,
+      "step": 1416
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.348824790578894e-06,
+      "loss": 1.8075,
+      "step": 1417
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.335579294464678e-06,
+      "loss": 1.9276,
+      "step": 1418
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.322336800984103e-06,
+      "loss": 1.7149,
+      "step": 1419
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.30909733402678e-06,
+      "loss": 1.477,
+      "step": 1420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.295860917476854e-06,
+      "loss": 1.7452,
+      "step": 1421
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.282627575212968e-06,
+      "loss": 1.5474,
+      "step": 1422
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.26939733110822e-06,
+      "loss": 1.634,
+      "step": 1423
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.256170209030118e-06,
+      "loss": 1.6142,
+      "step": 1424
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.242946232840544e-06,
+      "loss": 1.8452,
+      "step": 1425
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.229725426395693e-06,
+      "loss": 1.4931,
+      "step": 1426
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.216507813546054e-06,
+      "loss": 1.9225,
+      "step": 1427
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.203293418136345e-06,
+      "loss": 1.7824,
+      "step": 1428
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.190082264005484e-06,
+      "loss": 1.6474,
+      "step": 1429
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.176874374986543e-06,
+      "loss": 1.7429,
+      "step": 1430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.163669774906702e-06,
+      "loss": 1.762,
+      "step": 1431
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.15046848758721e-06,
+      "loss": 1.8568,
+      "step": 1432
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.137270536843331e-06,
+      "loss": 1.5841,
+      "step": 1433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.124075946484323e-06,
+      "loss": 1.4999,
+      "step": 1434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.110884740313369e-06,
+      "loss": 2.0805,
+      "step": 1435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.097696942127559e-06,
+      "loss": 1.6028,
+      "step": 1436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.084512575717826e-06,
+      "loss": 1.6116,
+      "step": 1437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.071331664868917e-06,
+      "loss": 1.7895,
+      "step": 1438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.058154233359338e-06,
+      "loss": 1.6542,
+      "step": 1439
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.04498030496133e-06,
+      "loss": 1.6345,
+      "step": 1440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.031809903440804e-06,
+      "loss": 1.6697,
+      "step": 1441
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.018643052557313e-06,
+      "loss": 1.8097,
+      "step": 1442
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.005479776064006e-06,
+      "loss": 1.5372,
+      "step": 1443
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.992320097707578e-06,
+      "loss": 1.7374,
+      "step": 1444
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.979164041228237e-06,
+      "loss": 1.9053,
+      "step": 1445
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.96601163035966e-06,
+      "loss": 1.7471,
+      "step": 1446
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.95286288882894e-06,
+      "loss": 1.6862,
+      "step": 1447
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.939717840356558e-06,
+      "loss": 1.7543,
+      "step": 1448
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.926576508656325e-06,
+      "loss": 1.732,
+      "step": 1449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.913438917435353e-06,
+      "loss": 1.7932,
+      "step": 1450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.900305090394004e-06,
+      "loss": 1.7887,
+      "step": 1451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.88717505122585e-06,
+      "loss": 1.6519,
+      "step": 1452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.87404882361763e-06,
+      "loss": 1.6273,
+      "step": 1453
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.860926431249202e-06,
+      "loss": 1.6383,
+      "step": 1454
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.847807897793515e-06,
+      "loss": 1.7255,
+      "step": 1455
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.834693246916546e-06,
+      "loss": 1.4699,
+      "step": 1456
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.821582502277272e-06,
+      "loss": 1.6295,
+      "step": 1457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.808475687527627e-06,
+      "loss": 1.7686,
+      "step": 1458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.795372826312451e-06,
+      "loss": 1.8048,
+      "step": 1459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.782273942269452e-06,
+      "loss": 1.7053,
+      "step": 1460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.769179059029165e-06,
+      "loss": 1.4565,
+      "step": 1461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.756088200214905e-06,
+      "loss": 1.8306,
+      "step": 1462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.743001389442726e-06,
+      "loss": 1.6023,
+      "step": 1463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.729918650321388e-06,
+      "loss": 1.7549,
+      "step": 1464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.716840006452297e-06,
+      "loss": 1.8226,
+      "step": 1465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.703765481429468e-06,
+      "loss": 1.7426,
+      "step": 1466
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.690695098839497e-06,
+      "loss": 1.9795,
+      "step": 1467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.677628882261499e-06,
+      "loss": 1.5771,
+      "step": 1468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.664566855267075e-06,
+      "loss": 1.7259,
+      "step": 1469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.651509041420271e-06,
+      "loss": 1.7394,
+      "step": 1470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.638455464277522e-06,
+      "loss": 1.6449,
+      "step": 1471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.62540614738764e-06,
+      "loss": 1.8974,
+      "step": 1472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.612361114291729e-06,
+      "loss": 1.6265,
+      "step": 1473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.599320388523179e-06,
+      "loss": 1.7549,
+      "step": 1474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5862839936076085e-06,
+      "loss": 1.9875,
+      "step": 1475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.573251953062812e-06,
+      "loss": 1.5487,
+      "step": 1476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5602242903987466e-06,
+      "loss": 1.4667,
+      "step": 1477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.547201029117456e-06,
+      "loss": 1.7492,
+      "step": 1478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5341821927130505e-06,
+      "loss": 1.7347,
+      "step": 1479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.521167804671663e-06,
+      "loss": 1.5531,
+      "step": 1480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.508157888471385e-06,
+      "loss": 1.8141,
+      "step": 1481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.495152467582261e-06,
+      "loss": 1.5604,
+      "step": 1482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.482151565466209e-06,
+      "loss": 1.6531,
+      "step": 1483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.46915520557701e-06,
+      "loss": 1.6647,
+      "step": 1484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4561634113602366e-06,
+      "loss": 1.5674,
+      "step": 1485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4431762062532345e-06,
+      "loss": 1.6765,
+      "step": 1486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4301936136850684e-06,
+      "loss": 1.8347,
+      "step": 1487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.417215657076477e-06,
+      "loss": 1.4813,
+      "step": 1488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.404242359839846e-06,
+      "loss": 1.8753,
+      "step": 1489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.391273745379144e-06,
+      "loss": 1.4878,
+      "step": 1490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.378309837089898e-06,
+      "loss": 1.614,
+      "step": 1491
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.365350658359147e-06,
+      "loss": 1.6103,
+      "step": 1492
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.352396232565391e-06,
+      "loss": 1.8221,
+      "step": 1493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.339446583078564e-06,
+      "loss": 1.4998,
+      "step": 1494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.326501733259972e-06,
+      "loss": 1.579,
+      "step": 1495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.313561706462279e-06,
+      "loss": 1.8494,
+      "step": 1496
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.300626526029434e-06,
+      "loss": 1.6291,
+      "step": 1497
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.287696215296642e-06,
+      "loss": 1.6989,
+      "step": 1498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.2747707975903405e-06,
+      "loss": 1.6337,
+      "step": 1499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.261850296228117e-06,
+      "loss": 1.8349,
+      "step": 1500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.248934734518709e-06,
+      "loss": 1.7647,
+      "step": 1501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.236024135761933e-06,
+      "loss": 1.7656,
+      "step": 1502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.223118523248649e-06,
+      "loss": 1.8396,
+      "step": 1503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.2102179202607335e-06,
+      "loss": 1.4456,
+      "step": 1504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.197322350071014e-06,
+      "loss": 1.8552,
+      "step": 1505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.1844318359432505e-06,
+      "loss": 1.6677,
+      "step": 1506
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.17154640113207e-06,
+      "loss": 1.7386,
+      "step": 1507
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.15866606888294e-06,
+      "loss": 1.6984,
+      "step": 1508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.1457908624321295e-06,
+      "loss": 1.7989,
+      "step": 1509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.132920805006651e-06,
+      "loss": 1.8641,
+      "step": 1510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.120055919824238e-06,
+      "loss": 1.68,
+      "step": 1511
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.107196230093281e-06,
+      "loss": 1.4582,
+      "step": 1512
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.094341759012813e-06,
+      "loss": 1.6949,
+      "step": 1513
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.081492529772438e-06,
+      "loss": 1.6823,
+      "step": 1514
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.06864856555231e-06,
+      "loss": 1.8283,
+      "step": 1515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.055809889523088e-06,
+      "loss": 1.4883,
+      "step": 1516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042976524845881e-06,
+      "loss": 1.6148,
+      "step": 1517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.030148494672233e-06,
+      "loss": 1.6684,
+      "step": 1518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.01732582214405e-06,
+      "loss": 1.7527,
+      "step": 1519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.004508530393572e-06,
+      "loss": 1.6968,
+      "step": 1520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.991696642543346e-06,
+      "loss": 1.7161,
+      "step": 1521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.978890181706153e-06,
+      "loss": 1.7872,
+      "step": 1522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.966089170985e-06,
+      "loss": 1.5203,
+      "step": 1523
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.95329363347305e-06,
+      "loss": 1.567,
+      "step": 1524
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.940503592253594e-06,
+      "loss": 1.4941,
+      "step": 1525
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.927719070400018e-06,
+      "loss": 1.9991,
+      "step": 1526
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.914940090975733e-06,
+      "loss": 1.6083,
+      "step": 1527
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.902166677034169e-06,
+      "loss": 1.7086,
+      "step": 1528
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.889398851618706e-06,
+      "loss": 1.8148,
+      "step": 1529
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.876636637762639e-06,
+      "loss": 1.8481,
+      "step": 1530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.863880058489156e-06,
+      "loss": 1.689,
+      "step": 1531
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.851129136811259e-06,
+      "loss": 1.7571,
+      "step": 1532
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.83838389573176e-06,
+      "loss": 2.2286,
+      "step": 1533
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.8256443582432176e-06,
+      "loss": 1.7282,
+      "step": 1534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.812910547327891e-06,
+      "loss": 1.4182,
+      "step": 1535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.800182485957731e-06,
+      "loss": 1.7051,
+      "step": 1536
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7874601970942936e-06,
+      "loss": 1.8554,
+      "step": 1537
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.774743703688737e-06,
+      "loss": 1.604,
+      "step": 1538
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.762033028681754e-06,
+      "loss": 1.5509,
+      "step": 1539
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.749328195003548e-06,
+      "loss": 1.5948,
+      "step": 1540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.736629225573783e-06,
+      "loss": 1.7312,
+      "step": 1541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.723936143301532e-06,
+      "loss": 1.7717,
+      "step": 1542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.7112489710852734e-06,
+      "loss": 1.3178,
+      "step": 1543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.698567731812795e-06,
+      "loss": 1.8681,
+      "step": 1544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.685892448361204e-06,
+      "loss": 1.686,
+      "step": 1545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6732231435968504e-06,
+      "loss": 1.5845,
+      "step": 1546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.660559840375298e-06,
+      "loss": 1.5759,
