diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.19469457288878073,
+  "eval_steps": 500,
+  "global_step": 800,
+  "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": 1.311,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.2832,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838709677419355e-07,
+      "loss": 1.2261,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.3032,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 1.252,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.2441,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1290322580645162e-06,
+      "loss": 1.2739,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.3179,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4516129032258066e-06,
+      "loss": 1.2461,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.1616,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.774193548387097e-06,
+      "loss": 1.1387,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.1543,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.096774193548387e-06,
+      "loss": 1.0835,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.1289,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 1.0815,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.0825,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7419354838709676e-06,
+      "loss": 1.0557,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 1.0674,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0645161290322584e-06,
+      "loss": 1.019,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 1.0522,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3870967741935484e-06,
+      "loss": 1.0103,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 1.0093,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7096774193548392e-06,
+      "loss": 0.8994,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9849,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 0.9639,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9517,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35483870967742e-06,
+      "loss": 0.9009,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.908,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.67741935483871e-06,
+      "loss": 0.9124,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9053,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9194,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.957,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.322580645161291e-06,
+      "loss": 0.8726,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.9224,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 0.8953,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8489,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.967741935483872e-06,
+      "loss": 0.9111,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.896,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.290322580645162e-06,
+      "loss": 0.9209,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8921,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.612903225806452e-06,
+      "loss": 0.9292,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.9124,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.935483870967743e-06,
+      "loss": 0.8887,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.9114,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 0.8796,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8706,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.580645161290323e-06,
+      "loss": 0.896,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8657,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.903225806451613e-06,
+      "loss": 0.8921,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.9199,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.225806451612904e-06,
+      "loss": 0.8433,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.8726,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.548387096774194e-06,
+      "loss": 0.8569,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.8643,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 0.8623,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.8892,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.193548387096775e-06,
+      "loss": 0.8669,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.8547,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.516129032258065e-06,
+      "loss": 0.8167,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.8284,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.838709677419356e-06,
+      "loss": 0.8489,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8596,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0161290322580646e-05,
+      "loss": 0.8657,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.8491,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.8496,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.8535,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0806451612903225e-05,
+      "loss": 0.8354,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.8955,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1129032258064516e-05,
+      "loss": 0.8369,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.8071,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1451612903225808e-05,
+      "loss": 0.8564,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.8494,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1774193548387098e-05,
+      "loss": 0.8325,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.9136,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 0.8345,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.8711,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2419354838709679e-05,
+      "loss": 0.8435,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.863,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.274193548387097e-05,
+      "loss": 0.8496,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.8477,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3064516129032258e-05,
+      "loss": 0.853,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.8325,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3387096774193548e-05,
+      "loss": 0.844,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.865,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 0.8159,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.8796,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4032258064516131e-05,
+      "loss": 0.7776,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.8303,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4354838709677421e-05,
+      "loss": 0.833,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.9224,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4677419354838712e-05,
+      "loss": 0.8237,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.8433,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8037,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.844,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 0.8789,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.8167,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5645161290322583e-05,
+      "loss": 0.8455,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.813,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.596774193548387e-05,
+      "loss": 0.8354,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.8,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6290322580645164e-05,
+      "loss": 0.8293,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.8396,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6612903225806454e-05,
+      "loss": 0.8596,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.8562,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6935483870967744e-05,
+      "loss": 0.8418,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.8579,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7258064516129035e-05,
+      "loss": 0.8162,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.8076,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7580645161290325e-05,
+      "loss": 0.7876,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.8203,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7903225806451612e-05,
+      "loss": 0.8108,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.8516,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8225806451612903e-05,
+      "loss": 0.8621,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.8184,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.8345,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.8628,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8870967741935487e-05,
+      "loss": 0.8584,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.8274,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9193548387096777e-05,
+      "loss": 0.8335,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.856,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9516129032258068e-05,
+      "loss": 0.8469,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.8784,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9838709677419358e-05,
+      "loss": 0.8586,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.834,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996892486132e-05,
+      "loss": 0.8394,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998756994645e-05,
+      "loss": 0.8481,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999972032386757e-05,
+      "loss": 0.8301,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999950279816705e-05,