+      "step": 1547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.647902561541297e-06,
+      "loss": 1.6674,
+      "step": 1548
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.635251329928714e-06,
+      "loss": 1.5743,
+      "step": 1549
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.622606168360516e-06,
+      "loss": 1.6893,
+      "step": 1550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.609967099648715e-06,
+      "loss": 1.8749,
+      "step": 1551
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.597334146594332e-06,
+      "loss": 1.8785,
+      "step": 1552
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.584707331987359e-06,
+      "loss": 1.4929,
+      "step": 1553
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.572086678606706e-06,
+      "loss": 1.5838,
+      "step": 1554
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.559472209220181e-06,
+      "loss": 1.8332,
+      "step": 1555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.546863946584425e-06,
+      "loss": 1.9658,
+      "step": 1556
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.534261913444882e-06,
+      "loss": 1.4424,
+      "step": 1557
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.521666132535765e-06,
+      "loss": 1.795,
+      "step": 1558
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.509076626580002e-06,
+      "loss": 1.6752,
+      "step": 1559
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.496493418289207e-06,
+      "loss": 1.8756,
+      "step": 1560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.483916530363627e-06,
+      "loss": 1.9803,
+      "step": 1561
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4713459854921054e-06,
+      "loss": 1.4724,
+      "step": 1562
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4587818063520525e-06,
+      "loss": 1.5906,
+      "step": 1563
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4462240156093815e-06,
+      "loss": 1.6771,
+      "step": 1564
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.433672635918496e-06,
+      "loss": 1.2961,
+      "step": 1565
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.421127689922218e-06,
+      "loss": 1.8783,
+      "step": 1566
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.40858920025178e-06,
+      "loss": 1.8338,
+      "step": 1567
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.396057189526752e-06,
+      "loss": 1.7909,
+      "step": 1568
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.3835316803550174e-06,
+      "loss": 1.5894,
+      "step": 1569
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.371012695332748e-06,
+      "loss": 1.8336,
+      "step": 1570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.35850025704432e-06,
+      "loss": 1.4437,
+      "step": 1571
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.345994388062324e-06,
+      "loss": 1.8256,
+      "step": 1572
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.333495110947483e-06,
+      "loss": 1.5398,
+      "step": 1573
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.3210024482486284e-06,
+      "loss": 1.6443,
+      "step": 1574
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.308516422502673e-06,
+      "loss": 1.5832,
+      "step": 1575
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.296037056234541e-06,
+      "loss": 1.7616,
+      "step": 1576
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.283564371957154e-06,
+      "loss": 1.7519,
+      "step": 1577
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2710983921713735e-06,
+      "loss": 1.556,
+      "step": 1578
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.258639139365959e-06,
+      "loss": 1.5676,
+      "step": 1579
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2461866360175546e-06,
+      "loss": 1.8558,
+      "step": 1580
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.233740904590608e-06,
+      "loss": 1.2815,
+      "step": 1581
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.221301967537363e-06,
+      "loss": 1.6872,
+      "step": 1582
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.208869847297802e-06,
+      "loss": 1.6558,
+      "step": 1583
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.196444566299602e-06,
+      "loss": 1.645,
+      "step": 1584
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.184026146958119e-06,
+      "loss": 1.6529,
+      "step": 1585
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.171614611676312e-06,
+      "loss": 1.9378,
+      "step": 1586
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1592099828447394e-06,
+      "loss": 1.7094,
+      "step": 1587
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.146812282841485e-06,
+      "loss": 1.6634,
+      "step": 1588
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.134421534032138e-06,
+      "loss": 1.4828,
+      "step": 1589
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.122037758769755e-06,
+      "loss": 1.7574,
+      "step": 1590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.109660979394795e-06,
+      "loss": 1.4817,
+      "step": 1591
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.097291218235117e-06,
+      "loss": 1.5346,
+      "step": 1592
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.084928497605902e-06,
+      "loss": 1.8912,
+      "step": 1593
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.072572839809645e-06,
+      "loss": 1.8409,
+      "step": 1594
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.060224267136084e-06,
+      "loss": 1.7326,
+      "step": 1595
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.047882801862184e-06,
+      "loss": 1.6717,
+      "step": 1596
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.035548466252089e-06,
+      "loss": 1.473,
+      "step": 1597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.023221282557076e-06,
+      "loss": 1.6401,
+      "step": 1598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.010901273015528e-06,
+      "loss": 1.5668,
+      "step": 1599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.998588459852877e-06,
+      "loss": 1.6484,
+      "step": 1600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.986282865281573e-06,
+      "loss": 1.7297,
+      "step": 1601
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.973984511501054e-06,
+      "loss": 1.6457,
+      "step": 1602
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.961693420697681e-06,
+      "loss": 1.7339,
+      "step": 1603
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9494096150447276e-06,
+      "loss": 1.6796,
+      "step": 1604
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.937133116702314e-06,
+      "loss": 1.5621,
+      "step": 1605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9248639478173745e-06,
+      "loss": 1.6619,
+      "step": 1606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.912602130523637e-06,
+      "loss": 1.5625,
+      "step": 1607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.90034768694155e-06,
+      "loss": 1.7602,
+      "step": 1608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.888100639178276e-06,
+      "loss": 1.9313,
+      "step": 1609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.875861009327622e-06,
+      "loss": 1.9211,
+      "step": 1610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.863628819470011e-06,
+      "loss": 1.7345,
+      "step": 1611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.851404091672464e-06,
+      "loss": 1.7456,
+      "step": 1612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.839186847988514e-06,
+      "loss": 1.89,
+      "step": 1613
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8269771104582165e-06,
+      "loss": 1.8085,
+      "step": 1614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.814774901108071e-06,
+      "loss": 1.8487,
+      "step": 1615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.802580241950997e-06,
+      "loss": 1.7769,
+      "step": 1616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.790393154986305e-06,
+      "loss": 1.8585,
+      "step": 1617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.77821366219963e-06,
+      "loss": 1.648,
+      "step": 1618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.76604178556291e-06,
+      "loss": 1.3967,
+      "step": 1619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.753877547034357e-06,
+      "loss": 1.5803,
+      "step": 1620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.741720968558386e-06,
+      "loss": 1.6377,
+      "step": 1621
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7295720720656074e-06,
+      "loss": 1.7743,
+      "step": 1622
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.717430879472761e-06,
+      "loss": 1.6719,
+      "step": 1623
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7052974126827e-06,
+      "loss": 1.734,
+      "step": 1624
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.693171693584332e-06,
+      "loss": 1.5833,
+      "step": 1625
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.681053744052587e-06,
+      "loss": 1.7899,
+      "step": 1626
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.66894358594839e-06,
+      "loss": 1.7985,
+      "step": 1627
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.656841241118593e-06,
+      "loss": 1.6248,
+      "step": 1628
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.644746731395971e-06,
+      "loss": 1.5707,
+      "step": 1629
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.63266007859915e-06,
+      "loss": 1.7934,
+      "step": 1630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.6205813045325855e-06,
+      "loss": 1.6613,
+      "step": 1631
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.608510430986527e-06,
+      "loss": 1.645,
+      "step": 1632
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.596447479736961e-06,
+      "loss": 1.5305,
+      "step": 1633
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.584392472545593e-06,
+      "loss": 1.9869,
+      "step": 1634
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.572345431159783e-06,
+      "loss": 1.3298,
+      "step": 1635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.560306377312538e-06,
+      "loss": 1.8374,
+      "step": 1636
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.548275332722442e-06,
+      "loss": 1.7055,
+      "step": 1637
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.53625231909363e-06,
+      "loss": 1.7806,
+      "step": 1638
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.524237358115763e-06,
+      "loss": 1.5743,
+      "step": 1639
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.512230471463955e-06,
+      "loss": 1.6908,
+      "step": 1640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.500231680798776e-06,
+      "loss": 1.6841,
+      "step": 1641
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.488241007766172e-06,
+      "loss": 1.8519,
+      "step": 1642
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.476258473997449e-06,
+      "loss": 1.6801,
+      "step": 1643
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.46428410110924e-06,
+      "loss": 1.892,
+      "step": 1644
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.452317910703441e-06,
+      "loss": 1.4952,
+      "step": 1645
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.440359924367201e-06,
+      "loss": 1.5208,
+      "step": 1646
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.428410163672855e-06,
+      "loss": 1.6726,
+      "step": 1647
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.416468650177911e-06,
+      "loss": 1.5319,
+      "step": 1648
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.404535405424991e-06,
+      "loss": 1.7071,
+      "step": 1649
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.392610450941798e-06,
+      "loss": 1.493,
+      "step": 1650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.380693808241093e-06,
+      "loss": 1.6776,
+      "step": 1651
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.368785498820626e-06,
+      "loss": 1.5058,
+      "step": 1652
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.356885544163124e-06,
+      "loss": 1.453,
+      "step": 1653
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.344993965736241e-06,
+      "loss": 1.5257,
+      "step": 1654
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.33311078499251e-06,
+      "loss": 1.9215,
+      "step": 1655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.321236023369332e-06,
+      "loss": 1.6453,
+      "step": 1656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.309369702288901e-06,
+      "loss": 1.5869,
+      "step": 1657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.297511843158203e-06,
+      "loss": 1.8622,
+      "step": 1658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.285662467368945e-06,
+      "loss": 1.612,
+      "step": 1659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.273821596297529e-06,
+      "loss": 1.4406,
+      "step": 1660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.261989251305028e-06,
+      "loss": 1.6153,
+      "step": 1661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.250165453737117e-06,
+      "loss": 1.4098,
+      "step": 1662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.238350224924069e-06,
+      "loss": 1.9057,
+      "step": 1663
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.2265435861806825e-06,
+      "loss": 1.3325,
+      "step": 1664
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.214745558806265e-06,
+      "loss": 1.4874,
+      "step": 1665
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.202956164084596e-06,
+      "loss": 1.5779,
+      "step": 1666
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.191175423283871e-06,
+      "loss": 1.2169,
+      "step": 1667
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.179403357656684e-06,
+      "loss": 1.6448,
+      "step": 1668
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.167639988439971e-06,
+      "loss": 1.3585,
+      "step": 1669
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1558853368549775e-06,
+      "loss": 1.5888,
+      "step": 1670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1441394241072355e-06,
+      "loss": 1.7078,
+      "step": 1671
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.132402271386495e-06,
+      "loss": 1.675,
+      "step": 1672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.120673899866719e-06,
+      "loss": 1.5931,
+      "step": 1673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.108954330706015e-06,
+      "loss": 1.4554,
+      "step": 1674
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0972435850466225e-06,
+      "loss": 1.6796,
+      "step": 1675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0855416840148585e-06,
+      "loss": 1.331,
+      "step": 1676
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.073848648721076e-06,
+      "loss": 1.3713,
+      "step": 1677
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.062164500259652e-06,
+      "loss": 1.7348,
+      "step": 1678
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.050489259708912e-06,
+      "loss": 1.5403,
+      "step": 1679
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.03882294813113e-06,
+      "loss": 1.2547,
+      "step": 1680
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.027165586572458e-06,
+      "loss": 1.6147,
+      "step": 1681
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.0155171960629055e-06,
+      "loss": 1.5699,
+      "step": 1682
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.003877797616306e-06,
+      "loss": 1.653,
+      "step": 1683
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.992247412230258e-06,
+      "loss": 1.5653,
+      "step": 1684
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.980626060886114e-06,
+      "loss": 1.5895,
+      "step": 1685
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.969013764548919e-06,
+      "loss": 1.53,
+      "step": 1686
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9574105441673825e-06,
+      "loss": 1.5876,
+      "step": 1687
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.945816420673851e-06,
+      "loss": 1.4225,
+      "step": 1688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.934231414984247e-06,
+      "loss": 1.3039,
+      "step": 1689
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.922655547998059e-06,
+      "loss": 1.731,
+      "step": 1690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.911088840598276e-06,
+      "loss": 1.6074,
+      "step": 1691
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.899531313651363e-06,
+      "loss": 1.6003,
+      "step": 1692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.88798298800724e-06,
+      "loss": 1.593,
+      "step": 1693
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.876443884499203e-06,
+      "loss": 1.6078,
+      "step": 1694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.864914023943937e-06,
+      "loss": 1.3766,
+      "step": 1695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.853393427141433e-06,
+      "loss": 1.5421,
+      "step": 1696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.841882114874975e-06,
+      "loss": 1.4816,
+      "step": 1697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.830380107911107e-06,
+      "loss": 1.6888,
+      "step": 1698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.818887426999569e-06,
+      "loss": 1.5108,
+      "step": 1699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.807404092873297e-06,
+      "loss": 1.6731,
+      "step": 1700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.795930126248344e-06,
+      "loss": 1.4262,
+      "step": 1701
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.784465547823885e-06,
+      "loss": 1.4915,
+      "step": 1702
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.773010378282143e-06,
+      "loss": 1.6782,
+      "step": 1703
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.76156463828837e-06,
+      "loss": 1.6383,
+      "step": 1704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7501283484908145e-06,
+      "loss": 1.4403,
+      "step": 1705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.738701529520666e-06,
+      "loss": 1.5249,
+      "step": 1706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.727284201992043e-06,
+      "loss": 1.5739,
+      "step": 1707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.715876386501925e-06,
+      "loss": 1.673,
+      "step": 1708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.70447810363014e-06,
+      "loss": 1.7918,
+      "step": 1709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.693089373939323e-06,
+      "loss": 1.7734,
+      "step": 1710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.681710217974862e-06,
+      "loss": 1.4695,
+      "step": 1711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.670340656264893e-06,
+      "loss": 1.546,
+      "step": 1712
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.658980709320226e-06,
+      "loss": 1.5291,
+      "step": 1713
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.64763039763433e-06,
+      "loss": 1.75,
+      "step": 1714
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.636289741683303e-06,
+      "loss": 1.4253,
+      "step": 1715
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6249587619258074e-06,
+      "loss": 1.341,
+      "step": 1716
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.613637478803066e-06,
+      "loss": 1.5447,
+      "step": 1717
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6023259127387965e-06,
+      "loss": 1.4075,
+      "step": 1718
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.591024084139188e-06,
+      "loss": 1.423,
+      "step": 1719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.579732013392874e-06,
+      "loss": 1.6123,
+      "step": 1720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.568449720870867e-06,
+      "loss": 1.4286,
+      "step": 1721
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.557177226926559e-06,
+      "loss": 1.5674,
+      "step": 1722
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.545914551895652e-06,
+      "loss": 1.4049,
+      "step": 1723
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.534661716096133e-06,
+      "loss": 1.8229,
+      "step": 1724
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5234187398282535e-06,
+      "loss": 1.404,
+      "step": 1725
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.51218564337446e-06,
+      "loss": 1.5884,
+      "step": 1726
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.500962446999393e-06,
+      "loss": 1.6361,
+      "step": 1727
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.489749170949817e-06,
+      "loss": 1.5398,
+      "step": 1728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.478545835454614e-06,
+      "loss": 1.6693,
+      "step": 1729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.467352460724726e-06,
+      "loss": 1.6895,
+      "step": 1730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.456169066953121e-06,
+      "loss": 1.3908,
+      "step": 1731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.444995674314776e-06,
+      "loss": 1.4884,
+      "step": 1732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.433832302966607e-06,
+      "loss": 1.5956,
+      "step": 1733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.422678973047471e-06,
+      "loss": 1.6552,
+      "step": 1734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.411535704678096e-06,
+      "loss": 1.6596,
+      "step": 1735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.400402517961059e-06,
+      "loss": 1.531,
+      "step": 1736
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.389279432980761e-06,
+      "loss": 1.5234,
+      "step": 1737
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.378166469803365e-06,
+      "loss": 1.5066,
+      "step": 1738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.36706364847679e-06,
+      "loss": 1.4412,
+      "step": 1739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.355970989030646e-06,
+      "loss": 1.3444,
+      "step": 1740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.3448885114762104e-06,
+      "loss": 1.4609,
+      "step": 1741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.333816235806406e-06,
+      "loss": 1.566,
+      "step": 1742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.322754181995734e-06,
+      "loss": 1.581,
+      "step": 1743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.311702370000272e-06,
+      "loss": 1.4622,
+      "step": 1744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.30066081975761e-06,
+      "loss": 1.3701,
+      "step": 1745
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.289629551186821e-06,
+      "loss": 1.3527,
+      "step": 1746
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.278608584188449e-06,
+      "loss": 1.6991,
+      "step": 1747
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.267597938644434e-06,
+      "loss": 1.6552,
+      "step": 1748
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2565976344181104e-06,
+      "loss": 1.7958,
+      "step": 1749
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.245607691354147e-06,
+      "loss": 1.3338,
+      "step": 1750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.234628129278522e-06,
+      "loss": 1.7431,
+      "step": 1751
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.223658967998494e-06,
+      "loss": 1.7078,
+      "step": 1752
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.212700227302547e-06,
+      "loss": 1.195,
+      "step": 1753
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.201751926960382e-06,
+      "loss": 1.3951,
+      "step": 1754
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.190814086722843e-06,
+      "loss": 1.6431,
+      "step": 1755
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1798867263219275e-06,
+      "loss": 1.7221,
+      "step": 1756
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.168969865470714e-06,
+      "loss": 1.638,
+      "step": 1757
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.158063523863335e-06,
+      "loss": 1.5043,
+      "step": 1758
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.147167721174963e-06,
+      "loss": 1.5438,
+      "step": 1759
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.13628247706174e-06,
+      "loss": 1.5941,
+      "step": 1760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.125407811160775e-06,
+      "loss": 1.7997,
+      "step": 1761
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1145437430900855e-06,
+      "loss": 1.6117,
+      "step": 1762
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1036902924485685e-06,
+      "loss": 1.5954,
+      "step": 1763
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.092847478815978e-06,
+      "loss": 1.7001,
+      "step": 1764
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.082015321752862e-06,
+      "loss": 1.811,
+      "step": 1765
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0711938408005645e-06,
+      "loss": 1.5515,
+      "step": 1766
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.060383055481153e-06,
+      "loss": 1.2041,
+      "step": 1767
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0495829852974015e-06,
+      "loss": 1.7794,
+      "step": 1768
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.038793649732766e-06,
+      "loss": 1.5443,
+      "step": 1769
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.028015068251322e-06,
+      "loss": 1.4973,
+      "step": 1770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.017247260297758e-06,
+      "loss": 1.7278,
+      "step": 1771
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.006490245297317e-06,
+      "loss": 1.4621,
+      "step": 1772
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.995744042655771e-06,
+      "loss": 1.6273,
+      "step": 1773
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.985008671759399e-06,
+      "loss": 1.6288,
+      "step": 1774
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9742841519749205e-06,
+      "loss": 1.426,
+      "step": 1775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.963570502649501e-06,
+      "loss": 1.6417,
+      "step": 1776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.952867743110676e-06,
+      "loss": 1.6365,
+      "step": 1777
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.942175892666343e-06,
+      "loss": 1.7426,
+      "step": 1778
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.931494970604728e-06,
+      "loss": 1.4853,
+      "step": 1779
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920824996194326e-06,
+      "loss": 1.6869,
+      "step": 1780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.910165988683897e-06,
+      "loss": 1.6548,
+      "step": 1781
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8995179673024045e-06,
+      "loss": 1.4459,
+      "step": 1782
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.888880951259005e-06,
+      "loss": 1.7443,
+      "step": 1783
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.878254959742992e-06,
+      "loss": 1.6786,
+      "step": 1784
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.86764001192377e-06,
+      "loss": 1.5259,
+      "step": 1785
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.85703612695083e-06,
+      "loss": 1.3669,
+      "step": 1786
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.846443323953694e-06,
+      "loss": 1.4886,
+      "step": 1787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.835861622041902e-06,
+      "loss": 1.5862,
+      "step": 1788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8252910403049625e-06,
+      "loss": 1.7249,
+      "step": 1789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.814731597812319e-06,
+      "loss": 1.6304,
+      "step": 1790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.804183313613332e-06,
+      "loss": 1.6188,
+      "step": 1791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.793646206737216e-06,
+      "loss": 1.4995,
+      "step": 1792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.783120296193039e-06,
+      "loss": 1.7555,
+      "step": 1793
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7726056009696567e-06,
+      "loss": 1.746,
+      "step": 1794
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.762102140035695e-06,
+      "loss": 1.6872,
+      "step": 1795
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7516099323395215e-06,
+      "loss": 1.6725,
+      "step": 1796
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.741128996809191e-06,
+      "loss": 1.4854,
+      "step": 1797
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.730659352352435e-06,
+      "loss": 1.5971,
+      "step": 1798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.720201017856605e-06,
+      "loss": 1.632,
+      "step": 1799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.709754012188652e-06,
+      "loss": 1.7826,
+      "step": 1800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.699318354195098e-06,
+      "loss": 1.4342,
+      "step": 1801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.688894062701982e-06,
+      "loss": 1.5885,
+      "step": 1802
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6784811565148472e-06,
+      "loss": 1.7204,
+      "step": 1803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6680796544186903e-06,
+      "loss": 1.3758,
+      "step": 1804
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.657689575177935e-06,
+      "loss": 1.8015,
+      "step": 1805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.647310937536408e-06,
+      "loss": 1.5624,
+      "step": 1806
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6369437602172807e-06,
+      "loss": 1.6496,
+      "step": 1807
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6265880619230655e-06,
+      "loss": 1.3669,
+      "step": 1808
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.61624386133555e-06,
+      "loss": 1.6368,
+      "step": 1809
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.605911177115796e-06,