+      "loss": 0.8267,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999922312249814e-05,
+      "loss": 0.821,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999888129703464e-05,
+      "loss": 0.8386,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998477321989e-05,
+      "loss": 0.8186,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999801119761233e-05,
+      "loss": 0.8503,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999748292419428e-05,
+      "loss": 0.7981,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999689250206324e-05,
+      "loss": 0.843,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999962399315861e-05,
+      "loss": 0.772,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999552521316842e-05,
+      "loss": 0.8628,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999474834725443e-05,
+      "loss": 0.8774,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999390933432696e-05,
+      "loss": 0.876,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999300817490745e-05,
+      "loss": 0.8496,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992044869556e-05,
+      "loss": 0.8882,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999101941887125e-05,
+      "loss": 0.8464,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998993182349057e-05,
+      "loss": 0.8472,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998878208408986e-05,
+      "loss": 0.8699,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998757020138372e-05,
+      "loss": 0.8755,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998629617612537e-05,
+      "loss": 0.8293,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998496000910653e-05,
+      "loss": 0.8521,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999835617011577e-05,
+      "loss": 0.842,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998210125314793e-05,
+      "loss": 0.8374,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998057866598487e-05,
+      "loss": 0.8633,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999789939406148e-05,
+      "loss": 0.8237,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999773470780227e-05,
+      "loss": 0.8472,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99975638079232e-05,
+      "loss": 0.8779,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999738669453049e-05,
+      "loss": 0.8723,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997203367734218e-05,
+      "loss": 0.8474,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997013827648314e-05,
+      "loss": 0.8269,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999681807439059e-05,
+      "loss": 0.8213,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996616108082703e-05,
+      "loss": 0.8074,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640792885017e-05,
+      "loss": 0.8362,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996193536822378e-05,
+      "loss": 0.854,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995972932132575e-05,
+      "loss": 0.8347,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995746114917866e-05,
+      "loss": 0.8301,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995513085319217e-05,
+      "loss": 0.8638,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999527384348146e-05,
+      "loss": 0.8411,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999502838955328e-05,
+      "loss": 0.8098,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994776723687228e-05,
+      "loss": 0.8169,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994518846039715e-05,
+      "loss": 0.8376,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999425475677102e-05,
+      "loss": 0.8689,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993984456045264e-05,
+      "loss": 0.7979,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993707944030448e-05,
+      "loss": 0.8376,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993425220898416e-05,
+      "loss": 0.8633,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993136286824894e-05,
+      "loss": 0.8442,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992841141989446e-05,
+      "loss": 0.8423,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992539786575505e-05,
+      "loss": 0.8428,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999223222077037e-05,
+      "loss": 0.8823,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999191844476519e-05,
+      "loss": 0.8528,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999159845875498e-05,
+      "loss": 0.8699,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991272262938606e-05,
+      "loss": 0.8064,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999093985751881e-05,
+      "loss": 0.8557,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990601242702174e-05,
+      "loss": 0.812,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990256418699154e-05,
+      "loss": 0.8242,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989905385724052e-05,
+      "loss": 0.793,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989548143995042e-05,
+      "loss": 0.8489,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998918469373415e-05,
+      "loss": 0.8633,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998881503516726e-05,
+      "loss": 0.8545,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998843916852412e-05,
+      "loss": 0.8562,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988057094038325e-05,
+      "loss": 0.9041,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998766881194734e-05,
+      "loss": 0.823,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987274322492477e-05,
+      "loss": 0.8452,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986873625918923e-05,
+      "loss": 0.832,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986466722475705e-05,
+      "loss": 0.8093,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986053612415717e-05,
+      "loss": 0.8174,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985634295995708e-05,
+      "loss": 0.802,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998520877347628e-05,
+      "loss": 0.8696,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99847770451219e-05,
+      "loss": 0.8584,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998433911120089e-05,
+      "loss": 0.8223,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983894971985425e-05,
+      "loss": 0.8301,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998344462775154e-05,
+      "loss": 0.8638,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982988078779125e-05,
+      "loss": 0.8823,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982525325351924e-05,
+      "loss": 0.822,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998205636775754e-05,
+      "loss": 0.8101,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981581206287437e-05,
+      "loss": 0.8235,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998109984123692e-05,
+      "loss": 0.8438,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980612272905165e-05,
+      "loss": 0.8481,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99801185015952e-05,
+      "loss": 0.8281,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99796185276139e-05,
+      "loss": 0.8618,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979112351271996e-05,
+      "loss": 0.8311,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978599972884085e-05,
+      "loss": 0.8384,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978081392768614e-05,
+      "loss": 0.8364,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977556611247872e-05,
+      "loss": 0.8564,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977025628648023e-05,
+      "loss": 0.8059,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976488445299066e-05,
+      "loss": 0.8147,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975945061534866e-05,
+      "loss": 0.8037,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997539547769314e-05,
+      "loss": 0.8123,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974839694115444e-05,
+      "loss": 0.8279,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974277711147216e-05,
+      "loss": 0.8342,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973709529137717e-05,
+      "loss": 0.8604,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973135148440077e-05,
+      "loss": 0.8567,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997255456941128e-05,
+      "loss": 0.833,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971967792412154e-05,
+      "loss": 0.8264,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997137481780738e-05,
+      "loss": 0.854,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970775645965493e-05,
+      "loss": 0.8601,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997017027725889e-05,
+      "loss": 0.8354,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969558712063795e-05,
+      "loss": 0.8306,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968940950760308e-05,
+      "loss": 0.8193,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968316993732366e-05,
+      "loss": 0.8445,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967686841367757e-05,
+      "loss": 0.8882,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996705049405813e-05,
+      "loss": 0.8306,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996640795219897e-05,