+      "loss": 1.7435,
+      "step": 1810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.595590027904078e-06,
+      "loss": 1.5219,
+      "step": 1811
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5852804323198608e-06,
+      "loss": 1.5955,
+      "step": 1812
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5749824089617803e-06,
+      "loss": 1.5761,
+      "step": 1813
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5646959764075763e-06,
+      "loss": 1.669,
+      "step": 1814
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.554421153214097e-06,
+      "loss": 1.4766,
+      "step": 1815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5441579579172356e-06,
+      "loss": 1.722,
+      "step": 1816
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5339064090319073e-06,
+      "loss": 1.6403,
+      "step": 1817
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5236665250520284e-06,
+      "loss": 1.5574,
+      "step": 1818
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5134383244504587e-06,
+      "loss": 1.5373,
+      "step": 1819
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.503221825678994e-06,
+      "loss": 1.7046,
+      "step": 1820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4930170471683067e-06,
+      "loss": 1.4539,
+      "step": 1821
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4828240073279316e-06,
+      "loss": 1.4367,
+      "step": 1822
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4726427245462325e-06,
+      "loss": 1.298,
+      "step": 1823
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4624732171903508e-06,
+      "loss": 1.2293,
+      "step": 1824
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4523155036061986e-06,
+      "loss": 1.6587,
+      "step": 1825
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4421696021184024e-06,
+      "loss": 1.595,
+      "step": 1826
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.432035531030279e-06,
+      "loss": 1.4745,
+      "step": 1827
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4219133086238133e-06,
+      "loss": 1.6928,
+      "step": 1828
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.411802953159601e-06,
+      "loss": 1.5004,
+      "step": 1829
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.401704482876844e-06,
+      "loss": 1.5912,
+      "step": 1830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.391617915993293e-06,
+      "loss": 1.6982,
+      "step": 1831
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3815432707052233e-06,
+      "loss": 1.594,
+      "step": 1832
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.371480565187416e-06,
+      "loss": 1.6493,
+      "step": 1833
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.361429817593098e-06,
+      "loss": 1.8924,
+      "step": 1834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.351391046053938e-06,
+      "loss": 1.6222,
+      "step": 1835
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3413642686799852e-06,
+      "loss": 1.3749,
+      "step": 1836
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.331349503559664e-06,
+      "loss": 1.4429,
+      "step": 1837
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.321346768759722e-06,
+      "loss": 1.7424,
+      "step": 1838
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3113560823252013e-06,
+      "loss": 1.5072,
+      "step": 1839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.301377462279418e-06,
+      "loss": 1.1762,
+      "step": 1840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.291410926623909e-06,
+      "loss": 1.3075,
+      "step": 1841
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.281456493338423e-06,
+      "loss": 1.6162,
+      "step": 1842
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.271514180380866e-06,
+      "loss": 1.557,
+      "step": 1843
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2615840056872784e-06,
+      "loss": 1.6116,
+      "step": 1844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2516659871718135e-06,
+      "loss": 1.6395,
+      "step": 1845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.241760142726683e-06,
+      "loss": 1.6226,
+      "step": 1846
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2318664902221476e-06,
+      "loss": 1.2204,
+      "step": 1847
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.221985047506463e-06,
+      "loss": 1.5783,
+      "step": 1848
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.21211583240586e-06,
+      "loss": 1.4065,
+      "step": 1849
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.20225886272452e-06,
+      "loss": 1.6738,
+      "step": 1850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.192414156244521e-06,
+      "loss": 1.6319,
+      "step": 1851
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.18258173072583e-06,
+      "loss": 1.2582,
+      "step": 1852
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1727616039062504e-06,
+      "loss": 1.5697,
+      "step": 1853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.162953793501398e-06,
+      "loss": 1.7176,
+      "step": 1854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1531583172046788e-06,
+      "loss": 1.5614,
+      "step": 1855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1433751926872347e-06,
+      "loss": 1.4517,
+      "step": 1856
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.133604437597941e-06,
+      "loss": 1.5378,
+      "step": 1857
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1238460695633445e-06,
+      "loss": 1.5163,
+      "step": 1858
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1141001061876484e-06,
+      "loss": 1.8981,
+      "step": 1859
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1043665650526868e-06,
+      "loss": 1.6189,
+      "step": 1860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.094645463717869e-06,
+      "loss": 1.5333,
+      "step": 1861
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.084936819720179e-06,
+      "loss": 1.4638,
+      "step": 1862
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0752406505741127e-06,
+      "loss": 1.5557,
+      "step": 1863
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.065556973771675e-06,
+      "loss": 1.8067,
+      "step": 1864
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.055885806782323e-06,
+      "loss": 1.6375,
+      "step": 1865
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0462271670529486e-06,
+      "loss": 1.6139,
+      "step": 1866
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0365810720078535e-06,
+      "loss": 1.5985,
+      "step": 1867
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0269475390486925e-06,
+      "loss": 1.4125,
+      "step": 1868
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0173265855544763e-06,
+      "loss": 1.7448,
+      "step": 1869
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.007718228881509e-06,
+      "loss": 1.4783,
+      "step": 1870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9981224863633684e-06,
+      "loss": 1.5603,
+      "step": 1871
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9885393753108916e-06,
+      "loss": 1.6574,
+      "step": 1872
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9789689130121113e-06,
+      "loss": 1.83,
+      "step": 1873
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9694111167322536e-06,
+      "loss": 1.6488,
+      "step": 1874
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.959866003713688e-06,
+      "loss": 1.3852,
+      "step": 1875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9503335911759045e-06,
+      "loss": 1.8802,
+      "step": 1876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9408138963154853e-06,
+      "loss": 1.5151,
+      "step": 1877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.931306936306063e-06,
+      "loss": 1.4682,
+      "step": 1878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9218127282983044e-06,
+      "loss": 1.7568,
+      "step": 1879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.912331289419866e-06,
+      "loss": 1.8746,
+      "step": 1880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9028626367753665e-06,
+      "loss": 1.5074,
+      "step": 1881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.893406787446368e-06,
+      "loss": 1.6227,
+      "step": 1882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8839637584913218e-06,
+      "loss": 1.5621,
+      "step": 1883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8745335669455642e-06,
+      "loss": 1.6621,
+      "step": 1884
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.865116229821262e-06,
+      "loss": 1.5247,
+      "step": 1885
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.855711764107403e-06,
+      "loss": 1.3613,
+      "step": 1886
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8463201867697445e-06,
+      "loss": 1.4345,
+      "step": 1887
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.836941514750794e-06,
+      "loss": 1.7005,
+      "step": 1888
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.82757576496979e-06,
+      "loss": 1.5753,
+      "step": 1889
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8182229543226412e-06,
+      "loss": 1.7639,
+      "step": 1890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8088830996819315e-06,
+      "loss": 1.3824,
+      "step": 1891
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7995562178968594e-06,
+      "loss": 1.5305,
+      "step": 1892
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.790242325793222e-06,
+      "loss": 1.5448,
+      "step": 1893
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7809414401733905e-06,
+      "loss": 1.6202,
+      "step": 1894
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.77165357781626e-06,
+      "loss": 1.4346,
+      "step": 1895
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7623787554772473e-06,
+      "loss": 1.4671,
+      "step": 1896
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7531169898882306e-06,
+      "loss": 1.3888,
+      "step": 1897
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.743868297757537e-06,
+      "loss": 1.529,
+      "step": 1898
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7346326957699175e-06,
+      "loss": 1.6593,
+      "step": 1899
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7254102005864944e-06,
+      "loss": 1.4141,
+      "step": 1900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7162008288447595e-06,
+      "loss": 1.9468,
+      "step": 1901
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7070045971585202e-06,
+      "loss": 1.6118,
+      "step": 1902
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.697821522117877e-06,
+      "loss": 1.5924,
+      "step": 1903
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.688651620289209e-06,
+      "loss": 1.4934,
+      "step": 1904
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6794949082151134e-06,
+      "loss": 1.6046,
+      "step": 1905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6703514024144096e-06,
+      "loss": 1.4308,
+      "step": 1906
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.66122111938208e-06,
+      "loss": 1.3559,
+      "step": 1907
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6521040755892567e-06,
+      "loss": 1.2404,
+      "step": 1908
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.643000287483194e-06,
+      "loss": 1.6553,
+      "step": 1909
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.633909771487221e-06,
+      "loss": 1.5677,
+      "step": 1910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6248325440007383e-06,
+      "loss": 1.5432,
+      "step": 1911
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6157686213991596e-06,
+      "loss": 1.6046,
+      "step": 1912
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.606718020033908e-06,
+      "loss": 1.6747,
+      "step": 1913
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5976807562323693e-06,
+      "loss": 1.4575,
+      "step": 1914
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.588656846297863e-06,
+      "loss": 1.2605,
+      "step": 1915
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5796463065096333e-06,
+      "loss": 1.4096,
+      "step": 1916
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5706491531227862e-06,
+      "loss": 1.5734,
+      "step": 1917
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.561665402368295e-06,
+      "loss": 1.7293,
+      "step": 1918
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.552695070452943e-06,
+      "loss": 1.6503,
+      "step": 1919
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.543738173559307e-06,
+      "loss": 1.2354,
+      "step": 1920
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5347947278457365e-06,
+      "loss": 1.652,
+      "step": 1921
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5258647494463006e-06,
+      "loss": 1.4209,
+      "step": 1922
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.516948254470788e-06,
+      "loss": 1.5474,
+      "step": 1923
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5080452590046522e-06,
+      "loss": 1.6182,
+      "step": 1924
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4991557791089927e-06,
+      "loss": 1.6506,
+      "step": 1925
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.490279830820539e-06,
+      "loss": 1.6627,
+      "step": 1926
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4814174301515947e-06,
+      "loss": 1.4332,
+      "step": 1927
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4725685930900355e-06,
+      "loss": 1.4928,
+      "step": 1928
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.463733335599261e-06,
+      "loss": 1.5617,
+      "step": 1929
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.454911673618171e-06,
+      "loss": 1.6301,
+      "step": 1930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.446103623061149e-06,
+      "loss": 1.76,
+      "step": 1931
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.437309199818012e-06,
+      "loss": 1.7775,
+      "step": 1932
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4285284197540025e-06,
+      "loss": 1.6469,
+      "step": 1933
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.419761298709745e-06,
+      "loss": 1.8643,
+      "step": 1934
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.4110078525012203e-06,
+      "loss": 1.6386,
+      "step": 1935
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.402268096919751e-06,
+      "loss": 1.6841,
+      "step": 1936
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.393542047731946e-06,
+      "loss": 1.5077,
+      "step": 1937
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.384829720679703e-06,
+      "loss": 1.5329,
+      "step": 1938
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3761311314801537e-06,
+      "loss": 1.6208,
+      "step": 1939
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.367446295825655e-06,
+      "loss": 1.6163,
+      "step": 1940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.358775229383744e-06,