+      "loss": 0.8044,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965759216189617e-05,
+      "loss": 0.8481,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996510428643327e-05,
+      "loss": 0.812,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996444316333696e-05,
+      "loss": 0.842,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996377584731159e-05,
+      "loss": 0.7874,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996310233877189e-05,
+      "loss": 0.895,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996242263813644e-05,
+      "loss": 0.9019,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996173674582769e-05,
+      "loss": 0.8486,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961044662271922e-05,
+      "loss": 0.802,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960346387899255e-05,
+      "loss": 0.7998,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959641923143686e-05,
+      "loss": 0.8853,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958931268443028e-05,
+      "loss": 0.8269,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958214424238962e-05,
+      "loss": 0.8513,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957491390977007e-05,
+      "loss": 0.8337,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956762169106527e-05,
+      "loss": 0.8472,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956026759080744e-05,
+      "loss": 0.7981,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955285161356708e-05,
+      "loss": 0.8152,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954537376395327e-05,
+      "loss": 0.843,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953783404661356e-05,
+      "loss": 0.824,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953023246623383e-05,
+      "loss": 0.8896,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952256902753856e-05,
+      "loss": 0.8318,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995148437352905e-05,
+      "loss": 0.8088,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950705659429107e-05,
+      "loss": 0.8259,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994992076093799e-05,
+      "loss": 0.7646,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949129678543516e-05,
+      "loss": 0.863,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994833241273735e-05,
+      "loss": 0.9019,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947528964014992e-05,
+      "loss": 0.782,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946719332875788e-05,
+      "loss": 0.8289,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945903519822927e-05,
+      "loss": 0.845,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945081525363435e-05,
+      "loss": 0.7922,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944253350008193e-05,
+      "loss": 0.8386,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99434189942719e-05,
+      "loss": 0.8794,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942578458673126e-05,
+      "loss": 0.8433,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941731743734258e-05,
+      "loss": 0.8503,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940878849981527e-05,
+      "loss": 0.8325,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940019777945017e-05,
+      "loss": 0.8271,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993915452815864e-05,
+      "loss": 0.8079,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938283101160154e-05,
+      "loss": 0.8027,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993740549749115e-05,
+      "loss": 0.8447,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936521717697066e-05,
+      "loss": 0.7971,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935631762327166e-05,
+      "loss": 0.8384,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934735631934568e-05,
+      "loss": 0.8796,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933833327076212e-05,
+      "loss": 0.8274,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932924848312888e-05,
+      "loss": 0.8352,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932010196209215e-05,
+      "loss": 0.8577,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931089371333653e-05,
+      "loss": 0.8413,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930162374258496e-05,
+      "loss": 0.8459,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992922920555988e-05,
+      "loss": 0.8425,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992828986581777e-05,
+      "loss": 0.8535,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927344355615963e-05,
+      "loss": 0.8457,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926392675542102e-05,
+      "loss": 0.8601,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925434826187657e-05,
+      "loss": 0.8367,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924470808147935e-05,
+      "loss": 0.8564,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992350062202207e-05,
+      "loss": 0.8413,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922524268413045e-05,
+      "loss": 0.8525,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992154174792766e-05,
+      "loss": 0.8435,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920553061176557e-05,
+      "loss": 0.7915,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919558208774204e-05,
+      "loss": 0.877,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991855719133891e-05,
+      "loss": 0.8955,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917550009492804e-05,
+      "loss": 0.8206,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991653666386186e-05,
+      "loss": 0.8062,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915517155075865e-05,
+      "loss": 0.825,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914491483768454e-05,
+      "loss": 0.8789,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913459650577082e-05,
+      "loss": 0.7766,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991242165614304e-05,
+      "loss": 0.8347,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911377501111438e-05,
+      "loss": 0.8662,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910327186131226e-05,
+      "loss": 0.8372,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909270711855177e-05,
+      "loss": 0.8552,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990820807893989e-05,
+      "loss": 0.7834,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907139288045796e-05,
+      "loss": 0.8665,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990606433983715e-05,
+      "loss": 0.7812,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990498323498204e-05,
+      "loss": 0.8025,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903895974152373e-05,
+      "loss": 0.8818,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902802558023883e-05,
+      "loss": 0.832,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901702987276133e-05,
+      "loss": 0.8145,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900597262592508e-05,
+      "loss": 0.8218,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989948538466022e-05,
+      "loss": 0.8254,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898367354170305e-05,
+      "loss": 0.8147,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989724317181762e-05,
+      "loss": 0.8308,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896112838300844e-05,
+      "loss": 0.8271,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894976354322494e-05,
+      "loss": 0.8201,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893833720588885e-05,
+      "loss": 0.8232,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892684937810176e-05,
+      "loss": 0.792,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891530006700335e-05,
+      "loss": 0.7866,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890368927977155e-05,
+      "loss": 0.8333,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988920170236225e-05,
+      "loss": 0.8608,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888028330581052e-05,
+      "loss": 0.844,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988684881336282e-05,
+      "loss": 0.8535,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988566315144062e-05,
+      "loss": 0.7595,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988447134555135e-05,
+      "loss": 0.8416,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883273396435722e-05,
+      "loss": 0.8647,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988206930483826e-05,
+      "loss": 0.8164,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988085907150731e-05,
+      "loss": 0.8125,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879642697195035e-05,
+      "loss": 0.8362,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987842018265742e-05,
+      "loss": 0.801,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987719152865426e-05,
+      "loss": 0.8772,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987595673594916e-05,
+      "loss": 0.8542,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874715805309557e-05,
+      "loss": 0.8518,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987346873750669e-05,
+      "loss": 0.8496,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872215533315607e-05,
+      "loss": 0.8196,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870956193515186e-05,
+      "loss": 0.844,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986969071888811e-05,