+      "loss": 1.4996,
+      "step": 1941
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3501179477971204e-06,
+      "loss": 1.7126,
+      "step": 1942
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.341474466683623e-06,
+      "loss": 1.6568,
+      "step": 1943
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3328448016361837e-06,
+      "loss": 1.5741,
+      "step": 1944
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3242289682228194e-06,
+      "loss": 1.2793,
+      "step": 1945
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3156269819865882e-06,
+      "loss": 1.3538,
+      "step": 1946
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.307038858445567e-06,
+      "loss": 1.7207,
+      "step": 1947
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2984646130928335e-06,
+      "loss": 1.4147,
+      "step": 1948
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2899042613964152e-06,
+      "loss": 1.5367,
+      "step": 1949
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2813578187992903e-06,
+      "loss": 1.6475,
+      "step": 1950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2728253007193325e-06,
+      "loss": 1.4783,
+      "step": 1951
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.264306722549299e-06,
+      "loss": 1.5815,
+      "step": 1952
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.255802099656805e-06,
+      "loss": 1.2929,
+      "step": 1953
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2473114473842804e-06,
+      "loss": 1.4325,
+      "step": 1954
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.238834781048964e-06,
+      "loss": 1.4254,
+      "step": 1955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2303721159428527e-06,
+      "loss": 1.6323,
+      "step": 1956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2219234673326874e-06,
+      "loss": 1.5993,
+      "step": 1957
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2134888504599296e-06,
+      "loss": 1.6437,
+      "step": 1958
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2050682805407187e-06,
+      "loss": 1.4735,
+      "step": 1959
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1966617727658613e-06,
+      "loss": 1.6537,
+      "step": 1960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.188269342300792e-06,
+      "loss": 1.6789,
+      "step": 1961
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1798910042855438e-06,
+      "loss": 1.6081,
+      "step": 1962
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.171526773834739e-06,
+      "loss": 1.5709,
+      "step": 1963
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1631766660375377e-06,
+      "loss": 1.8217,
+      "step": 1964
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.154840695957634e-06,
+      "loss": 1.4311,
+      "step": 1965
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.146518878633206e-06,
+      "loss": 1.6838,
+      "step": 1966
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1382112290769107e-06,
+      "loss": 1.4301,
+      "step": 1967
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1299177622758392e-06,
+      "loss": 1.6283,
+      "step": 1968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.121638493191496e-06,
+      "loss": 1.2111,
+      "step": 1969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.113373436759779e-06,
+      "loss": 1.4203,
+      "step": 1970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1051226078909405e-06,
+      "loss": 1.6565,
+      "step": 1971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0968860214695707e-06,
+      "loss": 1.4935,
+      "step": 1972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0886636923545646e-06,
+      "loss": 1.6589,
+      "step": 1973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0804556353790904e-06,
+      "loss": 1.248,
+      "step": 1974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.072261865350582e-06,
+      "loss": 1.5487,
+      "step": 1975
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0640823970506863e-06,
+      "loss": 1.6357,
+      "step": 1976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.055917245235263e-06,
+      "loss": 1.7831,
+      "step": 1977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.047766424634333e-06,
+      "loss": 1.7741,
+      "step": 1978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0396299499520676e-06,
+      "loss": 1.6315,
+      "step": 1979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0315078358667617e-06,
+      "loss": 1.5236,
+      "step": 1980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0234000970307966e-06,
+      "loss": 1.6701,
+      "step": 1981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.015306748070629e-06,
+      "loss": 1.458,
+      "step": 1982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0072278035867475e-06,
+      "loss": 1.5918,
+      "step": 1983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9991632781536584e-06,
+      "loss": 1.4779,
+      "step": 1984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9911131863198596e-06,
+      "loss": 1.4973,
+      "step": 1985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.983077542607802e-06,
+      "loss": 1.6933,
+      "step": 1986
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.975056361513883e-06,
+      "loss": 1.0907,
+      "step": 1987
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9670496575084006e-06,
+      "loss": 1.8284,
+      "step": 1988
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.959057445035536e-06,
+      "loss": 1.4099,
+      "step": 1989
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9510797385133352e-06,
+      "loss": 1.7728,
+      "step": 1990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9431165523336657e-06,
+      "loss": 1.3832,
+      "step": 1991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9351679008622094e-06,
+      "loss": 1.64,
+      "step": 1992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9272337984384183e-06,
+      "loss": 1.3968,
+      "step": 1993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9193142593755064e-06,
+      "loss": 1.3811,
+      "step": 1994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9114092979604106e-06,
+      "loss": 1.379,
+      "step": 1995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9035189284537648e-06,
+      "loss": 1.5132,
+      "step": 1996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.895643165089891e-06,
+      "loss": 1.6173,
+      "step": 1997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8877820220767483e-06,
+      "loss": 1.8622,
+      "step": 1998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.879935513595934e-06,
+      "loss": 1.4891,
+      "step": 1999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8721036538026339e-06,
+      "loss": 1.7258,
+      "step": 2000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.864286456825608e-06,
+      "loss": 1.8138,
+      "step": 2001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.856483936767174e-06,
+      "loss": 1.5287,
+      "step": 2002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8486961077031584e-06,
+      "loss": 1.3731,
+      "step": 2003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.840922983682899e-06,
+      "loss": 1.522,
+      "step": 2004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8331645787291964e-06,
+      "loss": 1.2103,
+      "step": 2005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8254209068382966e-06,
+      "loss": 1.7518,
+      "step": 2006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8176919819798765e-06,
+      "loss": 1.4095,
+      "step": 2007
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8099778180969984e-06,
+      "loss": 1.253,
+      "step": 2008
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8022784291061045e-06,
+      "loss": 1.6676,
+      "step": 2009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7945938288969778e-06,
+      "loss": 1.3718,
+      "step": 2010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7869240313327207e-06,
+      "loss": 1.8055,
+      "step": 2011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.779269050249738e-06,
+      "loss": 1.8702,
+      "step": 2012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.771628899457698e-06,
+      "loss": 1.5652,
+      "step": 2013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7640035927395217e-06,
+      "loss": 1.5194,
+      "step": 2014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7563931438513482e-06,
+      "loss": 1.2634,
+      "step": 2015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7487975665225077e-06,
+      "loss": 1.567,
+      "step": 2016
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7412168744555135e-06,
+      "loss": 1.6862,
+      "step": 2017
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7336510813260132e-06,
+      "loss": 1.6226,
+      "step": 2018
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7261002007827886e-06,
+      "loss": 1.2924,
+      "step": 2019
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.718564246447707e-06,
+      "loss": 1.4376,
+      "step": 2020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.71104323191572e-06,
+      "loss": 1.6592,
+      "step": 2021
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.703537170754821e-06,
+      "loss": 1.7992,
+      "step": 2022
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6960460765060238e-06,
+      "loss": 1.5656,
+      "step": 2023
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6885699626833541e-06,
+      "loss": 1.4141,
+      "step": 2024
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6811088427737976e-06,
+      "loss": 1.4633,
+      "step": 2025
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6736627302373055e-06,
+      "loss": 1.4501,
+      "step": 2026
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6662316385067434e-06,
+      "loss": 1.4007,
+      "step": 2027
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6588155809878847e-06,
+      "loss": 1.4313,
+      "step": 2028
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6514145710593842e-06,
+      "loss": 1.5618,
+      "step": 2029
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.644028622072742e-06,
+      "loss": 1.5912,
+      "step": 2030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6366577473522972e-06,
+      "loss": 1.5509,
+      "step": 2031
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6293019601951898e-06,
+      "loss": 1.6332,
+      "step": 2032
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6219612738713374e-06,
+      "loss": 1.4143,
+      "step": 2033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6146357016234283e-06,
+      "loss": 1.6039,
+      "step": 2034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6073252566668697e-06,
+      "loss": 1.7688,
+      "step": 2035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6000299521897933e-06,
+      "loss": 1.512,
+      "step": 2036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5927498013530062e-06,
+      "loss": 1.5758,
+      "step": 2037
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5854848172899817e-06,
+      "loss": 1.5803,
+      "step": 2038
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5782350131068359e-06,
+      "loss": 1.6576,
+      "step": 2039
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5710004018822921e-06,
+      "loss": 1.399,
+      "step": 2040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5637809966676765e-06,
+      "loss": 1.6549,
+      "step": 2041
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5565768104868728e-06,
+      "loss": 1.6819,
+      "step": 2042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5493878563363129e-06,
+      "loss": 1.7183,
+      "step": 2043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.542214147184955e-06,
+      "loss": 1.6696,
+      "step": 2044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.535055695974247e-06,
+      "loss": 1.6549,
+      "step": 2045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5279125156181185e-06,
+      "loss": 1.6603,
+      "step": 2046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.520784619002943e-06,
+      "loss": 1.5368,
+      "step": 2047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5136720189875309e-06,
+      "loss": 1.5132,
+      "step": 2048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5065747284030906e-06,
+      "loss": 1.388,
+      "step": 2049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4994927600532106e-06,
+      "loss": 1.626,
+      "step": 2050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4924261267138463e-06,
+      "loss": 1.3721,
+      "step": 2051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4853748411332803e-06,
+      "loss": 1.4854,
+      "step": 2052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4783389160321138e-06,
+      "loss": 1.568,
+      "step": 2053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4713183641032335e-06,
+      "loss": 1.3144,
+      "step": 2054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4643131980117908e-06,
+      "loss": 1.3429,
+      "step": 2055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4573234303951888e-06,
+      "loss": 1.4453,
+      "step": 2056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4503490738630433e-06,
+      "loss": 1.5241,
+      "step": 2057
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.443390140997174e-06,
+      "loss": 1.6426,
+      "step": 2058
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4364466443515745e-06,
+      "loss": 1.2128,
+      "step": 2059
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.429518596452386e-06,
+      "loss": 1.7411,
+      "step": 2060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.422606009797891e-06,
+      "loss": 1.4215,
+      "step": 2061
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4157088968584687e-06,
+      "loss": 1.3644,
+      "step": 2062
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4088272700765915e-06,
+      "loss": 1.4464,
+      "step": 2063
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.40196114186679e-06,
+      "loss": 1.6706,
+      "step": 2064
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3951105246156348e-06,
+      "loss": 1.4353,
+      "step": 2065
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3882754306817191e-06,
+      "loss": 1.7673,
+      "step": 2066
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3814558723956273e-06,
+      "loss": 1.5525,
+      "step": 2067
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3746518620599215e-06,
+      "loss": 1.7528,
+      "step": 2068
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3678634119491108e-06,
+      "loss": 1.5567,
+      "step": 2069
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3610905343096347e-06,
+      "loss": 1.509,
+      "step": 2070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3543332413598432e-06,
+      "loss": 1.3606,
+      "step": 2071
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3475915452899645e-06,