+      "loss": 0.7993,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868419110220873e-05,
+      "loss": 0.8042,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867141368303783e-05,
+      "loss": 0.7859,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986585749393096e-05,
+      "loss": 0.8406,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864567487900338e-05,
+      "loss": 0.8489,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986327135101366e-05,
+      "loss": 0.8147,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861969084076472e-05,
+      "loss": 0.811,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860660687898142e-05,
+      "loss": 0.8333,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985934616329184e-05,
+      "loss": 0.8337,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858025511074544e-05,
+      "loss": 0.7991,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985669873206705e-05,
+      "loss": 0.8066,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985536582709395e-05,
+      "loss": 0.7983,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854026796983652e-05,
+      "loss": 0.8613,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852681642568364e-05,
+      "loss": 0.8569,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98513303646841e-05,
+      "loss": 0.8384,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849972964170687e-05,
+      "loss": 0.8396,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848609441871752e-05,
+      "loss": 0.8394,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847239798634734e-05,
+      "loss": 0.8347,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984586403531086e-05,
+      "loss": 0.8655,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844482152755172e-05,
+      "loss": 0.8682,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843094151826522e-05,
+      "loss": 0.8171,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984170003338755e-05,
+      "loss": 0.8396,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840299798304708e-05,
+      "loss": 0.8503,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983889344744824e-05,
+      "loss": 0.8384,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837480981692202e-05,
+      "loss": 0.8386,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836062401914447e-05,
+      "loss": 0.8564,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983463770899662e-05,
+      "loss": 0.8157,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833206903824177e-05,
+      "loss": 0.8682,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831769987286368e-05,
+      "loss": 0.8618,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830326960276234e-05,
+      "loss": 0.8657,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828877823690623e-05,
+      "loss": 0.8374,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827422578430183e-05,
+      "loss": 0.8284,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982596122539935e-05,
+      "loss": 0.8582,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824493765506355e-05,
+      "loss": 0.8176,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982302019966323e-05,
+      "loss": 0.8101,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821540528785802e-05,
+      "loss": 0.8574,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820054753793693e-05,
+      "loss": 0.8062,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981856287561031e-05,
+      "loss": 0.8445,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981706489516287e-05,
+      "loss": 0.8486,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815560813382352e-05,
+      "loss": 0.8445,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981405063120357e-05,
+      "loss": 0.8245,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981253434956509e-05,
+      "loss": 0.8115,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811011969409297e-05,
+      "loss": 0.8359,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809483491682345e-05,
+      "loss": 0.8237,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807948917334193e-05,
+      "loss": 0.8364,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806408247318576e-05,
+      "loss": 0.8545,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804861482593038e-05,
+      "loss": 0.7869,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803308624118882e-05,
+      "loss": 0.8496,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801749672861223e-05,
+      "loss": 0.8423,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800184629788956e-05,
+      "loss": 0.8184,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798613495874754e-05,
+      "loss": 0.8486,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979703627209508e-05,
+      "loss": 0.8545,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795452959430187e-05,
+      "loss": 0.8936,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793863558864105e-05,
+      "loss": 0.874,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792268071384656e-05,
+      "loss": 0.835,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790666497983433e-05,
+      "loss": 0.8438,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789058839655824e-05,
+      "loss": 0.8606,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978744509740099e-05,
+      "loss": 0.8118,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785825272221876e-05,
+      "loss": 0.8259,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978419936512521e-05,
+      "loss": 0.8452,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97825673771215e-05,
+      "loss": 0.8481,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780929309225025e-05,
+      "loss": 0.896,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779285162453853e-05,
+      "loss": 0.8418,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777634937829822e-05,
+      "loss": 0.8586,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775978636378555e-05,
+      "loss": 0.8591,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774316259129444e-05,
+      "loss": 0.8352,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977264780711567e-05,
+      "loss": 0.8337,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770973281374168e-05,
+      "loss": 0.8259,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976929268294567e-05,
+      "loss": 0.8652,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767606012874664e-05,
+      "loss": 0.8618,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976591327220943e-05,
+      "loss": 0.7881,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764214462002006e-05,
+      "loss": 0.8083,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762509583308206e-05,
+      "loss": 0.8181,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976079863718762e-05,
+      "loss": 0.8218,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759081624703604e-05,
+      "loss": 0.8347,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757358546923286e-05,
+      "loss": 0.8335,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755629404917562e-05,
+      "loss": 0.8584,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97538941997611e-05,
+      "loss": 0.8376,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752152932532337e-05,
+      "loss": 0.8022,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975040560431347e-05,
+      "loss": 0.8337,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748652216190477e-05,
+      "loss": 0.79,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746892769253087e-05,
+      "loss": 0.8765,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745127264594802e-05,
+      "loss": 0.8503,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743355703312884e-05,
+      "loss": 0.8496,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741578086508373e-05,
+      "loss": 0.8406,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973979441528606e-05,
+      "loss": 0.8767,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738004690754495e-05,
+      "loss": 0.8484,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736208914026004e-05,
+      "loss": 0.8528,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973440708621666e-05,
+      "loss": 0.8499,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732599208446313e-05,
+      "loss": 0.8196,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730785281838554e-05,
+      "loss": 0.8052,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728965307520754e-05,
+      "loss": 0.8706,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727139286624026e-05,
+      "loss": 0.8462,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972530722028325e-05,
+      "loss": 0.7886,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972346910963705e-05,
+      "loss": 0.8394,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972162495582783e-05,
+      "loss": 0.8442,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719774760001734e-05,
+      "loss": 0.791,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717918523308656e-05,
+      "loss": 0.8306,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716056246902263e-05,
+      "loss": 0.8389,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971418793193996e-05,
+      "loss": 0.8608,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712313579582907e-05,
+      "loss": 0.8057,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971043319099602e-05,
+      "loss": 0.8286,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708546767347972e-05,
+      "loss": 0.8511,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706654309811174e-05,
+      "loss": 0.8127,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704755819561797e-05,
+      "loss": 0.853,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702851297779755e-05,
+      "loss": 0.8325,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700940745648717e-05,
+      "loss": 0.8428,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699024164356092e-05,
+      "loss": 0.802,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697101555093045e-05,
+      "loss": 0.7861,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695172919054482e-05,
+      "loss": 0.832,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693238257439056e-05,
+      "loss": 0.7766,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969129757144916e-05,
+      "loss": 0.8196,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689350862290943e-05,
+      "loss": 0.8552,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687398131174285e-05,
+      "loss": 0.8057,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685439379312815e-05,
+      "loss": 0.8462,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683474607923902e-05,
+      "loss": 0.8647,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968150381822866e-05,
+      "loss": 0.8564,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679527011451937e-05,
+      "loss": 0.8804,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677544188822322e-05,
+      "loss": 0.7947,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967555535157215e-05,
+      "loss": 0.8367,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967356050093749e-05,
+      "loss": 0.8528,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967155963815814e-05,
+      "loss": 0.8738,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669552764477644e-05,
+      "loss": 0.814,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667539881143284e-05,
+      "loss": 0.8262,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966552098940607e-05,
+      "loss": 0.8042,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966349609052075e-05,
+      "loss": 0.835,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96614651857458e-05,
+      "loss": 0.8311,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659428276343435e-05,
+      "loss": 0.7981,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96573853635796e-05,
+      "loss": 0.8162,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655336448723977e-05,
+      "loss": 0.8889,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965328153304996e-05,
+      "loss": 0.8389,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651220617834696e-05,
+      "loss": 0.8347,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649153704359042e-05,
+      "loss": 0.8359,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96470807939076e-05,
+      "loss": 0.803,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645001887768678e-05,
+      "loss": 0.8357,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642916987234327e-05,
+      "loss": 0.8376,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640826093600317e-05,
+      "loss": 0.8154,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963872920816615e-05,
+      "loss": 0.821,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963662633223504e-05,
+      "loss": 0.7957,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963451746711393e-05,
+      "loss": 0.8186,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963240261411349e-05,
+      "loss": 0.8059,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630281774548105e-05,
+      "loss": 0.8152,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628154949735882e-05,
+      "loss": 0.8479,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626022140998648e-05,
+      "loss": 0.791,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962388334966195e-05,
+      "loss": 0.8184,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621738577055056e-05,
+      "loss": 0.8857,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619587824510943e-05,
+      "loss": 0.8369,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617431093366312e-05,
+      "loss": 0.7593,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615268384961578e-05,
+      "loss": 0.9077,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961309970064087e-05,
+      "loss": 0.8154,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961092504175203e-05,
+      "loss": 0.873,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608744409646615e-05,
+      "loss": 0.8352,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606557805679893e-05,
+      "loss": 0.8181,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960436523121085e-05,
+      "loss": 0.8303,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960216668760217e-05,
+      "loss": 0.8276,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599962176220256e-05,
+      "loss": 0.8701,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959775169843522e-05,
+      "loss": 0.8328,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595535255620877e-05,
+      "loss": 0.8137,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959331284915476e-05,
+      "loss": 0.8777,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591084480418086e-05,
+      "loss": 0.8625,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958885015079581e-05,
+      "loss": 0.8225,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958660986167656e-05,
+      "loss": 0.7959,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958436361445269e-05,
+      "loss": 0.8379,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582111410520244e-05,
+      "loss": 0.79,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579853251278975e-05,
+      "loss": 0.8416,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577589138132338e-05,
+      "loss": 0.8032,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957531907248748e-05,
+      "loss": 0.7893,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957304305575526e-05,
+      "loss": 0.8503,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957076108935022e-05,
+      "loss": 0.7581,
+      "step": 497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956847317469062e-05,
+      "loss": 0.873,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566179313198393e-05,
+      "loss": 0.7935,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956387950629919e-05,
+      "loss": 0.8113,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561573755422338e-05,
+      "loss": 0.8066,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559262062000877e-05,
+      "loss": 0.8157,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556944427471524e-05,
+      "loss": 0.835,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554620853274698e-05,
+      "loss": 0.8159,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955229134085451e-05,
+      "loss": 0.8628,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954995589165875e-05,
+      "loss": 0.8196,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547614507138918e-05,
+      "loss": 0.8005,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954526718875018e-05,
+      "loss": 0.8738,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542913937951405e-05,
+      "loss": 0.877,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540554756205145e-05,
+      "loss": 0.8428,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953818964497764e-05,
+      "loss": 0.8391,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953581860573881e-05,
+      "loss": 0.8162,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533441639962266e-05,
+      "loss": 0.8677,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531058749125295e-05,
+      "loss": 0.8147,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952866993470887e-05,
+      "loss": 0.8088,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526275198197647e-05,
+      "loss": 0.8379,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523874541079964e-05,
+      "loss": 0.7754,
+      "step": 517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521467964847836e-05,
+      "loss": 0.7825,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519055470996954e-05,
+      "loss": 0.8813,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516637061026688e-05,
+      "loss": 0.7944,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514212736440094e-05,
+      "loss": 0.8342,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511782498743888e-05,
+      "loss": 0.8357,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509346349448473e-05,
+      "loss": 0.8081,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506904290067922e-05,
+      "loss": 0.7993,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950445632211998e-05,
+      "loss": 0.8115,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502002447126073e-05,
+      "loss": 0.8423,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499542666611284e-05,
+      "loss": 0.8376,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497076982104376e-05,
+      "loss": 0.8474,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494605395137778e-05,