+      "loss": 1.6173,
+      "step": 2072
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3408654582620995e-06,
+      "loss": 1.5706,
+      "step": 2073
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3341549924101794e-06,
+      "loss": 1.6493,
+      "step": 2074
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3274601598399638e-06,
+      "loss": 1.6605,
+      "step": 2075
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3207809726290055e-06,
+      "loss": 1.7615,
+      "step": 2076
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.314117442826629e-06,
+      "loss": 1.6245,
+      "step": 2077
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.307469582453923e-06,
+      "loss": 1.6912,
+      "step": 2078
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3008374035036963e-06,
+      "loss": 1.4867,
+      "step": 2079
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2942209179404797e-06,
+      "loss": 1.5677,
+      "step": 2080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2876201377004849e-06,
+      "loss": 1.3101,
+      "step": 2081
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2810350746915923e-06,
+      "loss": 1.3102,
+      "step": 2082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2744657407933313e-06,
+      "loss": 1.6354,
+      "step": 2083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2679121478568523e-06,
+      "loss": 1.7498,
+      "step": 2084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2613743077049146e-06,
+      "loss": 1.4503,
+      "step": 2085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2548522321318513e-06,
+      "loss": 1.554,
+      "step": 2086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2483459329035596e-06,
+      "loss": 1.5646,
+      "step": 2087
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2418554217574796e-06,
+      "loss": 1.5132,
+      "step": 2088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2353807104025628e-06,
+      "loss": 1.4132,
+      "step": 2089
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2289218105192646e-06,
+      "loss": 1.4698,
+      "step": 2090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2224787337595124e-06,
+      "loss": 1.2756,
+      "step": 2091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2160514917466847e-06,
+      "loss": 1.6272,
+      "step": 2092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2096400960756038e-06,
+      "loss": 1.4625,
+      "step": 2093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2032445583124953e-06,
+      "loss": 1.4888,
+      "step": 2094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1968648899949841e-06,
+      "loss": 1.3555,
+      "step": 2095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1905011026320634e-06,
+      "loss": 1.2659,
+      "step": 2096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1841532077040741e-06,
+      "loss": 1.5633,
+      "step": 2097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1778212166626934e-06,
+      "loss": 1.6282,
+      "step": 2098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1715051409309008e-06,
+      "loss": 1.6004,
+      "step": 2099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1652049919029717e-06,
+      "loss": 1.3732,
+      "step": 2100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1589207809444413e-06,
+      "loss": 1.45,
+      "step": 2101
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1526525193921024e-06,
+      "loss": 1.3553,
+      "step": 2102
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1464002185539647e-06,
+      "loss": 1.5483,
+      "step": 2103
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.140163889709248e-06,
+      "loss": 1.6369,
+      "step": 2104
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1339435441083656e-06,
+      "loss": 1.3021,
+      "step": 2105
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1277391929728844e-06,
+      "loss": 1.7775,
+      "step": 2106
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1215508474955283e-06,
+      "loss": 1.8968,
+      "step": 2107
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.11537851884014e-06,
+      "loss": 1.4951,
+      "step": 2108
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1092222181416678e-06,
+      "loss": 1.299,
+      "step": 2109
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1030819565061512e-06,
+      "loss": 1.5983,
+      "step": 2110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0969577450106872e-06,
+      "loss": 1.6404,
+      "step": 2111
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0908495947034258e-06,
+      "loss": 1.3522,
+      "step": 2112
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0847575166035385e-06,
+      "loss": 1.5463,
+      "step": 2113
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.078681521701197e-06,
+      "loss": 1.5545,
+      "step": 2114
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0726216209575714e-06,
+      "loss": 1.6517,
+      "step": 2115
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0665778253047864e-06,
+      "loss": 1.3116,
+      "step": 2116
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.06055014564592e-06,
+      "loss": 1.3956,
+      "step": 2117
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0545385928549701e-06,
+      "loss": 1.5204,
+      "step": 2118
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0485431777768506e-06,
+      "loss": 1.7434,
+      "step": 2119
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0425639112273545e-06,
+      "loss": 1.5303,
+      "step": 2120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.036600803993144e-06,
+      "loss": 1.5414,
+      "step": 2121
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0306538668317356e-06,
+      "loss": 1.4178,
+      "step": 2122
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0247231104714673e-06,
+      "loss": 1.2406,
+      "step": 2123
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0188085456114937e-06,
+      "loss": 1.6108,
+      "step": 2124
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0129101829217548e-06,
+      "loss": 1.2746,
+      "step": 2125
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0070280330429594e-06,
+      "loss": 1.5696,
+      "step": 2126
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0011621065865773e-06,
+      "loss": 1.5791,
+      "step": 2127
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.953124141348025e-07,
+      "loss": 1.5781,
+      "step": 2128
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89478966240548e-07,
+      "loss": 1.5212,
+      "step": 2129
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.836617734274178e-07,
+      "loss": 1.6055,
+      "step": 2130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.778608461896922e-07,
+      "loss": 1.6694,
+      "step": 2131
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.720761949923118e-07,
+      "loss": 1.5583,
+      "step": 2132
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.663078302708496e-07,
+      "loss": 1.5311,
+      "step": 2133
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.605557624315031e-07,
+      "loss": 1.614,
+      "step": 2134
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.548200018510633e-07,
+      "loss": 1.5797,
+      "step": 2135
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.49100558876912e-07,
+      "loss": 1.7868,
+      "step": 2136
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.433974438269855e-07,
+      "loss": 1.7742,
+      "step": 2137
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.377106669897651e-07,
+      "loss": 1.335,
+      "step": 2138
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.320402386242644e-07,
+      "loss": 1.8547,
+      "step": 2139
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.263861689599963e-07,
+      "loss": 1.4551,
+      "step": 2140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.207484681969691e-07,
+      "loss": 1.6797,
+      "step": 2141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.151271465056577e-07,
+      "loss": 1.1905,
+      "step": 2142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.095222140269855e-07,
+      "loss": 1.7518,
+      "step": 2143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.039336808723198e-07,
+      "loss": 1.3489,
+      "step": 2144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.983615571234317e-07,
+      "loss": 1.6033,
+      "step": 2145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.928058528325001e-07,
+      "loss": 1.2979,
+      "step": 2146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.872665780220724e-07,
+      "loss": 1.5736,
+      "step": 2147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.817437426850683e-07,
+      "loss": 1.6443,
+      "step": 2148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.762373567847416e-07,
+      "loss": 1.4559,
+      "step": 2149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.707474302546726e-07,
+      "loss": 1.5841,
+      "step": 2150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.652739729987536e-07,
+      "loss": 1.4749,
+      "step": 2151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.598169948911605e-07,
+      "loss": 1.7087,
+      "step": 2152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.543765057763465e-07,
+      "loss": 1.6256,
+      "step": 2153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.489525154690126e-07,
+      "loss": 1.5224,
+      "step": 2154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.435450337540974e-07,
+      "loss": 1.8942,
+      "step": 2155
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.381540703867608e-07,
+      "loss": 1.3847,
+      "step": 2156
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.327796350923589e-07,
+      "loss": 1.4485,
+      "step": 2157
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.274217375664362e-07,
+      "loss": 1.6773,
+      "step": 2158
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.220803874746964e-07,
+      "loss": 1.5805,
+      "step": 2159
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.167555944529993e-07,
+      "loss": 1.5481,
+      "step": 2160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.114473681073287e-07,
+      "loss": 1.4651,
+      "step": 2161
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.061557180137835e-07,
+      "loss": 1.6206,
+      "step": 2162
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.008806537185632e-07,
+      "loss": 1.4866,
+      "step": 2163
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.956221847379387e-07,
+      "loss": 1.1799,
+      "step": 2164
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.903803205582516e-07,
+      "loss": 1.3404,
+      "step": 2165
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.851550706358823e-07,
+      "loss": 1.5743,
+      "step": 2166
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.799464443972371e-07,
+      "loss": 1.4919,
+      "step": 2167
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.747544512387406e-07,
+      "loss": 1.4656,
+      "step": 2168
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.695791005268039e-07,
+      "loss": 1.4188,
+      "step": 2169
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.644204015978208e-07,
+      "loss": 1.5374,
+      "step": 2170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.592783637581369e-07,
+      "loss": 1.3037,
+      "step": 2171
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.541529962840522e-07,
+      "loss": 1.604,
+      "step": 2172
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.490443084217847e-07,
+      "loss": 1.4969,
+      "step": 2173
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.439523093874623e-07,
+      "loss": 1.4612,
+      "step": 2174
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.388770083671126e-07,
+      "loss": 1.6489,
+      "step": 2175
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.338184145166316e-07,
+      "loss": 1.1595,
+      "step": 2176
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.287765369617838e-07,
+      "loss": 1.4581,
+      "step": 2177
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.237513847981703e-07,
+      "loss": 1.5515,
+      "step": 2178
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.187429670912194e-07,
+      "loss": 1.6839,
+      "step": 2179
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.137512928761792e-07,
+      "loss": 1.7685,
+      "step": 2180
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.087763711580797e-07,
+      "loss": 1.5928,
+      "step": 2181
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.038182109117397e-07,
+      "loss": 1.4472,
+      "step": 2182
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.988768210817343e-07,
+      "loss": 1.8424,
+      "step": 2183
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.939522105823881e-07,
+      "loss": 1.6809,
+      "step": 2184
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.890443882977515e-07,
+      "loss": 1.822,
+      "step": 2185
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.8415336308159e-07,
+      "loss": 1.4371,
+      "step": 2186
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.79279143757372e-07,
+      "loss": 1.6891,
+      "step": 2187
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.744217391182406e-07,
+      "loss": 1.5112,
+      "step": 2188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.695811579270095e-07,
+      "loss": 1.3585,
+      "step": 2189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.647574089161424e-07,
+      "loss": 1.4165,
+      "step": 2190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.599505007877338e-07,
+      "loss": 1.4091,
+      "step": 2191
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.551604422135038e-07,
+      "loss": 1.5854,
+      "step": 2192
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.503872418347678e-07,
+      "loss": 1.5106,
+      "step": 2193
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.456309082624379e-07,
+      "loss": 1.6728,
+      "step": 2194
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.408914500769869e-07,
+      "loss": 1.647,
+      "step": 2195
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.361688758284568e-07,
+      "loss": 1.2241,
+      "step": 2196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.314631940364213e-07,
+      "loss": 1.534,
+      "step": 2197
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.267744131899822e-07,
+      "loss": 1.5349,
+      "step": 2198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.221025417477566e-07,
+      "loss": 1.8409,
+      "step": 2199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.174475881378483e-07,
+      "loss": 1.4193,
+      "step": 2200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.128095607578499e-07,
+      "loss": 1.5929,
+      "step": 2201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.081884679748141e-07,
+      "loss": 1.6358,
+      "step": 2202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.03584318125241e-07,
+      "loss": 1.9462,
+      "step": 2203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.989971195150723e-07,
+      "loss": 1.4787,
+      "step": 2204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.94426880419664e-07,
+      "loss": 1.5132,
+      "step": 2205
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.898736090837798e-07,
+      "loss": 1.4963,
+      "step": 2206
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.853373137215723e-07,
+      "loss": 1.3411,
+      "step": 2207
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.808180025165677e-07,
+      "loss": 1.3032,
+      "step": 2208
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.763156836216555e-07,
+      "loss": 1.4639,
+      "step": 2209
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.718303651590685e-07,
+      "loss": 1.9188,
+      "step": 2210
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.67362055220374e-07,
+      "loss": 1.6311,
+      "step": 2211
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.629107618664498e-07,
+      "loss": 1.6744,
+      "step": 2212
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.584764931274822e-07,
+      "loss": 1.6693,
+      "step": 2213
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.540592570029412e-07,
+      "loss": 1.5022,
+      "step": 2214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.496590614615682e-07,
+      "loss": 1.6234,
+      "step": 2215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.452759144413678e-07,
+      "loss": 1.3389,
+      "step": 2216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.409098238495836e-07,
+      "loss": 1.4389,
+      "step": 2217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.365607975626963e-07,
+      "loss": 1.471,
+      "step": 2218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.322288434263956e-07,
+      "loss": 1.4515,
+      "step": 2219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.279139692555735e-07,
+      "loss": 1.4639,
+      "step": 2220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.236161828343156e-07,
+      "loss": 1.5227,
+      "step": 2221
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.193354919158733e-07,
+      "loss": 1.3964,
+      "step": 2222
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.150719042226648e-07,
+      "loss": 1.5958,
+      "step": 2223
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.108254274462454e-07,
+      "loss": 1.3214,
+      "step": 2224
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.065960692473116e-07,
+      "loss": 1.6546,
+      "step": 2225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.02383837255671e-07,
+      "loss": 1.7047,
+      "step": 2226
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981887390702356e-07,
+      "loss": 1.4745,
+      "step": 2227
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940107822590135e-07,
+      "loss": 1.4574,
+      "step": 2228
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898499743590812e-07,
+      "loss": 1.404,
+      "step": 2229
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.857063228765868e-07,
+      "loss": 1.5467,
+      "step": 2230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.81579835286723e-07,
+      "loss": 1.359,
+      "step": 2231
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.774705190337181e-07,
+      "loss": 1.6816,
+      "step": 2232
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7337838153082684e-07,
+      "loss": 1.7762,
+      "step": 2233
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6930343016030946e-07,
+      "loss": 1.315,
+      "step": 2234
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6524567227342667e-07,
+      "loss": 1.6048,
+      "step": 2235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.612051151904151e-07,
+      "loss": 1.4604,
+      "step": 2236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.571817662004896e-07,
+      "loss": 1.372,
+      "step": 2237
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5317563256181416e-07,
+      "loss": 1.3716,
+      "step": 2238
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.491867215014989e-07,
+      "loss": 1.7412,
+      "step": 2239
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4521504021558637e-07,
+      "loss": 1.4137,
+      "step": 2240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.412605958690308e-07,
+      "loss": 1.561,
+      "step": 2241
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.373233955956968e-07,
+      "loss": 1.3989,
+      "step": 2242
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3340344649833834e-07,
+      "loss": 1.4222,
+      "step": 2243
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2950075564858415e-07,
+      "loss": 1.5909,
+      "step": 2244
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.256153300869359e-07,
+      "loss": 1.5047,
+      "step": 2245
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.217471768227432e-07,
+      "loss": 1.7538,
+      "step": 2246
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.178963028341987e-07,
+      "loss": 1.3186,
+      "step": 2247
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1406271506832183e-07,
+      "loss": 1.5048,
+      "step": 2248
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.102464204409473e-07,
+      "loss": 1.5036,
+      "step": 2249
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.064474258367157e-07,
+      "loss": 1.2821,
+      "step": 2250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0266573810905373e-07,
+      "loss": 1.3534,
+      "step": 2251
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.989013640801698e-07,
+      "loss": 1.5509,
+      "step": 2252
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9515431054103605e-07,
+      "loss": 1.4191,
+      "step": 2253
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9142458425137974e-07,
+      "loss": 1.2442,
+      "step": 2254
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.877121919396687e-07,
+      "loss": 1.6629,
+      "step": 2255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.840171403031001e-07,
+      "loss": 1.4533,
+      "step": 2256
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.803394360075896e-07,
+      "loss": 1.4371,
+      "step": 2257
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.766790856877567e-07,
+      "loss": 1.4733,
+      "step": 2258
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.730360959469126e-07,
+      "loss": 1.676,
+      "step": 2259
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.694104733570547e-07,
+      "loss": 1.5964,
+      "step": 2260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.658022244588444e-07,
+      "loss": 1.419,
+      "step": 2261
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.622113557616058e-07,
+      "loss": 1.6899,
+      "step": 2262
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.586378737433027e-07,
+      "loss": 1.5135,
+      "step": 2263
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5508178485054055e-07,
+      "loss": 1.5646,
+      "step": 2264
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5154309549854214e-07,
+      "loss": 1.6594,
+      "step": 2265
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4802181207114094e-07,
+      "loss": 1.713,
+      "step": 2266
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.445179409207744e-07,
+      "loss": 1.6405,
+      "step": 2267
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.41031488368464e-07,
+      "loss": 1.6267,
+      "step": 2268
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3756246070381084e-07,
+      "loss": 1.3974,
+      "step": 2269
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3411086418497884e-07,
+      "loss": 1.8485,
+      "step": 2270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.306767050386861e-07,
+      "loss": 1.6679,
+      "step": 2271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2725998946019575e-07,
+      "loss": 1.3318,
+      "step": 2272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2386072361329846e-07,
+      "loss": 1.6987,
+      "step": 2273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2047891363031104e-07,
+      "loss": 1.4544,
+      "step": 2274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.171145656120567e-07,
+      "loss": 1.5146,
+      "step": 2275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1376768562785375e-07,
+      "loss": 1.6514,
+      "step": 2276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1043827971551364e-07,
+      "loss": 1.5906,
+      "step": 2277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.071263538813218e-07,
+      "loss": 1.6442,
+      "step": 2278
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.038319141000301e-07,
+      "loss": 1.5862,
+      "step": 2279
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.0055496631484436e-07,
+      "loss": 1.4047,
+      "step": 2280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9729551643741473e-07,
+      "loss": 1.659,
+      "step": 2281
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9405357034782534e-07,
+      "loss": 1.5973,
+      "step": 2282
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9082913389458347e-07,
+      "loss": 1.374,
+      "step": 2283
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.876222128946105e-07,
+      "loss": 1.1938,
+      "step": 2284
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8443281313322637e-07,
+      "loss": 1.5455,
+      "step": 2285
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.812609403641431e-07,
+      "loss": 1.4637,
+      "step": 2286
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.781066003094557e-07,
+      "loss": 1.8138,
+      "step": 2287
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.749697986596289e-07,
+      "loss": 1.5964,
+      "step": 2288
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.718505410734895e-07,
+      "loss": 1.1939,
+      "step": 2289
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.687488331782118e-07,
+      "loss": 1.5828,
+      "step": 2290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6566468056931196e-07,
+      "loss": 1.5515,
+      "step": 2291
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.625980888106361e-07,
+      "loss": 1.2337,
+      "step": 2292
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.595490634343489e-07,
+      "loss": 1.6168,
+      "step": 2293
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5651760994092833e-07,
+      "loss": 1.5219,
+      "step": 2294
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.535037337991486e-07,
+      "loss": 1.2367,
+      "step": 2295
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5050744044607944e-07,
+      "loss": 1.5724,
+      "step": 2296
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4752873528706366e-07,
+      "loss": 1.5514,
+      "step": 2297
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.445676236957206e-07,
+      "loss": 1.7009,
+      "step": 2298
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4162411101392946e-07,
+      "loss": 1.6117,
+      "step": 2299
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.386982025518192e-07,
+      "loss": 1.4597,
+      "step": 2300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.35789903587762e-07,
+      "loss": 1.5742,
+      "step": 2301
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3289921936836214e-07,
+      "loss": 1.5487,
+      "step": 2302
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3002615510844484e-07,
+      "loss": 1.5736,
+      "step": 2303
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2717071599105188e-07,
+      "loss": 1.8752,
+      "step": 2304
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2433290716742716e-07,
+      "loss": 1.3747,
+      "step": 2305
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2151273375700999e-07,
+      "loss": 1.6527,
+      "step": 2306
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.187102008474251e-07,
+      "loss": 1.3545,
+      "step": 2307
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1592531349447054e-07,
+      "loss": 1.6654,
+      "step": 2308
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1315807672211752e-07,
+      "loss": 1.7542,
+      "step": 2309
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1040849552249165e-07,
+      "loss": 1.7238,
+      "step": 2310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0767657485586735e-07,
+      "loss": 1.6216,
+      "step": 2311
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0496231965066117e-07,
+      "loss": 1.4991,
+      "step": 2312
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.022657348034185e-07,
+      "loss": 1.6318,
+      "step": 2313
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9958682517881133e-07,
+      "loss": 1.242,
+      "step": 2314
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9692559560961944e-07,
+      "loss": 1.686,
+      "step": 2315
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9428205089673357e-07,
+      "loss": 1.5774,
+      "step": 2316
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9165619580913563e-07,
+      "loss": 1.5778,
+      "step": 2317
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8904803508389968e-07,
+      "loss": 1.5186,
+      "step": 2318
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8645757342617753e-07,
+      "loss": 1.3415,
+      "step": 2319
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.838848155091877e-07,
+      "loss": 1.5687,
+      "step": 2320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.813297659742186e-07,
+      "loss": 1.5469,
+      "step": 2321
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7879242943060426e-07,
+      "loss": 1.6031,
+      "step": 2322
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7627281045573208e-07,
+      "loss": 1.4229,
+      "step": 2323
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.737709135950205e-07,
+      "loss": 1.6439,
+      "step": 2324
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.712867433619192e-07,
+      "loss": 1.4898,
+      "step": 2325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6882030423790108e-07,
+      "loss": 1.3178,
+      "step": 2326
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6637160067244917e-07,
+      "loss": 1.2935,
+      "step": 2327
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.639406370830521e-07,
+      "loss": 1.5117,
+      "step": 2328
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6152741785519622e-07,
+      "loss": 1.5341,
+      "step": 2329
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.591319473423536e-07,
+      "loss": 1.5435,
+      "step": 2330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.56754229865983e-07,
+      "loss": 1.4882,
+      "step": 2331
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5439426971551207e-07,
+      "loss": 1.8462,
+      "step": 2332
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.520520711483353e-07,
+      "loss": 1.407,
+      "step": 2333
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4972763838980608e-07,