+      "loss": 0.8298,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492127907247588e-05,
+      "loss": 0.8149,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948964451997357e-05,
+      "loss": 0.8186,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487155234859165e-05,
+      "loss": 0.8489,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948466005345146e-05,
+      "loss": 0.7402,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482158977301217e-05,
+      "loss": 0.8208,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947965200796287e-05,
+      "loss": 0.842,
+      "step": 535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477139146994505e-05,
+      "loss": 0.7837,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474620395957867e-05,
+      "loss": 0.8276,
+      "step": 537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472095756418368e-05,
+      "loss": 0.8423,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469565229945083e-05,
+      "loss": 0.8083,
+      "step": 539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467028818110738e-05,
+      "loss": 0.835,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946448652249172e-05,
+      "loss": 0.8218,
+      "step": 541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461938344668075e-05,
+      "loss": 0.8013,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94593842862235e-05,
+      "loss": 0.7668,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945682434874535e-05,
+      "loss": 0.8596,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454258533824632e-05,
+      "loss": 0.8403,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451686843056013e-05,
+      "loss": 0.8337,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94491092780378e-05,
+      "loss": 0.8008,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446525840371957e-05,
+      "loss": 0.8271,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94439365316641e-05,
+      "loss": 0.8049,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944134135352349e-05,
+      "loss": 0.7988,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943874030756304e-05,
+      "loss": 0.8469,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436133395399305e-05,
+      "loss": 0.8176,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433520618652486e-05,
+      "loss": 0.8171,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430901978946436e-05,
+      "loss": 0.8086,
+      "step": 554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428277477908642e-05,
+      "loss": 0.8472,
+      "step": 555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425647117170245e-05,
+      "loss": 0.8628,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423010898366015e-05,
+      "loss": 0.8176,
+      "step": 557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420368823134374e-05,
+      "loss": 0.8687,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941772089311737e-05,
+      "loss": 0.7769,
+      "step": 559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415067109960716e-05,
+      "loss": 0.8611,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412407475313727e-05,
+      "loss": 0.8743,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409741990829384e-05,
+      "loss": 0.8789,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407070658164292e-05,
+      "loss": 0.8147,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404393478978688e-05,
+      "loss": 0.8149,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401710454936454e-05,
+      "loss": 0.8145,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939902158770509e-05,
+      "loss": 0.8206,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396326878955732e-05,
+      "loss": 0.8237,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393626330363152e-05,
+      "loss": 0.8752,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939091994360575e-05,
+      "loss": 0.8254,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388207720365558e-05,
+      "loss": 0.7712,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938548966232822e-05,
+      "loss": 0.8257,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938276577118302e-05,
+      "loss": 0.8748,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938003604862287e-05,
+      "loss": 0.8804,
+      "step": 573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937730049634429e-05,
+      "loss": 0.8091,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937455911604744e-05,
+      "loss": 0.8298,
+      "step": 575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371811909436097e-05,
+      "loss": 0.821,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936905887821765e-05,
+      "loss": 0.8108,
+      "step": 577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366300024103123e-05,
+      "loss": 0.8083,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936353534880715e-05,
+      "loss": 0.8154,
+      "step": 579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360764854047977e-05,
+      "loss": 0.8347,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357988541547486e-05,
+      "loss": 0.8201,
+      "step": 581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355206413031155e-05,
+      "loss": 0.8452,
+      "step": 582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352418470228085e-05,
+      "loss": 0.8215,
+      "step": 583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349624714870994e-05,
+      "loss": 0.8291,
+      "step": 584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346825148696204e-05,
+      "loss": 0.8159,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344019773443655e-05,
+      "loss": 0.8145,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341208590856898e-05,
+      "loss": 0.7983,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933839160268309e-05,
+      "loss": 0.8711,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335568810672995e-05,
+      "loss": 0.8088,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332740216580986e-05,
+      "loss": 0.8179,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329905822165042e-05,
+      "loss": 0.8369,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932706562918675e-05,
+      "loss": 0.8257,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324219639411297e-05,
+      "loss": 0.8386,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321367854607472e-05,
+      "loss": 0.8005,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318510276547664e-05,
+      "loss": 0.8091,
+      "step": 595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931564690700787e-05,
+      "loss": 0.8459,
+      "step": 596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312777747767685e-05,
+      "loss": 0.8472,
+      "step": 597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309902800610292e-05,
+      "loss": 0.8843,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307022067322484e-05,
+      "loss": 0.7932,
+      "step": 599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930413554969464e-05,
+      "loss": 0.7991,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301243249520746e-05,
+      "loss": 0.8379,
+      "step": 601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929834516859837e-05,
+      "loss": 0.8457,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295441308728676e-05,
+      "loss": 0.844,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292531671716423e-05,
+      "loss": 0.8352,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928961625936996e-05,
+      "loss": 0.7844,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928669507350122e-05,
+      "loss": 0.7625,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928376811592573e-05,
+      "loss": 0.8501,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92808353884626e-05,
+      "loss": 0.8354,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277896892934536e-05,
+      "loss": 0.8132,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274952631167813e-05,
+      "loss": 0.7798,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92720026049923e-05,
+      "loss": 0.8235,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269046816241447e-05,
+      "loss": 0.8313,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926608526675229e-05,
+      "loss": 0.8579,
+      "step": 613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263117958365428e-05,
+      "loss": 0.8364,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260144892925063e-05,
+      "loss": 0.8447,
+      "step": 615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257166072278957e-05,
+      "loss": 0.8064,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925418149827846e-05,
+      "loss": 0.8279,
+      "step": 617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251191172778487e-05,
+      "loss": 0.8213,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248195097637538e-05,
+      "loss": 0.7544,
+      "step": 619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924519327471768e-05,
+      "loss": 0.8618,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924218570588456e-05,
+      "loss": 0.8274,
+      "step": 621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923917239300738e-05,
+      "loss": 0.856,
+      "step": 622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236153337958928e-05,
+      "loss": 0.8706,
+      "step": 623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233128542615558e-05,
+      "loss": 0.8711,
+      "step": 624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230098008857185e-05,
+      "loss": 0.8298,
+      "step": 625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227061738567293e-05,
+      "loss": 0.8059,
+      "step": 626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224019733632938e-05,
+      "loss": 0.8367,
+      "step": 627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220971995944726e-05,
+      "loss": 0.7964,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217918527396842e-05,
+      "loss": 0.7949,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214859329887023e-05,
+      "loss": 0.8079,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211794405316566e-05,
+      "loss": 0.8582,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208723755590334e-05,
+      "loss": 0.8547,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920564738261674e-05,
+      "loss": 0.7827,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202565288307764e-05,
+      "loss": 0.8372,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919947747457893e-05,
+      "loss": 0.8472,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919638394334933e-05,
+      "loss": 0.7966,
+      "step": 636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919328469654159e-05,
+      "loss": 0.7805,
+      "step": 637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190179736081915e-05,
+      "loss": 0.8269,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918706906390004e-05,
+      "loss": 0.8245,
+      "step": 639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183952681929255e-05,
+      "loss": 0.842,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91808305921064e-05,
+      "loss": 0.8752,
+      "step": 641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917770279637187e-05,
+      "loss": 0.7739,
+      "step": 642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174569296669586e-05,
+      "loss": 0.8293,
+      "step": 643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171430094947036e-05,
+      "loss": 0.7954,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168285193155238e-05,
+      "loss": 0.7847,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916513459324876e-05,
+      "loss": 0.7339,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916197829718571e-05,
+      "loss": 0.7761,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158816306927736e-05,
+      "loss": 0.8484,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155648624440012e-05,
+      "loss": 0.8254,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152475251691275e-05,
+      "loss": 0.8521,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149296190653776e-05,
+      "loss": 0.812,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914611144330332e-05,
+      "loss": 0.8667,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142921011619225e-05,
+      "loss": 0.8076,
+      "step": 653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913972489758436e-05,
+      "loss": 0.7969,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913652310318512e-05,
+      "loss": 0.8503,
+      "step": 655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133315630411423e-05,
+      "loss": 0.8445,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130102481256725e-05,
+      "loss": 0.8411,
+      "step": 657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912688365771801e-05,
+      "loss": 0.8877,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912365916179578e-05,
+      "loss": 0.8643,
+      "step": 659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120428995494073e-05,
+      "loss": 0.8328,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117193160820445e-05,
+      "loss": 0.8098,
+      "step": 661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113951659785977e-05,
+      "loss": 0.8105,
+      "step": 662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110704494405268e-05,
+      "loss": 0.7974,
+      "step": 663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107451666696443e-05,
+      "loss": 0.8098,
+      "step": 664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104193178681146e-05,
+      "loss": 0.7734,
+      "step": 665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100929032384527e-05,
+      "loss": 0.7444,
+      "step": 666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909765922983527e-05,
+      "loss": 0.8501,
+      "step": 667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909438377306557e-05,
+      "loss": 0.8262,
+      "step": 668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909110266411112e-05,
+      "loss": 0.8201,
+      "step": 669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087815905011148e-05,
+      "loss": 0.8313,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908452349780838e-05,
+      "loss": 0.8137,
+      "step": 671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908122544454906e-05,
+      "loss": 0.7954,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077921747282932e-05,
+      "loss": 0.8113,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074612408063257e-05,
+      "loss": 0.8132,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071297428946795e-05,
+      "loss": 0.8152,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906797681199382e-05,
+      "loss": 0.8042,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064650559268097e-05,
+      "loss": 0.8613,
+      "step": 677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9061318672836907e-05,
+      "loss": 0.7991,
+      "step": 678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057981154771026e-05,
+      "loss": 0.8374,
+      "step": 679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905463800714473e-05,
+      "loss": 0.8093,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051289232035794e-05,
+      "loss": 0.8564,
+      "step": 681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047934831525492e-05,
+      "loss": 0.8706,
+      "step": 682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044574807698593e-05,
+      "loss": 0.8179,
+      "step": 683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041209162643365e-05,
+      "loss": 0.8411,
+      "step": 684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037837898451558e-05,
+      "loss": 0.8069,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034461017218424e-05,
+      "loss": 0.8113,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031078521042707e-05,
+      "loss": 0.8301,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027690412026638e-05,
+      "loss": 0.8423,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902429669227593e-05,
+      "loss": 0.8354,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020897363899802e-05,
+      "loss": 0.8379,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017492429010933e-05,
+      "loss": 0.8303,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014081889725508e-05,
+      "loss": 0.8359,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010665748163177e-05,
+      "loss": 0.8186,
+      "step": 693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007244006447093e-05,
+      "loss": 0.8162,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003816666703868e-05,
+      "loss": 0.8113,
+      "step": 695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900038373106361e-05,
+      "loss": 0.8127,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996945201659897e-05,
+      "loss": 0.8423,
+      "step": 697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993501080629784e-05,
+      "loss": 0.8149,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990051370113796e-05,
+      "loss": 0.834,
+      "step": 699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986596072255947e-05,
+      "loss": 0.7849,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983135189203708e-05,
+      "loss": 0.8008,
+      "step": 701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979668723108032e-05,
+      "loss": 0.8252,
+      "step": 702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897619667612333e-05,
+      "loss": 0.8054,
+      "step": 703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972719050407494e-05,
+      "loss": 0.8257,
+      "step": 704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969235848121883e-05,
+      "loss": 0.7725,
+      "step": 705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965747071431307e-05,
+      "loss": 0.8025,
+      "step": 706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896225272250406e-05,
+      "loss": 0.7703,
+      "step": 707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958752803511874e-05,
+      "loss": 0.7834,
+      "step": 708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955247316629975e-05,
+      "loss": 0.8381,
+      "step": 709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951736264037028e-05,
+      "loss": 0.8208,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948219647915157e-05,
+      "loss": 0.8018,
+      "step": 711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944697470449953e-05,
+      "loss": 0.8074,
+      "step": 712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894116973383046e-05,
+      "loss": 0.7783,
+      "step": 713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893763644024917e-05,
+      "loss": 0.8257,