+      "loss": 1.3782,
+      "step": 2334
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4742097563322565e-07,
+      "loss": 1.7772,
+      "step": 2335
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4513208703984094e-07,
+      "loss": 1.4266,
+      "step": 2336
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4286097673883226e-07,
+      "loss": 1.3279,
+      "step": 2337
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4060764882730783e-07,
+      "loss": 1.4482,
+      "step": 2338
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3837210737029706e-07,
+      "loss": 1.4172,
+      "step": 2339
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.361543564007406e-07,
+      "loss": 1.5208,
+      "step": 2340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3395439991948812e-07,
+      "loss": 1.6325,
+      "step": 2341
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.317722418952838e-07,
+      "loss": 1.5207,
+      "step": 2342
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.296078862647676e-07,
+      "loss": 1.5636,
+      "step": 2343
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2746133693245842e-07,
+      "loss": 1.1967,
+      "step": 2344
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2533259777075757e-07,
+      "loss": 1.729,
+      "step": 2345
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.232216726199331e-07,
+      "loss": 1.7718,
+      "step": 2346
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.211285652881189e-07,
+      "loss": 1.5244,
+      "step": 2347
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.190532795513033e-07,
+      "loss": 1.6034,
+      "step": 2348
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.169958191533238e-07,
+      "loss": 1.5455,
+      "step": 2349
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1495618780586581e-07,
+      "loss": 1.5076,
+      "step": 2350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1293438918844379e-07,
+      "loss": 1.6216,
+      "step": 2351
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1093042694840572e-07,
+      "loss": 1.3945,
+      "step": 2352
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.089443047009231e-07,
+      "loss": 1.5192,
+      "step": 2353
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0697602602898093e-07,
+      "loss": 1.4599,
+      "step": 2354
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.050255944833789e-07,
+      "loss": 1.9078,
+      "step": 2355
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0309301358271572e-07,
+      "loss": 1.4995,
+      "step": 2356
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0117828681338815e-07,
+      "loss": 1.7482,
+      "step": 2357
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.928141762958643e-08,
+      "loss": 1.6553,
+      "step": 2358
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.740240945328217e-08,
+      "loss": 1.5925,
+      "step": 2359
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.554126567422939e-08,
+      "loss": 1.7787,
+      "step": 2360
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.369798964995125e-08,
+      "loss": 1.4239,
+      "step": 2361
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.187258470573667e-08,
+      "loss": 1.3922,
+      "step": 2362
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.006505413464039e-08,
+      "loss": 1.3326,
+      "step": 2363
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.827540119746403e-08,
+      "loss": 1.8241,
+      "step": 2364
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.650362912276388e-08,
+      "loss": 1.5988,
+      "step": 2365
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.474974110683653e-08,
+      "loss": 1.4932,
+      "step": 2366
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.301374031371545e-08,
+      "loss": 1.494,
+      "step": 2367
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.129562987516549e-08,
+      "loss": 1.2914,
+      "step": 2368
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.959541289067618e-08,
+      "loss": 1.3611,
+      "step": 2369
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.791309242745959e-08,
+      "loss": 1.7675,
+      "step": 2370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.624867152043913e-08,
+      "loss": 1.4249,
+      "step": 2371
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.460215317225072e-08,
+      "loss": 1.283,
+      "step": 2372
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.297354035323057e-08,
+      "loss": 1.5511,
+      "step": 2373
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.136283600141514e-08,
+      "loss": 1.5122,
+      "step": 2374
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.977004302253231e-08,
+      "loss": 1.5209,
+      "step": 2375
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.819516428999917e-08,
+      "loss": 1.7948,
+      "step": 2376
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.663820264491528e-08,
+      "loss": 1.497,
+      "step": 2377
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.509916089605606e-08,
+      "loss": 1.8689,
+      "step": 2378
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.357804181986948e-08,
+      "loss": 1.3113,
+      "step": 2379
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.207484816047382e-08,
+      "loss": 1.4819,
+      "step": 2380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.058958262964432e-08,
+      "loss": 1.6037,
+      "step": 2381
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.912224790681875e-08,
+      "loss": 1.4453,
+      "step": 2382
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.767284663908412e-08,
+      "loss": 1.6482,
+      "step": 2383
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6241381441177745e-08,
+      "loss": 1.7168,
+      "step": 2384
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4827854895478374e-08,
+      "loss": 1.6233,
+      "step": 2385
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.343226955200287e-08,
+      "loss": 1.6044,
+      "step": 2386
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.20546279284051e-08,
+      "loss": 1.4804,
+      "step": 2387
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0694932509964825e-08,
+      "loss": 1.5586,
+      "step": 2388
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9353185749586586e-08,
+      "loss": 1.5941,
+      "step": 2389
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8029390067798606e-08,
+      "loss": 1.5264,
+      "step": 2390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.672354785274169e-08,
+      "loss": 1.4504,
+      "step": 2391
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.543566146017031e-08,
+      "loss": 1.5746,
+      "step": 2392
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.416573321344708e-08,
+      "loss": 1.5459,
+      "step": 2393
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.291376540353498e-08,
+      "loss": 1.6341,
+      "step": 2394
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.167976028899956e-08,
+      "loss": 1.6802,
+      "step": 2395
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.046372009599897e-08,
+      "loss": 1.4002,
+      "step": 2396
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9265647018283946e-08,
+      "loss": 1.3202,
+      "step": 2397
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.808554321719116e-08,
+      "loss": 1.2845,
+      "step": 2398
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.692341082164208e-08,
+      "loss": 1.4793,
+      "step": 2399
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.577925192813525e-08,
+      "loss": 1.5024,
+      "step": 2400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.465306860074624e-08,
+      "loss": 1.3246,
+      "step": 2401
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3544862871124354e-08,
+      "loss": 1.4726,
+      "step": 2402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.245463673848259e-08,
+      "loss": 1.8418,
+      "step": 2403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.138239216960326e-08,
+      "loss": 1.5771,
+      "step": 2404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.032813109882682e-08,
+      "loss": 1.287,
+      "step": 2405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9291855428051907e-08,
+      "loss": 1.7108,
+      "step": 2406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8273567026732006e-08,
+      "loss": 1.5953,
+      "step": 2407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.727326773186989e-08,
+      "loss": 1.7344,
+      "step": 2408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.629095934801651e-08,
+      "loss": 1.6269,
+      "step": 2409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5326643647268777e-08,
+      "loss": 1.6917,
+      "step": 2410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.43803223692618e-08,
+      "loss": 1.5118,
+      "step": 2411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3451997221171085e-08,
+      "loss": 1.2571,
+      "step": 2412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2541669877703677e-08,
+      "loss": 1.5099,
+      "step": 2413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1649341981101468e-08,
+      "loss": 1.5655,
+      "step": 2414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.077501514113456e-08,
+      "loss": 1.5892,
+      "step": 2415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9918690935097906e-08,
+      "loss": 1.7819,
+      "step": 2416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.908037090780912e-08,
+      "loss": 1.6177,
+      "step": 2417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8260056571607344e-08,
+      "loss": 1.4282,
+      "step": 2418
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.745774940634992e-08,
+      "loss": 1.3871,
+      "step": 2419
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.667345085940686e-08,
+      "loss": 1.4957,
+      "step": 2420
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.590716234566192e-08,
+      "loss": 1.5441,
+      "step": 2421
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5158885247508192e-08,
+      "loss": 1.5614,
+      "step": 2422
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.442862091484587e-08,
+      "loss": 1.4681,
+      "step": 2423
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3716370665081135e-08,
+      "loss": 1.5048,
+      "step": 2424
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3022135783118394e-08,
+      "loss": 1.7815,
+      "step": 2425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2345917521369154e-08,
+      "loss": 1.4894,
+      "step": 2426
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1687717099735374e-08,
+      "loss": 1.5056,
+      "step": 2427
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.104753570562056e-08,
+      "loss": 1.5098,
+      "step": 2428
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0425374493916452e-08,
+      "loss": 1.4588,
+      "step": 2429
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.821234587011896e-09,
+      "loss": 1.4183,
+      "step": 2430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.235117074779532e-09,
+      "loss": 1.8102,
+      "step": 2431
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.667023014583553e-09,
+      "loss": 1.4496,
+      "step": 2432
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.116953431270836e-09,
+      "loss": 1.4877,
+      "step": 2433
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.58490931717426e-09,
+      "loss": 1.4296,
+      "step": 2434
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.070891632106059e-09,
+      "loss": 1.7367,
+      "step": 2435
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.574901303361136e-09,
+      "loss": 1.5969,
+      "step": 2436
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.096939225711529e-09,
+      "loss": 1.7883,
+      "step": 2437
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.637006261405287e-09,
+      "loss": 1.5263,
+      "step": 2438
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.195103240167587e-09,
+      "loss": 1.2925,
+      "step": 2439
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.771230959195183e-09,
+      "loss": 1.4226,
+      "step": 2440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.365390183159734e-09,
+      "loss": 1.4501,
+      "step": 2441
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9775816442022555e-09,
+      "loss": 1.6656,
+      "step": 2442
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6078060419331152e-09,
+      "loss": 1.4476,
+      "step": 2443
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2560640434320388e-09,
+      "loss": 1.5475,
+      "step": 2444
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9223562832458863e-09,
+      "loss": 1.5669,
+      "step": 2445
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.606683363387541e-09,
+      "loss": 1.3882,
+      "step": 2446
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.309045853332581e-09,
+      "loss": 1.4476,
+      "step": 2447
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.029444290023719e-09,
+      "loss": 1.5478,
+      "step": 2448
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7678791778663606e-09,
+      "loss": 1.6366,
+      "step": 2449
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.524350988725276e-09,
+      "loss": 1.3639,
+      "step": 2450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2988601619279284e-09,
+      "loss": 1.5579,
+      "step": 2451
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0914071042633645e-09,
+      "loss": 1.5713,
+      "step": 2452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.019921899788842e-10,
+      "loss": 1.4798,
+      "step": 2453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.306157607811503e-10,
+      "loss": 1.4363,
+      "step": 2454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.772781258350791e-10,
+      "loss": 1.2962,
+      "step": 2455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.419795617627287e-10,
+      "loss": 1.2891,
+      "step": 2456
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2472031264552117e-10,
+      "loss": 1.6765,
+      "step": 2457
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2550059002091062e-10,
+      "loss": 1.8137,
+      "step": 2458
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4432057288016333e-10,
+      "loss": 1.745,
+      "step": 2459
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.118040767390867e-11,
+      "loss": 1.5827,
+      "step": 2460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.608020830880854e-11,
+      "loss": 1.2077,
+      "step": 2461
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.020056145336853e-12,
+      "loss": 1.3794,
+      "step": 2462
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.4622,
+      "step": 2463
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2463,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 3.421582143480922e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}