+      "step": 714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934097591902043e-05,
+      "loss": 0.8059,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893055319098848e-05,
+      "loss": 0.8088,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927003239711332e-05,
+      "loss": 0.7732,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923447740276907e-05,
+      "loss": 0.8191,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891988669489496e-05,
+      "loss": 0.8042,
+      "step": 719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916320105778686e-05,
+      "loss": 0.8372,
+      "step": 720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891274797514473e-05,
+      "loss": 0.8464,
+      "step": 721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909170305213184e-05,
+      "loss": 0.802,
+      "step": 722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905587098207584e-05,
+      "loss": 0.8171,
+      "step": 723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890199835635489e-05,
+      "loss": 0.8296,
+      "step": 724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898404081885527e-05,
+      "loss": 0.824,
+      "step": 725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894804277033345e-05,
+      "loss": 0.8459,
+      "step": 726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891198944035625e-05,
+      "loss": 0.8184,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88875880851331e-05,
+      "loss": 0.8691,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883971702569924e-05,
+      "loss": 0.8193,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880349798593695e-05,
+      "loss": 0.8418,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876722375455426e-05,
+      "loss": 0.8611,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887308943540958e-05,
+      "loss": 0.8076,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886945098071403e-05,
+      "loss": 0.8579,
+      "step": 733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88658070136301e-05,
+      "loss": 0.865,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886215753642251e-05,
+      "loss": 0.8074,
+      "step": 735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885850255135943e-05,
+      "loss": 0.8435,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854842060712438e-05,
+      "loss": 0.8362,
+      "step": 737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851176066756545e-05,
+      "loss": 0.8748,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847504571770167e-05,
+      "loss": 0.874,
+      "step": 739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884382757803516e-05,
+      "loss": 0.8162,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840145087836777e-05,
+      "loss": 0.8113,
+      "step": 741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836457103463703e-05,
+      "loss": 0.7952,
+      "step": 742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832763627208025e-05,
+      "loss": 0.8767,
+      "step": 743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829064661365247e-05,
+      "loss": 0.8589,
+      "step": 744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825360208234293e-05,
+      "loss": 0.8516,
+      "step": 745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821650270117485e-05,
+      "loss": 0.7959,
+      "step": 746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817934849320564e-05,
+      "loss": 0.8872,
+      "step": 747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814213948152675e-05,
+      "loss": 0.8477,
+      "step": 748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810487568926362e-05,
+      "loss": 0.8372,
+      "step": 749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806755713957588e-05,
+      "loss": 0.8325,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803018385565707e-05,
+      "loss": 0.7971,
+      "step": 751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799275586073477e-05,
+      "loss": 0.8157,
+      "step": 752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795527317807064e-05,
+      "loss": 0.8049,
+      "step": 753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879177358309602e-05,
+      "loss": 0.8474,
+      "step": 754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788014384273307e-05,
+      "loss": 0.7991,
+      "step": 755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784249723675273e-05,
+      "loss": 0.8354,
+      "step": 756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780479603641676e-05,
+      "loss": 0.842,
+      "step": 757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776704026515638e-05,
+      "loss": 0.8606,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772922994643707e-05,
+      "loss": 0.802,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769136510375797e-05,
+      "loss": 0.8037,
+      "step": 760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765344576065222e-05,
+      "loss": 0.8113,
+      "step": 761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761547194068674e-05,
+      "loss": 0.8198,
+      "step": 762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875774436674624e-05,
+      "loss": 0.8523,
+      "step": 763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753936096461388e-05,
+      "loss": 0.8401,
+      "step": 764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875012238558097e-05,
+      "loss": 0.8599,
+      "step": 765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746303236475212e-05,
+      "loss": 0.8503,
+      "step": 766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742478651517735e-05,
+      "loss": 0.8276,
+      "step": 767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738648633085522e-05,
+      "loss": 0.8237,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734813183558942e-05,
+      "loss": 0.8643,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730972305321738e-05,
+      "loss": 0.8159,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727126000761025e-05,
+      "loss": 0.7839,
+      "step": 771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723274272267295e-05,
+      "loss": 0.781,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719417122234405e-05,
+      "loss": 0.8176,
+      "step": 773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715554553059587e-05,
+      "loss": 0.843,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711686567143433e-05,
+      "loss": 0.8486,
+      "step": 775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707813166889915e-05,
+      "loss": 0.855,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703934354706358e-05,
+      "loss": 0.7393,
+      "step": 777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700050133003452e-05,
+      "loss": 0.8284,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696160504195253e-05,
+      "loss": 0.8252,
+      "step": 779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692265470699182e-05,
+      "loss": 0.842,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688365034936002e-05,
+      "loss": 0.7988,
+      "step": 781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684459199329856e-05,
+      "loss": 0.823,
+      "step": 782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680547966308224e-05,
+      "loss": 0.7939,
+      "step": 783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676631338301955e-05,
+      "loss": 0.7876,
+      "step": 784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672709317745233e-05,
+      "loss": 0.8142,
+      "step": 785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668781907075618e-05,
+      "loss": 0.8274,
+      "step": 786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664849108733994e-05,
+      "loss": 0.7935,
+      "step": 787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866091092516462e-05,
+      "loss": 0.7937,
+      "step": 788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656967358815073e-05,
+      "loss": 0.8704,
+      "step": 789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86530184121363e-05,
+      "loss": 0.8264,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864906408758258e-05,
+      "loss": 0.8022,
+      "step": 791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645104387611536e-05,
+      "loss": 0.813,
+      "step": 792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641139314684134e-05,
+      "loss": 0.7866,
+      "step": 793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863716887126468e-05,
+      "loss": 0.8589,
+      "step": 794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633193059820808e-05,
+      "loss": 0.8682,
+      "step": 795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629211882823502e-05,
+      "loss": 0.7974,
+      "step": 796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625225342747075e-05,
+      "loss": 0.8096,
+      "step": 797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862123344206917e-05,
+      "loss": 0.7632,
+      "step": 798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617236183270764e-05,
+      "loss": 0.8259,
+      "step": 799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613233568836164e-05,
+      "loss": 0.8159,
+      "step": 800
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4109,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 800,
+  "total_flos": 1881161810051072.0,
+  "train_batch_size": 24,
+  "trial_name": null,
+  "trial_params": null
+}